**Recognition and Resolution of "Comprehension Uncertainty" in** *AI*

Sukanto Bhattacharya1,\* and Kuldeep Kumar2 *1Deakin Graduate School of Business, Deakin University, 2School of Business, Bond University, Australia* 

#### **1. Introduction**

26 Will-be-set-by-IN-TECH

244 Intelligent Systems

[64] Rieˇcan, B., Lašová. L.: On the probability on the Kôpka D-posets. In: Developments in Fuzzy Sets, Intuitionistic Fuzzy Sets, Generalized Nets and related Topics I (K. Atanassov

[65] Rieˇcan, B., Král', P.: Probability on interval valued events. Proc. of the Eleventh International Workshop on Generalized Nets and the Second International Workshop on Generalized Nets, Intuitionistic Fuzzy Sets and Knowledge Engineering, London 9

[66] Rieˇcan B. and Mundici D.: Probability in MV-algebras. Handbook of Measure Theory II

[67] Rieˇcan B. and Neubrunn T.: Integral, Measure and Ordering. Kluwer, Dordrecht, 1997. [68] Rieˇcan, B., Petroviˇcová J.: On the Lukasiewicz probability theory on IF-sets. Tatra Mt.

[69] Szmidt, E., Kacprzyk, J.: Intuitionistic fuzzy sets in some medical applications. Notes IFS

[70] Valenˇcáková, V.: A note on the conditional probability on IF-events. Math. Slovaca 59,

[72] Vrábelová, M.: On the conditional probability in product MV-algebras. Soft Computing

[71] Vrábel P.: Integration on MV - algebras. Tatra Mt. Math. Publ. 12, 1997, 21 - 25.

[73] Zadeh L. A.: Fuzzy sets.Inform.and Control 8, 1965, 338 - 358.

et al. eds.), Warsaw 2010, 167 - 176.

Math. Publ. 46, 2010, 125 - 146.

( E. Pap ed.) Elsevier, Heidelberg, 2002, 869 - 910.

July 2010, 66 - 70.

7, 2001, No 4, 58 - 64.

2009, 251 - 260.

4, 2000, 58 - 61.

#### **1.1 Uncertainty resolution as an integral characteristic of intelligent systems**

Handling uncertainty is an important component of most intelligent behaviour – so uncertainty resolution is a key step in the design of an artificially intelligent decision system (Clark, 1990). Like other aspects of intelligent systems design, the aspect of uncertainty resolution is also typically sought to be handled by emulating natural intelligence (Halpern, 2003; Ball and Christensen, 2009). In this regard, a number of computational uncertainty resolution approaches have been proposed and tested by Artificial Intelligence (*AI*) researchers over the past several decades since birth of *AI* as a scientific discipline in early 1950s post- publication of Alan Turing's landmark paper (Turing, 1950).

The following chart categorizes various forms of uncertainty whose resolution ought to be a pertinent consideration in the design an artificial decision system that emulates natural intelligence:

Fig. 1. Broad classifications of "uncertainty" that intelligent systems are expected to resolve

<sup>\*</sup> Corresponding author

Recognition and Resolution of "Comprehension Uncertainty" in *AI* 247

(and past) decision states gleaned from different sources is a *set-valued* rather than *point-*

A three-valued extension of classical (i.e. binary) fuzzy logic was proposed by Smarandache (2002) when he coined the term "neutrosophic logic" as a generalization of fuzzy logic to such situations where it is impossible to *de-fuzzify* the original fuzzy–valued variables via some tractable membership function into either of set *T* or its complement *T*C where both *T* and *T*C are considered crisp sets. In these cases one has to allow for the possibility of a third unresolved state intermediate between *T* and *T*C. As an example one may cite the well known "thought experiment" in quantum metaphysics of *Schr*ö*dinger's cat* (Schrödinger, 1935) – the cat in a closed box is in limbo between two states "dead" and "alive" and it is impossible to tell which unless one opens the box at which point the effect of observer participation is said to intervene and cause that indeterminate state to collapse into a classical state of either a dead or an alive cat to be observed in the box. But as long as observer participation is completely absent one cannot in any way *disentangle* these two

This brings us to the final form of uncertainty that an artificially intelligent decision system ought to be able to resolve – something which we christened here as "comprehension uncertainty". While some elements of "comprehension uncertainty" is sought to be handled (often unknowingly) by the designers of intelligent systems by using one or more tools targeted to resolve either temporal or knowledge uncertainty, the concept of "comprehension uncertainty" has not yet been adequately described and addressed in contemporary *AI* literature. That is the reason we decided to depict this form of uncertainty using a *dashed rather than continuous connector* in the above chart. Also the question mark in the chart denotes the fact that there is no known repository of theoretical knowledge (not necessarily limited to the discipline of *AI*) that addresses such a form of uncertainty. The purpose of this chapter is to therefore posit a scientific theory of "comprehension

While all the other forms of uncertainty as discussed above necessarily *originates from and deals with the contents/specification of an elementary set of interest*, which is a subset of the universal set, by the term "comprehension uncertainty" we mean and include any form of uncertainty that *originates from and deals with the contents/specification of the universal set itself*. If the stock of our entire knowledge about a problem is *universal* (i.e. there is absolutely nothing else that is 'fundamentally unknown' about that problem) only then we can claim to fully comprehend the problem so that no "comprehension uncertainty" would then exist. There is a need here to distinguish between "complete knowledge" and "universal knowledge". The knowledge about a problem can be said to be complete if it consists of the *entire stock of current knowledge* that is pertinent to that particular problem. However the current stock of knowledge, even in its entirety, may not be the universal knowledge simply because ways of adding to that current stock of knowledge could be beyond the current limits of comprehension i.e. *the universal set could itself be ill-defined*. If intelligent systems are primarily intended to emulate natural intelligence and treat "functional comparability" with natural intelligence as the most desirable outcome, then the limits to comprehension for

natural intelligence should translate to similar limits for such systems as well.

*valued* feature (Sicilia, 2006).

crisp sets!

uncertainty".

**2. The meaning of "comprehension uncertainty"** 

Temporal uncertainty, as the name suggests, arises out of *imperfect foresight* – i.e. it concerns the general problem of determining the future decision state of a dynamic system the current and past decision states of which are known. As a sub-category of temporal uncertainty, parametric uncertainty is that form of uncertainty the resolution of which wholly depends on estimating a set of underlying parameters that determine a future decision state of a system given its current and/or past decision states. The fundamental premise is that there exist parameters, which if estimated accurately, would fully explain the temporal transition from current to a future decision state. In most practical *AI* applications it is handled by embedding an efficient parameter estimation *kernel* e.g. an asset price prediction kernel that is embedded within an intelligent financial trading system (Huang, Pasquier and Quek, 2009). On the other hand non-parametric uncertainty is that form of temporal uncertainty the resolution of which is either wholly or substantially independent of any parameters that can be statistically estimated from the current or past decision states of the system. That is, in resolving non-parametric uncertainty one cannot assume that there is a set of parameters whose accurate estimation can fully explain the dynamic system's time-path (Kosut, Lau and Boyd, 1992). To resolve non-parametric uncertainty, *AI* models are usually equipped with some feedback/learning mechanism coupled with a *performance measure index* that indicates when optimal learning has occurred so that predictive utility isn't lost on account of *overtraining* when predicting a future state using the current/past states as the *inputs* (Yang et al, 2010).

Knowledge uncertainty, again as the name suggests, arises out of *imperfect understanding* – i.e. it concerns the general problem of determining the future decision state of a dynamic system the knowledge about whose current and/or past states are either *incomplete*, *illdefined* or *inconsistent*. If there is incomplete information available about the current decision state of the system then the sub-category of knowledge uncertainty it would be categorized under is informational uncertainty. A common way of dealing with informational uncertainty is to try and *enhance* the current level of information by applying an appropriate information theoretic tool e.g. Ding *et al* (2008) applied rough sets theory coupled with a self-adaptive algorithm to separately "mine" consistent and inconsistent decision rules; along with experimental validation for large incomplete information systems. If the information available about the current decision state of the system is ill-defined i.e. it is subject to *interpretational ambiguity* then it would come under the sub-category of linguistic uncertainty. A large part of interpretational ambiguity arises as a direct result of statements made in natural language (Walley and Cooman, 2001). Lotfi Zadeh, the proponent of *fuzzy logic*, contended that *possibility measures* are best used to resolve linguistic uncertainty in decision systems (Zadeh, 1965). If the information available about the current decision state of the system is inconsistent i.e. it is fundamentally dependent on the origin, then the resulting uncertainty would come under the sub-category of paradigmatic uncertainty. If available information is dependent on its origin then it can be expected to materially change if one chooses a different source for the same information. For example, software agents have to reason and act on a domain in which the universe of possible scenarios is fundamentally prescribed by the available metadata records. But these metadata records can sometimes be found to be mutually inconsistent when compared. The paradigmatic uncertainty resulting from the inconsistency and imprecision is best addressed by building in enough flexibility in the system so that the cogency of information related to the current

Temporal uncertainty, as the name suggests, arises out of *imperfect foresight* – i.e. it concerns the general problem of determining the future decision state of a dynamic system the current and past decision states of which are known. As a sub-category of temporal uncertainty, parametric uncertainty is that form of uncertainty the resolution of which wholly depends on estimating a set of underlying parameters that determine a future decision state of a system given its current and/or past decision states. The fundamental premise is that there exist parameters, which if estimated accurately, would fully explain the temporal transition from current to a future decision state. In most practical *AI* applications it is handled by embedding an efficient parameter estimation *kernel* e.g. an asset price prediction kernel that is embedded within an intelligent financial trading system (Huang, Pasquier and Quek, 2009). On the other hand non-parametric uncertainty is that form of temporal uncertainty the resolution of which is either wholly or substantially independent of any parameters that can be statistically estimated from the current or past decision states of the system. That is, in resolving non-parametric uncertainty one cannot assume that there is a set of parameters whose accurate estimation can fully explain the dynamic system's time-path (Kosut, Lau and Boyd, 1992). To resolve non-parametric uncertainty, *AI* models are usually equipped with some feedback/learning mechanism coupled with a *performance measure index* that indicates when optimal learning has occurred so that predictive utility isn't lost on account of *overtraining* when predicting a future state using the current/past

Knowledge uncertainty, again as the name suggests, arises out of *imperfect understanding* – i.e. it concerns the general problem of determining the future decision state of a dynamic system the knowledge about whose current and/or past states are either *incomplete*, *illdefined* or *inconsistent*. If there is incomplete information available about the current decision state of the system then the sub-category of knowledge uncertainty it would be categorized under is informational uncertainty. A common way of dealing with informational uncertainty is to try and *enhance* the current level of information by applying an appropriate information theoretic tool e.g. Ding *et al* (2008) applied rough sets theory coupled with a self-adaptive algorithm to separately "mine" consistent and inconsistent decision rules; along with experimental validation for large incomplete information systems. If the information available about the current decision state of the system is ill-defined i.e. it is subject to *interpretational ambiguity* then it would come under the sub-category of linguistic uncertainty. A large part of interpretational ambiguity arises as a direct result of statements made in natural language (Walley and Cooman, 2001). Lotfi Zadeh, the proponent of *fuzzy logic*, contended that *possibility measures* are best used to resolve linguistic uncertainty in decision systems (Zadeh, 1965). If the information available about the current decision state of the system is inconsistent i.e. it is fundamentally dependent on the origin, then the resulting uncertainty would come under the sub-category of paradigmatic uncertainty. If available information is dependent on its origin then it can be expected to materially change if one chooses a different source for the same information. For example, software agents have to reason and act on a domain in which the universe of possible scenarios is fundamentally prescribed by the available metadata records. But these metadata records can sometimes be found to be mutually inconsistent when compared. The paradigmatic uncertainty resulting from the inconsistency and imprecision is best addressed by building in enough flexibility in the system so that the cogency of information related to the current

states as the *inputs* (Yang et al, 2010).

(and past) decision states gleaned from different sources is a *set-valued* rather than *pointvalued* feature (Sicilia, 2006).

A three-valued extension of classical (i.e. binary) fuzzy logic was proposed by Smarandache (2002) when he coined the term "neutrosophic logic" as a generalization of fuzzy logic to such situations where it is impossible to *de-fuzzify* the original fuzzy–valued variables via some tractable membership function into either of set *T* or its complement *T*C where both *T* and *T*C are considered crisp sets. In these cases one has to allow for the possibility of a third unresolved state intermediate between *T* and *T*C. As an example one may cite the well known "thought experiment" in quantum metaphysics of *Schr*ö*dinger's cat* (Schrödinger, 1935) – the cat in a closed box is in limbo between two states "dead" and "alive" and it is impossible to tell which unless one opens the box at which point the effect of observer participation is said to intervene and cause that indeterminate state to collapse into a classical state of either a dead or an alive cat to be observed in the box. But as long as observer participation is completely absent one cannot in any way *disentangle* these two crisp sets!

This brings us to the final form of uncertainty that an artificially intelligent decision system ought to be able to resolve – something which we christened here as "comprehension uncertainty". While some elements of "comprehension uncertainty" is sought to be handled (often unknowingly) by the designers of intelligent systems by using one or more tools targeted to resolve either temporal or knowledge uncertainty, the concept of "comprehension uncertainty" has not yet been adequately described and addressed in contemporary *AI* literature. That is the reason we decided to depict this form of uncertainty using a *dashed rather than continuous connector* in the above chart. Also the question mark in the chart denotes the fact that there is no known repository of theoretical knowledge (not necessarily limited to the discipline of *AI*) that addresses such a form of uncertainty. The purpose of this chapter is to therefore posit a scientific theory of "comprehension uncertainty".

#### **2. The meaning of "comprehension uncertainty"**

While all the other forms of uncertainty as discussed above necessarily *originates from and deals with the contents/specification of an elementary set of interest*, which is a subset of the universal set, by the term "comprehension uncertainty" we mean and include any form of uncertainty that *originates from and deals with the contents/specification of the universal set itself*. If the stock of our entire knowledge about a problem is *universal* (i.e. there is absolutely nothing else that is 'fundamentally unknown' about that problem) only then we can claim to fully comprehend the problem so that no "comprehension uncertainty" would then exist. There is a need here to distinguish between "complete knowledge" and "universal knowledge". The knowledge about a problem can be said to be complete if it consists of the *entire stock of current knowledge* that is pertinent to that particular problem. However the current stock of knowledge, even in its entirety, may not be the universal knowledge simply because ways of adding to that current stock of knowledge could be beyond the current limits of comprehension i.e. *the universal set could itself be ill-defined*. If intelligent systems are primarily intended to emulate natural intelligence and treat "functional comparability" with natural intelligence as the most desirable outcome, then the limits to comprehension for natural intelligence should translate to similar limits for such systems as well.

Recognition and Resolution of "Comprehension Uncertainty" in *AI* 249

mathematical lines. In that desired direction, we firstly posit and prove a fundamental theorem necessary for such an extension to the theory of probability. Subsequently we show

It is well known that much of modern theory of probability rests upon the three fundamental *Kolmogorov axioms* (Kolmogorov, 1956) which are conventionally stated as

1st axiom: The probability of any event is a non-negative real number i.e. P(E) ≥ 0 ∀ E ∈ U 2nd axiom: The probability of any one of the elementary events in the whole event space

3rd axiom: Any countable sequence of pair-wise *non-overlapping* events E1, E2, ... En satisfies the following relation: P(E1 E2 ... En) = ∑i P(Ei); i = 1, 2, ..., n.

It is basically Kolmogorov's second and third axioms as noted above that render any extensions of the probability concept to higher orders (i.e. "probability of probability") superfluous as the information content of any such higher order probability can be satisfactorily transmuted via existing set-theoretic constructs. So, extending to a higher order would arguably yield trivial information. However the Kolmogorov axioms by themselves are also open to 'extensions' – for instance there is previous research that has revisited the proofs of the well-known Bell inequality based on underlying assumptions of separability and noncontextuality and constructed a model of generalized "non-contextual contrapositive conditional probabilities" consistent with the results of the famous Aspect experiment showing in general such probabilities are not necessarily all positive (Atkinson, 2000). By themselves the Kolmogorov axioms do not unequivocally rule out an extension of the definition of the universal set U itself so as to make U possess a *time-dynamic* rather than a *timestatic* nature. So; in effect this means that if we were to consider a time-dynamic version of the universal set; then one would suddenly find that the information content of higher order probability no longer remains trivial i.e. an extension of the probability concept to higher orders (i.e. "probability of probability") is no longer superfluous – in fact it is logical! The good thing is that no new probability calculus needs to be formulated to describe such a theory of higher-order probabilities and this extended theory could still rest on the Kolmogorov axioms and could still draw fundamentally from the standard set-theoretic approach (as we will be demonstrating shortly); by merely using an extended definition of the universal set U which would now denote not merely an event space but a broader concept, which we christen as

*event-spacetime*, i.e. an event space that can *evolve* over a time dimension.

is also attainable but it's left for later.

Perhaps the only academic work preceding ours to have alluded that a higher-order probability theory is justifiable by an event space evolving over time was that by Haddawy and others (Haddawy, 1996; Lehner, Laskey and Dubois, 1996), where they provided "a logic that incorporates and integrates the concepts of subjective probability, objective probability, time and causality" (Lehner, Laskey and Dubois, 1996). We take a similar philosophical stance but go on to explicitly develop a logically tenable higher-order probability concept in discrete time. We have no doubt that an extension in continuous time

some computational 'tests' to illustrate the posited framework.

**3.1 A foray into higher order probabilities**

occurring is 1 i.e. P(U)=1

follows:

#### **2.1 How does natural intelligence resolve "comprehension uncertainty" in decision-making?**

As highly evolved, intelligent beings, humans have become adept at continually taking decisions based on information that is subject to various forms of uncertainty. We can negotiate a busy sidewalk more often than not without colliding with other pedestrians and can cross a road safely (again most of the times) without being flattened by a car although we have at best a very imprecise idea of the speed of an oncoming car. Human brain, as the highest seat of natural intelligence, has evolved unique ways of working with various uncertainties including "comprehension uncertainty". Humans are also dealing with "comprehension uncertainty", for example when designing an unmanned, deep-space probe. We design the space probe using our current stock of knowledge in astrophysics; thermodynamics etc., identifying, assessing and resolving the pertinent temporal and knowledge uncertainties. At the same time we are also cognisant of a *gap* in our knowledge. This is not because we haven't been able to fully utilize our current stock of knowledge; rather it is the gap that exists between our current knowledge of deep space etc. and the *universal* knowledge which is outside of our "limits" of comprehension i.e. primarily originating from an ill-defined universal set.

Artificially intelligent decision systems are typically programmed to inexorably seek a 'global' optimum while in reality, the presence of "comprehension uncertainty" will always negate that prospect. What an intelligent system returns as a 'global' optimum is thus at best only such within its current domain knowledge and not a "universal" optimum. But an artificially intelligent system will always terminate its search once it attains what it perceives as the "global" optimum; based on the underlying premise that its current stock of domainspecific knowledge is in fact the universal one! On the other hand, naturally intelligent beings recognize the fundamental gap between current and universal knowledge and so will endeavour to keep expanding their "limits of comprehension".

An artificially intelligent decision system ought to be designed to 'realize' that its current stock of knowledge may not be the universal knowledge pertinent to a decision problem it is invoked to work out. Emulating natural intelligence, *AI models* should aim to be 'autocognisant' of any fundamental knowledge gaps and therefore be able to reconcile any deviations of the "global" from the "universal" optimum. A first step towards that is *effective operationalization* of the "comprehension uncertainty" concept. In the following section we posit and develop a formal conceptualization of the "comprehension uncertainty" concept. This basically involves an extension of classical probability theory to a realm of *higher-order probabilities* in a manner that is computationally tractable and fully reconcilable with the classical theory. Finally we posit and defend a logical framework justifying the due consideration of "comprehension uncertainty" in the context of designing artificially intelligent systems for practical applications in business, industry and society.

#### **3. Developing some necessary theoretical groundwork**

The primary objective of our work here is to simply posit the logically conceivable underpinnings of a probability theory extended to formalize comprehension uncertainty – we believe that our main purpose here is to merely open the proverbial Pandora's Box and thereby spawn a healthy stream of new research along both philosophical as well as mathematical lines. In that desired direction, we firstly posit and prove a fundamental theorem necessary for such an extension to the theory of probability. Subsequently we show some computational 'tests' to illustrate the posited framework.

#### **3.1 A foray into higher order probabilities**

248 Intelligent Systems

As highly evolved, intelligent beings, humans have become adept at continually taking decisions based on information that is subject to various forms of uncertainty. We can negotiate a busy sidewalk more often than not without colliding with other pedestrians and can cross a road safely (again most of the times) without being flattened by a car although we have at best a very imprecise idea of the speed of an oncoming car. Human brain, as the highest seat of natural intelligence, has evolved unique ways of working with various uncertainties including "comprehension uncertainty". Humans are also dealing with "comprehension uncertainty", for example when designing an unmanned, deep-space probe. We design the space probe using our current stock of knowledge in astrophysics; thermodynamics etc., identifying, assessing and resolving the pertinent temporal and knowledge uncertainties. At the same time we are also cognisant of a *gap* in our knowledge. This is not because we haven't been able to fully utilize our current stock of knowledge; rather it is the gap that exists between our current knowledge of deep space etc. and the *universal* knowledge which is outside of our "limits" of comprehension i.e. primarily

Artificially intelligent decision systems are typically programmed to inexorably seek a 'global' optimum while in reality, the presence of "comprehension uncertainty" will always negate that prospect. What an intelligent system returns as a 'global' optimum is thus at best only such within its current domain knowledge and not a "universal" optimum. But an artificially intelligent system will always terminate its search once it attains what it perceives as the "global" optimum; based on the underlying premise that its current stock of domainspecific knowledge is in fact the universal one! On the other hand, naturally intelligent beings recognize the fundamental gap between current and universal knowledge and so

An artificially intelligent decision system ought to be designed to 'realize' that its current stock of knowledge may not be the universal knowledge pertinent to a decision problem it is invoked to work out. Emulating natural intelligence, *AI models* should aim to be 'autocognisant' of any fundamental knowledge gaps and therefore be able to reconcile any deviations of the "global" from the "universal" optimum. A first step towards that is *effective operationalization* of the "comprehension uncertainty" concept. In the following section we posit and develop a formal conceptualization of the "comprehension uncertainty" concept. This basically involves an extension of classical probability theory to a realm of *higher-order probabilities* in a manner that is computationally tractable and fully reconcilable with the classical theory. Finally we posit and defend a logical framework justifying the due consideration of "comprehension uncertainty" in the context of designing artificially

The primary objective of our work here is to simply posit the logically conceivable underpinnings of a probability theory extended to formalize comprehension uncertainty – we believe that our main purpose here is to merely open the proverbial Pandora's Box and thereby spawn a healthy stream of new research along both philosophical as well as

**2.1 How does natural intelligence resolve "comprehension uncertainty" in** 

**decision-making?** 

originating from an ill-defined universal set.

will endeavour to keep expanding their "limits of comprehension".

intelligent systems for practical applications in business, industry and society.

**3. Developing some necessary theoretical groundwork** 

It is well known that much of modern theory of probability rests upon the three fundamental *Kolmogorov axioms* (Kolmogorov, 1956) which are conventionally stated as follows:


It is basically Kolmogorov's second and third axioms as noted above that render any extensions of the probability concept to higher orders (i.e. "probability of probability") superfluous as the information content of any such higher order probability can be satisfactorily transmuted via existing set-theoretic constructs. So, extending to a higher order would arguably yield trivial information. However the Kolmogorov axioms by themselves are also open to 'extensions' – for instance there is previous research that has revisited the proofs of the well-known Bell inequality based on underlying assumptions of separability and noncontextuality and constructed a model of generalized "non-contextual contrapositive conditional probabilities" consistent with the results of the famous Aspect experiment showing in general such probabilities are not necessarily all positive (Atkinson, 2000). By themselves the Kolmogorov axioms do not unequivocally rule out an extension of the definition of the universal set U itself so as to make U possess a *time-dynamic* rather than a *timestatic* nature. So; in effect this means that if we were to consider a time-dynamic version of the universal set; then one would suddenly find that the information content of higher order probability no longer remains trivial i.e. an extension of the probability concept to higher orders (i.e. "probability of probability") is no longer superfluous – in fact it is logical! The good thing is that no new probability calculus needs to be formulated to describe such a theory of higher-order probabilities and this extended theory could still rest on the Kolmogorov axioms and could still draw fundamentally from the standard set-theoretic approach (as we will be demonstrating shortly); by merely using an extended definition of the universal set U which would now denote not merely an event space but a broader concept, which we christen as *event-spacetime*, i.e. an event space that can *evolve* over a time dimension.

Perhaps the only academic work preceding ours to have alluded that a higher-order probability theory is justifiable by an event space evolving over time was that by Haddawy and others (Haddawy, 1996; Lehner, Laskey and Dubois, 1996), where they provided "a logic that incorporates and integrates the concepts of subjective probability, objective probability, time and causality" (Lehner, Laskey and Dubois, 1996). We take a similar philosophical stance but go on to explicitly develop a logically tenable higher-order probability concept in discrete time. We have no doubt that an extension in continuous time is also attainable but it's left for later.

Recognition and Resolution of "Comprehension Uncertainty" in *AI* 251

*P3(E) = P[{E | (U0 = U1)} ∩ {E|(U1 = U2)} ∩ {E|(U2 = U3)}]* 

 *= P[[{E | (U0 = U1)} ∩ {E|(U1 = U2)}] ∩ {E|(U2 = U3)}]* 

 *P4(E) = P[{E | (U0 = U1)} ∩ {E|(U1 = U2)} ∩ {E|(U2 = U3)} ∩ {E|(U3 = U4)}]* 

 *= P[[{E | (U0 = U1)} ∩ {E|(U1 = U2)} ∩ {E|(U2 = U3)}] ∩ {E|(U3 = U4)}]* 

*(E) = P(E). P(t-2)+1(E).[P{(U(t-2)+1=Ut)|E}/P(U(t-2)+1 = Ut)* 

*(E)=P[{E | (U0 = U1)}∩{E|(U1 = U2)}∩{E|(U2 = U3)}∩{E|(U3 = U4)}∩...∩{E|(Ut-1 = Ut)}]* 

*=P[[{E|(U0=U1)}∩{E|(U1=U2)}∩{E|(U2=U3)}∩{E|(U3=U4)}∩...∩{E|(Ut-2 =Ut-1)]∩{E|(Ut-1= Ut)}]* 

*= Pt-1(E) . P{E|(Ut-1 = Ut)}* 

As (2) is identical to (3); by *principle of mathematical induction* the general case is proved for t= n.

Obviously then, if *P(Ut-1=Ut) =* P{*(Ut-1=Ut)|E}*, for all t = 1, 2, 3, ..., n; we will end up with *Pn(E) =*[*P(E)*]*n* which makes this approach to *H-O* probability fully consistent with classical probability theory and in fact a very natural extension thereof if one sees the fundamentally

*(E) = P(E)***.** *Pt-1(E).[P{(Ut-1= Ut)|E}/P(Ut-1= Ut)] for t = 1, 2,3, ..., n*

*P1(E) = P{P0(E)} = P(E). [P{(U0 = U1)|E}/P(U0 = U1)] from lemma 2*

*Pt-1(E) = P(E). P t-2(E).[P{(Ut-2= Ut-1)|E}/P(Ut-2= Ut-1)* (1)

*= P(E).Pt-1(E).[P{(Ut-1=Ut)|E}/P(Ut-1=Ut)* (2)

*= Pt-1(E). P(E). [P{(Ut-1=Ut)|E}/P(Ut-1=Ut)]* (3)

*QED* 

*P2(E) = P{P1(E)} = P(E) . P1(E) . [P{(U1 = U2)|E}/P(U1 = U2)] .. from lemma 3* 

**A fundamental theorem of higher order probabilities (in discrete time)** 

 *= P2(E) . P{E|(U2 = U3)}* 

 *= P3(E) . P{E|(U3 = U4)}* 

The expression for the t-th term is derived from (1) as follows:

Extending to the (t-1)-th term, we can therefore write:

 *Pt*

However we may also write:

time-dynamic characteristic of U.

*Pt*

 *= P(E). P2(E).[P{(U2 = U3)|E}/P(U2 = U3)* 

*= P(E). P3(E).[P{(U3 = U4)|E}/P(U3 = U4)]* 

*If we set P0(E)* 

**Proof** 

 *P(E), then Pt*

#### **Lemma 1**

*The probability that any one of the elementary events contained within the event space-time will occur between two successive time points t0 and t1 given that the contents/contours of the event space remains unchanged from t0 to t1 is unity i.e. P (U0| U0 = U1) = 1. By extension, P(Ut| Ut = Ut+1) = 1 for all t = 0, 1, 2, 3, ...* 

#### **Proof**

Lemma 1 results from a natural extension of Kolmogorov's second axiom if we allow the event space to be of a time-dynamic nature i.e. if U is allowed to evolve through time in discrete intervals.

*QED* 

#### **Lemma 2**

*If the classical probability of occurrence of a specific elementary event E contained within the event space-time is defined as P(E), then the first-order probability of occurrence of such event E becomes P{P(E)} = P1(E) = P{E | (U0 = U1)}* = *P(E).[P{(U0 = U1)|E}/P(U0 = U1)]* 

#### **Proof**

Applying the fundamental law of conditional probability we can write as follows:

P{E | (U0 = U1)} = P{E∩(U0 = U1)}/ P(U0 = U1)

P{E∩(U0 = U1)} = P{(U0 = U1)∩E} = P(E)**.**P{(U0 = U1)|E}; and thus the result follows.

*QED* 

#### **Lemma 3**

*Given the first-order probability of occurrence of elementary event E and assuming that (Ut = Ut+1) and (Ut+1 = Ut+2) are independent for all t = 0, 1, 2, 3, ..., the second-order probability of occurrence of E becomes P2(E) = P(E) . P1(E). [P{(U1 = U2)|E}/P(U1 = U2)].* 

#### **Proof**

By definition, P2(E) = P{P1(E)} = P[{E | (U0 = U1)} ∩ {E|(U1 = U2)}]

Since (Ut = Ut+1) and (Ut+1 = Ut+2) are assumed independent for t = 0, 1, 2, 3, ..., we can write:

$$\Pr\{\left|\mathbf{E}\mid\left(\mathbf{U}\_{0}=\mathbf{U}\_{1}\right)\right\} \cap \left\{\mathbf{E}\mid\left(\mathbf{U}\_{1}=\mathbf{U}\_{2}\right)\right\}\right] = \Pr\{\mathbf{E}\mid\left(\mathbf{U}\_{0}=\mathbf{U}\_{1}\right)\}\dots\Pr\{\mathbf{E}\mid\left(\mathbf{U}\_{1}=\mathbf{U}\_{2}\right)\}.$$

Substituting P{E | (U0 = U1)} with P1(E) and then applying the fundamental law of conditional probability; the result follows.

*QED* 

Thus, given the first-order probability of occurrence of an elementary event E, the secondorder probability is obtained as a "probability of the first-order probability" and is necessarily *either equal to or less than* the first-order probability, as is suggested by common intuition. This logic could then be extended to each of the subsequent higher order probability terms. Based on lemmas 1 – 3, we next propose and prove a fundamental theorem of higher order (hereafter *H*-*O*) probabilities.

*The probability that any one of the elementary events contained within the event space-time will occur between two successive time points t0 and t1 given that the contents/contours of the event space remains unchanged from t0 to t1 is unity i.e. P (U0| U0 = U1) = 1. By extension, P(Ut| Ut = Ut+1) =* 

Lemma 1 results from a natural extension of Kolmogorov's second axiom if we allow the event space to be of a time-dynamic nature i.e. if U is allowed to evolve through time in

*If the classical probability of occurrence of a specific elementary event E contained within the event space-time is defined as P(E), then the first-order probability of occurrence of such event E becomes* 

P{E | (U0 = U1)} = P{E∩(U0 = U1)}/ P(U0 = U1)

P{E∩(U0 = U1)} = P{(U0 = U1)∩E} = P(E)**.**P{(U0 = U1)|E}; and thus the result follows.

*Given the first-order probability of occurrence of elementary event E and assuming that (Ut = Ut+1) and (Ut+1 = Ut+2) are independent for all t = 0, 1, 2, 3, ..., the second-order probability of occurrence* 

Since (Ut = Ut+1) and (Ut+1 = Ut+2) are assumed independent for t = 0, 1, 2, 3, ..., we can write:

P[{E | (U0 = U1)} ∩ {E|(U1 = U2)}] = P{E | (U0 = U1)} **.** P{E|(U1 = U2)}. Substituting P{E | (U0 = U1)} with P1(E) and then applying the fundamental law of

Thus, given the first-order probability of occurrence of an elementary event E, the secondorder probability is obtained as a "probability of the first-order probability" and is necessarily *either equal to or less than* the first-order probability, as is suggested by common intuition. This logic could then be extended to each of the subsequent higher order probability terms. Based on lemmas 1 – 3, we next propose and prove a fundamental

Applying the fundamental law of conditional probability we can write as follows:

*P{P(E)} = P1(E) = P{E | (U0 = U1)}* = *P(E).[P{(U0 = U1)|E}/P(U0 = U1)]* 

*of E becomes P2(E) = P(E) . P1(E). [P{(U1 = U2)|E}/P(U1 = U2)].* 

conditional probability; the result follows.

theorem of higher order (hereafter *H*-*O*) probabilities.

By definition, P2(E) = P{P1(E)} = P[{E | (U0 = U1)} ∩ {E|(U1 = U2)}]

*QED* 

*QED* 

*QED* 

**Lemma 1** 

**Proof** 

**Lemma 2** 

**Proof** 

**Lemma 3** 

**Proof** 

*1 for all t = 0, 1, 2, 3, ...* 

discrete intervals.

#### **A fundamental theorem of higher order probabilities (in discrete time)**

*If we set P0(E) P(E), then Pt (E) = P(E)***.** *Pt-1(E).[P{(Ut-1= Ut)|E}/P(Ut-1= Ut)] for t = 1, 2,3, ..., n* **Proof** 

$$P(E) = P[P(E) | E) = P(E) \left[ P(\{ \mathcal{U}\_{0} = \mathcal{U}\_{1} \} \mid \mathcal{E}) \right] P(\{ \mathcal{U}\_{0} = \mathcal{U}\_{1} \}) \qquad \text{from Lemma 2}$$

$$P^{2}(E) = P[P(E)] = P(E) \cdot P(E) \cdot \left[ P(\{ \mathcal{U}\_{1} = \mathcal{U}\_{2} \} \mid \mathcal{E}) \{ P(\{ \mathcal{U}\_{1} = \mathcal{U}\_{2} \}) \} \right] \qquad \text{from Lemma 3}$$

$$P^{3}(E) = P[\{ E \mid \{ \mathcal{U}\_{0} = \mathcal{U}\_{3} \} \cap \left[ \mathcal{E} \mid \{ \mathcal{U}\_{1} = \mathcal{U}\_{2} \} \right] \cap \left[ \mathcal{E} \mid \{ \mathcal{U}\_{2} = \mathcal{U}\_{3} \} \right]]$$

$$= P[\{ \{ E \mid \: \{ \mathcal{U}\_{0} = \mathcal{U}\_{1} \} \} \cap \left[ \mathcal{E} \mid \{ \mathcal{U}\_{1} = \mathcal{U}\_{2} \} \} \} \cap \left[ \mathcal{E} \mid \{ \mathcal{U}\_{2} = \mathcal{U}\_{3} \} \right]]$$

$$= P^{2}(E) \cdot P(E) \cdot P(\mathcal{E}) \{ \mathcal{U}\_{2} = \mathcal{U}\_{3} \}$$

$$= P(E) \cdot P(E) \cdot \left[ \mathcal{U}\_{1} \{ \mathcal{U}\_{2} = \mathcal{U}\_{3} \} \right] \cap \left[ \mathcal{E} \mid \{ \mathcal{U}\_{1} = \mathcal{U}\_{2} \} \right] \cap \left[ \mathcal{E} \mid \{ \mathcal{U}\_{2} = \mathcal{U}\_{3} \} \right] \cap \left[ \mathcal{E$$

Extending to the (t-1)-th term, we can therefore write:

$$P^{t \cdot 1}(\mathcal{E}) = P(\mathcal{E}). \ P^{t \cdot 2}(\mathcal{E}). \left[ P\{ (\mathcal{L}\_{t \cdot 2} = \mathcal{L}\_{t \cdot 1}) \mid \mathcal{E} \} \middle| P(\mathcal{L}\_{t \cdot 2} = \mathcal{L}\_{t \cdot 1}) \right] \tag{1}$$

The expression for the t-th term is derived from (1) as follows:

$$P(E) = P(E). \ P^{(t \cdot 2) \ast 1}(E). \left[P\{(\mathcal{L}\_{(t \cdot 2) + 1} = \mathcal{U}\_t\} \mid E\}\right] P(\mathcal{L}\_{(t \cdot 2) + 1} = \mathcal{U}\_t)$$

$$= P(E). P^{t \cdot 1}(E). \left[P\{(\mathcal{L}\_{t \cdot 1} = \mathcal{U}\_t) \mid E\}\right] P(\mathcal{L}\_{t \cdot 1} = \mathcal{U}\_t) \tag{2}$$

However we may also write:

*Pt (E)=P[{E | (U0 = U1)}∩{E|(U1 = U2)}∩{E|(U2 = U3)}∩{E|(U3 = U4)}∩...∩{E|(Ut-1 = Ut)}] =P[[{E|(U0=U1)}∩{E|(U1=U2)}∩{E|(U2=U3)}∩{E|(U3=U4)}∩...∩{E|(Ut-2 =Ut-1)]∩{E|(Ut-1= Ut)}] = Pt-1(E) . P{E|(Ut-1 = Ut)} = Pt-1(E). P(E). [P{(Ut-1=Ut)|E}/P(Ut-1=Ut)]* (3)

As (2) is identical to (3); by *principle of mathematical induction* the general case is proved for t= n.

*QED* 

Obviously then, if *P(Ut-1=Ut) =* P{*(Ut-1=Ut)|E}*, for all t = 1, 2, 3, ..., n; we will end up with *Pn(E) =*[*P(E)*]*n* which makes this approach to *H-O* probability fully consistent with classical probability theory and in fact a very natural extension thereof if one sees the fundamentally time-dynamic characteristic of U.

Recognition and Resolution of "Comprehension Uncertainty" in *AI* 253

evolution of the current stock of domain knowledge, there is an erratic pattern in the "probability of probability" of the occurrence of the elementary event of interest because of the fact that some old knowledge that were 'replaced' by new knowledge make comebacks

(*E*); t = 1, 2, ..., 5 for P{(Ut -1 = Ut)/*E*} decreasing from 1 to 0 in steps of 0.05

(*E*); t = 1, 2, ..., 5 for P{(Ut -1 = Ut)/*E*} allowed to randomly oscillate about 0.50

following newer discoveries.

Fig. 3. Contracting Event-Spacetime, [P(*E*) = 0.10]

Fig. 4. Oscillating Event-Spacetime, [P(*E*) = 0.10]

Plot of Pt

Plot of Pt

#### **3.2 Simple computational 'tests' to better illustrate the above-posited concept of**  *H-O* **probability**

To provide a simple illustration of how the *H-O* probabilities would pan out in discrete event-spacetime we have done a series of computations the results of which are graphically represented below. The graphs show the temporal evolution of the event-spacetime in discrete "time steps" and the resulting Pt(*E*) values for t = 1, 2, ..., 5. We assume three temporal evolution forms – "*expanding event-spacetime*", "*contracting event-spacetime*" and "*oscillating event-spacetime*" and plot the Pt(*E*) values for each of these three forms starting with a pervading assumption that P(Ut-1 = Ut) = 1. This assumption simplifies a lot of the computations as Pt(*E*) then depends totally on P{(Ut-1 = Ut)/*E*}. When P{(Ut-1 = Ut)/*E*} = 1, we see that Pt (*E*) converges to P(*E*)t for all values of t. On the other hand, when P{(Ut-1 = Ut)/*E*} = 0, Pt (*E*) converges to zero for all values of t. So, holding P(*E*) = 0.10, in an "expanding event-spacetime", P1(*E*) = P(*E*) = 0.10, p2(*E*) = 0.102 = 0.01 and so on for P{(Ut-1=Ut)/*E*} = 1. For P{(Ut-1=Ut)/*E*} = 0, P1(*E*) = p2(*E*) = P3(*E*) = p4(*E*) = P5(*E*) = 0, while Pt (*E*) values are seen to oscillate for P{(Ut-1=Ut)/*E*} values randomly oscillating about 0.50 – the degree of oscillation decreasing with increasing order of probability i.e. P1(*E*) oscillates more than P2(*E*), P2(*E*) more than P3(*E*) and so on.

Fig. 2. Expanding Event-Spacetime, [P(*E*) = 0.10] Plot of Pt(*E*); t = 1, 2, ..., 5 for P{(Ut -1 = Ut)/*E*} increasing from 0 to 1 in steps of 0.05

The expanding event-spacetime represents the situation where, with passage of time and evolution of the current stock of domain knowledge, there is a steadily increasing "probability of probability" of the occurrence of the elementary event of interest. The contracting event-spacetime represents the situation where, with passage of time and evolution of the current stock of domain knowledge, there is a steadily decreasing "probability of probability" of the occurrence of the elementary event of interest. The oscillating event-spacetime represents the situation where, with passage of time and

To provide a simple illustration of how the *H-O* probabilities would pan out in discrete event-spacetime we have done a series of computations the results of which are graphically represented below. The graphs show the temporal evolution of the event-spacetime in discrete "time steps" and the resulting Pt(*E*) values for t = 1, 2, ..., 5. We assume three temporal evolution forms – "*expanding event-spacetime*", "*contracting event-spacetime*" and "*oscillating event-spacetime*" and plot the Pt(*E*) values for each of these three forms starting with a pervading assumption that P(Ut-1 = Ut) = 1. This assumption simplifies a lot of the computations as Pt(*E*) then depends totally on P{(Ut-1 = Ut)/*E*}. When P{(Ut-1 = Ut)/*E*} = 1,

"expanding event-spacetime", P1(*E*) = P(*E*) = 0.10, p2(*E*) = 0.102 = 0.01 and so on for P{(Ut-1=Ut)/*E*} = 1. For P{(Ut-1=Ut)/*E*} = 0, P1(*E*) = p2(*E*) = P3(*E*) = p4(*E*) = P5(*E*) = 0, while Pt

values are seen to oscillate for P{(Ut-1=Ut)/*E*} values randomly oscillating about 0.50 – the degree of oscillation decreasing with increasing order of probability i.e. P1(*E*) oscillates more

Plot of Pt(*E*); t = 1, 2, ..., 5 for P{(Ut -1 = Ut)/*E*} increasing from 0 to 1 in steps of 0.05

The expanding event-spacetime represents the situation where, with passage of time and evolution of the current stock of domain knowledge, there is a steadily increasing "probability of probability" of the occurrence of the elementary event of interest. The contracting event-spacetime represents the situation where, with passage of time and evolution of the current stock of domain knowledge, there is a steadily decreasing "probability of probability" of the occurrence of the elementary event of interest. The oscillating event-spacetime represents the situation where, with passage of time and

(*E*) converges to P(*E*)t for all values of t. On the other hand, when P{(Ut-1 =

(*E*) converges to zero for all values of t. So, holding P(*E*) = 0.10, in an

(*E*)

**3.2 Simple computational 'tests' to better illustrate the above-posited concept of** 

*H-O* **probability** 

we see that Pt

Ut)/*E*} = 0, Pt

than P2(*E*), P2(*E*) more than P3(*E*) and so on.

Fig. 2. Expanding Event-Spacetime, [P(*E*) = 0.10]

evolution of the current stock of domain knowledge, there is an erratic pattern in the "probability of probability" of the occurrence of the elementary event of interest because of the fact that some old knowledge that were 'replaced' by new knowledge make comebacks following newer discoveries.

Fig. 3. Contracting Event-Spacetime, [P(*E*) = 0.10] Plot of Pt(*E*); t = 1, 2, ..., 5 for P{(Ut -1 = Ut)/*E*} decreasing from 1 to 0 in steps of 0.05

Fig. 4. Oscillating Event-Spacetime, [P(*E*) = 0.10] Plot of Pt (*E*); t = 1, 2, ..., 5 for P{(Ut -1 = Ut)/*E*} allowed to randomly oscillate about 0.50

Recognition and Resolution of "Comprehension Uncertainty" in *AI* 255

In its current state, the design of artificially intelligent systems is pre-occupied with solving the "how" problems and as such do not quite recognize the need for resolving comprehension uncertainty. In fact, the concept of comprehension uncertainty was not even formally posited prior to this work by us although there have been a few takes on the mathematics of *H-O* probabilities. Earlier researchers mainly found the concept of *H-O* probabilities superfluous because they failed to view it in the context of formalizing

However, given that the exact emulation of human intelligence continues to remain the Holy Grail for *AI* researchers, they have to grapple with comprehension uncertainty at some point or the other. The reason for this is simple – a hallmark of human intelligence is that it recognizes the limitations of the current stock of knowledge from which it draws. Thus any artificial system that ultimately seeks to emulate that intelligence must also necessarily see the limitations in current domain knowledge and allow for the fact that the current domain knowledge can evolve over time so that the global optimum attained with the current stock of knowledge may not remain the same at a future time. Once an artificially intelligent system is hardwired to recognize the *time-dynamic* aspect of the relevant event space within which it has to calculate the probabilities of certain outcomes and take a decision so as to maximize the expected value of the most desirable outcome, it will not terminate its search as soon as global optimality is reached in terms of the contents/contours of the current event space. It would rather go into a 'dormant' mode and continue to monitor the evolution of the event space and 're-engage' in its search as soon as P{(Ut-1=Ut)/*E*} > 0 at any

With the formal hardwiring of comprehension uncertainty within the core design of an artificially intelligent system it can be trained to transcend from simply answering the "how" to ultimately formulating the "why" – firstly; why is the current body of knowledge an exhaustive source to draw from for finding the optimal solution to a particular problem and secondly; why that current body of knowledge may not be continue to remain an exhaustive source to draw from for all time in future. When it has been trained to formulate these "why" questions, only then can we expect an artificially intelligent system to take that

Atkinson, D. (2000). Bell's Inequalities and Kolmogorov's Axioms, *Pramana – Journal of* 

Ball, L. J. and B. T. Christensen (2009). Analogical reasoning and mental simulation in

Bhattacharya, S., Y. Wang and D. Xu (2010). Beyond Simon's Means-Ends Analysis: Natural

Clark, D. A. (1990). Numerical and symbolic approaches to uncertainty management in AI,

Decision-Making, *Minds and Machines*, Vol. 20, No. 3, pp.327–347

*Artificial Intelligence Review*, Vol. 4, pp. 109–146

design: two strategies linked to uncertainty resolution, *Design Studies*, Vol. 30, No.

Creativity and the Unanswered 'Why' in the Design of Intelligent Systems for

significant leap towards finally gaining parity with natural intelligence.

**4. Conclusion: "comprehending the incomprehensible" – the future of** 

comprehension uncertainty like we have done in this article.

*AI* **systems design** 

subsequent time point.

**5. References** 

*Physics*, Vol. 54, pp. 1–15

2, pp. 169–186

#### **3.3** *H-O* **probability implications for intelligent resolution of comprehension uncertainty**

Although we do not mathematically compute *H-O* probabilities while taking decisions (or for that matter even ordinary probabilities), human intelligence does enough 'background processing' of fringe information (mostly even without knowing) to 'see' a bigger picture of the likely scenarios. Going back to the example of crossing a busy road, we are continuously processing information (often unknowingly) from the environment in terms of the rapidly changing pertinent event space. As long as the pertinent event space is 'pre-populated' with likely forms of road hazards, an artificially intelligent system can be 'trained' to emulate human decision-making and cross the road. It is when the contents of the pertinent event space dynamically changes that would throw off even the most advanced of AI-based systems given the current state of design of such systems. This is pretty much what Bhattacharya, Wang and Xu (2010) identified as a 'gap' in the current state of design of intelligent systems. The current design paradigm is overwhelmingly concerned with the "how" rather than the "why" – and resolution of comprehension uncertainty involves more of the "why". Rather than trying to answer "how to avoid being hit by a vehicle or some other hazard while crossing" *AI* designers ought to be focusing on "why are we vulnerable while crossing a busy road".

As soon as the focus of the design shifts to the "why", the link with comprehension uncertainty becomes a very natural extension thereof. Then we are simply asking *why* a particular event space is a pertinent one for the problem at hand? The natural answer is that in a specified time window, it contains all the elementary events out of which one or a few are conducive for the desired outcome. Then the question naturally progresses to what would happen outside that specified time window? If we are pre-populating the pertinent event space and then assuming that it would hold good for all times, it would be at the cost of ignoring comprehension uncertainty which can defeat the AI design. At this point it is perhaps useful to again remind readers that it is not the vagueness or imprecision associated with some contents of an event space that is of importance here (existing uncertainty resolution methods like rough sets, fuzzy logic etc. are adequate for dealing with those) – it is a temporal instability of the event space itself that is crux of the comprehension uncertainty concept.

The mathematics of *H-O* probabilities then offers a plausible route towards formal incorporation of comprehension uncertainty within artificially intelligent systems designed to replicate naturally intelligent decision-making. As naturally intelligent beings, humans are capable of somehow grasping the "limits to comprehension" that result from a gap between current knowledge and universal knowledge. If this was not the case then 'research' as an intellectual endeavour would have ceased! In the current design paradigm the focus is on training *AI* models to 'search' for global optimality while, ideally, the focus ought to be on training such models to do 'research' rather than 'search'! Recognition and incorporation of comprehension uncertainty in their learning framework would at least allow future *AI* models to 'grasp' the limits to comprehension so as not to invariably terminate as soon as a 'globally optimal' decision point has been reached using the current domain knowledge.

Although we do not mathematically compute *H-O* probabilities while taking decisions (or for that matter even ordinary probabilities), human intelligence does enough 'background processing' of fringe information (mostly even without knowing) to 'see' a bigger picture of the likely scenarios. Going back to the example of crossing a busy road, we are continuously processing information (often unknowingly) from the environment in terms of the rapidly changing pertinent event space. As long as the pertinent event space is 'pre-populated' with likely forms of road hazards, an artificially intelligent system can be 'trained' to emulate human decision-making and cross the road. It is when the contents of the pertinent event space dynamically changes that would throw off even the most advanced of AI-based systems given the current state of design of such systems. This is pretty much what Bhattacharya, Wang and Xu (2010) identified as a 'gap' in the current state of design of intelligent systems. The current design paradigm is overwhelmingly concerned with the "how" rather than the "why" – and resolution of comprehension uncertainty involves more of the "why". Rather than trying to answer "how to avoid being hit by a vehicle or some other hazard while crossing" *AI* designers ought to be focusing on "why are we vulnerable

As soon as the focus of the design shifts to the "why", the link with comprehension uncertainty becomes a very natural extension thereof. Then we are simply asking *why* a particular event space is a pertinent one for the problem at hand? The natural answer is that in a specified time window, it contains all the elementary events out of which one or a few are conducive for the desired outcome. Then the question naturally progresses to what would happen outside that specified time window? If we are pre-populating the pertinent event space and then assuming that it would hold good for all times, it would be at the cost of ignoring comprehension uncertainty which can defeat the AI design. At this point it is perhaps useful to again remind readers that it is not the vagueness or imprecision associated with some contents of an event space that is of importance here (existing uncertainty resolution methods like rough sets, fuzzy logic etc. are adequate for dealing with those) – it is a temporal instability of the event space itself that is crux of the comprehension

The mathematics of *H-O* probabilities then offers a plausible route towards formal incorporation of comprehension uncertainty within artificially intelligent systems designed to replicate naturally intelligent decision-making. As naturally intelligent beings, humans are capable of somehow grasping the "limits to comprehension" that result from a gap between current knowledge and universal knowledge. If this was not the case then 'research' as an intellectual endeavour would have ceased! In the current design paradigm the focus is on training *AI* models to 'search' for global optimality while, ideally, the focus ought to be on training such models to do 'research' rather than 'search'! Recognition and incorporation of comprehension uncertainty in their learning framework would at least allow future *AI* models to 'grasp' the limits to comprehension so as not to invariably terminate as soon as a 'globally optimal' decision point has been reached using the current

**3.3** *H-O* **probability implications for intelligent resolution of comprehension** 

**uncertainty**

while crossing a busy road".

uncertainty concept.

domain knowledge.

#### **4. Conclusion: "comprehending the incomprehensible" – the future of**  *AI* **systems design**

In its current state, the design of artificially intelligent systems is pre-occupied with solving the "how" problems and as such do not quite recognize the need for resolving comprehension uncertainty. In fact, the concept of comprehension uncertainty was not even formally posited prior to this work by us although there have been a few takes on the mathematics of *H-O* probabilities. Earlier researchers mainly found the concept of *H-O* probabilities superfluous because they failed to view it in the context of formalizing comprehension uncertainty like we have done in this article.

However, given that the exact emulation of human intelligence continues to remain the Holy Grail for *AI* researchers, they have to grapple with comprehension uncertainty at some point or the other. The reason for this is simple – a hallmark of human intelligence is that it recognizes the limitations of the current stock of knowledge from which it draws. Thus any artificial system that ultimately seeks to emulate that intelligence must also necessarily see the limitations in current domain knowledge and allow for the fact that the current domain knowledge can evolve over time so that the global optimum attained with the current stock of knowledge may not remain the same at a future time. Once an artificially intelligent system is hardwired to recognize the *time-dynamic* aspect of the relevant event space within which it has to calculate the probabilities of certain outcomes and take a decision so as to maximize the expected value of the most desirable outcome, it will not terminate its search as soon as global optimality is reached in terms of the contents/contours of the current event space. It would rather go into a 'dormant' mode and continue to monitor the evolution of the event space and 're-engage' in its search as soon as P{(Ut-1=Ut)/*E*} > 0 at any subsequent time point.

With the formal hardwiring of comprehension uncertainty within the core design of an artificially intelligent system it can be trained to transcend from simply answering the "how" to ultimately formulating the "why" – firstly; why is the current body of knowledge an exhaustive source to draw from for finding the optimal solution to a particular problem and secondly; why that current body of knowledge may not be continue to remain an exhaustive source to draw from for all time in future. When it has been trained to formulate these "why" questions, only then can we expect an artificially intelligent system to take that significant leap towards finally gaining parity with natural intelligence.

#### **5. References**


**1. Introduction** 

area descriptions [O'Looney, 2000].

**12** 

*Czech Republic* 

**Intelligent Systems in Cartography** 

According to the recent progress and technical development in Geographic Information Science (GIScience) [Kraak, MacEachren, 1999], and in information technology we can trace the progressive significance of the role of maps, images, and computer graphics as mediators of collaboration - in a range of contexts including environmental and urban planning, resource management, scientific inquiry, and education [Brewer et al., 2000]. Maps became a tool for sharing knowledge around people. They are comprehended as a unique expression tool used for a variety of purposes that can be broadly grouped around two main roles: maps as tools for analysis, problem solving and decision making "visual thinking", [MacEachren, Kraak, 1997], and maps as tools for communication of ideas between people. Although the communicative role of maps seems to fully comply with the cartographic tradition, it should be borne in mind that the concept of cartographic communication has recently extended [Andrienko, Andrienko, Voss, 2002]. Maps are unique means for communication of adequate amount of spatial information. Visualizing allows us to grasp and retain larger amount of information compared to the usage of words. Without the visual image, recalling the same information would require memorizing a long list of

If the maps are processed correctly, they transmit spatial information accurately and quickly. If some of the rules of cartography are violated, communication of spatial information is inaccurate. The communication of spatial information is sometimes completely wrong. Subsequently, the map-reader can be significantly affected by the result of representation of information. From the other point of view, badly understood map may have fatal consequences in crisis management when transferring of the right information between collaborating people is necessary. In this context, map plays the role of symbolic operator able to act in such a decision making, characterized by urgency and criticality.

Thus, the good knowledge of all the rules for maps making is expected from the map maker. Knowledge of design principles can help the user create a highly specialized view on the data. Customized and right visualized data can help viewers identify patterns, which can be

Map making process can be done in two main ways. Firstly, the users make map from some datasets using adequate software. The opposite situation requires map server as end tool for visualizing of datasets. In both cases is necessary build-in acquired cartographical

lost when using the un-adequate method [O'Looney, 2000].

Zdena Dobesova and Jan Brus *Palacký University in Olomouc* 


### **Intelligent Systems in Cartography**

Zdena Dobesova and Jan Brus

*Palacký University in Olomouc Czech Republic* 

#### **1. Introduction**

256 Intelligent Systems

Ding, Z., X. Zhu, J. Zhao and H. Xu (2008). New knowledge acquisition method in

Haddawy, P. (1996). Believing change and changing belief, *IEEE Transactions on Systems, Man and Cybernetics, Part A: Systems and Humans*, Vol. 26, No. 3, pp.385–396 Halpern, J. Y. (2003). *Reasoning about uncertainty*, Cambridge, MA, and London: MIT Press Huang, H., M. Pasquier and C. Quek (2009). Financial Market Trading System with a

Kolmogorov, A. N. (1956). *Foundations of the Theory of Probability*, Second Ed. (English),

Kosut, R. L., M. K. Lau and S. P. Boyd (1992). Set-Membership Identification of Systems with

Lehner, P. E., K. B. Laskey and D. Dubois (1996). An introduction to issues in higher order

Schrödinger, E. (1935). Die gegenwärtige Situation in der Quantenmechanik (The present situation in quantum mechanics), *Naturwissenschaften*, Vol. 23, pp. 807–849 Smarandache, F. (2002). Preface: An Introduction to Neutrosophy, Neutrosophic Logic,

Sicilia, M-A. (2006). On Some Problems of Decision-Making under Uncertainty in the Semantic Web, *Studies in Fuzziness and Soft Computing*, Vol. 204, pp. 231–246 Turing, A. M. (1950). Computing machinery and intelligence, *Mind*, Vol. 59, pp. 433–460 Walley, P. and G. de Cooman (2001). A behavioral model for linguistic uncertainty,

Yang, Z., S. C. P. Yam, L. K. Li and Y. Wang (2010). Universal Repetitive Learning Control

Zadeh, L. A. (1965). Fuzzy sets, *Information and Control*, Vol. 8, pp. 338–353

for Nonparametric Uncertainty and Unknown State-Dependent Control Direction Matrix, *IEEE Transactions on Automatic Control*, Vol. 55, No. 7, pp. 1710 – 1715

Hangzhou, China (26th-28th August), pp. 196–200

*Evolutionary Computation*, Vol. 13, No. 1, pp. 56–70

Chelsea Publishing Company, NY

*Humans,* Vol. 26, No. 3, pp.289–293

*Information Sciences*, Vol. 134, pp. 1–37

Vol. 37, No. 7, pp. 929–941

AZ

incomplete information system based on rough set and self-adaptive genetic algorithm, In: *Proceedings of IEEE International Conference on Granular Computing*,

Hierarchical Co evolutionary Fuzzy Predictive Model, *IEEE Transactions on* 

Parametric and Nonparametric Uncertainty, *IEEE Transactions on Automatic Control*,

uncertainty, *IEEE Transactions on Systems, Man and Cybernetics, Part A: Systems and* 

Neutrosophic Set, and Neutrosophic Probability and Statistics, In: F. Smarandache (Ed.) *Proceedings of the First International Conference on Neutrosophy, Neutrosophic Logic, Neutrosophic Set, and Neutrosophic Probability and Statistics*, Xiquan, Phoenix, According to the recent progress and technical development in Geographic Information Science (GIScience) [Kraak, MacEachren, 1999], and in information technology we can trace the progressive significance of the role of maps, images, and computer graphics as mediators of collaboration - in a range of contexts including environmental and urban planning, resource management, scientific inquiry, and education [Brewer et al., 2000]. Maps became a tool for sharing knowledge around people. They are comprehended as a unique expression tool used for a variety of purposes that can be broadly grouped around two main roles: maps as tools for analysis, problem solving and decision making "visual thinking", [MacEachren, Kraak, 1997], and maps as tools for communication of ideas between people. Although the communicative role of maps seems to fully comply with the cartographic tradition, it should be borne in mind that the concept of cartographic communication has recently extended [Andrienko, Andrienko, Voss, 2002]. Maps are unique means for communication of adequate amount of spatial information. Visualizing allows us to grasp and retain larger amount of information compared to the usage of words. Without the visual image, recalling the same information would require memorizing a long list of area descriptions [O'Looney, 2000].

If the maps are processed correctly, they transmit spatial information accurately and quickly. If some of the rules of cartography are violated, communication of spatial information is inaccurate. The communication of spatial information is sometimes completely wrong. Subsequently, the map-reader can be significantly affected by the result of representation of information. From the other point of view, badly understood map may have fatal consequences in crisis management when transferring of the right information between collaborating people is necessary. In this context, map plays the role of symbolic operator able to act in such a decision making, characterized by urgency and criticality.

Thus, the good knowledge of all the rules for maps making is expected from the map maker. Knowledge of design principles can help the user create a highly specialized view on the data. Customized and right visualized data can help viewers identify patterns, which can be lost when using the un-adequate method [O'Looney, 2000].

Map making process can be done in two main ways. Firstly, the users make map from some datasets using adequate software. The opposite situation requires map server as end tool for visualizing of datasets. In both cases is necessary build-in acquired cartographical

value.

map (bottom)

Intelligent Systems in Cartography 259

appropriate scales according to visualization qualitative or quantitative data. When a quantity type of data is selected than the predefined color scales of tones based on one color with different saturation will be automatically offered in ArcGIS. This offer is cartographically correct. However, the user can make mistakes here because bad choice scales are also offered. This mistake of choosing wrong color ramp for expression quality or quantity is represented on Fig 1. This map visualizes the different six weeks of the student vacation in the Czech Republic. This qualitative phenomenon is correctly expressed with different tone of the color (yellow, orange, light blue, green, dark blue and violet) for every week on the upper map. Wrong usage of color for expression of six weeks by graduated color ramp (colours from yellow to brown) is in the map on top in Fig. 1. This graduated color ramp can be used only for quantitative data. Light color (yellow) expresses small value, dark color (brown) expresses big value. The week of vacation is not small or big

Fig. 1. Example of bad use of color ramp for qualitative data - wrong map (up) and correct

In developing an intelligent system, there are two related sets of problems. Transformation of existing cartographic practice into rule-based knowledge stands the first and the second is to guide the system through the map-making task. The knowledge in the domain is encoded

knowledge into these systems. There is a need for implementation of cartographic rules directly into the programs for the map making especially into GIS software.

The usage of intelligent systems has been enabled by the development in the field of artificial intelligence. Therefore, these systems find application in many sectors of cartography. Real cartographer can be partly substituted by the utilization of knowledge system (intelligent system).

#### **2. Cartography and intelligent systems**

Computer-assisted thematic cartography has been highlighted in the forefront of interest by the following development in the field of GIS and map making but also thanks to the expansion of improper map-making. Usage of different methods in thematic cartography is very dependent on the specific type of map, user and the resulting information. Cartographer accesses very often at this stage and determines what is appropriate and what is not. The possibility of intelligent system usage can be found in that stage.

Quantity of used thematic cartography methods, different types and quality of input data and other factors, however, can cause problems. The creating of a high-quality and comprehensive system for thematic cartography is extremely complex task. The main idea of designing the decision-making support system in thematic mapping is using all kinds of technologies and methods. The aim is to solve the decision-making problems in thematic mapping in order to make a perfect map through operating intelligent system by users [Quo, 1993]. Key decision-making issues referred to the thematic map design should be analyzed clearly at the beginning of designing a good intelligent system. Nevertheless, corresponding decision-making models and reasoning methods should be proposed according to different problems.

In order to transfer map information effectively, it must reduce the noise hidden behind the map information and prevent over much map information. In thematic cartography, there are more than 10 commonly known thematic map types, namely point diagram maps, linear diagram maps, chorochromatic mosaic maps, isoline maps, stereoscopic perspective methods, nominal point symbol maps, proportional symbol maps, dot methods (dot mapping), classification ratio method (choropleth maps), statistical maps (areal diagram methods), cartographic arrowhead methods, triangle charts law [Quo, Ren, 2003].

Various geographic data have a different structure of data. Every method should be corresponding to geographic data characteristics. Moreover, only some specific types of map graphics express specific geographic phenomena (Population Pyramid). This is the very important part of thematic cartography. Different methods will emphasize differently on different map data characteristics. Furthermore, some data characteristics can be only expressed by particular methods. When we can distinguish type of data and their structure, we will be able to know which method to choose [Andrienko, Andrienko, Voss, 2002]. We can select a different map representation according to the spatial distribution of quality, quantity, grade combined, compared, direction and temporal options.

Producers of GIS software try to incorporate sub-expert cartographic knowledge as part of the program functionality. For example, we can consider the offer of color scale as a specific program codified cartographic knowledge in ArcGIS software. A program shows

knowledge into these systems. There is a need for implementation of cartographic rules

The usage of intelligent systems has been enabled by the development in the field of artificial intelligence. Therefore, these systems find application in many sectors of cartography. Real cartographer can be partly substituted by the utilization of knowledge

Computer-assisted thematic cartography has been highlighted in the forefront of interest by the following development in the field of GIS and map making but also thanks to the expansion of improper map-making. Usage of different methods in thematic cartography is very dependent on the specific type of map, user and the resulting information. Cartographer accesses very often at this stage and determines what is appropriate and what

Quantity of used thematic cartography methods, different types and quality of input data and other factors, however, can cause problems. The creating of a high-quality and comprehensive system for thematic cartography is extremely complex task. The main idea of designing the decision-making support system in thematic mapping is using all kinds of technologies and methods. The aim is to solve the decision-making problems in thematic mapping in order to make a perfect map through operating intelligent system by users [Quo, 1993]. Key decision-making issues referred to the thematic map design should be analyzed clearly at the beginning of designing a good intelligent system. Nevertheless, corresponding decision-making models and reasoning methods should be proposed

In order to transfer map information effectively, it must reduce the noise hidden behind the map information and prevent over much map information. In thematic cartography, there are more than 10 commonly known thematic map types, namely point diagram maps, linear diagram maps, chorochromatic mosaic maps, isoline maps, stereoscopic perspective methods, nominal point symbol maps, proportional symbol maps, dot methods (dot mapping), classification ratio method (choropleth maps), statistical maps (areal diagram

Various geographic data have a different structure of data. Every method should be corresponding to geographic data characteristics. Moreover, only some specific types of map graphics express specific geographic phenomena (Population Pyramid). This is the very important part of thematic cartography. Different methods will emphasize differently on different map data characteristics. Furthermore, some data characteristics can be only expressed by particular methods. When we can distinguish type of data and their structure, we will be able to know which method to choose [Andrienko, Andrienko, Voss, 2002]. We can select a different map representation according to the spatial distribution of quality,

Producers of GIS software try to incorporate sub-expert cartographic knowledge as part of the program functionality. For example, we can consider the offer of color scale as a specific program codified cartographic knowledge in ArcGIS software. A program shows

methods), cartographic arrowhead methods, triangle charts law [Quo, Ren, 2003].

quantity, grade combined, compared, direction and temporal options.

directly into the programs for the map making especially into GIS software.

is not. The possibility of intelligent system usage can be found in that stage.

system (intelligent system).

according to different problems.

**2. Cartography and intelligent systems** 

appropriate scales according to visualization qualitative or quantitative data. When a quantity type of data is selected than the predefined color scales of tones based on one color with different saturation will be automatically offered in ArcGIS. This offer is cartographically correct. However, the user can make mistakes here because bad choice scales are also offered. This mistake of choosing wrong color ramp for expression quality or quantity is represented on Fig 1. This map visualizes the different six weeks of the student vacation in the Czech Republic. This qualitative phenomenon is correctly expressed with different tone of the color (yellow, orange, light blue, green, dark blue and violet) for every week on the upper map. Wrong usage of color for expression of six weeks by graduated color ramp (colours from yellow to brown) is in the map on top in Fig. 1. This graduated color ramp can be used only for quantitative data. Light color (yellow) expresses small value, dark color (brown) expresses big value. The week of vacation is not small or big value.

Fig. 1. Example of bad use of color ramp for qualitative data - wrong map (up) and correct map (bottom)

In developing an intelligent system, there are two related sets of problems. Transformation of existing cartographic practice into rule-based knowledge stands the first and the second is to guide the system through the map-making task. The knowledge in the domain is encoded

Intelligent Systems in Cartography 261

[Meng, 2003]. Generalization entails a number of different rules that must be correctly applied in a certain sequence. Different roles have different rules and different knowledge base. This compilation of a dynamic system is a possible solution to the automatic generalization. In the context of digital cartography and expert systems is therefore very necessary to examine and interpret the processes at manual generalization. The overwriting the procedure of the cartographer into a sequence of a procedure of very well defined

Implementation of the knowledge of experts to the programs for work with a map can greatly specify and simplify the whole process. Automatic generalization is interesting example. These intelligent guides can be found in different software such as ArcGIS, DynaGen and LaserScan. The development of intelligent systems is a major commercial application of artificial intelligence (AI) which is proposed to increase the quality and

For common users it is much more preferable to use freely available software resources. These resources can help with the creation of maps. In the following text, there are some of these applications. The "MapBrewer" system is named after the researcher and cartographer Cynthia Brewer. It is a new type of a system developed to encourage the creation of maps. It

Three versions, namely ColourBrewer [Harrower, Brewer, 2003] instrument for the correct choice of colour composition, SymbolBrewer [Schnabel, 2005] for the selection of appropriate map symbols, and TypeBrewer [Sheesley, 2006] to the appropriate font, are now available. These described systems can be rather referred as "digital teaching assistants".

processes is a key objective in creating a successful expert system [Lee, 1994].

availability of knowledge for automated decision-making [Boss, 1991].

Fig. 2. ColorBrewer 2.0 – Color advice for cartography

helps the user always with only one particular aspect in the production of maps.

in the form of rules, which constitute the building blocks of the knowledge base. The application logic and the procedural information of the system are described by rules and operated on objects, classes and slots [Stefanakis, Tsoulos 2005]. The structure and organization of the knowledge base is critical for efficiency and the overall performance of the system. Only features referred to the usage of a map can be presented and, on the other hand, only important elements can be shown when there are too many features [Hua, 1991].

That is why it is necessary to include all the potential factors to the database when designing such a system, or it is necessary to focus only on some issues in a map-making process.

In disrespect of the basic rules, there may be restrictions of expressing the ability of the map or the cartographic expression becomes unreadable. Intelligent system can assist to the correct selection of colour in accordance with the rules of cartography. E.g. conservation principle of conventionality (blue colour for waters, brown colour for contour lines), conservation principle of associativity (green forests for topographic maps), the right choice of colours for the qualitative data or the correct shade of colour for expressing the intensity of the phenomenon. They can take into consideration the type of imaging methods and the people suffering from daltonism etc. There is also an art to displaying information visually, and sometimes principles contradict each other [Andrienko, Andrienko, Voss, 2002].

The basic principle of the intelligent system is to divide the whole process into subsections, which affect the result. The resulting proposed system must be coherent and comprehensive. Good comprehensive intelligent system for thematic cartography should be able to propose appropriate solutions of the problem. Excellent intelligent system should be even able to offer not only one possible solution but also give the explanation and justification to the user.

#### **3. Cartographic intelligent systems with a specific knowledge**

With the development of digital cartography and transfer maps to digital form there is an increasing need to vectorize and generalize properly. Both processes are widely used in the last decade. This process, however, requires the presence of expert and correction of the process. Software that directly vectorize scanned image maps can be divided into automatic and semi-automatic, depending on the modes of information processing [Hori, Tanigawa, 1993], [Eikvil, Aas, Koren, 1995]. Most of current automatic vectorial systems apply the same method for all maps and do not take into consideration their different nature. It is expected from the user as the most accurate manual setting as possible, which presupposes good knowledge of the problems and knowledge of the system used [Hori, Tanigawa, 1993]. One option is to use the knowledge base and thus reduce the overall demand of cartographic literacy of the users and facilitate the whole process of vectorization. In conjunction with the knowledge base we get a system that is able to give results very similar to the of outputs highly sophisticated manual digitization. In addition, it provides more agreeable user interface which allows the selection of appropriate parameters in accordance with the visual information contained in the original map.

Even generalizing algorithms of existing systems often ignore the role of maps or fuzzy logic to optimize the process. There are thus not able to extract hidden information. The specific knowledge, which is not taken into account in so far known algorithms, is required

in the form of rules, which constitute the building blocks of the knowledge base. The application logic and the procedural information of the system are described by rules and operated on objects, classes and slots [Stefanakis, Tsoulos 2005]. The structure and organization of the knowledge base is critical for efficiency and the overall performance of the system. Only features referred to the usage of a map can be presented and, on the other hand, only important elements can be shown when there are too many features [Hua, 1991]. That is why it is necessary to include all the potential factors to the database when designing such a system, or it is necessary to focus only on some issues in a map-making process.

In disrespect of the basic rules, there may be restrictions of expressing the ability of the map or the cartographic expression becomes unreadable. Intelligent system can assist to the correct selection of colour in accordance with the rules of cartography. E.g. conservation principle of conventionality (blue colour for waters, brown colour for contour lines), conservation principle of associativity (green forests for topographic maps), the right choice of colours for the qualitative data or the correct shade of colour for expressing the intensity of the phenomenon. They can take into consideration the type of imaging methods and the people suffering from daltonism etc. There is also an art to displaying information visually,

and sometimes principles contradict each other [Andrienko, Andrienko, Voss, 2002].

**3. Cartographic intelligent systems with a specific knowledge** 

justification to the user.

information contained in the original map.

The basic principle of the intelligent system is to divide the whole process into subsections, which affect the result. The resulting proposed system must be coherent and comprehensive. Good comprehensive intelligent system for thematic cartography should be able to propose appropriate solutions of the problem. Excellent intelligent system should be even able to offer not only one possible solution but also give the explanation and

With the development of digital cartography and transfer maps to digital form there is an increasing need to vectorize and generalize properly. Both processes are widely used in the last decade. This process, however, requires the presence of expert and correction of the process. Software that directly vectorize scanned image maps can be divided into automatic and semi-automatic, depending on the modes of information processing [Hori, Tanigawa, 1993], [Eikvil, Aas, Koren, 1995]. Most of current automatic vectorial systems apply the same method for all maps and do not take into consideration their different nature. It is expected from the user as the most accurate manual setting as possible, which presupposes good knowledge of the problems and knowledge of the system used [Hori, Tanigawa, 1993]. One option is to use the knowledge base and thus reduce the overall demand of cartographic literacy of the users and facilitate the whole process of vectorization. In conjunction with the knowledge base we get a system that is able to give results very similar to the of outputs highly sophisticated manual digitization. In addition, it provides more agreeable user interface which allows the selection of appropriate parameters in accordance with the visual

Even generalizing algorithms of existing systems often ignore the role of maps or fuzzy logic to optimize the process. There are thus not able to extract hidden information. The specific knowledge, which is not taken into account in so far known algorithms, is required [Meng, 2003]. Generalization entails a number of different rules that must be correctly applied in a certain sequence. Different roles have different rules and different knowledge base. This compilation of a dynamic system is a possible solution to the automatic generalization. In the context of digital cartography and expert systems is therefore very necessary to examine and interpret the processes at manual generalization. The overwriting the procedure of the cartographer into a sequence of a procedure of very well defined processes is a key objective in creating a successful expert system [Lee, 1994].

Implementation of the knowledge of experts to the programs for work with a map can greatly specify and simplify the whole process. Automatic generalization is interesting example. These intelligent guides can be found in different software such as ArcGIS, DynaGen and LaserScan. The development of intelligent systems is a major commercial application of artificial intelligence (AI) which is proposed to increase the quality and availability of knowledge for automated decision-making [Boss, 1991].

For common users it is much more preferable to use freely available software resources. These resources can help with the creation of maps. In the following text, there are some of these applications. The "MapBrewer" system is named after the researcher and cartographer Cynthia Brewer. It is a new type of a system developed to encourage the creation of maps. It helps the user always with only one particular aspect in the production of maps.

Fig. 2. ColorBrewer 2.0 – Color advice for cartography

Three versions, namely ColourBrewer [Harrower, Brewer, 2003] instrument for the correct choice of colour composition, SymbolBrewer [Schnabel, 2005] for the selection of appropriate map symbols, and TypeBrewer [Sheesley, 2006] to the appropriate font, are now available. These described systems can be rather referred as "digital teaching assistants".

Intelligent Systems in Cartography 263

dialogue of expert with users – results are interaction between knowledge and way

The knowledge engineer should be aware that expert knowledge is more than one kind and not all this knowledge can be acquired from one person. An interview with only one expertcartographer can avoid some fail in expert system. Interview with group of cartographers is better. The suitable way of interview is brainstorming. There is necessary more punctually prepare interview and carefully lead interview with group of experts. There is also danger

Process of building expert system in cartography can involve certain steps. Knowledge acquisition step which involve individual expert interviews, the knowledge representation step which involve the creation of the knowledge base, knowledge validation occurred

Possible and appropriate method how to collect data can be usage of a modified Delphi method. The Delphi method [Okoli, Pawlowski, 2004] is a structured and iterative approach to collecting expert knowledge involving a series of interviews or questionnaires. As basement for building can be used ontologies. The plan for acquiring the knowledge and

based on the results of the free-form interviews, develop a questionnaire to collect

use the data collected from the questionnaires to create a preliminary knowledge base

distribute the preliminary knowledge base through the experts to fine-tune it, repeating

The first step in developing the cartographical knowledge base should be to contact experts with experience in cartography (mostly cartographers). Since this kind of work often involves a time commitment, it is important to develop a means of motivating experts to participate in this work [Booker, Meyer, 2001]. Motivation for the experts' participation in

Once their expertise is collected, it should be implemented into a draft of knowledge base rules and stored in an if-then format. This draft should be after fine-tuned by being passed back to the cartographers for further review. From collected results should be build final knowledge base and it is necessary to test whole knowledge base for errors after

use available data and statistical tools to further refine the knowledge base.

this work is necessary to use the results in the beta testing phase.

The cooperation with cartographers is considerable in some ways [Návrat et al., 2002]:

oriented interview - obtaining of facts,

monitoring - obtaining of global strategy,

of communication of user.

of conflicts between experts.

 structural interview - obtaining of terms and models, free association - obtaining of relation between knowledge,

during the testing and fine-tuning of the final knowledge base.

building the knowledge base had the following steps:

have initial free-form interviews with experts;

knowledge from a larger group of experts;

to store and represent knowledge;

this process if necessary;

finalization.

comment of steps - obtaining of derived strategy,

They offer important theoretical background, but also user-specialist can find out solutions in them. They differ from other forms of online assistance such as wizards, tutorials, guides, forums, and others. They do not effort the user only one solution and they do not work for him without an explanation. They rather propose a user the range of possible correct solutions and seek to encourage users to think of the problem as an expert does. This activity is similar to that offered by the expert consultation. Another system belonging to a group of research applications is the expert system developed in China for decision support in thematic cartography [Zhang, Guo, Jiao, 2008]. It is the kind of a geographic information system, which helps users with the process of creating thematic maps. The system has a single interface. Through this system, you can choose the thematic elements and then it is possible to create automatically thematic maps according to the type and characteristics of their elements. User can modify the design parameters of various charts and through the interface obtain satisfactory results. This system is the unique solution of a complex expert system in thematic cartography. Special distributed solution was developed in Switzerland [Iosifescu-Enescu, Hugentobler, Hurni, 2010]. QGIS mapserver is an open source WMS (Web Map Service) (1.3.0 and 1.1.1) implementation. In addition, it implements advanced cartographic features as specified in the Map and Diagram Service specifications. With QGIS mapserver the content of vector and raster data sources (e.g. shapefiles, gml, postgis, wfs, geotiff) can be visualized according to cartographic rules (specified as request parameters). The generated map is sent back to the client over the internet. The cartographic rules handle advanced filtering and symbolisation of features. For improved cartographic representation, the data should be enriched with attributes to control rotation, scale, size or even transparency.

As a cartographical guide we can consider also a knowledge-based software component, called task support guide, that proposes the users appropriate interactive techniques for accomplishing specific data analysis tasks and explains how to apply these techniques. The guide is integrated in mapping system CommonGIS [Andrienko, Andrienko, Voss, 2002].

In addition, there is a large number of systems as an outcome of research work. These systems come from number of the world's research places but they are mostly aimed at the individual field cartography. These systems also often end just as the output of research or as a springboard for further research. From most important we can choose, MAPAID [Robinson, Jackson, 1985], MAPKEY [Su, 1992], ACES [Pfefferkorn et al., 1985] and many others.

#### **3.1 Cartographical knowledge and their acquiring**

The first part of construction of cartographic intelligent system is transfer of expert knowledge from various sources to computer form. The sources in the area of cartography are cartographers - experts, cartographic books, maps and atlases.

Knowledge acquisition and building knowledge base is a complex and time-consuming stage of intelligent system development which is indispensable without collaborating between experts (cartographers) and knowledge engineers. An effectively deployed system must do more than embody expertise. Its rule base must be complete, non-contradictory, and reasonable. Knowledge engineers employ a variety of techniques for eliciting information from the expert in order to construct a complete and consistent rule base [Balch, Schrader, Ruan, 2007].

They offer important theoretical background, but also user-specialist can find out solutions in them. They differ from other forms of online assistance such as wizards, tutorials, guides, forums, and others. They do not effort the user only one solution and they do not work for him without an explanation. They rather propose a user the range of possible correct solutions and seek to encourage users to think of the problem as an expert does. This activity is similar to that offered by the expert consultation. Another system belonging to a group of research applications is the expert system developed in China for decision support in thematic cartography [Zhang, Guo, Jiao, 2008]. It is the kind of a geographic information system, which helps users with the process of creating thematic maps. The system has a single interface. Through this system, you can choose the thematic elements and then it is possible to create automatically thematic maps according to the type and characteristics of their elements. User can modify the design parameters of various charts and through the interface obtain satisfactory results. This system is the unique solution of a complex expert system in thematic cartography. Special distributed solution was developed in Switzerland [Iosifescu-Enescu, Hugentobler, Hurni, 2010]. QGIS mapserver is an open source WMS (Web Map Service) (1.3.0 and 1.1.1) implementation. In addition, it implements advanced cartographic features as specified in the Map and Diagram Service specifications. With QGIS mapserver the content of vector and raster data sources (e.g. shapefiles, gml, postgis, wfs, geotiff) can be visualized according to cartographic rules (specified as request parameters). The generated map is sent back to the client over the internet. The cartographic rules handle advanced filtering and symbolisation of features. For improved cartographic representation, the data should be enriched with attributes to control rotation, scale, size or even

As a cartographical guide we can consider also a knowledge-based software component, called task support guide, that proposes the users appropriate interactive techniques for accomplishing specific data analysis tasks and explains how to apply these techniques. The guide is integrated in mapping system CommonGIS [Andrienko, Andrienko, Voss, 2002]. In addition, there is a large number of systems as an outcome of research work. These systems come from number of the world's research places but they are mostly aimed at the individual field cartography. These systems also often end just as the output of research or as a springboard for further research. From most important we can choose, MAPAID [Robinson,

The first part of construction of cartographic intelligent system is transfer of expert knowledge from various sources to computer form. The sources in the area of cartography

Knowledge acquisition and building knowledge base is a complex and time-consuming stage of intelligent system development which is indispensable without collaborating between experts (cartographers) and knowledge engineers. An effectively deployed system must do more than embody expertise. Its rule base must be complete, non-contradictory, and reasonable. Knowledge engineers employ a variety of techniques for eliciting information from the expert in order to construct a complete and consistent rule base [Balch,

Jackson, 1985], MAPKEY [Su, 1992], ACES [Pfefferkorn et al., 1985] and many others.

**3.1 Cartographical knowledge and their acquiring** 

are cartographers - experts, cartographic books, maps and atlases.

transparency.

Schrader, Ruan, 2007].

The cooperation with cartographers is considerable in some ways [Návrat et al., 2002]:


The knowledge engineer should be aware that expert knowledge is more than one kind and not all this knowledge can be acquired from one person. An interview with only one expertcartographer can avoid some fail in expert system. Interview with group of cartographers is better. The suitable way of interview is brainstorming. There is necessary more punctually prepare interview and carefully lead interview with group of experts. There is also danger of conflicts between experts.

Process of building expert system in cartography can involve certain steps. Knowledge acquisition step which involve individual expert interviews, the knowledge representation step which involve the creation of the knowledge base, knowledge validation occurred during the testing and fine-tuning of the final knowledge base.

Possible and appropriate method how to collect data can be usage of a modified Delphi method. The Delphi method [Okoli, Pawlowski, 2004] is a structured and iterative approach to collecting expert knowledge involving a series of interviews or questionnaires. As basement for building can be used ontologies. The plan for acquiring the knowledge and building the knowledge base had the following steps:


The first step in developing the cartographical knowledge base should be to contact experts with experience in cartography (mostly cartographers). Since this kind of work often involves a time commitment, it is important to develop a means of motivating experts to participate in this work [Booker, Meyer, 2001]. Motivation for the experts' participation in this work is necessary to use the results in the beta testing phase.

Once their expertise is collected, it should be implemented into a draft of knowledge base rules and stored in an if-then format. This draft should be after fine-tuned by being passed back to the cartographers for further review. From collected results should be build final knowledge base and it is necessary to test whole knowledge base for errors after finalization.

Intelligent Systems in Cartography 265

The results of the evaluation confirm that most of the programs achieved satisfactory basic cartographic functions. Nine programs achieved more than 50 from the maximum possible score (100%). Tested programs were ArcGIS, MapInfo, Geomedia, GRASS, TopoL, AutoCAD Map, Kristýna GIS, MISYS and OCAD. Commercial programs are among the best because they are being developed for a long time, and thus have the chance to meet the requirements of expert cartographic outputs. The ArcGIS program was the bets in

Evaluation of programs also revealed some weak or missing cartographic functions. They are missing of some compound line (motivated line) and point symbol in symbol libraries. Programs also have insufficiencies in creating point and area diagram map (chart diagrams). Multi-parameters totalizing diagrams, comparative diagrams and dynamic diagrams are

Functionality of setting colours is acceptable. It is possible to select the color from a palette in different color models (RGB, HSV). Some color schemes (ramps) are, however, missing, in particular bipolar, gradation or hypsometric color schemes. Possibility to create, save and re-

GIS software is not only aimed for creation of cartographic outputs. Cartographic outputs are in the end of GIS analyses. The overlay analyses of spatial data (spatial clip, symmetrical difference, spatial union etc.) bring new results and new spatial data e.g. for urban planning (Dobesova, Krivka, 2011). Another example of spatial analysis is the field of the spreading of diseases (Absalon, Slesak, 2011). The results of analyses are necessary correctly express in the map. The process of analyzing and cartographic outputs can be automated by data flow

In fact, there is significant convergence of artificial intelligence and geographic information systems recently (Vozenilek, 2009). Artificial Intelligence (AI) takes many forms such as expert systems (ES), fuzzy logic, and neural networks (Ham, 1996). Two artificial intelligence methods are widely used in GIS - artificial neural networks and fuzzy logic. The

The development of intelligent (expert) system needs formalization of cartographical knowledge for computers "to understand" the map making process. Humans understand intuitively. On the contrary, computers need explicit coding. A design of ontology is way for coding the formal cartographic knowledge. Ontology is a formal specification of a shared understanding of a knowledge domain that facilitates accurate and effective communication

Ontologies are defined for purposes of sharing and re-use of knowledge across information systems. Specialized ontologies are aimed to design a common conceptual system thesaurus. Similarly, the cartographic ontology defines the basic conceptual system (conceptualization) for the cartography. Cartographic concepts (classes) are formed as a hierarchy of classes with simple constraints. The cartographic ontology had to capture also the context and constraints of classes using description logic. The final target was not only the creating of cartographical thesaurus but the usefulness of cartographic knowledge in the

missing. Cartograms methods (anamorphosis) are very seldom implemented.

diagrams or by programming language (Dobesova, 2011 a, b).

position of cartographic expert system in computer science is on Fig. 5.

evaluation.

use custom color schemes is very rare.

**4. Cartographical ontology** 

meaning (Gruber 1993).

There are also other methods which can be used. The best way is using cartographical literature and combined these results with interview methods. Methods strictly depend on the size of knowledge base and type of acquired cartographical knowledge. One intelligent system is not possible due to amount of rules and facts, which should be involved into database.

Fig. 3. The bases of expertise knowledge and expert decision making

#### **3.2 Evaluation of cartographic functionality in GIS software**

Starting point for design intelligent system was previous research at Palacký University in 2009 (Brus et al., 2010). This research compared possibilities of creation thematic map in various GIS software. Research was carried out to search the conditions and the possibilities of map making process in GIS software. The special evaluation method named "CartoEvaluation" has been proposed for finding out the GIS software cartography potential. New evaluation method is based on Goal-Question-Metric method. More than 13 GIS software of commercial production and Open Source Software (Czech and world-wide) were evaluated under this method. The evaluation results are summarized into complex tables and accessible at web pages of the scientific project (Dobesova, 2009).


Fig. 4. The part of evaluation table for evaluation of color scheme in GIS software

There are also other methods which can be used. The best way is using cartographical literature and combined these results with interview methods. Methods strictly depend on the size of knowledge base and type of acquired cartographical knowledge. One intelligent system is not possible due to amount of rules and facts, which should be involved into

**education experience intuition data**

**decision making**

**action**

Starting point for design intelligent system was previous research at Palacký University in 2009 (Brus et al., 2010). This research compared possibilities of creation thematic map in various GIS software. Research was carried out to search the conditions and the possibilities of map making process in GIS software. The special evaluation method named "CartoEvaluation" has been proposed for finding out the GIS software cartography potential. New evaluation method is based on Goal-Question-Metric method. More than 13 GIS software of commercial production and Open Source Software (Czech and world-wide) were evaluated under this method. The evaluation results are summarized into complex tables and accessible at web pages of the scientific project

Fig. 4. The part of evaluation table for evaluation of color scheme in GIS software

**information**

**expertise knowledge**

Fig. 3. The bases of expertise knowledge and expert decision making

**3.2 Evaluation of cartographic functionality in GIS software** 

database.

(Dobesova, 2009).

The results of the evaluation confirm that most of the programs achieved satisfactory basic cartographic functions. Nine programs achieved more than 50 from the maximum possible score (100%). Tested programs were ArcGIS, MapInfo, Geomedia, GRASS, TopoL, AutoCAD Map, Kristýna GIS, MISYS and OCAD. Commercial programs are among the best because they are being developed for a long time, and thus have the chance to meet the requirements of expert cartographic outputs. The ArcGIS program was the bets in evaluation.

Evaluation of programs also revealed some weak or missing cartographic functions. They are missing of some compound line (motivated line) and point symbol in symbol libraries. Programs also have insufficiencies in creating point and area diagram map (chart diagrams). Multi-parameters totalizing diagrams, comparative diagrams and dynamic diagrams are missing. Cartograms methods (anamorphosis) are very seldom implemented.

Functionality of setting colours is acceptable. It is possible to select the color from a palette in different color models (RGB, HSV). Some color schemes (ramps) are, however, missing, in particular bipolar, gradation or hypsometric color schemes. Possibility to create, save and reuse custom color schemes is very rare.

GIS software is not only aimed for creation of cartographic outputs. Cartographic outputs are in the end of GIS analyses. The overlay analyses of spatial data (spatial clip, symmetrical difference, spatial union etc.) bring new results and new spatial data e.g. for urban planning (Dobesova, Krivka, 2011). Another example of spatial analysis is the field of the spreading of diseases (Absalon, Slesak, 2011). The results of analyses are necessary correctly express in the map. The process of analyzing and cartographic outputs can be automated by data flow diagrams or by programming language (Dobesova, 2011 a, b).

#### **4. Cartographical ontology**

In fact, there is significant convergence of artificial intelligence and geographic information systems recently (Vozenilek, 2009). Artificial Intelligence (AI) takes many forms such as expert systems (ES), fuzzy logic, and neural networks (Ham, 1996). Two artificial intelligence methods are widely used in GIS - artificial neural networks and fuzzy logic. The position of cartographic expert system in computer science is on Fig. 5.

The development of intelligent (expert) system needs formalization of cartographical knowledge for computers "to understand" the map making process. Humans understand intuitively. On the contrary, computers need explicit coding. A design of ontology is way for coding the formal cartographic knowledge. Ontology is a formal specification of a shared understanding of a knowledge domain that facilitates accurate and effective communication meaning (Gruber 1993).

Ontologies are defined for purposes of sharing and re-use of knowledge across information systems. Specialized ontologies are aimed to design a common conceptual system thesaurus. Similarly, the cartographic ontology defines the basic conceptual system (conceptualization) for the cartography. Cartographic concepts (classes) are formed as a hierarchy of classes with simple constraints. The cartographic ontology had to capture also the context and constraints of classes using description logic. The final target was not only the creating of cartographical thesaurus but the usefulness of cartographic knowledge in the

Intelligent Systems in Cartography 267

of cartographical knowledge. There exist some attempts to design a comprehensive ontology. This effort nevertheless collides with different cartographical schools and

E. Pantaleáo (2003) presented a simple proposal of cartographic ontology in her dissertation work. This ontology concerns only basic class as map symbols, variables of symbols, shape of features and category of attribute data (nominal, ordinal and numeric). There are no classes about cartographic methods (graduated point method, choropleth method) and about main components (elements) of maps (map title, map area, legend, north arrow, scale,

Interesting results in cartographical ontology development can be found in the Institute of Cartography, EHT Zurich (Enescu & Hurni, 2007). Their cartographic ontology is centered on map concepts, graphic elements, visual variables and symbols. Furthermore, their cartographic domain ontology also focuses on the complexity of map semiotics because of the fact that different types of thematic maps (choropleth maps, graduated symbol maps, multi-variable graduated symbol maps, dot density maps, etc.) can be defined. Some details of the domain ontology such as thematic point symbols like diagrams (bar charts, pie charts, ring charts …) as well as some of their properties (divergent, divided, polar, proportional …) and some additional concepts - are arranged in the logical hierarchy. All these aspects were included in their proposed ontology. The latest research at the field of cartographical ontology can be traced at University of Georgia (Smith, 2010). The basic concept is similar to our CartoExpert ontology; however, there are several aspects which

Fig. 6. Detail of domain ontology from the Institute of Cartography, Zurich

nomenclatures.

and imprint).

differ.

process of machine inference. Protégé program is often used for building of ontology. Protégé allows the definition in the language OWL-DL. This language is currently the most commonly used ontological language. The cartography is a very extensiveness discipline. From that fact, two methods were chosen from thematic cartography – choropleth method (area quantitative method) and diagram maps (cartodiagram) for the pilot stage of scientific research. Built ontology was created the necessary basis for an intelligent system that supported the users in the creation of the cartography correct maps.

Fig. 5. The position of cartographic expert system in computer science

#### **4.1 Current state of the cartographical ontologies**

Well-known ontology can be found in literature and on websites for various fields of study, e.g. Protégé Ontologies Library. As a starting point, we tried to find some related works for cartography, geography, GIS and related sciences. Main concepts found in related works in the field of ontology for GIS data operability (Stanimirovic, 2010). GeoSpatial semantic web and geo-ontology should be also taken into consideration when designing a cartographical ontology.

After examination of accessible ontologies on the web and other ontological repository, we came to this conclusion: Only a few particular examples of domain ontology exist in the related field. There is no complex ontology which takes into consideration all aspects

process of machine inference. Protégé program is often used for building of ontology. Protégé allows the definition in the language OWL-DL. This language is currently the most commonly used ontological language. The cartography is a very extensiveness discipline. From that fact, two methods were chosen from thematic cartography – choropleth method (area quantitative method) and diagram maps (cartodiagram) for the pilot stage of scientific research. Built ontology was created the necessary basis for an intelligent system that

Computer Science

Artificial Intelligence

Expert Systems

Expert System in Cartography

Remake Wrong Map

Well-known ontology can be found in literature and on websites for various fields of study, e.g. Protégé Ontologies Library. As a starting point, we tried to find some related works for cartography, geography, GIS and related sciences. Main concepts found in related works in the field of ontology for GIS data operability (Stanimirovic, 2010). GeoSpatial semantic web and geo-ontology should be also taken into consideration when designing a cartographical

After examination of accessible ontologies on the web and other ontological repository, we came to this conclusion: Only a few particular examples of domain ontology exist in the related field. There is no complex ontology which takes into consideration all aspects

Planning of Steps in Map Design

xxxxx Expert Systems in Banking

Fuzzy Logic

Image and Speach Recognition

Mining xxxxx

Languages xxxxx

supported the users in the creation of the cartography correct maps.

Programming

Data

Diagnose of Suitable Cartographic Method

**4.1 Current state of the cartographical ontologies** 

ontology.

Fig. 5. The position of cartographic expert system in computer science

Expert Systems in Medicine

Operating Systems

Neural Networks of cartographical knowledge. There exist some attempts to design a comprehensive ontology. This effort nevertheless collides with different cartographical schools and nomenclatures.

E. Pantaleáo (2003) presented a simple proposal of cartographic ontology in her dissertation work. This ontology concerns only basic class as map symbols, variables of symbols, shape of features and category of attribute data (nominal, ordinal and numeric). There are no classes about cartographic methods (graduated point method, choropleth method) and about main components (elements) of maps (map title, map area, legend, north arrow, scale, and imprint).

Interesting results in cartographical ontology development can be found in the Institute of Cartography, EHT Zurich (Enescu & Hurni, 2007). Their cartographic ontology is centered on map concepts, graphic elements, visual variables and symbols. Furthermore, their cartographic domain ontology also focuses on the complexity of map semiotics because of the fact that different types of thematic maps (choropleth maps, graduated symbol maps, multi-variable graduated symbol maps, dot density maps, etc.) can be defined. Some details of the domain ontology such as thematic point symbols like diagrams (bar charts, pie charts, ring charts …) as well as some of their properties (divergent, divided, polar, proportional …) and some additional concepts - are arranged in the logical hierarchy. All these aspects were included in their proposed ontology. The latest research at the field of cartographical ontology can be traced at University of Georgia (Smith, 2010). The basic concept is similar to our CartoExpert ontology; however, there are several aspects which differ.

Fig. 6. Detail of domain ontology from the Institute of Cartography, Zurich

Intelligent Systems in Cartography 269

The creation of a thematic map, use of symbols and the use of cartographic methods are under theoretical principals. Additionally, creation of thematic maps also respects practical

The basic terms were also compared with terminological world lexical ontology WordNet.

The term as cartography, map, symbol, sign, choropleth map are included there.

experience (Vozenilkek, 2004).

Fig. 8. The classes in CartoExpert ontology in Protégé

#### **4.2 Ontology CartoExpert**

Our research team decided to create new cartographic ontology CartoExpert in 2010. Basic terms of the conceptualization of cartographic knowledge can be found in cartographical books. There are several important books that deal with cartography like "Thematic Cartography and Geographic Visualization" by Slocum et al. (2004), "Cartography, Visualization of Geospatial Data" by Kraak and Ormeling (2003) and "Elements of Cartography" by Robinson et al. (1995). Other resources are e.g. "How maps work? Representation, Visualization and Design" by MacEachren (2004) and "Mapping It Out: Expository Cartography for the Humanities and Social Sciences" by Monmonier (1993).

Some different cartographical concepts and methods exist in Central Europe. Other authors and their books like "Methods of map expression" by Pravda (2006), "Application of Cartography and Thematic Maps" by Vozenilek (2004) and "Quantitative method in cartography" by Kanok (1992) were also considered. All terms, rules and recommendations were collected from these books. Subsequently, they were used in the phase of ontology building and knowledge base design.

Fig. 7. The result of search for word "map" at the WordNet ontology

Maps are divided according to cartography to two main groups. There are thematic maps and topographic maps. Every thematic map contains a simple topographic base map. Thematic maps represent the distribution of one or more particular phenomena (Kraak, Ormeling, 2003). Census and statistical data are very often depicted on thematic maps. Data are divided into two types: qualitative and quantitative data. Quantitative data have absolute or relative value. Absolute and relative values are expressed by different cartographic methods in maps. Absolute values, which have a non-area related ratio, are expressed by diagrams in maps. All methods use cartographic symbols (point, line, area).

Our research team decided to create new cartographic ontology CartoExpert in 2010. Basic terms of the conceptualization of cartographic knowledge can be found in cartographical books. There are several important books that deal with cartography like "Thematic Cartography and Geographic Visualization" by Slocum et al. (2004), "Cartography, Visualization of Geospatial Data" by Kraak and Ormeling (2003) and "Elements of Cartography" by Robinson et al. (1995). Other resources are e.g. "How maps work? Representation, Visualization and Design" by MacEachren (2004) and "Mapping It Out: Expository Cartography for the Humanities and Social Sciences" by Monmonier (1993).

Some different cartographical concepts and methods exist in Central Europe. Other authors and their books like "Methods of map expression" by Pravda (2006), "Application of Cartography and Thematic Maps" by Vozenilek (2004) and "Quantitative method in cartography" by Kanok (1992) were also considered. All terms, rules and recommendations were collected from these books. Subsequently, they were used in the phase of ontology

Fig. 7. The result of search for word "map" at the WordNet ontology

Maps are divided according to cartography to two main groups. There are thematic maps and topographic maps. Every thematic map contains a simple topographic base map. Thematic maps represent the distribution of one or more particular phenomena (Kraak, Ormeling, 2003). Census and statistical data are very often depicted on thematic maps. Data are divided into two types: qualitative and quantitative data. Quantitative data have absolute or relative value. Absolute and relative values are expressed by different cartographic methods in maps. Absolute values, which have a non-area related ratio, are expressed by diagrams in maps. All methods use cartographic symbols (point, line, area).

**4.2 Ontology CartoExpert** 

building and knowledge base design.

The creation of a thematic map, use of symbols and the use of cartographic methods are under theoretical principals. Additionally, creation of thematic maps also respects practical experience (Vozenilkek, 2004).

The basic terms were also compared with terminological world lexical ontology WordNet. The term as cartography, map, symbol, sign, choropleth map are included there.

Fig. 8. The classes in CartoExpert ontology in Protégé

Intelligent Systems in Cartography 271

**is** or **has**. The relation is set as Domain *D(f)* and Range *H(f).* The fig. 9 shows the relation between class MapSymbol and class SymbolVariable. The name of property is

The main cartographic terms were necessary for the pilot project that concern only to two cartographic methods for thematic maps based on Quantitative data. The Choroplets map methods and Chart map methods are aimed. The system of the class hierarchy was designed more detailed for them than other part of the ontology. The names and division of these methods differ in the Czech and English version of the ontology. The last important class for method based on quantitative data is class Scale. This class expresses the **scale of values** (not scale of the map). This call contents two subclasses

The definition of hierarchy of classes and definition of properties represent the collection of stored knowledge for the domain of cartography. Ontology gathers mainly declarative knowledge. **Declarative knowledge** is the set of definitions of terms from the specific domain – cartography. The set is not only list of terms (thesaurus) but important is grouping terms to joint classes and creation of taxonomy. **Procedural knowledge** is the second type of knowledge. Procedural knowledge describes activities and processes in map creation. This type procedural knowledge can not be introduced to ontology. They can be record as rules

Intelligent systems have already covered a range of usages, the growing trend can be traced in their development especially in recent years. The possibility of their usage is increasing with the increasing power of computer technology. It is commendable that some attempts of

Within the development, it is necessary to require the presence of thematic cartographer in the role of the knowledge expert and equally important expert - the knowledge engineer who is able to incorporate this information into the intelligent system. Knowledge acquisition and building knowledge base is a complex and time-consuming stage of intelligent system development which is indispensable without collaborating between experts (cartographers) and knowledge engineers. An effectively deployed intelligent system must do more than embody expertise. Its rule base must be complete, noncontradictory, and reasonable. Knowledge engineers employ a variety of techniques for eliciting information from the expert in order to construct a complete and consistent rule

The situation in the field of professional software is still insufficient. Even the world's largest producers of GIS software do not implement tools that should increasingly guide the process of map-making in the accordance with the cartographical rules in their products. It is still necessary to have at least basic cartographic knowledge to visualize maps properly.

So far, there has been no comprehensive tool, which can easily deal with the problem of thematic cartography completely. The main reason is the complexity and comprehensiveness of a map-making process. To build a hierarchy of rules, affect all types

creation of intelligent system to force GIS have occurred recently.

"hasVariable".

**5. Conclusion** 

base.

FunctionalScale and IntervalScale.

and such some mathematical equations.

The base for the cartographical ontology was thesaurus – lexicon of cartographical terms. The lexicon also contained a list of synonyms. In the dictionary pruning stage, a pair wise comparison between the cartographic terms and their descriptions result to lexicon set. Synonyms of terms were grouped together. As a result, one description was chosen to represent all the synonym terms. The differences between the Central Europe and the English cartographic school were solved by the decision to design two ontologies – the Czech ontology and the English ontology. This chapter and figures describes only the English ontology for the better readability. The main classes are Data, MapColor, MapComposition, MapDescription, MapSymbol, Method, Phenomenon, Projection, SymbolVariables and Scale. These cartographic terms are expressed by **classes** in ontology in OWL language.

Fig. 9. List of object properties in Protégé (properties "hasVariable")

Very carefully was designed the **hierarchy** of classes. The relation of two classes is expressed by *subsumption*, *equivalence* or *disjunction*. The example of subsumption is upper class *AttributeData* and two sub class *QualitativeData* and *QuantitativeData*. The disjunction is also defined for these two classes. When data have qualitative value they can not have quantitative value. The terms isoline, isopleth and isochor are the example of equivalence (synonyms) (Penaz, 2010).

The important part of ontology is also the definition of **properties**. The property constructs relation between classes or individuals. The name of the property contains verb

The base for the cartographical ontology was thesaurus – lexicon of cartographical terms. The lexicon also contained a list of synonyms. In the dictionary pruning stage, a pair wise comparison between the cartographic terms and their descriptions result to lexicon set. Synonyms of terms were grouped together. As a result, one description was chosen to represent all the synonym terms. The differences between the Central Europe and the English cartographic school were solved by the decision to design two ontologies – the Czech ontology and the English ontology. This chapter and figures describes only the English ontology for the better readability. The main classes are Data, MapColor, MapComposition, MapDescription, MapSymbol, Method, Phenomenon, Projection, SymbolVariables and Scale. These cartographic terms are expressed by **classes** in ontology

Fig. 9. List of object properties in Protégé (properties "hasVariable")

Very carefully was designed the **hierarchy** of classes. The relation of two classes is expressed by *subsumption*, *equivalence* or *disjunction*. The example of subsumption is upper class *AttributeData* and two sub class *QualitativeData* and *QuantitativeData*. The disjunction is also defined for these two classes. When data have qualitative value they can not have quantitative value. The terms isoline, isopleth and isochor are the example of equivalence

The important part of ontology is also the definition of **properties**. The property constructs relation between classes or individuals. The name of the property contains verb

in OWL language.

(synonyms) (Penaz, 2010).

**is** or **has**. The relation is set as Domain *D(f)* and Range *H(f).* The fig. 9 shows the relation between class MapSymbol and class SymbolVariable. The name of property is "hasVariable".

The main cartographic terms were necessary for the pilot project that concern only to two cartographic methods for thematic maps based on Quantitative data. The Choroplets map methods and Chart map methods are aimed. The system of the class hierarchy was designed more detailed for them than other part of the ontology. The names and division of these methods differ in the Czech and English version of the ontology. The last important class for method based on quantitative data is class Scale. This class expresses the **scale of values** (not scale of the map). This call contents two subclasses FunctionalScale and IntervalScale.

The definition of hierarchy of classes and definition of properties represent the collection of stored knowledge for the domain of cartography. Ontology gathers mainly declarative knowledge. **Declarative knowledge** is the set of definitions of terms from the specific domain – cartography. The set is not only list of terms (thesaurus) but important is grouping terms to joint classes and creation of taxonomy. **Procedural knowledge** is the second type of knowledge. Procedural knowledge describes activities and processes in map creation. This type procedural knowledge can not be introduced to ontology. They can be record as rules and such some mathematical equations.

#### **5. Conclusion**

Intelligent systems have already covered a range of usages, the growing trend can be traced in their development especially in recent years. The possibility of their usage is increasing with the increasing power of computer technology. It is commendable that some attempts of creation of intelligent system to force GIS have occurred recently.

Within the development, it is necessary to require the presence of thematic cartographer in the role of the knowledge expert and equally important expert - the knowledge engineer who is able to incorporate this information into the intelligent system. Knowledge acquisition and building knowledge base is a complex and time-consuming stage of intelligent system development which is indispensable without collaborating between experts (cartographers) and knowledge engineers. An effectively deployed intelligent system must do more than embody expertise. Its rule base must be complete, noncontradictory, and reasonable. Knowledge engineers employ a variety of techniques for eliciting information from the expert in order to construct a complete and consistent rule base.

The situation in the field of professional software is still insufficient. Even the world's largest producers of GIS software do not implement tools that should increasingly guide the process of map-making in the accordance with the cartographical rules in their products. It is still necessary to have at least basic cartographic knowledge to visualize maps properly.

So far, there has been no comprehensive tool, which can easily deal with the problem of thematic cartography completely. The main reason is the complexity and comprehensiveness of a map-making process. To build a hierarchy of rules, affect all types

Intelligent Systems in Cartography 273

Brus, J., Dobesova, Z., Kanok, J., Pechanec, V. (2010) Design of intelligent system in

Brus J., Dobesova Z., Kanok J. (2009). Utilization of expert systems in thematic cartography

Brus J, Kanok, J. Dobesova, Z. (2010). Assisted cartography, vision or reality? [Asistovaná

Brewer, I., MacEachren, A. M., Abdo, H., Gundrumand, J., Otto, G. (2000) Collaborative

Dhaliwal, J. S., & Benbasat, I. (1996). The use and effects of knowledge-based system

Dobesova, Z., Brus, J. (2011). Coping with cartographical ontology. *Conference Proceedings* 

Dobesova, Z. (2011a). Visual programming language in geographic information systems,

Dobesova, Z. (2011b). Programming Language Python for Data Processing, *Proceedings of 2nd*

Dobesova, Z, Krivka, T. (2011). Walkability index in the urban planning: A case study in

Dobesova, Z. (2009). *Evaluation of Cartographic Functionality in Geographic Information* 

Eikvil, L., Aas, K., Koren, H. (1995) Tools for interactive map conversion and

Enescu, I. I., Hurni, L. (2007). Towards cartographic ontologies or how computers learn

Gruber, T. R. (1993). Toward Principles for the Design of Ontologies Used for Knowledge Sharing. *International Journal Human-Computer Studies* 43: pp. 907-928.

Olomouc city, *Urban Planning,* InTech, ISBN 979-953-307-412-1

*Information Systems Research*, 7, pp. 342–362.

WSEAS Press, pp. 276-280 , ISBN 978-1-61804-034-3

4869, ISBN 978-1-4244-8163-7

4244-7335-9

Barcelona

Czech)

137-141.

sgem2011

244-2353-1

Recognition: 927-930.

Moscow, Russia.

cartography. In Brad, R. (ed.): *Proceedings. 9 RoEduNet IEEE International Conference.* Sibiu, University of Sibiu, pp. 112-117 ISSN 2068-1038. ISBN 978-1-

*International Conference on Intelligent Networking and Collaborative Systems, INCoS*,

kartografie: vize nebo realita?], *Proceedings o 18th congress of Czech geographical society*, University of Ostrava, Ostrava, pp. 255-258, ISBN 978-80-7368-903-2 (in

geographic visualization: Enabling shared understanding of environmental processes. In: IEEE Information Visualization Symposium, Salt Lake City, Utah, pp.

explanations: theoretical foundations and a framework for empirical evaluation.

*SGEM 2011, 11th International Multidisciplinary Scientific GeoConfrence*. STEF92 Technology Ltd., Sofia, Bulgaria, pp. 377-384 ISSN 1314-2704, DOI:10.5593/

*Recent Researches in Applied Informatics*, Proceedings of the 2nd International Conference on Applied Informatics and Computing Theory, AICT `11, Prague,

*International Conference on Electrical and Control Engineering ICECE 2011,* Yichang, China*,* Volume 6, Institute of Electrical and Electronic Engineers (IEEE), pp. 4866-

*Systems*. Hodnoceni kartografické funkcionality geografických informačních systémů. Publishing house Palacký University, Olomouc, 132 p., ISBN 978-80-

vectorization. In: Third International Conference on Document Analysis and

cartography, In*: Proceedings of the 23rd International Cartographic Conference*,

of maps and the appropriate methods of thematic cartography in a single system, requires more than a comprehensive approach. Higher demands are put on the user's knowledge because he must be able to select correctly from the proposed system of options.

Despite the complexity of a map-making process the knowledge base of expert system is a solution how to help primarily non-cartographers in the production of maps according with the rules of thematic cartography towards to better decisions based on map output.

A great problem for those who tries to develop up-to-date knowledge-based software for computer mapping is the absence of systematized knowledge concerning building and use of interactive, dynamic maps. Replacing the human expert by a comprehensive intelligent system is a highly efficient objective for cartography as a whole. Not only reaching correct map, but also helping people to make right decisions is a main aim of whole cartography. Main objective will be to create a user-friendly expert system, simple and so comprehensive that will allow you to create the correct cartographic map without the need of combining more software. This software will become a popular tool for the broadest range of users. The educational potential of intelligent systems allows the extension of expertise among a large group of non-cartographers. Another advantage of intelligent system is the gradual insertion of further new expert knowledge of cartography into the knowledge base of expert system. This will quickly transfer expert knowledge between non-cartographers in the future. The elimination of the future inexpert and inaccurate maps will be achieved.

#### **6. Acknowledgment**

The research was supported by the project of the Czech Grant Science Foundation No. 205/09/1159 "Intelligent system for interactive support of thematic map creation".

#### **7. References**


of maps and the appropriate methods of thematic cartography in a single system, requires more than a comprehensive approach. Higher demands are put on the user's knowledge

Despite the complexity of a map-making process the knowledge base of expert system is a solution how to help primarily non-cartographers in the production of maps according with

A great problem for those who tries to develop up-to-date knowledge-based software for computer mapping is the absence of systematized knowledge concerning building and use of interactive, dynamic maps. Replacing the human expert by a comprehensive intelligent system is a highly efficient objective for cartography as a whole. Not only reaching correct map, but also helping people to make right decisions is a main aim of whole cartography. Main objective will be to create a user-friendly expert system, simple and so comprehensive that will allow you to create the correct cartographic map without the need of combining more software. This software will become a popular tool for the broadest range of users. The educational potential of intelligent systems allows the extension of expertise among a large group of non-cartographers. Another advantage of intelligent system is the gradual insertion of further new expert knowledge of cartography into the knowledge base of expert system. This will quickly transfer expert knowledge between non-cartographers in the

because he must be able to select correctly from the proposed system of options.

the rules of thematic cartography towards to better decisions based on map output.

future. The elimination of the future inexpert and inaccurate maps will be achieved.

205/09/1159 "Intelligent system for interactive support of thematic map creation".

*Environmental Safety* 74, pp. 967–973, Elsevier

Russia. St. Petersburg, pp. 114-117

Information Resources, Syracuse, NY, 1-3.

The research was supported by the project of the Czech Grant Science Foundation No.

Absalon D., Slesak B. (2011). The importance of time of exposure to harmful anthropogenic

Andrienko, G., Andrienko, N., Voss, H. (2002) Computer cartography and cartographic

Balch S. R., Schrader S. M., Ruan T. (2007) Collection, storage and application of

Booker, J.,M., Meyer, M. (2001) Eliciting and Analyzing Expert Judgment: A practical

Boss, R., W. (1991) What Is an Expert System? ERIC Digest. ERIC Clearing House on

Borst, W. N. (1997). Construction of engineering ontologies for knowledge sparing and

reuse. [Ph.D. thesis]. University of Twente, Enschede, 243 p.

factors as an element of cancer risk assessment in children. *Ecotoxicology and* 

knowledge. Proceedings: *Intercarto 8,* International Conference, Saint-Petersburg,

human knowledge in expert system development, Expert Systems, 24 no. 5, pp.

guide, ASA-SIAM Series on Statistics and Applied Probability, 459 p. ISBN: 0-

**6. Acknowledgment** 

346-355

89871-474-5

**7. References** 


Intelligent Systems in Cartography 275

Pantaleáo, E. (2003) Aplicacáo de técnicas de sistemas baseados em conhecimento em

Penaz, T. (2010) An Ontological Model Building for Application Use of Knowledge in

Pravda, J. (2006) Metódy mapového vyjadrovania, Klasifikácia a ukážky, [Methods of map

Guo Q, Ren, X. (2003) Intelligent geographic information processing. Wuhan University

Robinson, A. Morrison, J., Muehrcke, P., Kimerling, A., Guptill, S. (1995) *Elements of* 

Robinson, G., Jackson, M. (1985). Expert Systems in map design. Proceedings Auto Carto 7,

Schnabel, O. (2005). Map Symbol Brewer—A New Approach for a Cartographic Map

Slocum T., McMaster, R., Kesseler, F., Howard, H. (2004). *Thematic Cartography and* 

Smith, A. R. (2010). Designing a cartographic ontology for use with expert systems. A

Stanimirovic, A., Bogdanovic, M. et al. (2010) Mapping Ontologies to Object-Oriented

Su, B. (1992) Expert System in the application of cartographic. Journal of Geomatics, Vol. 1:

Vozenilek, V. (2004). *Aplikovaná kartografie I., Tematické mapy*, [Application of cartography,

Vozenilek V. (2009). Artificial intelligence and GIS: Mutual meeting and passing.

Multidisciplinary Scientific Geoconference: Sgem 2010, Vol I: 1127-1134 Stefanakis, K., Tsoulos, L. (2005) Structure and Development of a Knowledge. Base for

Symbol Generator. Paper read at 22nd International Cartographic Conference, 9-16

special joint symposium of ISPRS Technical Commission IV & AutoCarto in

Representation in Geonis Framework for Gis Interoperability. 10th International

Cartographic Composition, Proceedings of International Cartographic Conference,

Thematic maps], Publishing house of Palacký University, Olomouc, ISBN 80-224-

*International Conference on Intelligent Networking and Collaborative Systems*, INCoS

of sciences, Geographical institute, Bratislava, 127 p. (in Slovak) Protégé Ontologies Library. http://protege.stanford.edu/ontologies/ontologies.html. Guo Q. (1993) Design a decision-making support system of thematic mapping. Journal of

*Cartography*, John Wiley & Sons, INC., USA

*geographic Visualization*, Prentice Hall, 518 p.

conjuction with ASPRS/CaGIS 2010. Orlando, Florida.

Paraná, Curitiba, Brasil.

(in Czech)

9.

Press.

Washington D.C.

July, at Coruña, Spain

La Coruna, Spain.

0270-X (In Czech)

2009, Spain, pp. 279-284.

WordNet. http://wordnetweb.princeton.edu/perl/webwn

31-35

projeto cartográfico temático (Application of techniques for knowledge based systems in thematic cartography), disertation thesis, Universidade Federal de

Thematic Cartography Domain, Proceedings o 18th congress of Czech geographical society, University of Ostrava, pp. 259-265, ISBN 978-80-7368-903-2

expression, Classification and examples], Geographia Slovaca 21, Slovak academy

Wuhan University of Science and Technology Mapping, 1993, Vol.18 additional: 91-


Ham, J. (1996). Artificial intelligence: building on expert knowledge, InTech, 43 (3), 52-

Harrower, M., Brewer C. (2003). ColourBrewer.org: An Online Tool for Selecting Colour

Hori, O., Tanigawa, S. (1993). Raster-to-vector conversion by line based on contours and

Hua, Y., (1991) Determine the map symbol type of map element with expert system

Iosifescu-Enescu,I., Hugentobler, M., Hurni, L., (2010) Web cartography with open

Kaňok, J. (1992). *Kvantitativní metody v kartografii*, [Quantitative method in cartography]

Karvaš, P. (2011). Design of knowledge base for expert system in the realative data

Kraak, M., J., MacEachren, A., M., (1999) Visualization for exploration of spatial data. International Journal of Geographical Information Science 13:285-287. Kraak, M., J., Ormeling, F. (2003) *Cartography, Visualization of Geospatial data*, Second Edition,

Kasturi, R., Bow, S.T., Masri, W.E., Shah, J., Gattiker, J.R., Mokate, U.B. (1990). A system for

Lee D. (1994). "Knowledge Acquisition of Digital Cartographic Generalization". EGIS,

MacEachren, A.M., Kraak, M. J., (1997) Exploratory cartographic visualization: advancing

MacEachren A., M. (2004) *How maps work: representation, visualization, and design*, The

Meng L. (2003)."Cognitive Modelling of Cartographic Generalization". Strategies on

Monmonier, M. (1993) *Mapping It Out: Expository Cartography for the Humanities and Social* 

Návrat, P. Bieliková, M., Beňušková, L., Kapustník, I., Unger, M. (2002): Umelá inteligencia,

O'Looney, J. (2000) Beyond maps: GIS and decision making in local government. Redlands,

Okoli, C. & Pawlowski, S. D. (2004). The Delphi method as a research tool: an example,

design considerations and applications. Information & Management, 42,

Automated Generalization of Cartographic Data, Project Report.

Vydavateľstvo STU, Bratislava, 396 p. ISBN 80-227-1645-6

skeletons. In: Second International Conference on Document Analysis and

technology. Journal of the People's Liberation Army Institute of Surveying and

standards – A solution to cartographic challenges of environmental management, Environmental Modelling & Software, Volume 25, Issue 9, pp.

expresion, [diploma thesis], department of Geoinformatics, Palacky University in

interpretation of line drawings. IEEE Transactions on Pattern Analysis and

Schemes for Maps. Cartographic Journal 40 (1): 27-37.

Recognition. Tsukuba, Japan, 353-358.

Ostravská univerzita, Ostrava (in Czech)

Machine Intelligence 12 (10): 973-992.

*Sciences*, The Univesity of Chicago Press

the agenda. Computers & Geosciences, 23(4): 335-343.

Mapping, Vol. 3: 43-47.

Olomouc, 67 p., (in Czech)

Prentice Hall, London

55 pp.

988-999,

1-20.

15-29.

Guilford Press

California: ESRI Press.


**13** 

*Brazil* 

**Intelligent Expert System for Protection** 

The objective of this chapter consists of presenting an expert system that assists the procedures involved with the protection specification of transformers and equipments against atmospheric discharges, allowing also to analyze in a detailed and systematic way the behavior of the respective voltage transients that are generated at the supplying area.

For such purpose, the expert system developed makes efficient integration of approaches and techniques that take into account the characteristic aspects of the atmospheric discharges, the experimental analyses that represent the phenomenon and the mathematical

The results obtained from the experimental application of the expert system have contributed in a substantial way to optimize the processes involved with the efficient specification of protection devices associated with the transformers and equipments of the

The decision process taken into account by the expert system is based on information provided by the software "SimSurto", which was especially developed to simulate the voltage transients caused by atmospheric discharges in distribution lines, and its objective is the computation of several parameters related to the respective transients, considering the equipments already installed, the geographical location of the distribution line and the

The use of the developed tool has allowed the optimized specification for protection devices of equipments and transformers belonging to distribution system, enabling that differentiated protection strategies can be applied according to the particularities of each

Nerivaldo R. Santos2, Lucca Zamboni2, Leandro N. Soares3, José A. C. Ulson4, Rogério A. Flauzino1,

models that allow to map the process involved with the formation of the lightning.

respective incidence of atmospheric discharges in the distribution system.

Danilo H. Spatti1, Ricardo A. S. Fernandes1, Marcos M. Otsuji2 and Edison A. Goes 1*University of São Paulo (USP), São Carlos, SP, Brazil*

**1. Introduction** 

distribution system.

<sup>2</sup>*EDP Bandeirante, São Paulo, SP, Brazil* <sup>3</sup>*EDP ESCELSA, Vitória, ES, Brazil*

4*São Paulo State University, Bauru, SP*, *Brazil*

 \*

**Optimization Purposes in Electric** 

**Power Distribution Systems** 

*University of São Paulo (USP), São Carlos, SP* 

Ivan N. da Silva et al.\*

Zhang, L., Guo, Q., Jiao, L. (2008). Design and Implementation of Decision-making Support System for Thematic Map Cartography. The international archives of the photogrammetry, remote sensing and spatial information science. Volume XXXVII. Bejing China.

## **Intelligent Expert System for Protection Optimization Purposes in Electric Power Distribution Systems**

Ivan N. da Silva et al.\* *University of São Paulo (USP), São Carlos, SP Brazil* 

#### **1. Introduction**

276 Intelligent Systems

Zhang, L., Guo, Q., Jiao, L. (2008). Design and Implementation of Decision-making Support

Bejing China.

System for Thematic Map Cartography. The international archives of the photogrammetry, remote sensing and spatial information science. Volume XXXVII.

> The objective of this chapter consists of presenting an expert system that assists the procedures involved with the protection specification of transformers and equipments against atmospheric discharges, allowing also to analyze in a detailed and systematic way the behavior of the respective voltage transients that are generated at the supplying area.

> For such purpose, the expert system developed makes efficient integration of approaches and techniques that take into account the characteristic aspects of the atmospheric discharges, the experimental analyses that represent the phenomenon and the mathematical models that allow to map the process involved with the formation of the lightning.

> The results obtained from the experimental application of the expert system have contributed in a substantial way to optimize the processes involved with the efficient specification of protection devices associated with the transformers and equipments of the distribution system.

> The decision process taken into account by the expert system is based on information provided by the software "SimSurto", which was especially developed to simulate the voltage transients caused by atmospheric discharges in distribution lines, and its objective is the computation of several parameters related to the respective transients, considering the equipments already installed, the geographical location of the distribution line and the respective incidence of atmospheric discharges in the distribution system.

> The use of the developed tool has allowed the optimized specification for protection devices of equipments and transformers belonging to distribution system, enabling that differentiated protection strategies can be applied according to the particularities of each

 \* Nerivaldo R. Santos2, Lucca Zamboni2, Leandro N. Soares3, José A. C. Ulson4, Rogério A. Flauzino1,

Danilo H. Spatti1, Ricardo A. S. Fernandes1, Marcos M. Otsuji2 and Edison A. Goes 1*University of São Paulo (USP), São Carlos, SP, Brazil*

<sup>2</sup>*EDP Bandeirante, São Paulo, SP, Brazil*

<sup>3</sup>*EDP ESCELSA, Vitória, ES, Brazil*

<sup>4</sup>*São Paulo State University, Bauru, SP*, *Brazil*

Intelligent Expert System

conductor.

where:

estimation.

distribution line is 100 meters.

for Protection Optimization Purposes in Electric Power Distribution Systems 279

localization in relation to incidence point of the atmospheric discharge, that is, the induced voltage values produced in a line composed by several conductors of same height and with a small horizontal spacing, such as in distribution lines, would be equal in each

Measurements achieved with the reduced model technique (Paula et al., 2001; Salari & Portela, 2007), as well as measurements in fields made in South Africa, demonstrate that the results provided by Rusck's theory is coherent with those obtained by experimental results (Eriksson et al., 1982). Originally, Rusck proposed a current wave to the atmospheric discharge represented by a step function with amplitude *I*. The induced voltage produced

V x,t U x,t U x,t (1)

2

(4)

(3)

(2)

2 2 2 2 22 2

[y c t x ] ( c t) (1 )(x y )

5

I

( )

v 1 <sup>c</sup> 5.10 <sup>1</sup>

In this case, *V*(*x*,*t*) is the induced voltage (V) at a point *x* of the line; *t* is the time in seconds; *c* is the velocity of light in free space (m/s); *I* is the return-peak current value (A); *h* is the average height of the distribution line; *y* is the closest distance between the discharge

Equations (1), (2) and (3) express Rusck's theory basis. In (4) the expression for the

38.8 I h <sup>V</sup> y

From the previous expressions is possible to identify that they provide an analytic form for the computation of induced voltage in a distribution line, whereas other existent theories provide just iterative expressions that have high computational effort to perform the same

In Fig. 1 is presented the induced voltage at the point *x*=0m for an atmospheric discharge represented by a step function with amplitude *I*=10 kA in relation to an infinite line with 10 meters of height, where the distance between the atmospheric discharge from the

In order to illustrate how the proposed formulation in this section is efficient for induced voltage estimation in overhead distribution lines, the induced voltage profile for different

by this discharge in relation to an infinite line can be computed by:

incidence point and the distribution line (m) and *x* is a point along the line (m).

max

maximum induced voltage at the point *x*=0m is given by:

positions along the distribution line is presented in Fig. 2.

(c t x) x (c t x) U(x,t) 30 I h . <sup>1</sup>

region, contributing then for value aggregation to services provided by the distribution company, since the available tools proportionate more optimized analyses in relation to the procedures involved with the protection specification.

Therefore, in this chapter, the particularities for estimation of induced voltages in real distribution networks, such as the network discontinuity, the phase conductor arrangement, the intrinsic characteristics of the incident atmospheric discharges in each region of the considered distribution system, are taken into account by the expert system. Performance evaluations indicate that the expert system provides coherent results and its practical application contributes to optimize the processes involved with parameters specification related to the protection of equipments and transformers.

For such purpose, this paper is organized as follows. In Section 2, a brief summary about induced voltage estimation techniques are presented. In Section 3, the achieved modifications in relation to the conventional techniques are introduced in order to produce a greater accuracy when compared to the results obtained from real situations. The expert system for protection specification against atmospheric discharges, named by Protection Plus, is briefly described in Section 4. The expert system for optimized design of grounding systems is presented in Section 5. Finally, in Section 6, the key issues raised in the paper are summarized and conclusions are drawn.

#### **2. Rusck's conventional model for induced voltage estimation in overhead distribution lines**

In this section the main aspects concerning to Rusck's methodology for induced voltage estimation in distribution lines caused by atmospheric discharges are presented.

Therefore, it is achieved a general study regarding induced voltage estimation in distribution lines through use of conventional methods discussed in the technical literature.

Although the methodology originally developed in Rusck (1957) has some limitations to areas with soil resistivity less than 100 m, it is still widely used for induced voltage estimation in overhead distribution and transmission lines generated from indirect atmospheric discharges occurred near to the respective line.

The induced voltage estimation methodology presented in Rusck (1957) has as start point the modeling of the return current imposed by the atmospheric discharge in the distribution line. Rusck's method calculates the electric field generated by this return current in the ground surface and, from this electric field and from the line multi-wire arrangement, the theory provides the resultant values of induced voltages along the distribution line.

In Rubinstein & Uman (1989) is mathematically demonstrated that the studies presented in Rusck (1957) for resultant electric field computation of return current is correct. This fact has contributed to increase the reliability in relation to method developed by Rusck. Other additional procedures involved with models for induced voltage are also found in Cooray (2003).

An existent question related to this theory is that it estimates induced voltage values for conductors of a multi-wire line taking just into account the conductor geometric localization in relation to incidence point of the atmospheric discharge, that is, the induced voltage values produced in a line composed by several conductors of same height and with a small horizontal spacing, such as in distribution lines, would be equal in each conductor.

Measurements achieved with the reduced model technique (Paula et al., 2001; Salari & Portela, 2007), as well as measurements in fields made in South Africa, demonstrate that the results provided by Rusck's theory is coherent with those obtained by experimental results (Eriksson et al., 1982). Originally, Rusck proposed a current wave to the atmospheric discharge represented by a step function with amplitude *I*. The induced voltage produced by this discharge in relation to an infinite line can be computed by:

$$\mathbf{V(x,t) = U(x,t) + U(-x,t)}\tag{1}$$

where:

278 Intelligent Systems

region, contributing then for value aggregation to services provided by the distribution company, since the available tools proportionate more optimized analyses in relation to the

Therefore, in this chapter, the particularities for estimation of induced voltages in real distribution networks, such as the network discontinuity, the phase conductor arrangement, the intrinsic characteristics of the incident atmospheric discharges in each region of the considered distribution system, are taken into account by the expert system. Performance evaluations indicate that the expert system provides coherent results and its practical application contributes to optimize the processes involved with parameters specification

For such purpose, this paper is organized as follows. In Section 2, a brief summary about induced voltage estimation techniques are presented. In Section 3, the achieved modifications in relation to the conventional techniques are introduced in order to produce a greater accuracy when compared to the results obtained from real situations. The expert system for protection specification against atmospheric discharges, named by Protection Plus, is briefly described in Section 4. The expert system for optimized design of grounding systems is presented in Section 5. Finally, in Section 6, the key issues raised in the paper are

**2. Rusck's conventional model for induced voltage estimation in overhead** 

estimation in distribution lines caused by atmospheric discharges are presented.

In this section the main aspects concerning to Rusck's methodology for induced voltage

Therefore, it is achieved a general study regarding induced voltage estimation in distribution lines through use of conventional methods discussed in the technical

Although the methodology originally developed in Rusck (1957) has some limitations to areas with soil resistivity less than 100 m, it is still widely used for induced voltage estimation in overhead distribution and transmission lines generated from indirect

The induced voltage estimation methodology presented in Rusck (1957) has as start point the modeling of the return current imposed by the atmospheric discharge in the distribution line. Rusck's method calculates the electric field generated by this return current in the ground surface and, from this electric field and from the line multi-wire arrangement, the

In Rubinstein & Uman (1989) is mathematically demonstrated that the studies presented in Rusck (1957) for resultant electric field computation of return current is correct. This fact has contributed to increase the reliability in relation to method developed by Rusck. Other additional procedures involved with models for induced voltage are also found in Cooray

An existent question related to this theory is that it estimates induced voltage values for conductors of a multi-wire line taking just into account the conductor geometric

theory provides the resultant values of induced voltages along the distribution line.

procedures involved with the protection specification.

related to the protection of equipments and transformers.

atmospheric discharges occurred near to the respective line.

summarized and conclusions are drawn.

**distribution lines** 

literature.

(2003).

$$\mathbf{U}(\mathbf{x},\mathbf{t}) = \Re \mathbf{0} \cdot \mathbf{I} \cdot \mathbf{h} \cdot \boldsymbol{\mathfrak{R}} \frac{(\mathbf{c} \cdot \mathbf{t} - \mathbf{x})}{\left[\mathbf{y}^2 + \boldsymbol{\mathfrak{R}}^2 \left(\mathbf{c} \cdot \mathbf{t} - \mathbf{x}\right)^2\right]} \cdot \left(1 + \frac{\mathbf{x} + \boldsymbol{\mathfrak{R}}^2 (\mathbf{c} \cdot \mathbf{t} - \mathbf{x})}{\sqrt{\left(\mathbf{\bar{c}} \cdot \mathbf{c} \cdot \mathbf{t}\right)^2 + \left(1 - \boldsymbol{\mathfrak{R}}^2\right) \left(\mathbf{x}^2 + \mathbf{y}^2\right)}}\right) \tag{2}$$

$$\beta = \frac{\mathbf{v}}{\mathbf{c}} \approx \sqrt{\frac{1}{1 + \frac{5.10^5}{\mathbf{I}}}} \tag{3}$$

In this case, *V*(*x*,*t*) is the induced voltage (V) at a point *x* of the line; *t* is the time in seconds; *c* is the velocity of light in free space (m/s); *I* is the return-peak current value (A); *h* is the average height of the distribution line; *y* is the closest distance between the discharge incidence point and the distribution line (m) and *x* is a point along the line (m).

Equations (1), (2) and (3) express Rusck's theory basis. In (4) the expression for the maximum induced voltage at the point *x*=0m is given by:

$$\mathbf{V}\_{\text{max}} \approx \frac{\mathbf{38.8 \cdot I \cdot h}}{\mathbf{y}} \tag{4}$$

From the previous expressions is possible to identify that they provide an analytic form for the computation of induced voltage in a distribution line, whereas other existent theories provide just iterative expressions that have high computational effort to perform the same estimation.

In Fig. 1 is presented the induced voltage at the point *x*=0m for an atmospheric discharge represented by a step function with amplitude *I*=10 kA in relation to an infinite line with 10 meters of height, where the distance between the atmospheric discharge from the distribution line is 100 meters.

In order to illustrate how the proposed formulation in this section is efficient for induced voltage estimation in overhead distribution lines, the induced voltage profile for different positions along the distribution line is presented in Fig. 2.

Intelligent Expert System

time.

conductor.

3

Fig. 3. Maximum induced voltage variation along the line.

energy dissipation in relation to the metallic conductors.

practically constant along the distribution line.

more smooth along the line.

3.1

3.2

voltage (V)

3.3

3.4

3.5 x 104

for Protection Optimization Purposes in Electric Power Distribution Systems 281

It is observed from Fig. 2 that the induced voltage waveform modifies in relation to the distance between the maximum voltage point and the measurement point. The alterations in the induced voltage waveforms along the distribution line can be better verified through their parameters, such as maximum induced voltage, rising time, peak time and half-wave

For comparative effects, it is assumed as rising time that necessary time for the voltage wavefront to reach 90% of its maximum value, considering half-wave time as that necessary time for the voltage wavefront to reach 50% of peak value after the occurrence of its maximum value. Therefore, Fig. 3 to 6 presents how these parameters are altered in relation to the distance between the maximum voltage point and a point along this

<sup>0</sup> <sup>100</sup> <sup>200</sup> <sup>300</sup> <sup>400</sup> <sup>500</sup> <sup>600</sup> <sup>700</sup> <sup>800</sup> <sup>900</sup> <sup>1000</sup> 2.9

distance (m)

From Fig. 3, it is observed that the voltage along the line length reduces at a rate practically linear in relation to the distance from the atmospheric discharge occurrence point. This observation indicates that the voltage wave along the distribution line suffers an attenuation generated from high frequencies involved with the propagation process as well as from

Figs. 4 to 6 illustrate how rising time, peak time and half-wave time alter along the distribution line. We can certify that these three parameters tend to increase at a rate

This fact indicates that the voltage waveform loses energy in relation to the distance along the line since that rising time, peak time and half-wave time higher cause voltage gradients

Fig. 1. Induced voltage in relation to the maximum voltage point in infinite line with 10m of height and atmospheric discharge of 10kA in perpendicular distance of 100m from the line.

Fig. 2. Induced voltages in infinite distribution line with 10m of height and atmospheric discharge of 10kA in perpendicular distance of 100m from the line in relation to several points along the line.

0 1 2 3 4 5 6

time (s)

0 1 2 3 4 5 6

time (s)

Fig. 2. Induced voltages in infinite distribution line with 10m of height and atmospheric discharge of 10kA in perpendicular distance of 100m from the line in relation to several

Fig. 1. Induced voltage in relation to the maximum voltage point in infinite line with 10m of height and atmospheric discharge of 10kA in perpendicular distance of 100m from the line.

0


points along the line.


0

1

voltage (V)

2

3

4 x 104

0.5

1

1.5

voltage (V)

2

2.5

3

3.5 x 104

x 10-5

x=0 x=100 x=200 x=300 x=400 x=500 x=600 x=700 x=800 x=900 x=1000

x 10-5

It is observed from Fig. 2 that the induced voltage waveform modifies in relation to the distance between the maximum voltage point and the measurement point. The alterations in the induced voltage waveforms along the distribution line can be better verified through their parameters, such as maximum induced voltage, rising time, peak time and half-wave time.

For comparative effects, it is assumed as rising time that necessary time for the voltage wavefront to reach 90% of its maximum value, considering half-wave time as that necessary time for the voltage wavefront to reach 50% of peak value after the occurrence of its maximum value. Therefore, Fig. 3 to 6 presents how these parameters are altered in relation to the distance between the maximum voltage point and a point along this conductor.

Fig. 3. Maximum induced voltage variation along the line.

From Fig. 3, it is observed that the voltage along the line length reduces at a rate practically linear in relation to the distance from the atmospheric discharge occurrence point. This observation indicates that the voltage wave along the distribution line suffers an attenuation generated from high frequencies involved with the propagation process as well as from energy dissipation in relation to the metallic conductors.

Figs. 4 to 6 illustrate how rising time, peak time and half-wave time alter along the distribution line. We can certify that these three parameters tend to increase at a rate practically constant along the distribution line.

This fact indicates that the voltage waveform loses energy in relation to the distance along the line since that rising time, peak time and half-wave time higher cause voltage gradients more smooth along the line.

Intelligent Expert System

0.9

considered in Rusck's method.

(2007) and Borghetti et al. (2007).

Fig. 6. Half-wave time variation along the distribution line.

0.95

1

1.05

1.1

half-wave time (s)

& Portela, 2007).

1.15

1.2

1.25

1.3

1.35 x 10-5

for Protection Optimization Purposes in Electric Power Distribution Systems 283

<sup>0</sup> <sup>100</sup> <sup>200</sup> <sup>300</sup> <sup>400</sup> <sup>500</sup> <sup>600</sup> <sup>700</sup> <sup>800</sup> <sup>900</sup> <sup>1000</sup> 0.85

distance (m)

From simulations accomplished and presented in this section and taking also into account the formulation proposed in Rusck (1957), it is verified that the obtained results by Rusck's method are coherent with those obtained through field experiments (Eriksson et al., 1982) or even with those results produced using reduced model techniques (Paula et al., 2001; Salari

However, some modifications in this method are necessary in order to transpose this methodology to practical situations involved with real distribution systems. Basically, it is necessary the consideration of current waveforms for the atmospheric discharge similar to those found in the nature. It is needed due to the atmospheric discharge characteristics

As presented at the beginning of this section, the waveform for the atmospheric discharge current used in Rusck's methodology has been a step function. In the next section, the necessary modifications in the approach proposed in Rusck (1957) are conducted in order to

Other works involving practical extension of Rusck's formula for maximum lightninginduced voltages that accounts for ground resistivity and improved procedures for the assessment of overhead line indirect lightning performance can also be found in Darveniza

complement the existent theory, becoming it appropriate for practical applications.

Fig. 4. Rising time variation along the distribution line.

Fig. 5. Peak time variation along the distribution line.

0 100 200 300 400 500 600 700 800 900 1000

distance (m)

<sup>0</sup> <sup>100</sup> <sup>200</sup> <sup>300</sup> <sup>400</sup> <sup>500</sup> <sup>600</sup> <sup>700</sup> <sup>800</sup> <sup>900</sup> <sup>1000</sup> <sup>2</sup>

distance (m)

1.5

2.5

Fig. 5. Peak time variation along the distribution line.

3

3.5

4

4.5

peak time (s)

5

5.5

6

6.5 x 10-6

Fig. 4. Rising time variation along the distribution line.

2

2.5

3

3.5

rising time (s)

4

4.5

5 x 10-6

Fig. 6. Half-wave time variation along the distribution line.

From simulations accomplished and presented in this section and taking also into account the formulation proposed in Rusck (1957), it is verified that the obtained results by Rusck's method are coherent with those obtained through field experiments (Eriksson et al., 1982) or even with those results produced using reduced model techniques (Paula et al., 2001; Salari & Portela, 2007).

However, some modifications in this method are necessary in order to transpose this methodology to practical situations involved with real distribution systems. Basically, it is necessary the consideration of current waveforms for the atmospheric discharge similar to those found in the nature. It is needed due to the atmospheric discharge characteristics considered in Rusck's method.

As presented at the beginning of this section, the waveform for the atmospheric discharge current used in Rusck's methodology has been a step function. In the next section, the necessary modifications in the approach proposed in Rusck (1957) are conducted in order to complement the existent theory, becoming it appropriate for practical applications.

Other works involving practical extension of Rusck's formula for maximum lightninginduced voltages that accounts for ground resistivity and improved procedures for the assessment of overhead line indirect lightning performance can also be found in Darveniza (2007) and Borghetti et al. (2007).

Intelligent Expert System

in Fig. 7, can be provided as shown in Fig. 8.

0


Fig. 8. Discrete current waveform composed by steps.

0

10

20

30

40

discrete current waveform (A)

50

60

70

80

90

Fig. 7. Current waveform for atmospheric discharge modeling.

2000

4000

6000

8000

current (A)

10000

12000

14000

for Protection Optimization Purposes in Electric Power Distribution Systems 285

steps. Thus, the value of each one of them, which represent the current waveform presented

0 0.2 0.4 0.6 0.8 1 1.2

time (s)

0 0.2 0.4 0.6 0.8 1 1.2

time (s)

x 10-5

x 10-5

#### **3. Modification of the conventional theory for induced voltage estimation in practical applications**

#### **3.1 Generalization of rusck's methodology for generic discharge current waveform**

Rusck's formulation presupposes that the atmospheric discharge can be represented by a waveform represented by a step function. However, measurements achieved in field have evidenced that the current waveform characteristics influence in the induced voltage in distribution lines located nearby the discharge occurrence point.

More specifically, parameters such as rising time and current waveform peak time have high correlation with the voltage induction process in distribution lines. Therefore, it is suggested that the induced voltage estimation in distribution lines to be achieved considering a waveform for discharge current near to that found in nature.

An approach often adopted for the atmospheric discharge current modeling can be provided as in (5), that is:

$$\mathbf{i(t) = i\_{h1}(t) + i\_{h2}(t) + i\_{de}(t)}\tag{5}$$

where:

$$\mathbf{i}\_{\rm hm}\left(\mathbf{t}\right) = \frac{\mathbf{I}\_{0\rm m}}{\eta\_{\rm m}} \frac{\left(\frac{\mathbf{t}}{\tau\_{\rm m1}}\right)^{\rm nm}}{\mathbf{1} + \left(\frac{\mathbf{t}}{\tau\_{\rm m1}}\right)^{\rm nm}} \exp\left(-\frac{\mathbf{t}}{\tau\_{\rm m2}}\right) \tag{6}$$

$$\mathbf{i}\_{\rm de}\left(\mathbf{t}\right) = \left[\left(1 - \exp(\alpha)\right) - \left(1 - \exp(\beta)\right)\right] \tag{7}$$

and:

$$\eta\_{\rm m} = \exp\left[-\left(\frac{\tau\_{\rm m1}}{\tau\_{\rm m2}}\right) \cdot \left(\mathbf{nm} \frac{\tau\_{\rm m2}}{\tau\_{\rm m1}}\right)^{\frac{1}{\rm mm}}\right] \tag{8}$$

Equation (6) is an example of Heidler's functions. An alternative frequently employed in atmospheric discharge modeling is double exponential.

Nevertheless, the modeling through two Heidler's function, as presented in (5), provides a more appropriate approximation for representation of the real phenomenon since the derivative of current at the instant t=0s is null. This fact is proved by innumerous practical cases.

In Fig. 7 is illustrated the current waveform results from modeling presented in this section, where the current has a peak value near to 12kA with a time of 0,81x10-6 s.

Supposing that the system to be linear, it is possible the use of Duhamel's integral (Greenwood, 1992) in order to represent the current waveform through a successive series of

**3. Modification of the conventional theory for induced voltage estimation in** 

**3.1 Generalization of rusck's methodology for generic discharge current waveform**  Rusck's formulation presupposes that the atmospheric discharge can be represented by a waveform represented by a step function. However, measurements achieved in field have evidenced that the current waveform characteristics influence in the induced voltage in

More specifically, parameters such as rising time and current waveform peak time have high correlation with the voltage induction process in distribution lines. Therefore, it is suggested that the induced voltage estimation in distribution lines to be achieved

An approach often adopted for the atmospheric discharge current modeling can be

nm

m1

 

nm m1 m2 <sup>m</sup>

Equation (6) is an example of Heidler's functions. An alternative frequently employed in

Nevertheless, the modeling through two Heidler's function, as presented in (5), provides a more appropriate approximation for representation of the real phenomenon since the derivative of current at the instant t=0s is null. This fact is proved by innumerous practical

In Fig. 7 is illustrated the current waveform results from modeling presented in this section,

Supposing that the system to be linear, it is possible the use of Duhamel's integral (Greenwood, 1992) in order to represent the current waveform through a successive series of

where the current has a peak value near to 12kA with a time of 0,81x10-6 s.

exp nm

m2 m1

0m m1 hm nm <sup>m</sup> m2

t <sup>I</sup> <sup>t</sup> i t exp <sup>t</sup> <sup>1</sup>

 

h1 h2 de i(t) i (t) i (t) i (t) (5)

i t [(1 exp( )) (1 exp( ))] de (7)

1

(6)

(8)

distribution lines located nearby the discharge occurrence point.

atmospheric discharge modeling is double exponential.

considering a waveform for discharge current near to that found in nature.

**practical applications** 

provided as in (5), that is:

where:

and:

cases.

steps. Thus, the value of each one of them, which represent the current waveform presented in Fig. 7, can be provided as shown in Fig. 8.

Fig. 7. Current waveform for atmospheric discharge modeling.

Fig. 8. Discrete current waveform composed by steps.

Intelligent Expert System

for Protection Optimization Purposes in Electric Power Distribution Systems 287

*U*(-*x*0,*t*) *U*(*x*0,*t*)

*x*0

As there is no line located at the right part of the point *x*1, there is no contribution of loads coming from the right of *x*1, that is, the voltage contribution *U*(*x*1,*t*) is null. As the line has

> f L f L R Z R Z

where *ZL* is the characteristic impedance of the distribution line. The numeric value for *ZL* is

<sup>h</sup> Z 138 log 2

Supposing that the discontinuity at the point *x*1 to be substituted by a compensation source,

r

termination impedance, the voltage at the point *x*1 can be computed as follows:

where is the reflection coefficient. The expression to obtain is given by:

L

where *h* is the height of the distribution line and *r* is the conductor diameter.

the value of this source can be computed according to the following development:

*x*1

*Rf* 

V(x ,t) U(x ,t) U( x ,t) 11 1 (10)

V(x ,t) U( x ,t) U( x ,t) 11 1 (11)

(12)

(13)

V(x ,t) V U( x ,t) U( x ,t) 1 11 (14)

Fig. 10. Induced voltage composition at the point x0 of the line.

Fig. 11. Induced voltage in a finite line.

provided by:

*U*(-*x*1,*t*)

If the line was infinite, the voltage at the point *x*1 would be given by:

From this modification, the induced voltage at any point *x* in the distribution line can be given by the sum of individual contribution in relation to each discrete current component.

Supposing an atmospheric discharge characterized as in Fig. 7, occurring in a distance of 100m from infinite distribution line with 10m of height, the voltage waveform at the point x=0m (point of maximum voltage value) can be represented as in Fig. 9.

Fig. 9. Induced voltage at the point of maximum voltage value for a current waveform expressed in terms of Heidler's function.

#### **3.2 Considerations for induced voltage estimation in finite lines**

Rusck's expression for induced voltage calculation in distribution lines is composed of two parcels, which can be observed through the following equation:

$$\mathbf{V(x\_0, t) = U(x\_0, t) + U(-x\_0, t)}\tag{9}$$

where *V*(*x*0,*t*) is the induced voltage at the point *x* of the line; *U*(*x*0,*t*) is the induced voltage component due to the load contribution located at the right part of this point and *U*(*-x*0,*t*) is the induced voltage component due to the load contribution located at the left part of *x*0 .

In Fig. 10 is presented the interpretation of induced voltage proposed in the formulation suggested by Rusck.

In case of a finite line, some modifications in Rusck's theory must be incorporated in order to enable that the induced voltage estimation in any point of the distribution line to be modeled adequately according to real situations.

Hence, we assume a line with termination in *x*1 with impedance of termination *Rf* as indicated in Fig. 11.

Fig. 10. Induced voltage composition at the point x0 of the line.

Fig. 11. Induced voltage in a finite line.

286 Intelligent Systems

From this modification, the induced voltage at any point *x* in the distribution line can be given by the sum of individual contribution in relation to each discrete current component. Supposing an atmospheric discharge characterized as in Fig. 7, occurring in a distance of 100m from infinite distribution line with 10m of height, the voltage waveform at the point

0 1 2

time (s)

Rusck's expression for induced voltage calculation in distribution lines is composed of two

where *V*(*x*0,*t*) is the induced voltage at the point *x* of the line; *U*(*x*0,*t*) is the induced voltage component due to the load contribution located at the right part of this point and *U*(*-x*0,*t*) is the induced voltage component due to the load contribution located at the left part of *x*0 .

In Fig. 10 is presented the interpretation of induced voltage proposed in the formulation

In case of a finite line, some modifications in Rusck's theory must be incorporated in order to enable that the induced voltage estimation in any point of the distribution line to be

Hence, we assume a line with termination in *x*1 with impedance of termination *Rf* as

V(x ,t) U(x ,t) U( x ,t) 00 0 (9)

Fig. 9. Induced voltage at the point of maximum voltage value for a current waveform

**3.2 Considerations for induced voltage estimation in finite lines** 

parcels, which can be observed through the following equation:

x 10-4

x=0m (point of maximum voltage value) can be represented as in Fig. 9.

0

suggested by Rusck.

indicated in Fig. 11.

expressed in terms of Heidler's function.

modeled adequately according to real situations.

0.5

1

1.5

voltage (V)

2

2.5

3

3.5 x 104

If the line was infinite, the voltage at the point *x*1 would be given by:

$$\mathbf{V(x\_1, t) = U(x\_1, t) + U(-x\_1, t)}\tag{10}$$

As there is no line located at the right part of the point *x*1, there is no contribution of loads coming from the right of *x*1, that is, the voltage contribution *U*(*x*1,*t*) is null. As the line has termination impedance, the voltage at the point *x*1 can be computed as follows:

$$\mathbf{V(x\_1, t)} = \mathbf{U(-x\_1, t)} + \Gamma \mathbf{U(-x\_1, t)}\tag{11}$$

where is the reflection coefficient. The expression to obtain is given by:

$$\Gamma = \frac{\mathbf{R\_f} - \left| Z\_{\rm L} \right|}{\mathbf{R\_f} + \left| Z\_{\rm L} \right|} \tag{12}$$

where *ZL* is the characteristic impedance of the distribution line. The numeric value for *ZL* is provided by:

$$Z\_{\rm L} = 138 \cdot \log \left( 2 \frac{\rm h}{\rm r} \right) \tag{13}$$

where *h* is the height of the distribution line and *r* is the conductor diameter.

Supposing that the discontinuity at the point *x*1 to be substituted by a compensation source, the value of this source can be computed according to the following development:

$$\mathbf{V}(\mathbf{x}\_1, \mathbf{t}) + \boldsymbol{\Delta V} = \mathbf{U}(-\mathbf{x}\_1, \mathbf{t}) + \boldsymbol{\Gamma}\mathbf{U}(-\mathbf{x}\_1, \mathbf{t}) \tag{14}$$

Intelligent Expert System

distribution transformer of 75kVA.

Fig. 12. Transformer registration window.

well as it emphasizes the preliminary results of simulations.

region.

for Protection Optimization Purposes in Electric Power Distribution Systems 289

Therefore, the transformers and equipments protection designs, through this expert system, consider the induced voltage in distribution line where the transformer is installed, the distribution network topology, as well as the atmospheric discharge characteristics of the

As the system operates through these databases, before the specification of a determined design, it is necessary that each one of the system elements to be adequately registered. Then, it is presented in Fig. 12, the transformer registration window adequately filled for a

After registration of each component of the electric system, inclusively of the arresters, the protection design can be registered. Fig. 13 illustrates the project registration window, as

$$\mathbf{U}(\mathbf{x}\_1, \mathbf{t}) + \mathbf{U}(-\mathbf{x}\_1, \mathbf{t}) + \Delta \mathbf{V} = \mathbf{U}(-\mathbf{x}\_1, \mathbf{t}) + \Gamma \mathbf{U}(-\mathbf{x}\_1, \mathbf{t}) \tag{15}$$

$$
\Delta \mathbf{V} = \Gamma \mathbf{U}(-\mathbf{x}\_1, \mathbf{t}) - \mathbf{U}(\mathbf{x}\_1, \mathbf{t}) \tag{16}
$$

The compensation source of value *V* is applied in the point *x*1, but its effect must be propagated throughout the line, since the non existence of line in the right of *x*1 alters the induced voltage values along whole line.

In order to compute the voltage at any point *x*, we can sum the induced voltage computed for the point *x*, assuming an infinite line, to the value of the compensation source applied in *x*1.

However, the compensation source located at *x*1 suffers a time delay during the trajectory between *x* and *x*1, that is:

$$\mathbf{V}(\mathbf{x},\mathbf{t}) = \mathbf{U}(\mathbf{x},\mathbf{t}) + \mathbf{U}(-\mathbf{x},\mathbf{t}) + [\Gamma \mathbf{U}(-\mathbf{x}\_1,\mathbf{t}) - \mathbf{U}(\mathbf{x}\_1,\mathbf{t})] \mathbf{u}(\mathbf{t} - \mathbf{t}\_1) \tag{17}$$

The function *u*(*t – tf*) is a unit step function and *tf* is the travel time between the point *x* and *x*1, i.e.:

$$\mathbf{t}\_{\mathbf{f}} = \frac{|\mathbf{x} - \mathbf{x}\_{1}|}{\mathbf{v}\_{0}} \tag{18}$$

where *v*0 is the propagation velocity, which for the simulations in question will be assumed as being equal to the velocity of light.

The same procedure can be achieved supposing a discontinuity at the left of *x*. Then, assuming a finite line, with the origin at the point *x*0 and termination in *xf* , the induced voltage at a point *x* along the distribution line can be estimated according to the following expression:

$$\begin{aligned} \mathbf{V}(\mathbf{x}, \mathbf{t}) &= \mathbf{U}(\mathbf{x}, \mathbf{t}) + \mathbf{U}(-\mathbf{x}, \mathbf{t}) + [\Gamma\_1 \mathbf{U}(-\mathbf{x}\_\mathbf{t}, \mathbf{t}) - \mathbf{U}(\mathbf{x}\_\mathbf{t}, \mathbf{t})] \cdot \mathbf{u}(\mathbf{t} - \mathbf{t}\_\mathbf{t}) + \cdots \\ &\cdots + [\Gamma\_0 \mathbf{U}(\mathbf{x}\_0, \mathbf{t}) - \mathbf{U}(-\mathbf{x}\_0, \mathbf{t})] \cdot \mathbf{u}(\mathbf{t} - \mathbf{t}\_0) \end{aligned} \tag{19}$$

The replacement of the line discontinuity effect by a voltage compensation source is an effective procedure, mainly when is desired to produce computational algorithms.

#### **4. Expert system for specification of transformers and equipments protection against atmospheric discharges**

The expert system proposed in this work, which was developed in order to help in the arresters specification for equipments and distribution transformer protection, has its implementation aspects based on the studies about induced voltages presented in previous sections. Besides using those suggested modifications, the expert system incorporates in a integrated way the databases referent to equipments installed on the distribution lines of Bandeirante Energy, as well as the databases involved with arresters and atmospheric discharge characteristics incident in its concession region.

The compensation source of value *V* is applied in the point *x*1, but its effect must be propagated throughout the line, since the non existence of line in the right of *x*1 alters the

In order to compute the voltage at any point *x*, we can sum the induced voltage computed for the point *x*, assuming an infinite line, to the value of the compensation source applied in

However, the compensation source located at *x*1 suffers a time delay during the trajectory

The function *u*(*t – tf*) is a unit step function and *tf* is the travel time between the point *x* and

f

t

1

0 x x

v

where *v*0 is the propagation velocity, which for the simulations in question will be assumed

The same procedure can be achieved supposing a discontinuity at the left of *x*. Then, assuming a finite line, with the origin at the point *x*0 and termination in *xf* , the induced voltage at a point *x* along the distribution line can be estimated according to the following

00 0 0 V(x,t) U(x,t) U( x,t) [ U( x ,t) U(x ,t)] u(t t )

The replacement of the line discontinuity effect by a voltage compensation source is an

**4. Expert system for specification of transformers and equipments protection** 

The expert system proposed in this work, which was developed in order to help in the arresters specification for equipments and distribution transformer protection, has its implementation aspects based on the studies about induced voltages presented in previous sections. Besides using those suggested modifications, the expert system incorporates in a integrated way the databases referent to equipments installed on the distribution lines of Bandeirante Energy, as well as the databases involved with arresters and atmospheric

effective procedure, mainly when is desired to produce computational algorithms.

[ U(x ,t) U( x ,t)] u(t t )

induced voltage values along whole line.

as being equal to the velocity of light.

**against atmospheric discharges** 

discharge characteristics incident in its concession region.

between *x* and *x*1, that is:

*x*1.

*x*1, i.e.:

expression:

U(x ,t) U( x ,t) V U( x ,t) U( x ,t) 11 1 1 (15)

V(x,t) U(x,t) U( x,t) [ U( x ,t) U(x ,t)]u(t t ) 11 f (17)

ff f f

(19)

V U( x ,t) U(x ,t) 1 1 (16)

(18)

Therefore, the transformers and equipments protection designs, through this expert system, consider the induced voltage in distribution line where the transformer is installed, the distribution network topology, as well as the atmospheric discharge characteristics of the region.

As the system operates through these databases, before the specification of a determined design, it is necessary that each one of the system elements to be adequately registered. Then, it is presented in Fig. 12, the transformer registration window adequately filled for a distribution transformer of 75kVA.


Fig. 12. Transformer registration window.

After registration of each component of the electric system, inclusively of the arresters, the protection design can be registered. Fig. 13 illustrates the project registration window, as well as it emphasizes the preliminary results of simulations.

Intelligent Expert System

flowing to ground.

can be seen in Fig. 15.

improving it.

**Distribution Feeders**

*Expert System*

for Protection Optimization Purposes in Electric Power Distribution Systems 291

Other expert system treated in this chapter involves the creation of a computational platform that helps in the specification and decision making regarding the optimized design of grounding systems, which must take into account the particularities of the distribution system under consideration, such as extension of the network, installed equipment and even

However, the effects of lightning should still be considered, since the voltages induced on the line are higher than those where the surge arresters operate, which imply in current

Thus, analyzing in terms of optimizing the grounding system, the best arrangement must be defined according to the desired type of grounding, which is characterized as a problem of structural optimization. Furthermore, it is of fundamental importance to determine the parameters of the chosen arrangement, such as distance, depth, number of stems etc. The search of these variables characterizes a parametric adjustment problem, whose objective is

A representation of the operation of the expert system for optimization of grounding design

**Resistivity Tests Structural Paramters**

**Frequency Response Grounding Efficiency**

**GA Optimization**

**Optimized Ground System**

Fig. 15. Diagram representing the expert system for optimized design of grounding systems.

The efficiency of the grounding system must be checked every iteration of the optimization process with the purpose of verifying how the parametric and structural adjustments are

**Induced Voltage Estimation Atmospheric**

**Discharges**

to determine a grounding system where impedance, and not resistance, is minimal.

**5. Expert system for optimized design of grounding systems** 

the performance requirements expected for such system.


Fig. 13. Project registration window presenting preliminary results.

Fig. 14 presents the window where indicates how each selected arrester can be also employed for protection of distribution network nearby the transformer.


Fig. 14. Window indicating installation distance between arresters aiming the distribution line protection against atmospheric discharges.

Fig. 13. Project registration window presenting preliminary results.

line protection against atmospheric discharges.

employed for protection of distribution network nearby the transformer.

Fig. 14 presents the window where indicates how each selected arrester can be also

Fig. 14. Window indicating installation distance between arresters aiming the distribution

### **5. Expert system for optimized design of grounding systems**

Other expert system treated in this chapter involves the creation of a computational platform that helps in the specification and decision making regarding the optimized design of grounding systems, which must take into account the particularities of the distribution system under consideration, such as extension of the network, installed equipment and even the performance requirements expected for such system.

However, the effects of lightning should still be considered, since the voltages induced on the line are higher than those where the surge arresters operate, which imply in current flowing to ground.

Thus, analyzing in terms of optimizing the grounding system, the best arrangement must be defined according to the desired type of grounding, which is characterized as a problem of structural optimization. Furthermore, it is of fundamental importance to determine the parameters of the chosen arrangement, such as distance, depth, number of stems etc. The search of these variables characterizes a parametric adjustment problem, whose objective is to determine a grounding system where impedance, and not resistance, is minimal.

A representation of the operation of the expert system for optimization of grounding design can be seen in Fig. 15.

Fig. 15. Diagram representing the expert system for optimized design of grounding systems.

The efficiency of the grounding system must be checked every iteration of the optimization process with the purpose of verifying how the parametric and structural adjustments are improving it.

Intelligent Expert System

approximately 35 kV.

grounding system.

induced voltage (V)

for Protection Optimization Purposes in Electric Power Distribution Systems 293

The graph, in Fig. 17, highlights the behavior of the induced voltage in a distribution system, considering an optimized grounding system and using those same lightning data used for obtaining the results illustrated in Fig. 16. Depending on the optimization of the grounding system, there is a reduction of the peak value of induced voltage, which is now

Besides reduction of four times to the peak value, it is possible to verify that the duration of this electromagnetic event was less than that situation characterized by non-optimized

Fig. 17. Behavior of induced voltages for an optimized grounding system.

contemplate the evolution of the analysis of lightning, as illustrated in Fig. 18.

as information on lightning that occurred in the analyzed region.

as indicated by a rectangle highlighted in Fig. 19.

right side of the B-TERRA software.

The expert system for optimizing new designs of grounding systems, called B-TERRA, uses, to start its calculations, the feeders database, which contains structural information, as well

time (s)

While the database, with information about the feeders, are loaded, the operator can

The procedures performed by the B-TERRA software allow the identification of the area of influence of lightning on the devices registered in the database of EDP Bandeirante circuits,

In this figure, it is possible to see an example where the device 2047568 is selected and, consequently, the area of influence of lightning has been highlighted on the screen at the

Thus, at the end of the process, the result should be an optimized grounding system. To evaluate this, the electrical parameters of the grounding system will be initially estimated, taking into account its own structural parameters, such as number of stems, distance between them and their depths.

Once the electrical parameters of the grounding system are known, the characteristic impedance of the grounding can then be calculated. This characteristic impedance is a value that allows relating the propagation of voltage induced by the distribution system to the propagation by the grounding system. To relate both modes of propagation, it is necessary to consider the electrical data of the feeder.

Once the parameters to model the distribution system at high frequencies are known, as well as the characteristic impedance of the grounding system, it is then possible to conduct simulations to verify impulsive voltage in the system.

To illustrate this evaluation procedure, a feeder in which the grounding system is not optimally implemented will be considered. This feeder belongs to the substation BIR of EDP Bandeirante. The induced voltage in the distribution system was calculated by assuming standard data for the lightning. The temporal behavior of the induced voltages is presented in Figure 16.

Fig. 16. Behavior of induced voltages for a non-optimized grounding system.

Fig. 16 shows that the peak value of induced voltage to a non-optimized grounding system was above 140 kV (in module). These magnitudes can be compared with those obtained when considering an optimized grounding system.

Thus, at the end of the process, the result should be an optimized grounding system. To evaluate this, the electrical parameters of the grounding system will be initially estimated, taking into account its own structural parameters, such as number of stems, distance

Once the electrical parameters of the grounding system are known, the characteristic impedance of the grounding can then be calculated. This characteristic impedance is a value that allows relating the propagation of voltage induced by the distribution system to the propagation by the grounding system. To relate both modes of propagation, it is necessary

Once the parameters to model the distribution system at high frequencies are known, as well as the characteristic impedance of the grounding system, it is then possible to conduct

To illustrate this evaluation procedure, a feeder in which the grounding system is not optimally implemented will be considered. This feeder belongs to the substation BIR of EDP Bandeirante. The induced voltage in the distribution system was calculated by assuming standard data for the lightning. The temporal behavior of the induced voltages is presented

Fig. 16. Behavior of induced voltages for a non-optimized grounding system.

when considering an optimized grounding system.

Fig. 16 shows that the peak value of induced voltage to a non-optimized grounding system was above 140 kV (in module). These magnitudes can be compared with those obtained

time (s)

between them and their depths.

in Figure 16.

induced voltage (V)

to consider the electrical data of the feeder.

simulations to verify impulsive voltage in the system.

The graph, in Fig. 17, highlights the behavior of the induced voltage in a distribution system, considering an optimized grounding system and using those same lightning data used for obtaining the results illustrated in Fig. 16. Depending on the optimization of the grounding system, there is a reduction of the peak value of induced voltage, which is now approximately 35 kV.

Besides reduction of four times to the peak value, it is possible to verify that the duration of this electromagnetic event was less than that situation characterized by non-optimized grounding system.

Fig. 17. Behavior of induced voltages for an optimized grounding system.

The expert system for optimizing new designs of grounding systems, called B-TERRA, uses, to start its calculations, the feeders database, which contains structural information, as well as information on lightning that occurred in the analyzed region.

While the database, with information about the feeders, are loaded, the operator can contemplate the evolution of the analysis of lightning, as illustrated in Fig. 18.

The procedures performed by the B-TERRA software allow the identification of the area of influence of lightning on the devices registered in the database of EDP Bandeirante circuits, as indicated by a rectangle highlighted in Fig. 19.

In this figure, it is possible to see an example where the device 2047568 is selected and, consequently, the area of influence of lightning has been highlighted on the screen at the right side of the B-TERRA software.

Intelligent Expert System

selected device.

**6. Conclusion** 

**7. References** 

for Protection Optimization Purposes in Electric Power Distribution Systems 295

Fig. 20. Output of the B-TERRA software related to the best grounding design for the

waveform, as well as assuming a finite distribution line.

protection of the distribution line where these equipments are inserted.

specification related to the equipments and transformers protection.

observed in a non-optimized system, as shown in the charts in Section 5.

In this chapter, it has been presented the theoretical development employed to estimate induced voltages in overhead distribution lines supposing a generic discharge current

Taking into account the results provided by the developed technique, an expert system to help in the equipments and distribution transformers protection specification was implemented in order to provide indicatives about the best protection to be adopted to transformers, as well as the best installation distance between arresters, aiming the full

Performance evaluations indicate that the expert system provides coherent results and its practical application contributes to optimize the processes involved with parameters

Additionally, an expert system to assist in the specification of parameters for grounding designs was also implemented. Regarding the grounding system optimized with this tool, we can state that energy to be dissipated on the distribution system is lower than that

The difference in energy between the two cases implies in the energy that flows through the grounding system, i.e., for being better designed the optimized grounding system allows more energy to flow by itself compared to that case of non-optimized system. Experimental evaluations indicate that they provide very consistent results and their practical applications help for optimizing the processes involved with the protection of distribution systems.

Borghetti, A.; Nucci, C. A. & Paolone, M. (2007). An improved procedure for the assessment

Cooray, G. V. (2003). *The Lightning Flash*. IEE Power & Energy Series, London, UK.

of overhead line indirect lightning performance and its comparison with the IEEE Std. 1410 method. *IEEE Transactions on Power Delivery*, Vol. 22, No.1, pp. 684-692.

Fig. 18. Results of evolution analysis of lightning intensity.

Fig. 19. Area identification where the devices are influenced by the lightning.

After accomplishment of all optimization procedures, through genetic algorithms, B-TERRA software provides as one of its answers the configuration of the best grounding design for the selected device, as shown in Figure 20.

Fig. 20. Output of the B-TERRA software related to the best grounding design for the selected device.

#### **6. Conclusion**

294 Intelligent Systems

D:\atmospheric\_discharges\_2006.txt Search

Atmospheric discharges intensity distribution histogram

Cancel Import

Atmospheric discharges data input in the B-Terra database

Atmospheric discharges file

Fig. 18. Results of evolution analysis of lightning intensity.

LV point service

the selected device, as shown in Figure 20.

Switches

Fig. 19. Area identification where the devices are influenced by the lightning.

After accomplishment of all optimization procedures, through genetic algorithms, B-TERRA software provides as one of its answers the configuration of the best grounding design for

Capacitors

Transformer unity

MV construction

Power transformers

In this chapter, it has been presented the theoretical development employed to estimate induced voltages in overhead distribution lines supposing a generic discharge current waveform, as well as assuming a finite distribution line.

Taking into account the results provided by the developed technique, an expert system to help in the equipments and distribution transformers protection specification was implemented in order to provide indicatives about the best protection to be adopted to transformers, as well as the best installation distance between arresters, aiming the full protection of the distribution line where these equipments are inserted.

Performance evaluations indicate that the expert system provides coherent results and its practical application contributes to optimize the processes involved with parameters specification related to the equipments and transformers protection.

Additionally, an expert system to assist in the specification of parameters for grounding designs was also implemented. Regarding the grounding system optimized with this tool, we can state that energy to be dissipated on the distribution system is lower than that observed in a non-optimized system, as shown in the charts in Section 5.

The difference in energy between the two cases implies in the energy that flows through the grounding system, i.e., for being better designed the optimized grounding system allows more energy to flow by itself compared to that case of non-optimized system. Experimental evaluations indicate that they provide very consistent results and their practical applications help for optimizing the processes involved with the protection of distribution systems.

#### **7. References**

Borghetti, A.; Nucci, C. A. & Paolone, M. (2007). An improved procedure for the assessment of overhead line indirect lightning performance and its comparison with the IEEE Std. 1410 method. *IEEE Transactions on Power Delivery*, Vol. 22, No.1, pp. 684-692. Cooray, G. V. (2003). *The Lightning Flash*. IEE Power & Energy Series, London, UK.

**14** 

Majid Tolouei-Rad

*Australia* 

**Intelligent Analysis of Utilization of Special** 

Drilling and drilling-related operations constitute more than 60% of all machining processes in manufacturing industries. Consequently, it is important to know how to perform these operations properly. With availability of many machining processes capable of performing drilling operations sometimes it is difficult to decide which process would result in a higher profit or a lower unit cost for a given task. Due to increasing global competition, manufacturing industries are now more concerned with their productivity and are more sensitive than ever to their investments with respect to flexibility and efficiency of production equipment (Boothroyd and Knight, 2005, Wecka and Staimer, 2002). Researchers (Ko *et al*., 2005) believe that increasing the quality of production and reducing cost and time of production are very important factors in achieving higher productivity. Achieving this goal requires reconsidering current production methods that could lead to introduction of

In traditional drilling processes a sharp cutting tool with multiple cutting edges is used to cut a round hole in the workpiece material. In non-traditional drilling processes various forms of energy other than sharp cutting tools or abrasive particles are used to remove the material. The energy forms include mechanical, electrochemical, thermal and chemical (Groover, 2010). Generally non-traditional processes incorporate high capital and operating costs. Therefore, when machining economy is of concern manufacturing companies focus on traditional processes. Even within this category, a machining specialist has the choice of using conventional drilling machines, CNC machines, and special purpose machines (SPMs). According to the literature (Groover 2008) when production quantity and variety are low, universal machine tools give the best result. When various components should be produced, CNC is the best option. For the condition of high production quantity with low variety, SPM gives the highest productivity and is considered as the most economic production method. Accordingly, Tolouei-Rad and Zolfaghari (2009) believe that SPMs are superior to computer numerical control (CNC) machines for producing large quantities of similar parts; however, most manufacturers still rely on well-known CNCs for large volume production tasks. This is mainly attributable to the fact that both SPMs and CNCs incorporate high capital costs; SPMs are more productive and CNCs are more flexible. When the part in production is no longer in demand due to frequent market changes, SPMs

new production techniques and more advanced technologies.

**1. Introduction** 

**Purpose Machines for Drilling Operations** 

*School of Engineering, Edith Cowan University, Perth,* 


### **Intelligent Analysis of Utilization of Special Purpose Machines for Drilling Operations**

Majid Tolouei-Rad

*School of Engineering, Edith Cowan University, Perth, Australia* 

#### **1. Introduction**

296 Intelligent Systems

Darveniza, M. (2007). A practical extension of Rusck's formula for maximum lightning-

Eriksson, A. J.; Stringfellow, M. F. & Meal, M. F. (1982). Lightning induced overvoltages on

Greenwood, A. (1992). *Electrical Transients in Power Systems*. John Wiley & Sons, New York,

Paula, S. C. M.; Mendonça, R. G.; Neto, L. M.; Medeiros, C. A. G. & Silva, R. V. R. (2001).

Rubinstein, M. & Uman, M. A. (1989). Methods for calculating the electromagnetic fields

Rusck, S. (1957). *Induced Lightning Over-voltage on Power Transmission Lines with Special* 

Salari, J. C. & Portela, C. (2007). A methodology for electromagnetic transients calculation –

*Electromagnetic Compatibility*, Vol. 31, No. 2, pp. 183-189.

*IEEE Transactions on Power Delivery*, Vol. 22, No. 1, pp. 527-536.

Royal Institute of Technology, Sweden.

*Delivery*, Vol. 22, No.1, pp. 605-612.

101, No. 4, pp. 960-968.

Canada, pp. 737-742.

USA.

induced voltages that accounts for ground resistivity. *IEEE Transactions on Power* 

overhead transmission lines. *IEEE Transactions on Power Apparatus and Systems*, Vol.

Evaluation of performance of groundings electrics in conditions of lightning current. *Canadian Conference on Electrical and Computer Engineering*, Toronto,

from a known source distribution: Application to lightning. *IEEE Transactions on* 

*Reference to Over-voltage Protection of Low Voltage Networks*. Ph.D. Thesis, Stockholm

an application for the calculation of lightning propagation in transmission lines.

Drilling and drilling-related operations constitute more than 60% of all machining processes in manufacturing industries. Consequently, it is important to know how to perform these operations properly. With availability of many machining processes capable of performing drilling operations sometimes it is difficult to decide which process would result in a higher profit or a lower unit cost for a given task. Due to increasing global competition, manufacturing industries are now more concerned with their productivity and are more sensitive than ever to their investments with respect to flexibility and efficiency of production equipment (Boothroyd and Knight, 2005, Wecka and Staimer, 2002). Researchers (Ko *et al*., 2005) believe that increasing the quality of production and reducing cost and time of production are very important factors in achieving higher productivity. Achieving this goal requires reconsidering current production methods that could lead to introduction of new production techniques and more advanced technologies.

In traditional drilling processes a sharp cutting tool with multiple cutting edges is used to cut a round hole in the workpiece material. In non-traditional drilling processes various forms of energy other than sharp cutting tools or abrasive particles are used to remove the material. The energy forms include mechanical, electrochemical, thermal and chemical (Groover, 2010). Generally non-traditional processes incorporate high capital and operating costs. Therefore, when machining economy is of concern manufacturing companies focus on traditional processes. Even within this category, a machining specialist has the choice of using conventional drilling machines, CNC machines, and special purpose machines (SPMs). According to the literature (Groover 2008) when production quantity and variety are low, universal machine tools give the best result. When various components should be produced, CNC is the best option. For the condition of high production quantity with low variety, SPM gives the highest productivity and is considered as the most economic production method. Accordingly, Tolouei-Rad and Zolfaghari (2009) believe that SPMs are superior to computer numerical control (CNC) machines for producing large quantities of similar parts; however, most manufacturers still rely on well-known CNCs for large volume production tasks. This is mainly attributable to the fact that both SPMs and CNCs incorporate high capital costs; SPMs are more productive and CNCs are more flexible. When the part in production is no longer in demand due to frequent market changes, SPMs

Intelligent Analysis of Utilization of Special Purpose Machines for Drilling Operations 299

Groover (2008) has defined the term "production automation" as the application of electrical, electronic, mechanical, hydraulic and pneumatic systems for rapid and quality productions in large volumes. Automated production techniques are widely used in manufacturing industries for dealing with issues such as high cost of labour, shortage of skilled people, low interest of labour to work in production firms, safety, high cost of raw materials, improved quality, uniformity in the quality of products, low inventory, customers satisfaction, and performing difficult operations. Figure 1 shows a SPM as an example of utilization of automated production techniques in manufacturing. This machine has two work stations, one for drilling and one for tapping. The machine is used for machining the

Fig. 1. A two station SPM for drilling and tapping operations with parts being produced

Generally SPMs lack the high rigidity found in conventional and CNC machines. Consequently, majority of these machines are used for performing drilling and drillingrelated operations such as tapping, reaming, counterboring and countersinking on machinable materials where the magnitude of machining forces is relatively low. This eliminates excessive vibrations of the machine tool due to high cutting forces. However, it should be noted that SPMs are also capable of performing milling and some other machining operations that would result in high cutting forces. In such cases there is a need for stronger chassis, stronger machining and sliding units, and use of special accessories in

The units used in SPMs can be divided into two main groups: machining and sliding. A machining unit is equipped with an electro-motor that revolves the spindle by means of pulley and belt systems in order to rotate the cutting tool. Like other machine tools, the connection of cutting tools to the machining unit is accomplished by standard tool holders. Machining units

**2. Fundamentals of SPM technology** 

parts shown in the Figure.

(Photo: Suhner, 2001)

order to eliminate vibrations when possible.

**2.1 Machining and sliding units** 

become idle while CNCs can be easily reprogrammed for producing other parts. Yet the concluding statement could be different when modular SPMs are utilized.

The field of machine tools for generating singular products is well documented; however, the area of specialist machines for dedicated tasks has received less attention (Allen *et al.*, 2010). This is particularly true for modular SPMs that are a new addition to the family of SPMs (Tolouei-Rad and Tabatabaei, 2005). Proper design and utilization of these machines depend upon knowledge, experience, and creativity of SPM designers and machining specialists. Because of modularity in structure, these machines can be applied to the production of a range of parts upon modification. The specific advantages of utilization of this technology have placed them in a superior position in comparison with other machine tools. These advantages include mass production of parts in shorter time, high accuracy of products, uniformity and repeatability of production, elimination of some quality control steps, simultaneous machining of a number of parts, and reduced labour and overhead costs.

The modular principle is very popular in the design of many products such as automobile, home appliances, information devices, industrial equipment, etc. This trend can be considered as one of the great contributions of modular design of machine tools to those working in other industries (Yoshimi, 2008). This article focuses on modular SPMs and for simplicity in the rest of this article modular SPM is referred to as SPM. SPMs do not have a rigid bulky configuration and the machine can be rapidly set up by putting together a number of machining and sliding units, chassis, and other equipment. This is achieved by making use of various types of mechanical fasteners. Once the part in production is no longer in demand, SPMs can be dis-assembled and re-assembled in a different configuration to be used for producing other parts. Properly utilization of SPMs could have a significant impact on the productivity of manufacturing industries; and production improvements of up to 25:1 have been reported (Suhner, 2001). However, the extent of the application of SPM technology in industry is not proportional to its potential impact on productivity improvement. This is mainly attributed to the fact that machining specialists find it difficult to decide when to use SPMs. Making the right decision is a time-consuming task and requires a techno-economical analysis to be performed by expert people. This article addresses a methodology developed to tackle this vital problem. It investigates the possibility and effectiveness of employing artificial intelligent techniques to assist manufacturing firms in feasibility analysis of utilizing SPMs in order to improve productivity. It is important to note that in spite of many publications on production technologies and machine tool design; publications on design and utilization of SPMs are very limited.

Intelligent systems have been extensively used to effectively tackle some real engineering problems in the last three decades. Yet researchers explore new application areas for utilization of various artificial intelligence techniques. Knowledge-based expert systems (KBESs) have proven to be effective for decision making when dealing with qualitative information, hard to capture in a computer program. Accordingly, in the current work a KBES has been developed and used for utilization feasibility analysis of SPMs in different manufacturing settings.

#### **2. Fundamentals of SPM technology**

298 Intelligent Systems

become idle while CNCs can be easily reprogrammed for producing other parts. Yet the

The field of machine tools for generating singular products is well documented; however, the area of specialist machines for dedicated tasks has received less attention (Allen *et al.*, 2010). This is particularly true for modular SPMs that are a new addition to the family of SPMs (Tolouei-Rad and Tabatabaei, 2005). Proper design and utilization of these machines depend upon knowledge, experience, and creativity of SPM designers and machining specialists. Because of modularity in structure, these machines can be applied to the production of a range of parts upon modification. The specific advantages of utilization of this technology have placed them in a superior position in comparison with other machine tools. These advantages include mass production of parts in shorter time, high accuracy of products, uniformity and repeatability of production, elimination of some quality control steps, simultaneous machining of a number of parts, and reduced labour and overhead

The modular principle is very popular in the design of many products such as automobile, home appliances, information devices, industrial equipment, etc. This trend can be considered as one of the great contributions of modular design of machine tools to those working in other industries (Yoshimi, 2008). This article focuses on modular SPMs and for simplicity in the rest of this article modular SPM is referred to as SPM. SPMs do not have a rigid bulky configuration and the machine can be rapidly set up by putting together a number of machining and sliding units, chassis, and other equipment. This is achieved by making use of various types of mechanical fasteners. Once the part in production is no longer in demand, SPMs can be dis-assembled and re-assembled in a different configuration to be used for producing other parts. Properly utilization of SPMs could have a significant impact on the productivity of manufacturing industries; and production improvements of up to 25:1 have been reported (Suhner, 2001). However, the extent of the application of SPM technology in industry is not proportional to its potential impact on productivity improvement. This is mainly attributed to the fact that machining specialists find it difficult to decide when to use SPMs. Making the right decision is a time-consuming task and requires a techno-economical analysis to be performed by expert people. This article addresses a methodology developed to tackle this vital problem. It investigates the possibility and effectiveness of employing artificial intelligent techniques to assist manufacturing firms in feasibility analysis of utilizing SPMs in order to improve productivity. It is important to note that in spite of many publications on production technologies and machine tool design; publications on design and utilization of SPMs are

Intelligent systems have been extensively used to effectively tackle some real engineering problems in the last three decades. Yet researchers explore new application areas for utilization of various artificial intelligence techniques. Knowledge-based expert systems (KBESs) have proven to be effective for decision making when dealing with qualitative information, hard to capture in a computer program. Accordingly, in the current work a KBES has been developed and used for utilization feasibility analysis of SPMs in different

concluding statement could be different when modular SPMs are utilized.

costs.

very limited.

manufacturing settings.

Groover (2008) has defined the term "production automation" as the application of electrical, electronic, mechanical, hydraulic and pneumatic systems for rapid and quality productions in large volumes. Automated production techniques are widely used in manufacturing industries for dealing with issues such as high cost of labour, shortage of skilled people, low interest of labour to work in production firms, safety, high cost of raw materials, improved quality, uniformity in the quality of products, low inventory, customers satisfaction, and performing difficult operations. Figure 1 shows a SPM as an example of utilization of automated production techniques in manufacturing. This machine has two work stations, one for drilling and one for tapping. The machine is used for machining the parts shown in the Figure.

Fig. 1. A two station SPM for drilling and tapping operations with parts being produced (Photo: Suhner, 2001)

Generally SPMs lack the high rigidity found in conventional and CNC machines. Consequently, majority of these machines are used for performing drilling and drillingrelated operations such as tapping, reaming, counterboring and countersinking on machinable materials where the magnitude of machining forces is relatively low. This eliminates excessive vibrations of the machine tool due to high cutting forces. However, it should be noted that SPMs are also capable of performing milling and some other machining operations that would result in high cutting forces. In such cases there is a need for stronger chassis, stronger machining and sliding units, and use of special accessories in order to eliminate vibrations when possible.

#### **2.1 Machining and sliding units**

The units used in SPMs can be divided into two main groups: machining and sliding. A machining unit is equipped with an electro-motor that revolves the spindle by means of pulley and belt systems in order to rotate the cutting tool. Like other machine tools, the connection of cutting tools to the machining unit is accomplished by standard tool holders. Machining units

Intelligent Analysis of Utilization of Special Purpose Machines for Drilling Operations 301

CNC machining units are also used for drilling, taping and milling operations precisely as they are equipped with servomotors. CNC units can be programmed for very accurate machining operations when used in conjunction with a controller. Figure 4 shows a two-axis CNC tapping unit. The tapping unit is mounted on the sliding unit where both the units are equipped with servomotors. The servomotor of the tapping unit provides the rotational motion of the cutting tool whereas the servomotor of the sliding unit provides feed motion. When integrated with a control unit, this assembly can be programmed similar to CNC

There exist special stands, adjustable bases, and supports used for positioning and supporting basic machine components. These are also used for preventing or reducing vibrations at the time of machining. Figure 5 shows some of the assembly equipment used to

Indexing table is one of the important accessories used in SPMs. Figure 6 shows an indexing table used for positioning the workpiece in different machining stations where the workpiece is machined in a number of rotary stations. After determination of machining

accurately position and support machining units in any position and at any angle.

Fig. 5. Special stands, adjustable bases, and supports (Photo: Suhner)

machines.

**2.2 Accessories** 

Fig. 4. Two-axis CNC tapping unit (Photo: Suhner)

are of three types: quill, power, and CNC. Quill units are used for light drilling and drillingrelated operations as they also provide the spindle with a linear feed motion necessary for penetration of the cutting tool into the workpiece. Both the linear and the rotational motions necessary for performing operation are provided simultaneously. Power units are used for drilling, drilling related, and milling operations where large cutting forces exist. Unlike quill units, power units lack the linear feed motion due to presence of significantly larger cutting forces that may cause deflection in the rotating spindle. Consequently, these units are mounted on the sliding units providing them with necessary linear feed motion. Figure 2 shows quill and power units together with tool holders and cutting tools.

Fig. 2. A pneumatic sliding unit with mechanical course adjustment (Photo: Tolouei-Rad and Zolfaghari, 2009)

Sliding units may carry machining units and provide necessary feed motion of the tool by means of hydraulic/pneumatic actuators, or servomotors. Adjusting the course of motion is provided by use of micro-switches or mechanical limits. Figure 3 shows a pneumatic sliding unit with a mechanical course adjustment device. The sliding plate that carries the machining unit is fastened to the connecting rod of the piston, and therefore, is capable of moving back and forth on the base. Depending on the nature of machining operation and cutting tool motion requirements, machining units can be mounted on the sliding unit such that spindle axis is either along or perpendicular to the sliding direction.

Fig. 3. A pneumatic sliding unit with mechanical course adjustment

CNC machining units are also used for drilling, taping and milling operations precisely as they are equipped with servomotors. CNC units can be programmed for very accurate machining operations when used in conjunction with a controller. Figure 4 shows a two-axis CNC tapping unit. The tapping unit is mounted on the sliding unit where both the units are equipped with servomotors. The servomotor of the tapping unit provides the rotational motion of the cutting tool whereas the servomotor of the sliding unit provides feed motion. When integrated with a control unit, this assembly can be programmed similar to CNC machines.

Fig. 4. Two-axis CNC tapping unit (Photo: Suhner)

#### **2.2 Accessories**

300 Intelligent Systems

are of three types: quill, power, and CNC. Quill units are used for light drilling and drillingrelated operations as they also provide the spindle with a linear feed motion necessary for penetration of the cutting tool into the workpiece. Both the linear and the rotational motions necessary for performing operation are provided simultaneously. Power units are used for drilling, drilling related, and milling operations where large cutting forces exist. Unlike quill units, power units lack the linear feed motion due to presence of significantly larger cutting forces that may cause deflection in the rotating spindle. Consequently, these units are mounted on the sliding units providing them with necessary linear feed motion. Figure 2 shows quill

and power units together with tool holders and cutting tools.

Fig. 2. A pneumatic sliding unit with mechanical course adjustment

that spindle axis is either along or perpendicular to the sliding direction.

Fig. 3. A pneumatic sliding unit with mechanical course adjustment

Sliding units may carry machining units and provide necessary feed motion of the tool by means of hydraulic/pneumatic actuators, or servomotors. Adjusting the course of motion is provided by use of micro-switches or mechanical limits. Figure 3 shows a pneumatic sliding unit with a mechanical course adjustment device. The sliding plate that carries the machining unit is fastened to the connecting rod of the piston, and therefore, is capable of moving back and forth on the base. Depending on the nature of machining operation and cutting tool motion requirements, machining units can be mounted on the sliding unit such

(Photo: Tolouei-Rad and Zolfaghari, 2009)

There exist special stands, adjustable bases, and supports used for positioning and supporting basic machine components. These are also used for preventing or reducing vibrations at the time of machining. Figure 5 shows some of the assembly equipment used to accurately position and support machining units in any position and at any angle.

Fig. 5. Special stands, adjustable bases, and supports (Photo: Suhner)

Indexing table is one of the important accessories used in SPMs. Figure 6 shows an indexing table used for positioning the workpiece in different machining stations where the workpiece is machined in a number of rotary stations. After determination of machining

Intelligent Analysis of Utilization of Special Purpose Machines for Drilling Operations 303

The table and chassis of the machine are very important considerations in SPMs. Based on technical considerations and machining properties of the workpiece material, the table and chassis are properly designed or selected from the standardized SPM chassis. Due to high machining forces resulting from machining operations the machine table and chassis should be sufficiently rigid to avoid vibrations. It is also very important to consider appropriate coolant and chip removal mechanisms in design of machine table

Because production process is systematic, planning for design and manufacturing has an effective influence on the success of any project (Lutters *et al.,* 2004). The flowchart shown in Figure 8 represents all necessary steps for proper analysis, design and manufacture of SPMs. These steps should be followed in order to achieve feasible results in SPM design and

As the cost of SPM design and manufacturing is relatively high, critical technical and economic justification of utilization of these machines should be made before any attempt to design and manufacture them. This includes an analysis of machinability of the workpiece, and a comparison of the production costs with other production alternatives considering production volume and machine amortisation period. For technical feasibility analysis a number of questions will be asked and the user needs to answer these questions interactively. These questions investigate quality of workpiece material and its physical and geometrical characteristics to determine whether or not it can be machined with SPMs. The flowchart shown in Figure 9 describes the type of questions asked for technical feasibility analysis. If the answer to any of the questions is "No" then the workpiece is considered to be "Not Suitable" for machining with SPM and its processing will be

Upon completion of technical feasibility analysis, an economical feasibility analysis is performed. To do so a detailed computation is needed in order to determine the cost of machining a unit of product using SPM. Then the same computation is repeated for traditional and CNC machines in order to achieve a unit cost comparison for different methods, and to find the one that results in a lower cost. For determination of unit cost so many factors are taken into consideration including machining time, production volume, machine cost, cutting tool cost, labour cost, overhead costs, depreciation cost, interest rate, etc. A case study is presented in Section 5 that provides a detailed economic analysis for a sample part. It is noteworthy that sometimes it is necessary to repeat the economic analysis before the final approval of SPM design. This happens when more accurate information on the cost of SPM and required accessories become available. This is represented by a dashed line in the

**2.3 Machine table, chip removal and coolant system** 

and chassis.

manufacturing.

terminated.

flowchart of Figure 8.

**3. Design and manufacturing** 

**3.1 Technical and economic analysis** 

steps and the number of working stations, fixtures could be placed on the indexing table. Number of stations could be anything between two and twelve, and is determined on the basis of production volume and technical considerations.

Fig. 6. An indexing table (Photo: Suhner, 2001)

Multi-drill heads provide the possibility of drilling many holes on the same plane simultaneously; thus, reducing machining time significantly. Multi-drill heads are divided into fixed and adjustable types. In fixed multi-drill heads the position of tools are fixed, but in adjustable ones the position of the tools could be adjusted as needed. Angle heads are spindle attachments used to alter the orientation of cutting tool axis relative to the spindle axis. These attachments are used in milling operations. Figure 7 shows different types of multi-drill heads and angle heads used in SPMs.

Fig. 7. Various types of multi-drill heads and angle heads (Photo: Suhner)

#### **2.3 Machine table, chip removal and coolant system**

The table and chassis of the machine are very important considerations in SPMs. Based on technical considerations and machining properties of the workpiece material, the table and chassis are properly designed or selected from the standardized SPM chassis. Due to high machining forces resulting from machining operations the machine table and chassis should be sufficiently rigid to avoid vibrations. It is also very important to consider appropriate coolant and chip removal mechanisms in design of machine table and chassis.

#### **3. Design and manufacturing**

302 Intelligent Systems

steps and the number of working stations, fixtures could be placed on the indexing table. Number of stations could be anything between two and twelve, and is determined on the

Multi-drill heads provide the possibility of drilling many holes on the same plane simultaneously; thus, reducing machining time significantly. Multi-drill heads are divided into fixed and adjustable types. In fixed multi-drill heads the position of tools are fixed, but in adjustable ones the position of the tools could be adjusted as needed. Angle heads are spindle attachments used to alter the orientation of cutting tool axis relative to the spindle axis. These attachments are used in milling operations. Figure 7 shows different types of

basis of production volume and technical considerations.

Fig. 6. An indexing table (Photo: Suhner, 2001)

multi-drill heads and angle heads used in SPMs.

Fig. 7. Various types of multi-drill heads and angle heads (Photo: Suhner)

Because production process is systematic, planning for design and manufacturing has an effective influence on the success of any project (Lutters *et al.,* 2004). The flowchart shown in Figure 8 represents all necessary steps for proper analysis, design and manufacture of SPMs. These steps should be followed in order to achieve feasible results in SPM design and manufacturing.

#### **3.1 Technical and economic analysis**

As the cost of SPM design and manufacturing is relatively high, critical technical and economic justification of utilization of these machines should be made before any attempt to design and manufacture them. This includes an analysis of machinability of the workpiece, and a comparison of the production costs with other production alternatives considering production volume and machine amortisation period. For technical feasibility analysis a number of questions will be asked and the user needs to answer these questions interactively. These questions investigate quality of workpiece material and its physical and geometrical characteristics to determine whether or not it can be machined with SPMs. The flowchart shown in Figure 9 describes the type of questions asked for technical feasibility analysis. If the answer to any of the questions is "No" then the workpiece is considered to be "Not Suitable" for machining with SPM and its processing will be terminated.

Upon completion of technical feasibility analysis, an economical feasibility analysis is performed. To do so a detailed computation is needed in order to determine the cost of machining a unit of product using SPM. Then the same computation is repeated for traditional and CNC machines in order to achieve a unit cost comparison for different methods, and to find the one that results in a lower cost. For determination of unit cost so many factors are taken into consideration including machining time, production volume, machine cost, cutting tool cost, labour cost, overhead costs, depreciation cost, interest rate, etc. A case study is presented in Section 5 that provides a detailed economic analysis for a sample part. It is noteworthy that sometimes it is necessary to repeat the economic analysis before the final approval of SPM design. This happens when more accurate information on the cost of SPM and required accessories become available. This is represented by a dashed line in the flowchart of Figure 8.

Intelligent Analysis of Utilization of Special Purpose Machines for Drilling Operations 305

Start

Machineable material?

Size & weight within limits?

Material thickness ≥ 3 mm?

All drilling related machining features?

Yes

Yes

Yes

Yes

Yes

Accuracy > 0.02 mm?

Surface roughness ≥ 0.08 µm?

End

Yes

Suitable for SPM

All hole dia between 3 to 40 mm?

Yes

No

No

No

No

No

No

No

Not suitable for SPM

Fig. 9. Flowchart for technical feasibility analysis

Properly determination of machining sequence is a key point in successful SPM utilization. A poor machining sequence plan leads to lower quality of production and/or increased machining times and consequently higher production costs. Often it is possible to combine and perform a number of operations in a single setup lowering machining times and costs while also improving production quality. Indeed machining sequence planning determines the overall configuration of the machine and required machining

**3.2 Machining sequence planning** 

Machineable materials include no heat treated low carbon steels, and alloys of Al, Brass, Bronze, Copper, Nickel and non-metallic materials

Size and weight limits to be defined

Limit could be revised by user.

Limits could be revised by user.

Limit could be revised by user.

Limit could be revised by user.

excluding ceramics.

by user.

units and accessories.

Fig. 8. Flowchart for SPM design and manufacture

Fig. 8. Flowchart for SPM design and manufacture

Fig. 9. Flowchart for technical feasibility analysis

#### **3.2 Machining sequence planning**

Properly determination of machining sequence is a key point in successful SPM utilization. A poor machining sequence plan leads to lower quality of production and/or increased machining times and consequently higher production costs. Often it is possible to combine and perform a number of operations in a single setup lowering machining times and costs while also improving production quality. Indeed machining sequence planning determines the overall configuration of the machine and required machining units and accessories.

Intelligent Analysis of Utilization of Special Purpose Machines for Drilling Operations 307

Generally there are two layouts for SPMs; single-station and multi-station. In the former method the workpiece is held in a fixed position where machining and sliding units are positioned around it such that they can process the part from different directions. The part is machined by a single machining unit or by multiple machining units. In the case of multiple machining units they may process the part simultaneously or in sequence depending on the geometry of the workpiece and machining features. This layout is shown in Figure 10(a). In latter method the workpiece is transferred from one station to another until it is processed in all stations. The number of machining stations varies from two to twelve. Transferring workpiece between stations is performed by rotational or linear motions. The rotational motion is provided by indexing tables and the linear motion can be performed by use of sliding units or other methods. Figures 10 (b) to 10(e) illustrate different multi-station layouts. The layout of the machine and positioning all the machining and sliding units, the number of stations in case of multi-station processing, and workpiece transferring method between stations are decided by machine designers considering technical and productive measures. In general a higher production rate is achieved in the multi-station method because of simultaneous machining of several workpieces in multiple machining stations.

(a) (b) (c)

Fig. 10. Different layouts for SPMs; (a) single-station, (b) special application, (c) transfer

Before designing the control system, the unit motion diagrams representing reciprocating motions of all units should be prepared. These diagrams explicitly represent speed and magnitude of motion of each unit, exact start/stop times, and its position at any time. As described earlier, the motion of units is often provided by hydraulic and pneumatic cylinders, or servo-motors. Start and stop signals of motion are usually issued by a programmable logic controller (PLC) that is programmed based on the unit motion

machine, (d) rotary machine, and (e) in-line drilling machine (Photos: Suhner)

(d) (e)

**3.7 Control system** 

diagrams.

**3.6 SPM layout** 

#### **3.3 Cutting conditions**

Properly selection of cutting tools and cutting conditions such as cutting speed, feed, and depth of cut is of great importance in the success of any machining operation. When SPM is in use, due to the stability requirements of the production process in order to produce high quantities of the product, appropriate cutting tools and cutting conditions should be employed. As frequent tool changes influence the productivity of the machine tool, it is suggested to employ long lasting hard material cutting tools made from tungsten carbides and ceramics for high production rates. These tool materials provide longer tool lives and higher production rates. Other important considerations in the selection of cutting tools are the shape and geometry of the tool. Cutting tools are generally divided into standard and special groups. By use of specially designed cutting tools sometimes it is possible to combine different machining operations in a single operation.

#### **3.4 Setup and clamping**

Machining jigs and fixtures are frequently used to increase the speed and quality of production and to reduce production times and required skill level of machinists. Uniformity of production due to use of jigs and fixtures has an important effect on production quality. Accordingly, properly design and application of jigs and fixtures is very important in SPM utilization. Fixtures used in SPMs are complex as normally a number of machining operations are performed in a single part setup. Fixtures should be designed such that (a) there is adequate tool access to the workpiece in all work stations; (b) the part is easily, quickly, and accurately positioned inside the fixture, and removed from it, and (c) the fixture is rigid enough to withstand large cutting forces applied by multiple cutting tools working on the part simultaneously. In locating the part in the fixture, the most difficult and accurate operation should be considered first in order to achieve the best result. Because there are different machining operations, locating surfaces need to be machined accurately before the workpiece is placed in the fixture. Appropriate measures should be taken for free flow of coolant and chip removal from the fixture.

#### **3.5 Machining and sliding units**

As described in the previous sections, machining and sliding units are the most important components of SPMs that make the cutting tool capable of rotational and linear motions necessary for cutting. Consequently, the selection of machining units, sliding units and accessories should be accomplished such that following three conditions are met.


In is important to note that selection of machining and sliding units should always be accomplished after selection of cutting tools and cutting conditions. This is due to the fact that cutting tools' geometry and cutting conditions dictate required powers, velocities, and motions of machining and sliding units.

#### **3.6 SPM layout**

306 Intelligent Systems

Properly selection of cutting tools and cutting conditions such as cutting speed, feed, and depth of cut is of great importance in the success of any machining operation. When SPM is in use, due to the stability requirements of the production process in order to produce high quantities of the product, appropriate cutting tools and cutting conditions should be employed. As frequent tool changes influence the productivity of the machine tool, it is suggested to employ long lasting hard material cutting tools made from tungsten carbides and ceramics for high production rates. These tool materials provide longer tool lives and higher production rates. Other important considerations in the selection of cutting tools are the shape and geometry of the tool. Cutting tools are generally divided into standard and special groups. By use of specially designed cutting tools sometimes it is possible to

Machining jigs and fixtures are frequently used to increase the speed and quality of production and to reduce production times and required skill level of machinists. Uniformity of production due to use of jigs and fixtures has an important effect on production quality. Accordingly, properly design and application of jigs and fixtures is very important in SPM utilization. Fixtures used in SPMs are complex as normally a number of machining operations are performed in a single part setup. Fixtures should be designed such that (a) there is adequate tool access to the workpiece in all work stations; (b) the part is easily, quickly, and accurately positioned inside the fixture, and removed from it, and (c) the fixture is rigid enough to withstand large cutting forces applied by multiple cutting tools working on the part simultaneously. In locating the part in the fixture, the most difficult and accurate operation should be considered first in order to achieve the best result. Because there are different machining operations, locating surfaces need to be machined accurately before the workpiece is placed in the fixture. Appropriate measures should be taken for free

As described in the previous sections, machining and sliding units are the most important components of SPMs that make the cutting tool capable of rotational and linear motions necessary for cutting. Consequently, the selection of machining units, sliding units and

1. Previously determined cutting tools are capable of performing all rotational and linear

In is important to note that selection of machining and sliding units should always be accomplished after selection of cutting tools and cutting conditions. This is due to the fact that cutting tools' geometry and cutting conditions dictate required powers, velocities, and

2. Proper cutting conditions such as spindle speed, feed, and depth of cut are provided.

accessories should be accomplished such that following three conditions are met.

motions necessary for performing corresponding machining operations.

combine different machining operations in a single operation.

flow of coolant and chip removal from the fixture.

3. Required machining power is provided.

motions of machining and sliding units.

**3.5 Machining and sliding units** 

**3.3 Cutting conditions** 

**3.4 Setup and clamping** 

Generally there are two layouts for SPMs; single-station and multi-station. In the former method the workpiece is held in a fixed position where machining and sliding units are positioned around it such that they can process the part from different directions. The part is machined by a single machining unit or by multiple machining units. In the case of multiple machining units they may process the part simultaneously or in sequence depending on the geometry of the workpiece and machining features. This layout is shown in Figure 10(a). In latter method the workpiece is transferred from one station to another until it is processed in all stations. The number of machining stations varies from two to twelve. Transferring workpiece between stations is performed by rotational or linear motions. The rotational motion is provided by indexing tables and the linear motion can be performed by use of sliding units or other methods. Figures 10 (b) to 10(e) illustrate different multi-station layouts. The layout of the machine and positioning all the machining and sliding units, the number of stations in case of multi-station processing, and workpiece transferring method between stations are decided by machine designers considering technical and productive measures. In general a higher production rate is achieved in the multi-station method because of simultaneous machining of several workpieces in multiple machining stations.

Fig. 10. Different layouts for SPMs; (a) single-station, (b) special application, (c) transfer machine, (d) rotary machine, and (e) in-line drilling machine (Photos: Suhner)

#### **3.7 Control system**

Before designing the control system, the unit motion diagrams representing reciprocating motions of all units should be prepared. These diagrams explicitly represent speed and magnitude of motion of each unit, exact start/stop times, and its position at any time. As described earlier, the motion of units is often provided by hydraulic and pneumatic cylinders, or servo-motors. Start and stop signals of motion are usually issued by a programmable logic controller (PLC) that is programmed based on the unit motion diagrams.

Intelligent Analysis of Utilization of Special Purpose Machines for Drilling Operations 309

The most common obstacle in utilization of SPMs in manufacturing industries is inadequate knowledge of manufacturing engineers and machining specialists with this technology, and the lack of a solid foundation for technical and economic feasibility analysis. This is not an easy task and requires engagement of qualified personnel with reasonable expertise and experience in this field. One needs to do a lot of computations and use various handbooks and assumptions in order to accomplish this task. In recent years artificial intelligence techniques have proven to be capable of restoring human's logic and expertise and efficiently applying this expertise to tackle complicated engineering problems. For example, KBESs have been used to restore human's logic and expertise and efficiently applying this expertise to tackle complicated engineering problems including product design (Myung and Han, 2001), design for assembly (Sanders *et al.,* 2009), and process planning (Patil and Pande, 2002). Accordingly, a KBES has been developed in order to capture the knowledge of SPM specialists in a computer program, and integrate it with a large amount of machining and tooling data restored in the database. This allows less experienced people to use the system developed in order to perform a detailed and accurate analysis of SPM utilization for production tasks. A rulebase has been developed that restores knowledge in the rule-base in the form of *if-then*

**4.1 Knowledge acquisition** 

rules. An example rule is presented here:

*if there are multiple holes of the same diameter and on the same plane,* 

A number of expertise rules have been developed in order to restore qualitative information in the rule-base as shown in Figure 11. One group of rules is specific to determination of workpiece setup such that there is tool access to all machining features in a single setup if possible. Another group of rules determine proper clamping method such that workpiece is securely held in place during machining. A group of rules determine the number of machining stations such that the total number of stations is kept minimal. Determination of required cutting tools and cutting conditions, and required machine power are performed by other groups of rules. Some rules are developed for selection of machining units, sliding units, chassis, and accessories such as multi-drill

As can be seen in Figure 11, the KBES developed in this work is also equipped with a database. It contains quantitative information of available cutting tools and corresponding cutting conditions extracted from handbooks, together with characteristics of standard SPM components. Machining and sliding units restored in the database include CNC units (CNCmasters), quill units (MONOmasters), small drilling units with flexible power transmission mechanism (MULTImasters), power units (POWERmasters), and tapping units (TAPmasters). Table 1 represents characteristics of eight MONOmasters restored in the database which include designation, maximum drill diameter when used for drilling low carbon steels, working stroke that determines maximum hole depth, available power and thrust, spindle speeds, and weight for each unit. Other information restored in the database

*and the minimum centre-to-centre distance is 30 mm, then a multi-drill head can be used in a combined operation, else the holes are to be machined in multiple operations.* 

*Rule 121:* 

heads, angle heads, etc.

#### **3.8 Approval**

Upon completion of preceding steps, it is necessary that all design steps be controlled and inspected by experienced SPM specialists to correct possible errors before sending the machine design to the workshop for manufacturing. These points deserve special consideration at this stage: a) control system and PLC programming, b) types and specifications of machining and sliding units, c) motion diagrams, d) hydraulic and pneumatic systems and servo-systems, e) performance of the machine, and g) possible collision of the moving parts with other moving or fixed parts. As mentioned before, sometimes it becomes necessary to repeat economic considerations of the feasibility of SPM utilization before the machine is built. This is attributable to the fact that initial economic analysis has been made based on the initial estimation. However, when detailed machine design is available a more precise machine cost becomes available that could be different.

#### **3.9 Manufacturing and testing**

Chassis and table of the machine should be made and assembled considering technical issues. These parts should be sufficiently rigid and equipped with special dampers in order to minimize vibrations resulting from the operation of cutting tools. Generally, thick steel plates and cast iron are used for machine table. Cast irons have good damping character, and therefore, are used for making the machine table to reduce vibrations. Chip removal could be a huge problem in SPMs that cannot be appreciated before the machine is made. The volume of chips produced in SPMs is high and this could reduce effective machining time by half or even less when a proper chip removal mechanism is not considered. In addition, a properly designed coolant system should be used to enhance the lives of cutting tools as frequent tool changes increase machining costs. Then, based on detailed engineering drawings, installation of stands, supports, machining units, sliding units, indexing table and coolant system are performed. Installation of hydraulic and pneumatic systems, wiring, electric power supply to electro-motors, and finally, the control systems are all performed at this stage.

Upon completion of previous steps, machine performance is measured considering required product quality and production volume. Possible issues at this stage are detected and resolved to bring the machine to a more productive state. Producing a reasonable number of quality products is necessary before actual production begins.

#### **4. Knowledge-Based Expert System (KBES)**

KBESs use rules as the knowledge representation for knowledge coded into the system. The definitions of KBES depend almost entirely on expert systems, which are system that mimic the reasoning of human expert in solving a knowledge intensive problem. Instead of representing knowledge in a declarative, static way as a set of things which are true, KBESs represent knowledge in terms of a set of rules that tells what to do or what to conclude in different situations (Grosan and Abraham, 2011). In this work a KBES has been developed to perform the analysis of SPM utilization and determination of machine layout and its basic components. Its development has been described in this Section.

#### **4.1 Knowledge acquisition**

308 Intelligent Systems

Upon completion of preceding steps, it is necessary that all design steps be controlled and inspected by experienced SPM specialists to correct possible errors before sending the machine design to the workshop for manufacturing. These points deserve special consideration at this stage: a) control system and PLC programming, b) types and specifications of machining and sliding units, c) motion diagrams, d) hydraulic and pneumatic systems and servo-systems, e) performance of the machine, and g) possible collision of the moving parts with other moving or fixed parts. As mentioned before, sometimes it becomes necessary to repeat economic considerations of the feasibility of SPM utilization before the machine is built. This is attributable to the fact that initial economic analysis has been made based on the initial estimation. However, when detailed machine design is available a more precise machine cost becomes available that could be different.

Chassis and table of the machine should be made and assembled considering technical issues. These parts should be sufficiently rigid and equipped with special dampers in order to minimize vibrations resulting from the operation of cutting tools. Generally, thick steel plates and cast iron are used for machine table. Cast irons have good damping character, and therefore, are used for making the machine table to reduce vibrations. Chip removal could be a huge problem in SPMs that cannot be appreciated before the machine is made. The volume of chips produced in SPMs is high and this could reduce effective machining time by half or even less when a proper chip removal mechanism is not considered. In addition, a properly designed coolant system should be used to enhance the lives of cutting tools as frequent tool changes increase machining costs. Then, based on detailed engineering drawings, installation of stands, supports, machining units, sliding units, indexing table and coolant system are performed. Installation of hydraulic and pneumatic systems, wiring, electric power supply to electro-motors, and finally, the control systems are all performed at

Upon completion of previous steps, machine performance is measured considering required product quality and production volume. Possible issues at this stage are detected and resolved to bring the machine to a more productive state. Producing a reasonable number of

KBESs use rules as the knowledge representation for knowledge coded into the system. The definitions of KBES depend almost entirely on expert systems, which are system that mimic the reasoning of human expert in solving a knowledge intensive problem. Instead of representing knowledge in a declarative, static way as a set of things which are true, KBESs represent knowledge in terms of a set of rules that tells what to do or what to conclude in different situations (Grosan and Abraham, 2011). In this work a KBES has been developed to perform the analysis of SPM utilization and determination of machine layout and its basic

quality products is necessary before actual production begins.

components. Its development has been described in this Section.

**4. Knowledge-Based Expert System (KBES)** 

**3.8 Approval** 

this stage.

**3.9 Manufacturing and testing** 

The most common obstacle in utilization of SPMs in manufacturing industries is inadequate knowledge of manufacturing engineers and machining specialists with this technology, and the lack of a solid foundation for technical and economic feasibility analysis. This is not an easy task and requires engagement of qualified personnel with reasonable expertise and experience in this field. One needs to do a lot of computations and use various handbooks and assumptions in order to accomplish this task. In recent years artificial intelligence techniques have proven to be capable of restoring human's logic and expertise and efficiently applying this expertise to tackle complicated engineering problems. For example, KBESs have been used to restore human's logic and expertise and efficiently applying this expertise to tackle complicated engineering problems including product design (Myung and Han, 2001), design for assembly (Sanders *et al.,* 2009), and process planning (Patil and Pande, 2002). Accordingly, a KBES has been developed in order to capture the knowledge of SPM specialists in a computer program, and integrate it with a large amount of machining and tooling data restored in the database. This allows less experienced people to use the system developed in order to perform a detailed and accurate analysis of SPM utilization for production tasks. A rulebase has been developed that restores knowledge in the rule-base in the form of *if-then* rules. An example rule is presented here:

*Rule 121: if there are multiple holes of the same diameter and on the same plane, and the minimum centre-to-centre distance is 30 mm, then a multi-drill head can be used in a combined operation, else the holes are to be machined in multiple operations.* 

A number of expertise rules have been developed in order to restore qualitative information in the rule-base as shown in Figure 11. One group of rules is specific to determination of workpiece setup such that there is tool access to all machining features in a single setup if possible. Another group of rules determine proper clamping method such that workpiece is securely held in place during machining. A group of rules determine the number of machining stations such that the total number of stations is kept minimal. Determination of required cutting tools and cutting conditions, and required machine power are performed by other groups of rules. Some rules are developed for selection of machining units, sliding units, chassis, and accessories such as multi-drill heads, angle heads, etc.

As can be seen in Figure 11, the KBES developed in this work is also equipped with a database. It contains quantitative information of available cutting tools and corresponding cutting conditions extracted from handbooks, together with characteristics of standard SPM components. Machining and sliding units restored in the database include CNC units (CNCmasters), quill units (MONOmasters), small drilling units with flexible power transmission mechanism (MULTImasters), power units (POWERmasters), and tapping units (TAPmasters). Table 1 represents characteristics of eight MONOmasters restored in the database which include designation, maximum drill diameter when used for drilling low carbon steels, working stroke that determines maximum hole depth, available power and thrust, spindle speeds, and weight for each unit. Other information restored in the database

*Designation*  *Drill Dia (mm)*  *Working Stroke (mm)* 

*Total Storke (mm)* 

BEM03 3 25 40 940 to 10,270 380 9

BEM06 6 50 80 550 to 7,730 0.37 700 16

BEM06D 6 50 80 1450 to 11,600 0.37 700 12

BEM12 12 50 80 35 to 7,730 0.75 1,470 26

BEM12D 12 50 80 90 to 2,900 0.75 1,470 20

BEM20 20 125 125 360 to 10,000 1.5 73

BEM25H 25 125 125 360 to 10,000 1.5 108

BEM28 28 200 200 400 to 2,580 2.2 8,200 150

Table 1. Database of MONOmasters restored in the database (Photos: Suhner)

Intelligent Analysis of Utilization of Special Purpose Machines for Drilling Operations 311

*Motor Power (kW)* 

*Thrust at 85 psi (N)* 

*Unit Weigh t (kg)* 

*Configuration* 

*Spindle Speed (rpm)* 

includes characteristics of assembly components for accurately positioning and orienting the units; multi-drill heads (POLYdrills) and angle heads, tool holders, and machine components or standardized chassis. It is noteworthy that the database contains full characteristics of SPM components and three-dimensional (3D) solid models of these components are restored in a feature library of a computer-aided design (CAD) system integrated with the KBES.

Fig. 11. KBES architecture


includes characteristics of assembly components for accurately positioning and orienting the units; multi-drill heads (POLYdrills) and angle heads, tool holders, and machine components or standardized chassis. It is noteworthy that the database contains full characteristics of SPM components and three-dimensional (3D) solid models of these components are restored in a feature library of a computer-aided design (CAD) system

Machining and sliding units

Poly drill heads

Rules to determine SPM layout

integrated with the KBES.

Fig. 11. KBES architecture


Table 1. Database of MONOmasters restored in the database (Photos: Suhner)

Intelligent Analysis of Utilization of Special Purpose Machines for Drilling Operations 313

The user consults with the KBES for determination of appropriate machining units and then s/he uses the CAD system for designing the required SPM. The CAD system used in this work is SolidWorks which provides user with a 3D modelling environment. It is customized for SPM design by developing a feature library containing 3D models of standardized SPM components. As shown in Figure 12 the feature library contains a number of folders, each containing a group of SPM components. When the user wishes to insert a component, s/he simply opens the corresponding folder and double clicks on the desired component. Component's model is extracted from the library and can be easily placed in the desired position and orientation within modelling environment. Figure 13 shows different 3D solid models of quill units (MONOmasters) restored in the feature library, and Figure 14 represents the major steps of processing a typical drilling operation and the way that

various components of the system are used in different activities.

BEM03 BEM06 BEM06D

BEM12 BEM012D BEM20

Fig. 13. 3D solid models of eight quill units (MONOmasters) restored in the feature library

BEM25H BEM28

The KBES developed is capable of integrating qualitative information of the rule-base with quantitative data of the database and the feature library. It uses forward chaining approach for firing the rules in the rule-base and to achieve the goal. Forward chaining starts with the data available (for example the plane of holes, size of holes, and centre-to centre distance between holes) and uses the inference rules to extract more data until a desired goal (for example the possibility of using multi-drill head) is reached. An inference engine searches the inference rules until it finds one in which the "if" clause is known to be true. It then concludes the "then" clause and adds this information to its data. It continues to do this until a "goal" is reached. The system stores input and output information of the processed workpieces in the database for future use. Therefore, it adds to the extent of its knowledge.

To determine the feasibility of utilization of SPM for a new workpiece, the inference engine first searches the database to find out whether it has been processed before. If so, it uses previously restored information. If not processed before then the inference engine searches for similar workpieces. When a similar workpiece is found then the system provides user with possibility of interactive modification if necessary. When a similar workpiece is not found then it is processed by the system.

Fig. 12. Developed feature library containing 3D solid models of standardized SPM components

The KBES developed is capable of integrating qualitative information of the rule-base with quantitative data of the database and the feature library. It uses forward chaining approach for firing the rules in the rule-base and to achieve the goal. Forward chaining starts with the data available (for example the plane of holes, size of holes, and centre-to centre distance between holes) and uses the inference rules to extract more data until a desired goal (for example the possibility of using multi-drill head) is reached. An inference engine searches the inference rules until it finds one in which the "if" clause is known to be true. It then concludes the "then" clause and adds this information to its data. It continues to do this until a "goal" is reached. The system stores input and output information of the processed workpieces in the database for future use. Therefore, it adds to the extent of its knowledge. To determine the feasibility of utilization of SPM for a new workpiece, the inference engine first searches the database to find out whether it has been processed before. If so, it uses previously restored information. If not processed before then the inference engine searches for similar workpieces. When a similar workpiece is found then the system provides user with possibility of interactive modification if necessary. When a similar workpiece is not

Fig. 12. Developed feature library containing 3D solid models of standardized SPM components

found then it is processed by the system.

The user consults with the KBES for determination of appropriate machining units and then s/he uses the CAD system for designing the required SPM. The CAD system used in this work is SolidWorks which provides user with a 3D modelling environment. It is customized for SPM design by developing a feature library containing 3D models of standardized SPM components. As shown in Figure 12 the feature library contains a number of folders, each containing a group of SPM components. When the user wishes to insert a component, s/he simply opens the corresponding folder and double clicks on the desired component. Component's model is extracted from the library and can be easily placed in the desired position and orientation within modelling environment. Figure 13 shows different 3D solid models of quill units (MONOmasters) restored in the feature library, and Figure 14 represents the major steps of processing a typical drilling operation and the way that various components of the system are used in different activities.

Fig. 13. 3D solid models of eight quill units (MONOmasters) restored in the feature library

Intelligent Analysis of Utilization of Special Purpose Machines for Drilling Operations 315

and orientation in a 3D modelling environment. All the models are placed in a similar method that leads to the completion of machine design with many components where any

Figure 16 shows a rotational part 50 mm in diameter and 75 mm in length. As shown in the Figure this part has three machining features: counterboring, drilling, and tapping. The workpiece material is low carbon steel and it has not been subjected to heat treatment processes before. The annual production quantity is 1,500,000 and production will be running for five years. Manufacturer of this part faces three options for production: traditional machines, CNC Chiron machining centre, and SPM. As the part size is small, on the CNC machining centre it is possible to use a pallet carrying 50 parts. Once the pallet is loaded the machine begins processing 50 parts in one setup. Once processing of all 50 parts is completed the pallet will be exchanged with another one that is already loaded with 50 new parts ready for processing. This would significantly reduce machine idle time for

Fig. 16. (a) The part with three machining features, (b) machining operations of the part

Table 2 compares the times required for performing machining operations on the traditional lathe, CNC Chiron machining centre, and SPM. Total time of machining on traditional lathe and CNC machine are equal to the sum of cutting times plus non-cutting times that include tool changing between processes, loading/unloading, and free movements of cutting tool. As schematically shown in Figure 17, the multi-station SPM for this part has an indexing table with four stations, one for loading/unloading and three for processing. This makes it possible to perform all machining operations simultaneously, one process at each station. Machining units are arranged such that all of the operations can be performed at a single part setup. Accordingly, the total machining time for each part is equal to the longest time needed for a single operation, plus one indexing time. As represented in Table 2, the total time per part on traditional lathe is 50 seconds, on the CNC machine 15.12 seconds, and it is only 6.8 seconds for SPM. Therefore, SPM produces 529.41 parts/hour, a figure remarkably higher than 238.10 for the CNC machine, and significantly higher than 72 for the lathe. Yet it is possible to multiply the output of the SPM when all machining stations are equipped with

possible part collisions will be detected early at design stage.

(a) (b)

from left to right: counterboring, drilling, and tapping

**5. Case study** 

loading and unloading.

multi-drill heads.

Fig. 14. Different steps in processing a drilling operation

Figure 15 illustrates a BEM12 quill unit extracted from the feature library. SolidWorks provides the user with full freedom in placing the selected models in the desired position

Fig. 15. 3D solid model of a BEM12 quill unit extracted from the feature library while it is being positioned in the modelling environment of SolidWorks

and orientation in a 3D modelling environment. All the models are placed in a similar method that leads to the completion of machine design with many components where any possible part collisions will be detected early at design stage.

#### **5. Case study**

314 Intelligent Systems

Figure 15 illustrates a BEM12 quill unit extracted from the feature library. SolidWorks provides the user with full freedom in placing the selected models in the desired position

Fig. 15. 3D solid model of a BEM12 quill unit extracted from the feature library while it is

being positioned in the modelling environment of SolidWorks

Fig. 14. Different steps in processing a drilling operation

Figure 16 shows a rotational part 50 mm in diameter and 75 mm in length. As shown in the Figure this part has three machining features: counterboring, drilling, and tapping. The workpiece material is low carbon steel and it has not been subjected to heat treatment processes before. The annual production quantity is 1,500,000 and production will be running for five years. Manufacturer of this part faces three options for production: traditional machines, CNC Chiron machining centre, and SPM. As the part size is small, on the CNC machining centre it is possible to use a pallet carrying 50 parts. Once the pallet is loaded the machine begins processing 50 parts in one setup. Once processing of all 50 parts is completed the pallet will be exchanged with another one that is already loaded with 50 new parts ready for processing. This would significantly reduce machine idle time for loading and unloading.

Fig. 16. (a) The part with three machining features, (b) machining operations of the part from left to right: counterboring, drilling, and tapping

Table 2 compares the times required for performing machining operations on the traditional lathe, CNC Chiron machining centre, and SPM. Total time of machining on traditional lathe and CNC machine are equal to the sum of cutting times plus non-cutting times that include tool changing between processes, loading/unloading, and free movements of cutting tool. As schematically shown in Figure 17, the multi-station SPM for this part has an indexing table with four stations, one for loading/unloading and three for processing. This makes it possible to perform all machining operations simultaneously, one process at each station. Machining units are arranged such that all of the operations can be performed at a single part setup. Accordingly, the total machining time for each part is equal to the longest time needed for a single operation, plus one indexing time. As represented in Table 2, the total time per part on traditional lathe is 50 seconds, on the CNC machine 15.12 seconds, and it is only 6.8 seconds for SPM. Therefore, SPM produces 529.41 parts/hour, a figure remarkably higher than 238.10 for the CNC machine, and significantly higher than 72 for the lathe. Yet it is possible to multiply the output of the SPM when all machining stations are equipped with multi-drill heads.

Intelligent Analysis of Utilization of Special Purpose Machines for Drilling Operations 317

 Parts required per year (*D*) 1,500,000 1,500,000 1,500,000 Production cycle (*t*) 5 years 5 years 5 years Interest rate (*r*) 6% 6% 6%

per year (*H*) 3,600 3,600 3,600

 Parts per hour (*p*) 72 238.10 529.41 Machine availability (*a*) 90% 95% 90% Effective parts per hour (*E*) *E* = *pa* 64.8 226.2 476.47 Working hours per year (*h*) *h* = *D*/*E* 23148.15 6,637.17 3,148.15 Machines required (*M*) *M* = *h*/*H* 6.43 => 7 1.84 => 2 0.87 => 1

 Wage rate (*w*) \$45/h \$45/h \$45/h Machinists required (*R*) 7 2 1 Wage per hour (*W*) *W* = *wR* \$315 \$90 \$45 **Wage cost per part (***Cw***)** *Cw* **=** *W***/***E* **\$4.4811 \$0.3979 \$0.0944** 

Tool cost per process (*T*) \$0.0168 \$0.0168 \$0.0168

per part (*n*) 3 3 3

Electricity cost per kWh (*k*) \$0.15 \$0.15 \$0.15

consumption (*e*) 9 kW 11 kW 36 kW Total consumption (*d*) *d* = *eR* 63 kW 22 kW 36 kW Electricity cost per h (*c*) *c* = *kd* \$9.45 \$3.30 \$5.40 **Electricity cost per part (***Ce***)** *Ce* **=** *c* **/** *E* **\$0.1456 \$0.0146 \$0.0113** 

per unit (*u*) \$35,900 \$124,800 \$264,678

cost (*U*) *U* = *M<sup>u</sup>* \$251,300 \$249,600 \$264,678

per year (*f*) *f* = *U*/*t* \$50,260 \$49,920 \$52,935.60 **Depreciation cost/part (***Cm***)** *Cm* **=** *f***/***D* **\$0.0335 \$0.0333 \$0.0353** 

 **=** *nT* **\$0.0504 \$0.0504 \$0.0504** 

**)** *Ct*

*Production data* 

*Machine tool data* 

*Wage costs* 

*Cutting tool consumption* 

Number of processes

*Electricity consumption costs* 

Machine electricity

*Machine depreciation costs*  Machine investment cost

Total machine investment

Machine depreciation cost

**Cutting tool cost per part (***Ct*

Max. working hours

*Lathe CNC SPM* 


*1*: On the SPM the longest operation time determines the time required for each operation

*2*: Tool changing time for the CNC machine is 3 times of 2 seconds each for 50 pieces (0.12 sec/part)

*3*: Free tool traveling for the CNC machine is 30 seconds for 50 pieces (0.6 sec/ part)

*4*: Loading/unloading time of one pallet carrying 50 pieces is 2 minutes (2.4 sec/part)

*5*: Loading/unloading on the SPM will be performed by an automated system and at the same time machining is in progress in other stations

Table 2. Comparison of machining times for traditional lathe, CNC, and SPM

Fig. 17. Part exchange time on traditional lathe, CNC, and SPM.

Table 3 represents machining unit cost for all of the three methods and provides all cost components. When traditional lathe is used there is a need to use seven machines in order to achieve required annual output. This significantly increases labour and overhead costs that would result in a unit cost of \$4.7423. In the case of CNC machine there is a need to use two machines in order to achieve the required output. This would reduce the unit cost to \$0.5211 that is significantly lower. Yet SPM would further decrease this figure. Due to high productivity of SPMs only one machine with a single operator is needed to achieve the required output. This decreases most cost components including labour and overhead costs. Consequently the cost per part is reduced to only \$0.2138. In other words, the use of SPM results in a significant 59% reduction of unit cost in comparison with CNC, and an amazing 95.5% cost reduction is achieved when compared to traditional lathe.


**(sec) Time (sec)** 

**Time (sec) Time** 

*Lathe* **CNC SPM** 

Counterboring time 5.0 3.0 3.0 Drilling time 8.0 4.0 4.0 Tapping time 10.0 5.0 5.0 Cutting time **23.0 12.0 5.6***<sup>1</sup>*

Indexing time per part 1.2 Loading/unloading 15.0 2.40*4* 5.0*<sup>5</sup>* Non-cutting time **27.0 3.12 1.2 Total time per part 50.0 15.12 6.8 Parts per hour 72 238.10 529.41**  *1*: On the SPM the longest operation time determines the time required for each operation

*3*: Free tool traveling for the CNC machine is 30 seconds for 50 pieces (0.6 sec/ part) *4*: Loading/unloading time of one pallet carrying 50 pieces is 2 minutes (2.4 sec/part)

Table 2. Comparison of machining times for traditional lathe, CNC, and SPM

Fig. 17. Part exchange time on traditional lathe, CNC, and SPM.

95.5% cost reduction is achieved when compared to traditional lathe.

*2*: Tool changing time for the CNC machine is 3 times of 2 seconds each for 50 pieces (0.12 sec/part)

*5*: Loading/unloading on the SPM will be performed by an automated system and at the same time

Table 3 represents machining unit cost for all of the three methods and provides all cost components. When traditional lathe is used there is a need to use seven machines in order to achieve required annual output. This significantly increases labour and overhead costs that would result in a unit cost of \$4.7423. In the case of CNC machine there is a need to use two machines in order to achieve the required output. This would reduce the unit cost to \$0.5211 that is significantly lower. Yet SPM would further decrease this figure. Due to high productivity of SPMs only one machine with a single operator is needed to achieve the required output. This decreases most cost components including labour and overhead costs. Consequently the cost per part is reduced to only \$0.2138. In other words, the use of SPM results in a significant 59% reduction of unit cost in comparison with CNC, and an amazing

Tool changing per part 6.0 0.12*<sup>2</sup>* Free tool traveling per part 6.0 0.6*<sup>3</sup>*

machining is in progress in other stations


Intelligent Analysis of Utilization of Special Purpose Machines for Drilling Operations 319

3D components of SPM including machining and sliding units and other accessories has been constructed on Solidworks software platform. This assists SPM designers in the design task, and helps standardization of SPM designs that is of great importance to

Allen J., Axinte D., Roberts P., and Anderson R. (2010) "A review of recent developments in

Boothroyd, G., and Knight, W.A. (2005) *Fundamentals of Machining and Machine Tools*, *Third* 

Groover, M.P. (2008) *Automation, production systems, and computer integrated manufacturing*,

Groover, M.P. (2010) *"Fundamentals of Modern Manufacturing,"* John Wiley and Sons Inc, 4th

Grosan, C., and Abraham, A., (2011) "Rule-based expert systems," *Intelligent Systems,* Vol.

Ko, J., Hu, S J., and Huang, T. (2005) "Reusability assessment for manufacturing systems,"

Lutters, D., Vaneker, T.H.J. and Van Houten, F.J.A.M. (2004) "What-if design: a synthesis

Myung, S., and Han, S. (2001) "Knowledge-based parametric design of mechanical products

Patil, L., and Pande, S. S, (2002) "An intelligent feature process planning system for

Sanders, D., Tan, Y. C., Rogers, I., Tewkesbury, G. E. (2009) "An expert system for automatic design-for-assembly", *Assembly Automation,* Vol. 29 Iss: 4, pp.378 – 388.

Tolouei-Rad, M. (2009) "Intelligent design of reconfigureable machines," *World Academy of* 

Tolouei-Rad, M., and Tabatabaei, S. M. (2005) "Design and manufacture of modular special

Tolouei-Rad, M. & Zolfaghari, S. (2009) "Productivity improvement using Special-Purpose

Wecka M & Staimer D (2002) Parallel Kinematic Machine Tools – Current State and Future Potentials, *CIRP Annals – Manufac. Tech.,* Vol 51, No 2, 2002, pp 671-683

method in the design process", *College International pour la Recherche en Productique* 

based on configuration design method," *Expert Systems with Applications,* Vol. 21,

prismatic parts," International Journal of Production Research, Vol. 40, No. 17, pp.

purpose machine tools", *CD-ROM Proceedings of International Conference on Achievements in Manufacturing and Materials Science*, 6-9 Dec 2005, Giliwice-

Modular machine tools," *International Journal of Manufacturing Research,* Vol 4, No 2,

*Manufacturing Technology*, Vol 50, pp. 50:843–857.

*edition,* Taylor & Francis, UK.

*CIRP Annals,* Vol 54, No 1, pp. 113-114.

*(CIRP) Annals*, Vol. 53, No. 1, pp. 113-114.

Suhner (2001) "*Automation Expert Catalogue*," Switzerland (Suhner.com)

*Science, Engineering and Technology*, Vol. 59, 2009, pp. 278-282.

Prentice Hall, NJ, USA.

edition.

17, pp. 149-185.

Issue 2, pp. 99-107.

Zakopane, Poland.

2009, pp 219-235.

4431-4447.

the design of special-purpose machine tools with a view to identification of solutions for portable in situ machining systems*," International Journal of Advanced* 

industries.

**7. References** 


Table 3. Machining costs for traditional lathe, CNC, and SPM

#### **6. Conclusions**

Production quality and low production cost are essential for the success of manufacturers in today's competitive market. SPMs are very useful for producing large quantities of high quality products at low costs. These machines can also be altered to produce similar components when necessary. High accuracy, uniform quality, and large production quantities are important characteristics of SPMs. However, the inadequate knowledge of machining specialists with this technology has resulted in its low utilization in manufacturing firms. In this article a detailed discussion of SPMs, their capabilities and accessories have been described. It also explained the development of a KBES to assist SPM users in deciding whether or not to make use of SPMs for a given production task. An analysis was made on the basis of technical and economical considerations. The case study presented clarified the method of analysis between three methods for producing a typical part. After a detailed discussion and extensive computations it has been concluded that for the given production task SPM would result in a significant 59% reduction of costs when compared to CNC, and an unbelievable 95.5% cost reduction was achieved when compared to traditional lathe. The system described in this work significantly reduces the time and effort needed for decision making on utilization of SPMs and determination of machine layout. In addition, the system developed minimizes the level of expertise required to perform the analysis and eliminates possible human errors.

The current system focuses on drilling and drilling-related operations. More work is needed to cover other machining operations including milling. Also the KBES developed currently works on a standalone basis. Work is in progress to integrate it with the 3D CAD modelling system such that the information could be directly extracted from the CAD system, eliminating the need for manual data input by user. A database of standard 3D components of SPM including machining and sliding units and other accessories has been constructed on Solidworks software platform. This assists SPM designers in the design task, and helps standardization of SPM designs that is of great importance to industries.

#### **7. References**

318 Intelligent Systems

to interest (*A*) *A* = *U* \$251,300 \$249,600 \$264,678

**(***I***)** *I* **=** *i***/***<sup>D</sup>* **\$0.0117 \$0.0116 \$0.0124** 

(trans., rent, etc.) (*v*) \$30,000 \$20,000 \$15,000 **Overhead cost per part (***O***)** *O* **=** *v***/***D* **\$0.02 \$0.0133 \$0.01** 

Production quality and low production cost are essential for the success of manufacturers in today's competitive market. SPMs are very useful for producing large quantities of high quality products at low costs. These machines can also be altered to produce similar components when necessary. High accuracy, uniform quality, and large production quantities are important characteristics of SPMs. However, the inadequate knowledge of machining specialists with this technology has resulted in its low utilization in manufacturing firms. In this article a detailed discussion of SPMs, their capabilities and accessories have been described. It also explained the development of a KBES to assist SPM users in deciding whether or not to make use of SPMs for a given production task. An analysis was made on the basis of technical and economical considerations. The case study presented clarified the method of analysis between three methods for producing a typical part. After a detailed discussion and extensive computations it has been concluded that for the given production task SPM would result in a significant 59% reduction of costs when compared to CNC, and an unbelievable 95.5% cost reduction was achieved when compared to traditional lathe. The system described in this work significantly reduces the time and effort needed for decision making on utilization of SPMs and determination of machine layout. In addition, the system developed minimizes the level of expertise required to

The current system focuses on drilling and drilling-related operations. More work is needed to cover other machining operations including milling. Also the KBES developed currently works on a standalone basis. Work is in progress to integrate it with the 3D CAD modelling system such that the information could be directly extracted from the CAD system, eliminating the need for manual data input by user. A database of standard

*Interest costs* 

*Overhead costs* 

**per part** 

of material)

**6. Conclusions** 

Annual amount subject

Annual overhead costs

**Total production cost** 

(excluding the cost

(*i*) *i* = *<sup>A</sup>*

Table 3. Machining costs for traditional lathe, CNC, and SPM

perform the analysis and eliminates possible human errors.

Interest per year

**Interest per part** 

*Lathe CNC SPM* 

*r* \$17,591 \$17,472 \$18,527.46

**\$4.7423 \$0.5211 \$0.2138** 


**Intelligent Biosystems and the Idea of the** 

Immediately after the appearance of first computers more than sixty years ago, the idea of the creation of artificial intelligence similar to that of humans has inspired activity of thousands of outstanding individuals. Interesting results have been achieved in this endeavor but the situation still seems unsatisfactory. With exception of very narrow domains such as chess playing, intelligent systems cannot compete with humans or even animals. Several factors may be the reason of this situation. In this chapter, I consider one of the most important reasons: there are serious problems in the theory of intelligent systems and, accordingly, in theoretical approaches to the building of artificial intelligence. Therefore, the consideration of fundamental principles of intelligence may be an appropriate

Although there is no clear definition for the term 'intelligence', it is intuitively understandable that intelligence is an attribute of goal-directed systems. A goal-directed system has various goals, which the system attempts to achieve through interactions with the environment based on diverse methods or means. Intelligence characterizes the efficacy of such systems in the achievement of its goals (Russell & Norvig, 2003). Humans and animals undoubtedly are goal-directed systems and observations of their activities reveal

One class that may underlie the activity of nonhuman animals contains goal-directed systems in which basic goals and means are determined jointly in the moment of the creation of a system. A system belonging to this class functions as follows: one or several basic goals are activated along with a broad diversity of means innately associated with those goals. In accordance with the requirements of the situation, one or several of such means are performed and then associations between those goals and means are changed through feedback loops using hard-wired relations between goals and the result of performance or/and those relations generate new means, which are consequences of some changes in ongoing means. It seems that various systems in neural nets, evolutionary computing, reinforcement learning, etc correspond to this class of goal-directed systems

(Haykin, 1998; Bäck, 1996; Holland, 1975; Leslie et al, 1996; Sutton & Barto, 1998).

The observation of human actions and introspection allow us to define the other class in which goals and means can be constructed arbitrarily and independently from each other. If

**1. Introduction** 

method for constructing effective systems of AI.

two obvious classes of such systems.

**Joint Synthesis of Goals and Means** 

Pavel N. Prudkov *Ecomon Ltd., Moscow,* 

*Russia* 

Yoshimi, I., (2008) *"Modular Design for Machine Tools,"* McGraw-Hill Professional Publishing, Blacklick, OH, USA. **15** 

### **Intelligent Biosystems and the Idea of the Joint Synthesis of Goals and Means**

Pavel N. Prudkov *Ecomon Ltd., Moscow, Russia* 

#### **1. Introduction**

320 Intelligent Systems

Yoshimi, I., (2008) *"Modular Design for Machine Tools,"* McGraw-Hill Professional Publishing,

Immediately after the appearance of first computers more than sixty years ago, the idea of the creation of artificial intelligence similar to that of humans has inspired activity of thousands of outstanding individuals. Interesting results have been achieved in this endeavor but the situation still seems unsatisfactory. With exception of very narrow domains such as chess playing, intelligent systems cannot compete with humans or even animals. Several factors may be the reason of this situation. In this chapter, I consider one of the most important reasons: there are serious problems in the theory of intelligent systems and, accordingly, in theoretical approaches to the building of artificial intelligence. Therefore, the consideration of fundamental principles of intelligence may be an appropriate method for constructing effective systems of AI.

Although there is no clear definition for the term 'intelligence', it is intuitively understandable that intelligence is an attribute of goal-directed systems. A goal-directed system has various goals, which the system attempts to achieve through interactions with the environment based on diverse methods or means. Intelligence characterizes the efficacy of such systems in the achievement of its goals (Russell & Norvig, 2003). Humans and animals undoubtedly are goal-directed systems and observations of their activities reveal two obvious classes of such systems.

One class that may underlie the activity of nonhuman animals contains goal-directed systems in which basic goals and means are determined jointly in the moment of the creation of a system. A system belonging to this class functions as follows: one or several basic goals are activated along with a broad diversity of means innately associated with those goals. In accordance with the requirements of the situation, one or several of such means are performed and then associations between those goals and means are changed through feedback loops using hard-wired relations between goals and the result of performance or/and those relations generate new means, which are consequences of some changes in ongoing means. It seems that various systems in neural nets, evolutionary computing, reinforcement learning, etc correspond to this class of goal-directed systems (Haykin, 1998; Bäck, 1996; Holland, 1975; Leslie et al, 1996; Sutton & Barto, 1998).

The observation of human actions and introspection allow us to define the other class in which goals and means can be constructed arbitrarily and independently from each other. If

Intelligent Biosystems and the Idea of the Joint Synthesis of Goals and Means 323

accordance with this opinion, the dual-processes models (Stanovich & West, 2000; Evans, 2003) have supposed that the mind includes two components. One component uses searching procedures and accordingly is responsible for deliberate actions. The other component, which complies with the systems of the first class, underlies routine, automatic actions. It is suggested that in routine everyday situations, in which, according to such theories, the vast majority of actions is performed, the routine component effectively selects an appropriate action. Searching and planning are involved only in unusual situations. Some unspecified mechanisms constrain searching in those rare cases when the latter is

In my opinion, however, nonroutine and routine situations are intertwined more strongly than it may be consciously acknowledged. The mind cannot be separated into the two components. For example, if an individual is hungry, she may open her home refrigerator without the clear awareness of this process. However, people usually do not open somebody's refrigerators automatically even if they are hungry. No special intention to inhibit the wish "to open somebody's refrigerator" is necessary in such situations. Obviously, there is no universal routine "not to open somebody's refrigerator", simply because it is very difficult to define unequivocally and finally what refrigerators are permitted to open. Therefore, it is necessary to suggest that ongoing goals somehow control activity when an individual attends to the refrigerator, allowing or forbidding the opening of the latter. In the same manner, ongoing goals unconsciously involve in most of routine situations even when the individual believes that some component of her activity is automatic. With the involvement of ongoing goals in most of everyday situations, the problem of combinatorial explosion becomes unresolved for the dual-processes models.

Whereas, AI research is not intended directly to imitate human intelligence but it seems obvious that a certain view on human intelligence is a very important tacit heuristic to AI researchers and strongly influences AI studies. In my opinion, the analysis of the two conventional classes of goal-directed systems demonstrates that human activity hardly can be derived from these classes and this may be a very serious factor constraining AI research. I suggest that the standard view on possible classes of goal-directed systems is incomplete and consider a more complex categorization below. I present, based on this classification, a new view on human goal-directed activity as a characteristic of a particular class of goaldirected systems. Some ideas on how this class can be represented in the brain are considered. These ideas form the basis for the simulation of simple models of goal-directed activity. Some proposals on how this novel understanding of humans as goal-directed

systems can be used to create intelligent systems are also considered in the article.

**2. Two-dimensional classification of goal-directed systems and the idea of** 

The two classes of goal-directed systems are usually considered as two poles of one axis and as a result, it seems that there are no other classes. However, a more profound view on the classes demonstrates that the situation may be more complex. Indeed, the first class contains goal-directed systems in which basic goals and means are constructed innately and together. In the systems of the second-class goals and means can be constructed arbitrarily and separately from each other. It is easy to discern that the words "innately" and "separately"

necessary.

**joint synthesis** 

a goal is constructed arbitrarily, then searching through all of possible means is the only method for selecting one or several means appropriate to achieve the goal. The efficacy of those means may increase the probability of their usage in similar situations; however, this class does not suggest unequivocal methods based on feedback loops to construct novel means. Various, largely symbolic, systems can be related to this class (Bertino, Piero & Zarria, 2001; Jackson, 1998; Newell, 1990).

Because these classes are so obvious, it is very reasonable to assume that any AI project can be attributed to one of the classes (though some projects may combine characteristics of both). Like other technical systems, AI projects are not intended to imitate their natural counterparts but rather attempt to achieve "natural functionality". The fact that the objective of Artificial Intelligence as a scientific and engineering activity is the full-scale functionality of human intelligence means AI researchers implicitly suggest that humans can be attributed to one or both of these classes. However, in my opinion this supposition is doubtful.

Undoubtedly, like other animals humans have a complex structure of innate goals associated with survival and reproduction. As a result, some scholars attribute humans to the first class of goal-directed systems. For example, behaviorism suggested an innate motivation mechanism in order to establish connections between goals and means through reward and punishment (Heckhausen, 1980). Currently, evolutionary psychology is very explicit in supposing that humans have an innate repertoire of goals and domain-specific modules (Tooby &Cosmides 1992; Tooby, Cosmides & Barrett, 2005). However, the attribution of humans to the first class system is unable to explain the diversity and rapid alterations of actions either at the level of a single individual, or at that of a whole society (Buller, 1998).

This inability hints that humans belong to the second-class systems. The main problem, which faces such systems, is a combinatorial explosion owing to the need to search through the potentially infinite number of possible means. However, people regularly make effective and flexible decisions without being overwhelmed by their decision-making processes. Some ideas to explain how the mind avoids a combinatorial explosion have been suggested (Newell, 1990), but they do not seem satisfactory (Cooper & Shallice, 1995). Moreover, although people are able to apply the strategy of deliberately searching among several conscious alternatives, some problems demonstrate that the thinking system is reluctant to use searching.

Consider, for example the following simplest chess riddle: White: Ke1, Rf2, Rh1; Black: Ka1. White to play and mate in one. When the author (P.P.) was acquainted with the problem he found that many poor chess players (and P.P. himself), who were, of course, familiar with the chess rules, could not solve the riddle or solved it after many attempts. However, any chess program immediately finds the solution: castling O-O. Indeed, since White should mate in one, in order to solve this problem it is necessary to generate each formally possible move for White in the given position, and to test whether this move is the solution. Such a searching procedure is available for the computer program but often not available for a human.

Everyday experience seems to demonstrate that people seldom use searching among possible alternatives. Instead, they prefer (often unconsciously) a routine action. In

a goal is constructed arbitrarily, then searching through all of possible means is the only method for selecting one or several means appropriate to achieve the goal. The efficacy of those means may increase the probability of their usage in similar situations; however, this class does not suggest unequivocal methods based on feedback loops to construct novel means. Various, largely symbolic, systems can be related to this class (Bertino, Piero &

Because these classes are so obvious, it is very reasonable to assume that any AI project can be attributed to one of the classes (though some projects may combine characteristics of both). Like other technical systems, AI projects are not intended to imitate their natural counterparts but rather attempt to achieve "natural functionality". The fact that the objective of Artificial Intelligence as a scientific and engineering activity is the full-scale functionality of human intelligence means AI researchers implicitly suggest that humans can be attributed to one or both of these classes. However, in my opinion this supposition is

Undoubtedly, like other animals humans have a complex structure of innate goals associated with survival and reproduction. As a result, some scholars attribute humans to the first class of goal-directed systems. For example, behaviorism suggested an innate motivation mechanism in order to establish connections between goals and means through reward and punishment (Heckhausen, 1980). Currently, evolutionary psychology is very explicit in supposing that humans have an innate repertoire of goals and domain-specific modules (Tooby &Cosmides 1992; Tooby, Cosmides & Barrett, 2005). However, the attribution of humans to the first class system is unable to explain the diversity and rapid alterations of actions either at the level of a single individual, or at that of a whole society

This inability hints that humans belong to the second-class systems. The main problem, which faces such systems, is a combinatorial explosion owing to the need to search through the potentially infinite number of possible means. However, people regularly make effective and flexible decisions without being overwhelmed by their decision-making processes. Some ideas to explain how the mind avoids a combinatorial explosion have been suggested (Newell, 1990), but they do not seem satisfactory (Cooper & Shallice, 1995). Moreover, although people are able to apply the strategy of deliberately searching among several conscious alternatives, some problems demonstrate that the thinking system is reluctant to

Consider, for example the following simplest chess riddle: White: Ke1, Rf2, Rh1; Black: Ka1. White to play and mate in one. When the author (P.P.) was acquainted with the problem he found that many poor chess players (and P.P. himself), who were, of course, familiar with the chess rules, could not solve the riddle or solved it after many attempts. However, any chess program immediately finds the solution: castling O-O. Indeed, since White should mate in one, in order to solve this problem it is necessary to generate each formally possible move for White in the given position, and to test whether this move is the solution. Such a searching procedure is available for the computer program but often not available for a

Everyday experience seems to demonstrate that people seldom use searching among possible alternatives. Instead, they prefer (often unconsciously) a routine action. In

Zarria, 2001; Jackson, 1998; Newell, 1990).

doubtful.

(Buller, 1998).

use searching.

human.

accordance with this opinion, the dual-processes models (Stanovich & West, 2000; Evans, 2003) have supposed that the mind includes two components. One component uses searching procedures and accordingly is responsible for deliberate actions. The other component, which complies with the systems of the first class, underlies routine, automatic actions. It is suggested that in routine everyday situations, in which, according to such theories, the vast majority of actions is performed, the routine component effectively selects an appropriate action. Searching and planning are involved only in unusual situations. Some unspecified mechanisms constrain searching in those rare cases when the latter is necessary.

In my opinion, however, nonroutine and routine situations are intertwined more strongly than it may be consciously acknowledged. The mind cannot be separated into the two components. For example, if an individual is hungry, she may open her home refrigerator without the clear awareness of this process. However, people usually do not open somebody's refrigerators automatically even if they are hungry. No special intention to inhibit the wish "to open somebody's refrigerator" is necessary in such situations. Obviously, there is no universal routine "not to open somebody's refrigerator", simply because it is very difficult to define unequivocally and finally what refrigerators are permitted to open. Therefore, it is necessary to suggest that ongoing goals somehow control activity when an individual attends to the refrigerator, allowing or forbidding the opening of the latter. In the same manner, ongoing goals unconsciously involve in most of routine situations even when the individual believes that some component of her activity is automatic. With the involvement of ongoing goals in most of everyday situations, the problem of combinatorial explosion becomes unresolved for the dual-processes models.

Whereas, AI research is not intended directly to imitate human intelligence but it seems obvious that a certain view on human intelligence is a very important tacit heuristic to AI researchers and strongly influences AI studies. In my opinion, the analysis of the two conventional classes of goal-directed systems demonstrates that human activity hardly can be derived from these classes and this may be a very serious factor constraining AI research.

I suggest that the standard view on possible classes of goal-directed systems is incomplete and consider a more complex categorization below. I present, based on this classification, a new view on human goal-directed activity as a characteristic of a particular class of goaldirected systems. Some ideas on how this class can be represented in the brain are considered. These ideas form the basis for the simulation of simple models of goal-directed activity. Some proposals on how this novel understanding of humans as goal-directed systems can be used to create intelligent systems are also considered in the article.

#### **2. Two-dimensional classification of goal-directed systems and the idea of joint synthesis**

The two classes of goal-directed systems are usually considered as two poles of one axis and as a result, it seems that there are no other classes. However, a more profound view on the classes demonstrates that the situation may be more complex. Indeed, the first class contains goal-directed systems in which basic goals and means are constructed innately and together. In the systems of the second-class goals and means can be constructed arbitrarily and separately from each other. It is easy to discern that the words "innately" and "separately"

Intelligent Biosystems and the Idea of the Joint Synthesis of Goals and Means 325

synthesis is not a method to create the best action (this is impossible due to combinatorial explosion) but a method to create any action (because the number of possible actions is infinite, in principle). To a certain degree, an alternative to the action constructed by the ongoing joint synthesis is not another action but rather its absence. Therefore, the idea of joint synthesis is not hurt by the fact that people are able to imagine, plan, or pursue completely arbitrary even unachievable goals. Because even when the individual thinks that there is no method to achieve the goal, nevertheless an inappropriate method is chosen because the selection of a certain aspect of reality among the infinite number of other

Second, experience teaches us that one goal can be achieved by various methods, ways (this is the principle of equifinality (Bertalanffy, 1968)) and that one method can be applied to achieve various goals. These obvious facts, which underlie one of the two conventional classes, seem inconsistent with the joint synthesis hypothesis (referred to as the JSH hereinafter). In my opinion, the idea that goals and means can be constructed separately is correct at the level of social practice but a psychological illusion at the level of psychological

In order to clear this idea, imagine that one needs to achieve the 35th floor of a skyscraper. Firstly, this can be made by means of an elevator. If no elevator can be used (e.g. there is no voltage), it is possible to go upstairs. Finally, if the staircase is destroyed then one can climb on the wall using necessary tools. It seems one invariable goal can be combined with various methods to achieve it. However, the first method is available for everyone because it requires no concentration of mental recourses. The second one can be accepted when there is a serious need to reach the goal. In addition, the last one can be used only under extreme circumstances requiring the strongest concentration of will and energy. In other words, from the position of internal processes each way requires a certain psychological arrangement with special goals and this arrangement is acknowledged by any individual as distinctive from the others. Therefore, a change in the situation results in the alteration of goals at a particular level of the hierarchy of goals. It is reasonable to assume that the interaction between goals and means in the process of the construction of a goal-directed activity is a

In my opinion, like other psychological illusions, such as, for example, the illusion of the instantaneous reaction to an external stimulus (the understanding that the reaction is not instant, occurred in 1823 only (Corsini & Auerbach, 1998)), the illusion of the separate construction of goals and means results from the fact that it is very difficult to combine the involvement in a particular activity with the simultaneous introspective monitoring of this activity. Indeed, when an individual pursues a particular everyday goal (e.g., shopping at the supermarket) she usually does not pay attention to all variations in the intermediate goals and means necessary for this multi-stage pursuit. As a result, the complex interplay of these intermediate processes is reflected by consciousness and memory only partially, while the success or failure in the achievement of the main goal is usually in the focus of consciousness. In addition, the detailed awareness of each stage in a multi-staged activity is merely impossible because this is able to destroy the activity itself. The result of these circumstances is, in my opinion, a false feeling of the separate formation and change of goals

possible aspects occurred.

mechanisms of a particular action.

characteristic of any such activity.

and means.

are not antonyms neither are the words "together" and "arbitrarily". This may mean that the two classes are only an apparent projection of a two-dimensional structure, in which one dimension can be characterized as "innate" versus "arbitrary" or "learned" and another dimension as "constructed together" versus "constructed separately". With this assumption, a representation of this structure can be given as the following table.


Table 1. Classification of goal-directed systems

This results in a more complicated structure with four classes. Prior to the consideration of this structure, it seems useful to raise an issue of whether this classification is fundamental enough. Undoubtedly, there may be many various sources of classification: for example, the diversity of goals or the number of levels in the system can be used to classify. However, obviously, the most important classification should be based on key characteristics of goaldirected systems. In my view, the table reflects such fundamental characteristics because one axis is the capability of a goal-directed system to change and adjust and the second dimension is the relationship between the main components of any goal-directed system, i.e. its goals and means.

It is easy to discern that two cells in the table correspond to the conventional classes but two new classes emerge from the other cells. One new class is goal-directed systems, in which goals and means are constructed innately and separately. Such architecture is, however, logically impossible. Indeed, if basic goals and means of a certain goal-directed system are defined at the moment of the creation of the system, then a common configuration undoubtedly underlies them and they cannot be constructed separately.

The other new class is goal-directed systems, in which goals and means can be constructed arbitrarily and jointly. If one suggests that the construction of a goal and means in such a system is a self-organizing process, which is based on an extremal principle, e.g. that the costs on the synthesis should be minimal, then particular advantages of this class can be easily revealed. Indeed, because the goal and means in a system of this class are constructed jointly, there is no need to search among a potentially infinite set of means to satisfy the given goal; this is a simple solution to the problem of combinatorial explosion. On the other hand, the possibility to synthesize goals and means arbitrarily indicates the actions of the systems belonging to this class may be very flexible and adaptive. With such characteristics of this class, my main idea is that human beings are goal-directed systems in which arbitrary goals and means are synthesized jointly.

One may propose some objections to this hypothesis. First, if a goal and means are constructed together then the means ought to be appropriate for achieving the goal. However, people often understand what goal must be achieved but they cannot suggest appropriate means to achieve the goal. However, it is necessary to note that the joint

are not antonyms neither are the words "together" and "arbitrarily". This may mean that the two classes are only an apparent projection of a two-dimensional structure, in which one dimension can be characterized as "innate" versus "arbitrary" or "learned" and another dimension as "constructed together" versus "constructed separately". With this assumption,

constructed innately and

constructed arbitrarily and

This results in a more complicated structure with four classes. Prior to the consideration of this structure, it seems useful to raise an issue of whether this classification is fundamental enough. Undoubtedly, there may be many various sources of classification: for example, the diversity of goals or the number of levels in the system can be used to classify. However, obviously, the most important classification should be based on key characteristics of goaldirected systems. In my view, the table reflects such fundamental characteristics because one axis is the capability of a goal-directed system to change and adjust and the second dimension is the relationship between the main components of any goal-directed system, i.e.

It is easy to discern that two cells in the table correspond to the conventional classes but two new classes emerge from the other cells. One new class is goal-directed systems, in which goals and means are constructed innately and separately. Such architecture is, however, logically impossible. Indeed, if basic goals and means of a certain goal-directed system are defined at the moment of the creation of the system, then a common configuration

The other new class is goal-directed systems, in which goals and means can be constructed arbitrarily and jointly. If one suggests that the construction of a goal and means in such a system is a self-organizing process, which is based on an extremal principle, e.g. that the costs on the synthesis should be minimal, then particular advantages of this class can be easily revealed. Indeed, because the goal and means in a system of this class are constructed jointly, there is no need to search among a potentially infinite set of means to satisfy the given goal; this is a simple solution to the problem of combinatorial explosion. On the other hand, the possibility to synthesize goals and means arbitrarily indicates the actions of the systems belonging to this class may be very flexible and adaptive. With such characteristics of this class, my main idea is that human beings are goal-directed systems in which arbitrary

One may propose some objections to this hypothesis. First, if a goal and means are constructed together then the means ought to be appropriate for achieving the goal. However, people often understand what goal must be achieved but they cannot suggest appropriate means to achieve the goal. However, it is necessary to note that the joint

Goals and means are constructed innately and

Goals and means are constructed arbitrarily and

separately

separately

a representation of this structure can be given as the following table.

Innately Goals and means are

Arbitrarily Goals and means are

Table 1. Classification of goal-directed systems

goals and means are synthesized jointly.

its goals and means.

Together Separately

together

together

undoubtedly underlies them and they cannot be constructed separately.

synthesis is not a method to create the best action (this is impossible due to combinatorial explosion) but a method to create any action (because the number of possible actions is infinite, in principle). To a certain degree, an alternative to the action constructed by the ongoing joint synthesis is not another action but rather its absence. Therefore, the idea of joint synthesis is not hurt by the fact that people are able to imagine, plan, or pursue completely arbitrary even unachievable goals. Because even when the individual thinks that there is no method to achieve the goal, nevertheless an inappropriate method is chosen because the selection of a certain aspect of reality among the infinite number of other possible aspects occurred.

Second, experience teaches us that one goal can be achieved by various methods, ways (this is the principle of equifinality (Bertalanffy, 1968)) and that one method can be applied to achieve various goals. These obvious facts, which underlie one of the two conventional classes, seem inconsistent with the joint synthesis hypothesis (referred to as the JSH hereinafter). In my opinion, the idea that goals and means can be constructed separately is correct at the level of social practice but a psychological illusion at the level of psychological mechanisms of a particular action.

In order to clear this idea, imagine that one needs to achieve the 35th floor of a skyscraper. Firstly, this can be made by means of an elevator. If no elevator can be used (e.g. there is no voltage), it is possible to go upstairs. Finally, if the staircase is destroyed then one can climb on the wall using necessary tools. It seems one invariable goal can be combined with various methods to achieve it. However, the first method is available for everyone because it requires no concentration of mental recourses. The second one can be accepted when there is a serious need to reach the goal. In addition, the last one can be used only under extreme circumstances requiring the strongest concentration of will and energy. In other words, from the position of internal processes each way requires a certain psychological arrangement with special goals and this arrangement is acknowledged by any individual as distinctive from the others. Therefore, a change in the situation results in the alteration of goals at a particular level of the hierarchy of goals. It is reasonable to assume that the interaction between goals and means in the process of the construction of a goal-directed activity is a characteristic of any such activity.

In my opinion, like other psychological illusions, such as, for example, the illusion of the instantaneous reaction to an external stimulus (the understanding that the reaction is not instant, occurred in 1823 only (Corsini & Auerbach, 1998)), the illusion of the separate construction of goals and means results from the fact that it is very difficult to combine the involvement in a particular activity with the simultaneous introspective monitoring of this activity. Indeed, when an individual pursues a particular everyday goal (e.g., shopping at the supermarket) she usually does not pay attention to all variations in the intermediate goals and means necessary for this multi-stage pursuit. As a result, the complex interplay of these intermediate processes is reflected by consciousness and memory only partially, while the success or failure in the achievement of the main goal is usually in the focus of consciousness. In addition, the detailed awareness of each stage in a multi-staged activity is merely impossible because this is able to destroy the activity itself. The result of these circumstances is, in my opinion, a false feeling of the separate formation and change of goals and means.

Intelligent Biosystems and the Idea of the Joint Synthesis of Goals and Means 327

It is usually suggested that a goal-directed activity pursues a clear and unequivocal goal and when the individual acknowledges that the outcome of the process meets its goal then the activity completes. However, in my opinion, the idea of a clear and unequivocal goal seems doubtful. Consider, for example, the situation with Experimenter and Subject above. Obviously, that Subject unconsciously converted the goal "to find a pen" into the goal "to find a pen in the pockets" and as a result, he is astonished by the proposal "to search a pen in New York", though this proposal is consistent with the initial request. Obviously, the supposition "to search for a pen in another room" could astonish Subject to a lesser degree. Similarly, Experimenter would be stunned, if Subject could pull a giant pen (for example, 50 centimeters in length) out of his bag though such a pen meets his request. On the other hand, a pen of a very unusual design but a standard size could wonder Experimenter less. Therefore, it can be assumed that Experimenter and Subject have some distributions of anticipations regarding the result of their goal-directed activities rather than unambiguous

I suggest that any goal-directed activity is a distribution of anticipations regarding the goal and means of the activity. The activation of some components of this distribution is determined by particular aspects of the situation and the changes in the situation results in the activation of slightly other components of the distribution. The construction and changes

A suggestion that the goal and means of a goal-directed process are some distributions leads to two fundamental conclusions. First, this means that there is no simple procedure to define when the goal is achieved because it may be difficult to find an unequivocal compliance between the distributed representation of the goal and the output of the activity. Therefore, the completion of an ongoing process is the result of the interaction between this process, the situation, and the hierarchy of other processes. In other words, there is no special comparator always able to compare the goal and the output of the activity and as a result, people sometimes do not acknowledge that the result of an ongoing activity does not respond to its initial goal. In my opinion, everyday experience is consistent with this suggestion. Consider, for example, an individual who plans to buy necessary goods at the supermarket. Sometimes the result of such activity is that an individual misses several objects planned. Instead, she purchases other goods but thinks that the goal of the action is

Second, the vague representation of the goal and method implies that the sustainability of a goal-directed activity can be considered as its relatively autonomous attribute. Indeed, sustainability seems a one-dimensional parameter and hence less variable than multivariate distributions of goals and means that ought to meet the very complex structure of the situation. A proposal of the autonomy of sustainability seems unusual enough but perseveration, i.e., the involuntary and uncontrollable repetitions of a particular action, which is a very frequent attribute of disturbances in goal-directed behavior (Luria, 1966, 1972,1983; Joseph, 1999), clearly favors this proposal. Indeed, perseveration can be considered as the activation of a sustainable component, which, if the goal-directed system is damaged, persists regardless the influence of the situation or

goals, but they acknowledge those anticipations only partially.

achieved.

other processes.

in the distribution are based on the criterion of minimal construction costs.

It is necessary to note that the hypothesis that the mind constructs the goal and means together does not imply that an individual deliberately cannot search through possible options as a method to determine an appropriate means. Indeed, the conscious idea to apply searching along with the awareness of several possible options may be the result of the ongoing synthesis.

The validation of the JSH is easy. Indeed, because the hypothesis suggests that the mind constructs the goal and means of an action jointly following the criterion of minimal construction costs. This means that if there are no explicit preferences to choose among several possible actions then an action requiring minimal mental costs to be constructed is preferable. This action should be selected without intensive searching among probable alternatives. On the other hand, this choice should not be a result of the activation of a routine procedure and can be changed deliberately. A real experimentation to test these suppositions is possible but beyond the scope of this article (Prudkov &Rodina, 1999, Rodina &Prudkov, 2005). Instead, I consider a thought experiment, which, in my opinion, is sufficient to demonstrate the relevance of the JSH.

Imagine that two individuals participate in this experiment, one of them is Experimenter, the other is Subject, accordingly, and the experiment takes place in London. The participants are discussing some problem and at a certain moment, Experimenter asks Subject to give him a pen without specifying the location of the pen. Many people have a pen in their pockets, and it is very probable that Subject is among them. Subject takes the pen out of the pocket and gives it to Experimenter. It is very reasonable to suggest that the construction of this action needs minimal mental costs. In response, however, Experimenter asks, "why did Subject take the pen out of the pocket instead of calling New York?" Subject is astonished by this question and then Experimenter says that there are many pens in New York and Subject could find a pen there. The astonishment of Subject means that his mind did not find among possible alternatives of the pen's location but one may argue that this reflects the fact that Experimenter's request is performed by the activation of a corresponding routine. It is obvious, however, that if Experimenter would merely ask Subject to find a pen in New York then Subject could easily convert this idea to a sequence of actions. Such a rapid adjustment to the situation cannot be provided by a routine. This is the result of a special goal-directed process. In my opinion, this simple situation, which can be easily repeated in reality, demonstrates the appropriateness of the idea of joint synthesis.

Although the joint synthesis is a basic attribute of humans as goal-directed systems, the consideration of this characteristic alone may be insufficient to understand the whole diversity of human actions. Humans, of course, have innate mechanisms necessary for survival and reproduction and those, although are under control from more modern systems, influence actions and therefore, to a certain degree, humans can be considered as the goal-directed systems of the first class. On the other hand, using language and complex social skills, an individual can "emulate" the separation between goals and means. Indeed, by discussing some ideas with other people or by writing the ideas down and afterwards thinking about them, an individual can concentrate either on the goals or on the means of a goal-directed activity. The fact implies, to some extent, humans can be considered as systems with the separate and arbitrary construction of goals and means. However, it is the joint synthesis that determines the involvement of the other classes of goal-directed systems in human actions.

It is necessary to note that the hypothesis that the mind constructs the goal and means together does not imply that an individual deliberately cannot search through possible options as a method to determine an appropriate means. Indeed, the conscious idea to apply searching along with the awareness of several possible options may be the result of the

The validation of the JSH is easy. Indeed, because the hypothesis suggests that the mind constructs the goal and means of an action jointly following the criterion of minimal construction costs. This means that if there are no explicit preferences to choose among several possible actions then an action requiring minimal mental costs to be constructed is preferable. This action should be selected without intensive searching among probable alternatives. On the other hand, this choice should not be a result of the activation of a routine procedure and can be changed deliberately. A real experimentation to test these suppositions is possible but beyond the scope of this article (Prudkov &Rodina, 1999, Rodina &Prudkov, 2005). Instead, I consider a thought experiment, which, in my opinion, is

Imagine that two individuals participate in this experiment, one of them is Experimenter, the other is Subject, accordingly, and the experiment takes place in London. The participants are discussing some problem and at a certain moment, Experimenter asks Subject to give him a pen without specifying the location of the pen. Many people have a pen in their pockets, and it is very probable that Subject is among them. Subject takes the pen out of the pocket and gives it to Experimenter. It is very reasonable to suggest that the construction of this action needs minimal mental costs. In response, however, Experimenter asks, "why did Subject take the pen out of the pocket instead of calling New York?" Subject is astonished by this question and then Experimenter says that there are many pens in New York and Subject could find a pen there. The astonishment of Subject means that his mind did not find among possible alternatives of the pen's location but one may argue that this reflects the fact that Experimenter's request is performed by the activation of a corresponding routine. It is obvious, however, that if Experimenter would merely ask Subject to find a pen in New York then Subject could easily convert this idea to a sequence of actions. Such a rapid adjustment to the situation cannot be provided by a routine. This is the result of a special goal-directed process. In my opinion, this simple situation, which can be easily repeated in reality,

Although the joint synthesis is a basic attribute of humans as goal-directed systems, the consideration of this characteristic alone may be insufficient to understand the whole diversity of human actions. Humans, of course, have innate mechanisms necessary for survival and reproduction and those, although are under control from more modern systems, influence actions and therefore, to a certain degree, humans can be considered as the goal-directed systems of the first class. On the other hand, using language and complex social skills, an individual can "emulate" the separation between goals and means. Indeed, by discussing some ideas with other people or by writing the ideas down and afterwards thinking about them, an individual can concentrate either on the goals or on the means of a goal-directed activity. The fact implies, to some extent, humans can be considered as systems with the separate and arbitrary construction of goals and means. However, it is the joint synthesis that determines the involvement of the other classes of goal-directed systems

ongoing synthesis.

in human actions.

sufficient to demonstrate the relevance of the JSH.

demonstrates the appropriateness of the idea of joint synthesis.

It is usually suggested that a goal-directed activity pursues a clear and unequivocal goal and when the individual acknowledges that the outcome of the process meets its goal then the activity completes. However, in my opinion, the idea of a clear and unequivocal goal seems doubtful. Consider, for example, the situation with Experimenter and Subject above. Obviously, that Subject unconsciously converted the goal "to find a pen" into the goal "to find a pen in the pockets" and as a result, he is astonished by the proposal "to search a pen in New York", though this proposal is consistent with the initial request. Obviously, the supposition "to search for a pen in another room" could astonish Subject to a lesser degree. Similarly, Experimenter would be stunned, if Subject could pull a giant pen (for example, 50 centimeters in length) out of his bag though such a pen meets his request. On the other hand, a pen of a very unusual design but a standard size could wonder Experimenter less. Therefore, it can be assumed that Experimenter and Subject have some distributions of anticipations regarding the result of their goal-directed activities rather than unambiguous goals, but they acknowledge those anticipations only partially.

I suggest that any goal-directed activity is a distribution of anticipations regarding the goal and means of the activity. The activation of some components of this distribution is determined by particular aspects of the situation and the changes in the situation results in the activation of slightly other components of the distribution. The construction and changes in the distribution are based on the criterion of minimal construction costs.

A suggestion that the goal and means of a goal-directed process are some distributions leads to two fundamental conclusions. First, this means that there is no simple procedure to define when the goal is achieved because it may be difficult to find an unequivocal compliance between the distributed representation of the goal and the output of the activity. Therefore, the completion of an ongoing process is the result of the interaction between this process, the situation, and the hierarchy of other processes. In other words, there is no special comparator always able to compare the goal and the output of the activity and as a result, people sometimes do not acknowledge that the result of an ongoing activity does not respond to its initial goal. In my opinion, everyday experience is consistent with this suggestion. Consider, for example, an individual who plans to buy necessary goods at the supermarket. Sometimes the result of such activity is that an individual misses several objects planned. Instead, she purchases other goods but thinks that the goal of the action is achieved.

Second, the vague representation of the goal and method implies that the sustainability of a goal-directed activity can be considered as its relatively autonomous attribute. Indeed, sustainability seems a one-dimensional parameter and hence less variable than multivariate distributions of goals and means that ought to meet the very complex structure of the situation. A proposal of the autonomy of sustainability seems unusual enough but perseveration, i.e., the involuntary and uncontrollable repetitions of a particular action, which is a very frequent attribute of disturbances in goal-directed behavior (Luria, 1966, 1972,1983; Joseph, 1999), clearly favors this proposal. Indeed, perseveration can be considered as the activation of a sustainable component, which, if the goal-directed system is damaged, persists regardless the influence of the situation or other processes.

Intelligent Biosystems and the Idea of the Joint Synthesis of Goals and Means 329

and at the cognitive level this means some similar functions. After this, flexible elements with new functions start interacting with each other also exchanging its characteristics. It is reasonable to suggest that the more similar characteristics shared by some elements, the more probability of its interactions. For example, if neuron A has a synapse with neuron B then a probability that the discharge of neuron A results in the discharge of neuron B seems more than the same probability for two neurons that do not share a common synapse. The relationship between the similarity of elements and the probability of its interaction is the

It is reasonable to expect that owing to interactions between elements, the resemblance of elements can be increased. As a result, a pattern joining many elements with similar characteristics gradually emerges and this pattern becomes sustainable. This indicates that the construction of a new process is completed. Although, elements in the pattern have something in common but there are some distinctions among them and this is a prerequisite

It is suggested each pattern can be considered as a construction with two interconnected components: one component is responsible for the goal and the other for the means. Such a separation is based on the idea that some neurons in the pattern have mainly local connections within the prefrontal cortex (they comprise the goal component). Other neurons in the pattern are linked to other brain structures (those are the means component). Because the activity of neurons within the PFC is likely more reverberatory and self-sustained than that of neurons linked to other structures, the goal component can be more stable and

Once a goal-directed process is constructed, some activation from the means component propagates to other brain structures, which are able to carry out the process, and its performance is initiated (B.T. Miller& D'Esposito 2005). Simultaneously, the components interact with each other; this stabilizes the means component while it receives feedback because of performing the process. Therefore, the fact that the process pursues the goal is a result of the stability in the goal component produced by self-sustainable characteristics of the PFC. It is possible to say that goal-directed processes are self-sustained gates, which amplify appropriate information and diminish inappropriate one. The components are constructed together but their architecture is slightly different. The functioning of components gradually increases these differences and this change may be a basis for an

As is emphasized above I do not suggest that the brain includes a special comparator, which monitors when the outcome of the process meets the goal, and then turns the process off. Simply, with the achievement of the goal, the current situation undergoes changes, thus not being able to support the ongoing process with appropriate information. To meet novel requirements of the situation, the construction of another process begins. Probably, more stable processes at a higher level of the goal-directed hierarchy supervising short-term ones also participate in the completion of the ongoing process. Free neural ensembles again become flexible components of the blackboard. It can be hypothesized that a real goal-directed process is a hierarchical multilevel structure

substantiation of the criterion of minimal construction costs.

for the distributed representation for the goal and means.

autonomous representation of goals and means in consciousness.

persistent than the means component.

joining many of such ideal processes.

#### **3. Neural basis for the joint synthesis**

If the goal and means of a goal-directed activity are constructed together then it is of great importance to understand how this can be implemented in the brain because similar mechanisms can be used to create artificial goal-directed systems. Undoubtedly, human goal-directed activity is very complex and a detailed understanding of it is the beyond scope of this article. Instead, I consider the neural basis of a certain "ideal" goal-directed process suggesting it includes three obvious stages, i.e. initiation, execution, and termination. My approach meets most of the contemporary hypotheses, which consider that the prefrontal cortex (PFC) plays a key role in goal-directed processes (E.K. Miller& Cohen, 2001; Wood& Grafman, 2003). In accordance with this position, I propose that the prefrontal cortex is heavily involved in the construction and maintenance of neural patterns representing goals and means.

It is suggested that the capacity of the PFC to construct and maintain sustainable neural patterns is based on possible reverberatory characteristics of neurons in this structure (Fuster 1997). It can be supposed that owing to such reverberatory properties the emergence of sustainable characteristics of a neural pattern is, to a certain extent, autonomous from the emergence of its other characteristics. In other words, relatively weak changes in neurons of the PFC may be sufficient to make a pattern sustainable but more serious alterations are necessary to form its other characteristics. This underlines a relative autonomy of the sustainability of goal-directed processes at the cognitive level.

It is suggested that the prefrontal cortex can be considered as blackboard architecture. Blackboard architecture consists of a set of specialized or stable processors that interact with each other using a blackboard, consisting of less stable, flexible elements. Some authors (van der Velde& de Kamps 2003, 2006) have suggested the idea that the prefrontal cortex uses this sort of architecture. This idea is consistent with the neural data. For example, this means that most of prefrontal neurons must flexibly adapt its activity to the ongoing task. And 30- 80 percents of prefrontal neurons of the monkey show selective responses to some aspect of that task's events (Asaad et al 2000). However, it is necessary to emphasize a distinction between conventional views on blackboard architecture used in AI (Corkill 1991; Craig 1995) and that used in this text. Unlike conventional models, the given model does not suggest an absolute difference between stable processors and flexible elements, i.e. stable processors can be converted to flexible elements and vice versa because both groups comprise of similar neurons and only the level of stability distinguishes them.

It is reasonable to assume that a new goal-directed process emerges from the integration of various sources of information associated with the ongoing situation. So, it is hypothesized that prior to the construction of a new goal-directed process the prefrontal cortex can be considered as a blackboard system in which incoming sensory information and/or ongoing internal processes (emotions, innate drives, other goal-directed processes, especially those at higher levels, etc.) presented as spatiotemporal patterns of neural activity in the PFC and other brain structures are stable processors. Moreover, other ensembles of the PFC comprise a bulletin board with flexible elements. The construction of a new process started from interactions between stable processors and flexible elements and owing to such interactions, the characteristics of flexible elements become similar to some characteristics of stable processors. At the neural level, this means similar frequency or distribution of firing, etc.

If the goal and means of a goal-directed activity are constructed together then it is of great importance to understand how this can be implemented in the brain because similar mechanisms can be used to create artificial goal-directed systems. Undoubtedly, human goal-directed activity is very complex and a detailed understanding of it is the beyond scope of this article. Instead, I consider the neural basis of a certain "ideal" goal-directed process suggesting it includes three obvious stages, i.e. initiation, execution, and termination. My approach meets most of the contemporary hypotheses, which consider that the prefrontal cortex (PFC) plays a key role in goal-directed processes (E.K. Miller& Cohen, 2001; Wood& Grafman, 2003). In accordance with this position, I propose that the prefrontal cortex is heavily involved in the construction and maintenance of neural patterns representing goals

It is suggested that the capacity of the PFC to construct and maintain sustainable neural patterns is based on possible reverberatory characteristics of neurons in this structure (Fuster 1997). It can be supposed that owing to such reverberatory properties the emergence of sustainable characteristics of a neural pattern is, to a certain extent, autonomous from the emergence of its other characteristics. In other words, relatively weak changes in neurons of the PFC may be sufficient to make a pattern sustainable but more serious alterations are necessary to form its other characteristics. This underlines a relative autonomy of the

It is suggested that the prefrontal cortex can be considered as blackboard architecture. Blackboard architecture consists of a set of specialized or stable processors that interact with each other using a blackboard, consisting of less stable, flexible elements. Some authors (van der Velde& de Kamps 2003, 2006) have suggested the idea that the prefrontal cortex uses this sort of architecture. This idea is consistent with the neural data. For example, this means that most of prefrontal neurons must flexibly adapt its activity to the ongoing task. And 30- 80 percents of prefrontal neurons of the monkey show selective responses to some aspect of that task's events (Asaad et al 2000). However, it is necessary to emphasize a distinction between conventional views on blackboard architecture used in AI (Corkill 1991; Craig 1995) and that used in this text. Unlike conventional models, the given model does not suggest an absolute difference between stable processors and flexible elements, i.e. stable processors can be converted to flexible elements and vice versa because both groups

It is reasonable to assume that a new goal-directed process emerges from the integration of various sources of information associated with the ongoing situation. So, it is hypothesized that prior to the construction of a new goal-directed process the prefrontal cortex can be considered as a blackboard system in which incoming sensory information and/or ongoing internal processes (emotions, innate drives, other goal-directed processes, especially those at higher levels, etc.) presented as spatiotemporal patterns of neural activity in the PFC and other brain structures are stable processors. Moreover, other ensembles of the PFC comprise a bulletin board with flexible elements. The construction of a new process started from interactions between stable processors and flexible elements and owing to such interactions, the characteristics of flexible elements become similar to some characteristics of stable processors. At the neural level, this means similar frequency or distribution of firing, etc.

comprise of similar neurons and only the level of stability distinguishes them.

sustainability of goal-directed processes at the cognitive level.

**3. Neural basis for the joint synthesis** 

and means.

and at the cognitive level this means some similar functions. After this, flexible elements with new functions start interacting with each other also exchanging its characteristics. It is reasonable to suggest that the more similar characteristics shared by some elements, the more probability of its interactions. For example, if neuron A has a synapse with neuron B then a probability that the discharge of neuron A results in the discharge of neuron B seems more than the same probability for two neurons that do not share a common synapse. The relationship between the similarity of elements and the probability of its interaction is the substantiation of the criterion of minimal construction costs.

It is reasonable to expect that owing to interactions between elements, the resemblance of elements can be increased. As a result, a pattern joining many elements with similar characteristics gradually emerges and this pattern becomes sustainable. This indicates that the construction of a new process is completed. Although, elements in the pattern have something in common but there are some distinctions among them and this is a prerequisite for the distributed representation for the goal and means.

It is suggested each pattern can be considered as a construction with two interconnected components: one component is responsible for the goal and the other for the means. Such a separation is based on the idea that some neurons in the pattern have mainly local connections within the prefrontal cortex (they comprise the goal component). Other neurons in the pattern are linked to other brain structures (those are the means component). Because the activity of neurons within the PFC is likely more reverberatory and self-sustained than that of neurons linked to other structures, the goal component can be more stable and persistent than the means component.

Once a goal-directed process is constructed, some activation from the means component propagates to other brain structures, which are able to carry out the process, and its performance is initiated (B.T. Miller& D'Esposito 2005). Simultaneously, the components interact with each other; this stabilizes the means component while it receives feedback because of performing the process. Therefore, the fact that the process pursues the goal is a result of the stability in the goal component produced by self-sustainable characteristics of the PFC. It is possible to say that goal-directed processes are self-sustained gates, which amplify appropriate information and diminish inappropriate one. The components are constructed together but their architecture is slightly different. The functioning of components gradually increases these differences and this change may be a basis for an autonomous representation of goals and means in consciousness.

As is emphasized above I do not suggest that the brain includes a special comparator, which monitors when the outcome of the process meets the goal, and then turns the process off. Simply, with the achievement of the goal, the current situation undergoes changes, thus not being able to support the ongoing process with appropriate information. To meet novel requirements of the situation, the construction of another process begins. Probably, more stable processes at a higher level of the goal-directed hierarchy supervising short-term ones also participate in the completion of the ongoing process. Free neural ensembles again become flexible components of the blackboard. It can be hypothesized that a real goal-directed process is a hierarchical multilevel structure joining many of such ideal processes.

Intelligent Biosystems and the Idea of the Joint Synthesis of Goals and Means 331

from 0 to 1. With the idea of a relative autonomy of sustainability above, this parameter reflects the stability and activity of the module, i.e., as AL increases the functioning of the

A fundamental characteristic of modules is that they interact with each other. The functional proximity between two elements is calculated as follows. First, the following characteristic

)

(3)

*<sup>z</sup>* (6)

*<sup>n</sup>* (1)

*<sup>n</sup>* (2)

(4)

(5)

*k k im im*

, ,1

*IV CV*

(, ) | <sup>|</sup> \* \*

The interaction between module i and module j at iteration m occurs if fp(i,m) is less than a threshold (p1) plus a small noise. The fact that modules interacts only if its functional proximity is less than a threshold is an implementation of the idea of minimal construction

,1 ,1 , , 1

,1 ,1 , , 1

It is suggested that modules interact in parallel and the formulae reflect this. Owing to interactions, the activation level of each module (for example, module i at iteration m) is also

> , , 1 , , 1 \* 4 \* (1 ) 3 \* *i m i m*

where both p3 and p4 <1 and ti,m is the number of interactions between module i and the

It is easy to see that as the AL of a module increases, the components of the module become less prone to change. In addition, if a module did not interact with other modules at the last

*CV CV IV AL p CV CV*

*CV CV IV AL p CV CV*

*k k im jm jm i m*

*k k jm im im j m*

*t p AL AL p AL*

*i m i m*

*fp i j AL me AL me*

costs. The result of the interaction between module i and module jis as follows:

*k k im im im*

 <sup>2</sup> , ,1 ,

*IV CV me*

, , ,1 . ,1 ,

*sd sd*

*i m j m*

*im im j m j m*

( ( ) \* (1 )) \* 2 *k kk*

*z*

*z*

( ( ) \* (1 )) \* 2 *k kk*

*k n*

*k*

*i m*

*k n*

*k*

*me*

*i m*

and then the functional proximity between i and j, fp(i,j) is

, ,1

, ,1

*jm jm*

other modules of the system at iteration m.

changed :

*im im*

*m*

*sd*

<sup>1</sup> ,

<sup>1</sup> ,

module becomes more stable.

afterward another parameter is calculated

for module i (and j, accordingly) at iteration m is computed

#### **4. Simulation of a goal-directed activity based on joint synthesis**

The hypothesis of joint synthesis and its possible neural implementation can be considered a basis for computer models of goal-directed activity. It is necessary to point out that these models are neither models of a certain aspect of human or animal activity nor implementations of goal-directed activity in the brain. They are simply intended to demonstrate how a goal-directed process can be constructed. The models share a common basis but have certain distinctive characteristics.

#### **4.1 A simple model of goal-directed activity (model 1)**

The architecture of the model is presented in Figure 1.

Fig. 1. Architecture of model 1

The model consists of two fractions; one is the system in which a goal-directed activity should be constructed and the other is the environment influencing the state of the system. At the beginning, a new goal-directed process emerges within the system under a certain state of the environment and after changing the environment, the process pursues its goal associated with the initial state of the environment using the means constructed.

The system includes one layer consisting of z autonomous modules and the output of the system is a summation of the outputs of its modules. Each module contains several ndimension vectors with real numbers as its components. These vectors are an input vector (IV), which is filled by information from the environment (its kth component is IVk, accordingly), a vector of coefficients (CV), and an output vector (OV). The functioning of the vectors is described below. Also, each module has an activation level (AL), a real number from 0 to 1. With the idea of a relative autonomy of sustainability above, this parameter reflects the stability and activity of the module, i.e., as AL increases the functioning of the module becomes more stable.

A fundamental characteristic of modules is that they interact with each other. The functional proximity between two elements is calculated as follows. First, the following characteristic for module i (and j, accordingly) at iteration m is computed

$$mc\_{i,m} = \frac{\sum\_{k=1}^{k=n} |IV\_{i,m}|^k - CV\_{i,m-1}^k}{n} \tag{1}$$

afterward another parameter is calculated

330 Intelligent Systems

The hypothesis of joint synthesis and its possible neural implementation can be considered a basis for computer models of goal-directed activity. It is necessary to point out that these models are neither models of a certain aspect of human or animal activity nor implementations of goal-directed activity in the brain. They are simply intended to demonstrate how a goal-directed process can be constructed. The models share a common

**Environment** 

**SYSTEM** 

M O D U L E i

M O D U L E j

M O D U L E z

M O D U L E 3

The model consists of two fractions; one is the system in which a goal-directed activity should be constructed and the other is the environment influencing the state of the system. At the beginning, a new goal-directed process emerges within the system under a certain state of the environment and after changing the environment, the process pursues its goal

The system includes one layer consisting of z autonomous modules and the output of the system is a summation of the outputs of its modules. Each module contains several ndimension vectors with real numbers as its components. These vectors are an input vector (IV), which is filled by information from the environment (its kth component is IVk, accordingly), a vector of coefficients (CV), and an output vector (OV). The functioning of the vectors is described below. Also, each module has an activation level (AL), a real number

associated with the initial state of the environment using the means constructed.

**4. Simulation of a goal-directed activity based on joint synthesis** 

basis but have certain distinctive characteristics.

M O D U L E 2

Fig. 1. Architecture of model 1

M O D U L E 1

**4.1 A simple model of goal-directed activity (model 1)**  The architecture of the model is presented in Figure 1.

$$\text{ssd}\_{i,m} = \sqrt{\frac{\sum\_{k=1}^{k=n} \left( \left| IV\_{i,m} \right. \right. \left. - \left. \text{CV}\_{i,m-1}^{k} \right| \left. - m e\_{i,m} \right) \right.}{n}} \tag{2}$$

and then the functional proximity between i and j, fp(i,j) is

$$|fp(i,j)\_m = |\frac{sd\_{i,m}}{AL\_{i,m-1} \ast me\_{i,m}} - \frac{sd\_{j,m}}{AL\_{j,m-1} \ast me\_{j,m}}|\tag{3}$$

The interaction between module i and module j at iteration m occurs if fp(i,m) is less than a threshold (p1) plus a small noise. The fact that modules interacts only if its functional proximity is less than a threshold is an implementation of the idea of minimal construction costs. The result of the interaction between module i and module jis as follows:

$$\text{CC}\_{i,m}^{k} = \text{CV}\_{i,m-1}^{k} + \frac{(\text{CV}\_{i,m-1}^{k} - (\text{CV}\_{j,m-1}^{k} - \text{IV}\_{j,m}^{k})^{\*} (1 - AL\_{i,m-1}))^{\*} p2}{z} \tag{4}$$

$$\text{CC}\_{j,m}^{k} = \text{CV}\_{j,m-1}^{k} + \frac{(\text{CV}\_{j,m-1}^{k} - (\text{CV}\_{i,m-1}^{k} - \text{IV}\_{i,m}^{k})^{\*} (1 - AL\_{j,m-1}))^{\*} p2}{z} \tag{5}$$

It is suggested that modules interact in parallel and the formulae reflect this. Owing to interactions, the activation level of each module (for example, module i at iteration m) is also changed :

$$AL\_{i,m} = p\mathfrak{Z}^\* \operatorname{AL}\_{i,m-1} + \frac{t\_{i,m} \, ^\*p\mathfrak{A}^\* \left(1 - AL\_{i,m-1}\right)}{z} \tag{6}$$

where both p3 and p4 <1 and ti,m is the number of interactions between module i and the other modules of the system at iteration m.

It is easy to see that as the AL of a module increases, the components of the module become less prone to change. In addition, if a module did not interact with other modules at the last

Intelligent Biosystems and the Idea of the Joint Synthesis of Goals and Means 333

Afterwards, the constants were 80 again. Because at any moment all constants were identical, the components of input and coefficients vectors in modules could be averaged within each vector and across all modules. As a result, one number was sufficient to describe the state of coefficients vectors at any iteration. In addition, Al averaged across all

It was suggested that owing to stable processors filled by 80 the coefficient vectors of the model with stable processors (CV-sp) should exceed those of the model without stable processors (CV-wsp) at stage 2. Moreover, CV-sp should be more stable after a sudden fluctuation in the constants of the environment. Also, AL-sp should be more than AL-wsp. The results of both simulations are in figure 2, where for convenience, ALs were multiplied

AL-wsp 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16

Though model 1 is able to demonstrate some characteristics of goal-directed activity, it seems too primitive for serious actions. Model 2 is more complex, its system has a perceptive spot, which includes the modules whose input vectors are filled by a useful signal from the environment while the input vectors of the other modules are filled by noise. Both a useful signal and noise are real numbers but the amplitude of noise is considerably less. The system is able to move the center of the spot but cannot change its size. The

Fluctuation

Iterations

 CV-sp Al-sp CV-wsp

modules also was used as a characteristic of the process.

Stage 1 Stage 2

Fig. 2. The comparison of the results of two simulations.

**4.2 A model with perceptive spot (model 2)** 

by 100.

0

10

20

30

40

50

60

70

80

90

100

iteration, its AL should be decreased. Module i is able to influence the environment only if its AL exceeds a threshold (p5) at iteration m, then

$$COV\_{i,m}^k = \frac{\{IV\_{i,m}^k - CV\_{i,m}^k\}}{z}, \text{ otherwise } \ OV\_{i,m}^k = 0 \tag{7}$$

The environment is also a n-dimension vector (E) and its kth component at iteration m is changed by the following formula:

$$E\_m^k = \cos \tan t\_m^k + noise + \frac{\sum\_{l=1}^{t=z} OV\_{i,m-1}^k}{z} \tag{8}$$

It is not difficult to see that, unlike the analysis of neural mechanisms above, this model does not include special layers to form output. The objective of such a design is to avoid unnecessary difficulties conditioned by complex relations between such layers. These difficulties are able to complicate the understanding of the model's functioning without clearing its main ideas. However, because each module has a complex structure with internal vectors such as CV and OV the model can be useful to understand the functioning of various goal-directed systems.

In all simulations, the number of modules in the system (z) was 300 and the vectors in each module were three-dimensional. Real numbers were used as the stuff of all vectors in the system. A goal-directed activity was constructed as follows. First, 40 modules were considered as stable processors. Its coefficient vectors were filled by a constant plus small noise and its active levels was more than p5 (0.3). The other modules of the system were flexible elements. Its coefficient vectors were randomly filled by numbers from 0 to 100 and its ALs were randomly established at 0,06 plus small noise. Following this initialization, interactions between stable processors and flexible elements started. Five iterations of this process took place and p1 was 0,3. At this stage (stage 1), no outputs from the system influenced the environment. This corresponded to the construction of a goal-directed activity. After this, novel constants were established and the interaction between the system and the environment became possible. This stage (stage 2) meant the functioning of a goaldirected activity.

It is necessary to emphasize that the architecture of the model means stable processors are not a necessary condition for the interaction between the system and the environment. In principle, flexible elements are sufficient to provide the functioning of the system but in this case, the activity of the system must be less stable and persistent. To test this suggestion a special simulation was carried out. In this simulation no stable processors were formed but five iterations similar to those in simulation 1 were performed (stage 1, accordingly). After this, certain constants were selected and the interaction between the system and the environment became possible (stage 2).

In both simulations all constants were 80, in other words, the goal of the activity in simulation 1 entirely met the initial state of its stable processors. After three iterations with such constants at stage 2, in both simulations the constants were forcefully established at 50 during one iteration to estimate the stability of the models to random fluctuations.

iteration, its AL should be decreased. Module i is able to influence the environment only if

The environment is also a n-dimension vector (E) and its kth component at iteration m is

<sup>1</sup> tan

It is not difficult to see that, unlike the analysis of neural mechanisms above, this model does not include special layers to form output. The objective of such a design is to avoid unnecessary difficulties conditioned by complex relations between such layers. These difficulties are able to complicate the understanding of the model's functioning without clearing its main ideas. However, because each module has a complex structure with internal vectors such as CV and OV the model can be useful to understand the functioning

In all simulations, the number of modules in the system (z) was 300 and the vectors in each module were three-dimensional. Real numbers were used as the stuff of all vectors in the system. A goal-directed activity was constructed as follows. First, 40 modules were considered as stable processors. Its coefficient vectors were filled by a constant plus small noise and its active levels was more than p5 (0.3). The other modules of the system were flexible elements. Its coefficient vectors were randomly filled by numbers from 0 to 100 and its ALs were randomly established at 0,06 plus small noise. Following this initialization, interactions between stable processors and flexible elements started. Five iterations of this process took place and p1 was 0,3. At this stage (stage 1), no outputs from the system influenced the environment. This corresponded to the construction of a goal-directed activity. After this, novel constants were established and the interaction between the system and the environment became possible. This stage (stage 2) meant the functioning of a goal-

It is necessary to emphasize that the architecture of the model means stable processors are not a necessary condition for the interaction between the system and the environment. In principle, flexible elements are sufficient to provide the functioning of the system but in this case, the activity of the system must be less stable and persistent. To test this suggestion a special simulation was carried out. In this simulation no stable processors were formed but five iterations similar to those in simulation 1 were performed (stage 1, accordingly). After this, certain constants were selected and the interaction between the system and the

In both simulations all constants were 80, in other words, the goal of the activity in simulation 1 entirely met the initial state of its stable processors. After three iterations with such constants at stage 2, in both simulations the constants were forcefully established at 50 during one iteration to estimate the stability of the models to random fluctuations.

*k k <sup>i</sup> m m*

*E cons t noise*

*i z*

, 1

*z*

*OV*

*k i m*

otherwise , <sup>0</sup> *<sup>k</sup> OVi m* (7)

(8)

its AL exceeds a threshold (p5) at iteration m, then

*OV*

changed by the following formula:

of various goal-directed systems.

environment became possible (stage 2).

directed activity.

*i m*

 , , , ( ) , *k k k im im*

*IV CV*

*z*

Afterwards, the constants were 80 again. Because at any moment all constants were identical, the components of input and coefficients vectors in modules could be averaged within each vector and across all modules. As a result, one number was sufficient to describe the state of coefficients vectors at any iteration. In addition, Al averaged across all modules also was used as a characteristic of the process.

It was suggested that owing to stable processors filled by 80 the coefficient vectors of the model with stable processors (CV-sp) should exceed those of the model without stable processors (CV-wsp) at stage 2. Moreover, CV-sp should be more stable after a sudden fluctuation in the constants of the environment. Also, AL-sp should be more than AL-wsp. The results of both simulations are in figure 2, where for convenience, ALs were multiplied by 100.

Fig. 2. The comparison of the results of two simulations.

#### **4.2 A model with perceptive spot (model 2)**

Though model 1 is able to demonstrate some characteristics of goal-directed activity, it seems too primitive for serious actions. Model 2 is more complex, its system has a perceptive spot, which includes the modules whose input vectors are filled by a useful signal from the environment while the input vectors of the other modules are filled by noise. Both a useful signal and noise are real numbers but the amplitude of noise is considerably less. The system is able to move the center of the spot but cannot change its size. The

Intelligent Biosystems and the Idea of the Joint Synthesis of Goals and Means 335

,1 ,1 ,1 ,1

,1 ,1 ,1 ,1

Owing to interactions, the characteristics of the vector of differences (DV) of module i at

it is possible to say that CV is the long-term memory of a module and DV is its short-term

The modules with AL exceeding a threshold ( p8 in this model or p5 in the previous one) , also, influence the position of the center of perceptive spot. This position (center position or

*CP CP p T CP*

 1 ,1 1

and *T iz i m*, / if , 8 *AL p i m* , otherwise *T CP im m* , 1 . (14)

It is suggested the position of the left boundary of the system is 0 and that of the right

As is emphasized above, only the modules, which are within perceptive spot, are filled by information from the environment, i.e. if |CPm –i/z| ≤ p11, then for module i at iteration

, 1 *k k IV E noise im m*

, 1 *<sup>k</sup> IV noise i m* .

The formulae for computing activation level (AL), output vector OV, and the environment

An idea underlying the usage of model is that under certain circumstances the input vectors of modules i.e. the state of the environment and the vectors of coefficients in the system ought to converge to each other. The results of a simulation intended to test this assumption are presented in table 2. In this simulation as well as in the simulations below, the number of modules in the system (z) was 300 and the vectors in each module were three-dimensional. The constants of environment vector in this simulation were 10, 50, and 90, accordingly. Perceptive spot covered all modules, and each module was able to interact with all of the rest i.e. p2 and p11 were 0,95. The threshold for interactions (p3) was 4,2. The other parameters are in Appendix, they were kept invariable through the other simulations. The values averaged across the components of input vectors were used as the description of the

and for modules which do not meet this inequality (15)

*i z m m im m i*

10 \* ( )

*CV CV DV AL p CV CV*

*CV CV DV AL p CV CV*

*k k im jm jm i m*

*k k jm im im j m*

, ,1

, ,1

*jm jm*

iteration m are also changed as follows :

CP) is determined at iteration m as follows:

memory.

boundary is 1.

are identical those in model 1.

m+1

*im im*

( ( ) \* (1 )) \* 4 *kkk*

*z*

*z*

, , , , 1 5\*( ) (1 5) , *k kk <sup>k</sup> DV p CV IV p DV i m im im i m* p5<1 (13)

( ( ) \* (1 )) \* 4 *k kk*

(11)

(12)

behavior of the system in model 1 is similar to the activity of an inseсt, which moves in the environment filled by a nutrient with variable concentration. Model 2 is, to some extent, similar to the action of an eye of an animal.

Fig. 3. Architecture of model 2

The modules in this model are similar to those in model 1 but also include a vector of differences (DV), its functioning is described below. In order to avoid the spurious activity of modules filled by noise, a threshold of perception is used, i.e. module i is able to participate in the activity of the system at iteration m only if

$$\frac{AL\_{i,m-1} \* \sum\_{k=1}^{k=n} IV\_{i,m}^k}{n} \ge p\mathbf{1} \tag{9}$$

The interaction between module *i* and module *j* at iteration *m* occurs if the distance between the modules, i.e. |(i-j)/z| is less than a certain parameter (p2) and the functional proximity i.e.

$$\frac{\sum\_{k=1}^{k=n} |\left| IV\_{i,m}{}^k - \mathbb{C}V\_{i,m-1}^k \right| - \left| IV\_{j,m}^k - \mathbb{C}V\_{j,m-1}^k \right|}{n} \tag{10}$$

is less than a threshold (p3) plus a small noise. The result of the interaction between module *i* and module *j* is as follows:

$$\text{CC}\_{i,m}^{k} = \text{CV}\_{i,m-1}^{k} + \frac{(\text{CV}\_{i,m-1}^{k} - (\text{CV}\_{j,m-1}^{k} - \text{DV}\_{j,m-1}^{k})^{\*}(1 - AL\_{i,m-1}))^{\*}p4}{z} \tag{11}$$

$$\text{CC}\_{j,m}^{k} = \text{CV}\_{j,m-1}^{k} + \frac{(\text{CV}\_{j,m-1}^{k} - (\text{CV}\_{i,m-1}^{k} - \text{DV}\_{i,m-1}^{k})^{\*}(1 - AL\_{j,m-1}))^{\*} \, p4}{z} \tag{12}$$

Owing to interactions, the characteristics of the vector of differences (DV) of module i at iteration m are also changed as follows :

$$DV\_{i,m}^{k} = p\mathbf{5}^{\*} \left(\mathbf{C}V\_{i,m}^{k} - IV\_{i,m}^{k}\right) + (1 - p\mathbf{5})^{\*}DV\_{i,m-1'}^{k} \text{ p5} \mathbf{<1} \tag{13}$$

it is possible to say that CV is the long-term memory of a module and DV is its short-term memory.

The modules with AL exceeding a threshold ( p8 in this model or p5 in the previous one) , also, influence the position of the center of perceptive spot. This position (center position or CP) is determined at iteration m as follows:

$$\text{CP}\_m = \text{CP}\_{m-1} + p\mathbf{1} \bullet \sum\_{i=1}^{i=z} (T\_{i,m} - \text{CP}\_{m-1})$$
 
$$\text{and} \qquad T\_{i,m} = \text{i } / z \text{ if } \text{AL}\_{i,m} \rangle p\mathbf{8} \text{, otherwise } T\_{i,m} = \text{CP}\_{m-1} \text{.} \tag{14}$$

334 Intelligent Systems

behavior of the system in model 1 is similar to the activity of an inseсt, which moves in the environment filled by a nutrient with variable concentration. Model 2 is, to some extent,

**Environment** 

The modules in this model are similar to those in model 1 but also include a vector of differences (DV), its functioning is described below. In order to avoid the spurious activity of modules filled by noise, a threshold of perception is used, i.e. module i is able to

> ,1 , 1 \*

> > *n*

*i m i m k AL IV*

*k n*

The interaction between module *i* and module *j* at iteration *m* occurs if the distance between the modules, i.e. |(i-j)/z| is less than a certain parameter (p2) and the functional proximity

, ,1 , ,1

is less than a threshold (p3) plus a small noise. The result of the interaction between module

*IV CV IV CV*

*kk k k im im jm jm*

*k*

M O D U L E i

M O D U L E j

M O D U L E z

1

(9)

*n* (10)

*p*

similar to the action of an eye of an animal.

SYSTEM

M O D U L E 3

Perceptive spot

M O D U L E 2

Fig. 3. Architecture of model 2

M O D U L E 1

*i* and module *j* is as follows:

i.e.

participate in the activity of the system at iteration m only if

*k n*

*k*

1

It is suggested the position of the left boundary of the system is 0 and that of the right boundary is 1.

As is emphasized above, only the modules, which are within perceptive spot, are filled by information from the environment, i.e. if |CPm –i/z| ≤ p11, then for module i at iteration m+1

$$IV\_{i,m+1}^k = E\_m^k + noise$$
 
$$\text{and for modules which do not meet this inequality} \tag{15}$$
 
$$IV\_{i,m+1}^k = noise \ .$$

The formulae for computing activation level (AL), output vector OV, and the environment are identical those in model 1.

An idea underlying the usage of model is that under certain circumstances the input vectors of modules i.e. the state of the environment and the vectors of coefficients in the system ought to converge to each other. The results of a simulation intended to test this assumption are presented in table 2. In this simulation as well as in the simulations below, the number of modules in the system (z) was 300 and the vectors in each module were three-dimensional. The constants of environment vector in this simulation were 10, 50, and 90, accordingly. Perceptive spot covered all modules, and each module was able to interact with all of the rest i.e. p2 and p11 were 0,95. The threshold for interactions (p3) was 4,2. The other parameters are in Appendix, they were kept invariable through the other simulations. The values averaged across the components of input vectors were used as the description of the

Intelligent Biosystems and the Idea of the Joint Synthesis of Goals and Means 337

First, consider the simulation of a goal-directed process with a simple goal. At stage 1, the constants of the environment were 80, 50, and 20. The center of perceptive spot was established at 0,85, the size of perceptive spot (p11) was 0,2. In addition, at this stage, p2 was 0,2 and p3 was 4,2. This stage lasted until the mean AL exceeded 0,2. At stage 2 all constants were 50 and the center of the spot was at 0,5 while p2 and p11 were 0,95 and p3 became 3 at this and last stages. At the start of the last stage all constants were 30, and the center was

It was suggested that at stage 3 the process was to move perceptive spot to the right where there were stable processors, thus increasing CP. The state of the vectors of coefficients in the modules within the spot should meet the relationship between the components of coefficient vectors of stable processors caused by the different constants of the environment at stage 1. The components of input and coefficients vectors averaged across the modules within perceptive spot were used to describe the state of the process along with AL

The table shows that at stage 3, the process was increasing CP and the relationship between the components of the vectors of coefficients gradually became similar to that between constants at stage 1. The opposite relationship between the components of input vectors results from formula 7, after inserting a constant as an input vector in it and taking the relationship between the components of CVs into account. Because the constants of the environment were equal at stage 3, the coefficient vectors of the system were influenced by these constants and, as a result, the relationship formed at stage 1 tended to disappear. This corresponds to the completion of the process owing to the influence of the situation. It is important to note that the action of the system cannot be explained by combination of the perseveratory activity of trained modules and the inactivity of untrained ones. The fact that at stage 3 the relationship between the components of the vectors of coefficients was already weakly present at 0,48, considerably beyond the area of modules changed at stage 1 means that a process including most modules indeed was formed at stage 2, while increasing the

 In another simulation, a process with a complex goal, including two constituents, was formed. In this simulation, stage 1 was divided in two phases. At the first phase, all constants were 20, the center of perceptive spot was at 0, 85, p3 was 4,2 while p2 and p11 were 0,2. After eight iterations this phase was completed, all constants became 80 and the center of perceptive spot was moved to 0,65 without changing p2, p3, and p11. This was the second phase of stage 1 and four iterations were performed. Stage 2 in this simulation was the same as in the previous one. At the beginning of the last stage all constants were 10, and

It was suggested that the process was to move the center of spot to the right and because there could be two groups of stable processors. The components of the vectors of coefficients within the spot could firstly increase and later decrease but the components of input vectors might change in the opposite direction following formula 7. To some extent, this can be

Because at any moment all constants were identical, the components of input and coefficients vectors in modules could be averaged within each vector and across all modules in the spot. As a result, one number was sufficient to describe the state of input or

established at 0,25 while p2 and p11 were 0,2 again.

averaged across all modules. The results are in table 2

mean AL at stage 3 implies activity in modules untrained at stage 1.

the center was established at 0, 25 while p2 and p11 were 0, 2 again.

considered as a very primitive form of multilevel activity.


influence of the environment and the averaged components of the vectors of coefficients were considered as the characteristic of change in coefficients. The AL averaged across all modules reflected activity in the whole system.

IV1, IV2, IV3 are the first, second, and third averaged components of input vectors; CV1, CV2, CV3 are the same components of the vectors of coefficients.

Table 2. Simulation of the convergence between input vectors and vectors of coefficients.

It is easy to see that input vectors and the vectors of coefficients indeed converged, though this process was incomplete probably because, as the mean AL approached to 1, changes in the system became practically impossible.

A goal-directed process was constructed as follows. First, one region of modules (or several regions consecutively) was considered as perceptive spot and the state of modules within this region was changed under a certain state of the environment. The position of the spot could not be changed during this stage. This corresponds to the formation of stable processors. And the modules beyond the spot were considered as flexible elements. At the second stage, perceptive spot covered all modules of the system, which obtained information from a neutral state of the environment. This stage, which corresponds to the interaction between stable processors and flexible elements and the formation of a goaldirected activity, is suggested to be rapid without interacting with the situation. Therefore, at the second stage there was no feedback loop between the system and the environment. A certain distribution of goals and means encoded by coefficient vectors and activation levels resulted from this stage. At the last stage, a new, local perceptive spot was established and the goal-directed process pursued its goal through interactions with the environment. The position of spot was able to change within the third stage.

influence of the environment and the averaged components of the vectors of coefficients were considered as the characteristic of change in coefficients. The AL averaged across all

Iteration AL IV1 CV1 IV2 CV2 IV3 CV3 1 0,3 15,03 -3,5 55,04 -3,78 94,87 -3,59 2 0,44 23,96 1,43 80,62 11,16 137,56 21,05 3 0,52 31,49 4,64 104,35 20,66 177,83 36,8 4 0,63 35,97 8,55 119,68 32,02 204,08 55,69 5 0,68 36,76 10,23 124,31 37,03 212,5 64,1 6 0,73 36,18 11,76 124,02 41,71 212,28 71,98 7 0,78 34,34 12,75 120,1 44,83 205,74 77,26 8 0,81 32,18 13,44 114,51 47,06 196,08 81,01 9 0,84 30,03 13,89 108,31 48,58 186,22 83,57 10 0,86 27,87 14,19 102,51 49,68 176,43 85,41 11 0,89 25,8 14,36 96,86 50,36 167,03 86,55 12 0,9 24,24 14,36 91,99 50,58 158,47 86,93 13 0,92 23,05 14,31 87,67 50,6 151,7 86,97 14 0,94 22,23 14,27 84,33 50,55 146,17 86,91 15 0,94 21,12 14,25 81,75 50,55 141,95 86,91 16 0,95 20,48 14,25 79,46 50,57 138,81 86,94 17 0,95 19,87 14,25 78,13 50,59 135,9 86,97 18 0,95 19,75 14,26 76,47 50,61 133,62 87,01 IV1, IV2, IV3 are the first, second, and third averaged components of input vectors; CV1, CV2, CV3 are the

Table 2. Simulation of the convergence between input vectors and vectors of coefficients.

It is easy to see that input vectors and the vectors of coefficients indeed converged, though this process was incomplete probably because, as the mean AL approached to 1, changes in

A goal-directed process was constructed as follows. First, one region of modules (or several regions consecutively) was considered as perceptive spot and the state of modules within this region was changed under a certain state of the environment. The position of the spot could not be changed during this stage. This corresponds to the formation of stable processors. And the modules beyond the spot were considered as flexible elements. At the second stage, perceptive spot covered all modules of the system, which obtained information from a neutral state of the environment. This stage, which corresponds to the interaction between stable processors and flexible elements and the formation of a goaldirected activity, is suggested to be rapid without interacting with the situation. Therefore, at the second stage there was no feedback loop between the system and the environment. A certain distribution of goals and means encoded by coefficient vectors and activation levels resulted from this stage. At the last stage, a new, local perceptive spot was established and the goal-directed process pursued its goal through interactions with the environment. The

modules reflected activity in the whole system.

same components of the vectors of coefficients.

the system became practically impossible.

position of spot was able to change within the third stage.

First, consider the simulation of a goal-directed process with a simple goal. At stage 1, the constants of the environment were 80, 50, and 20. The center of perceptive spot was established at 0,85, the size of perceptive spot (p11) was 0,2. In addition, at this stage, p2 was 0,2 and p3 was 4,2. This stage lasted until the mean AL exceeded 0,2. At stage 2 all constants were 50 and the center of the spot was at 0,5 while p2 and p11 were 0,95 and p3 became 3 at this and last stages. At the start of the last stage all constants were 30, and the center was established at 0,25 while p2 and p11 were 0,2 again.

It was suggested that at stage 3 the process was to move perceptive spot to the right where there were stable processors, thus increasing CP. The state of the vectors of coefficients in the modules within the spot should meet the relationship between the components of coefficient vectors of stable processors caused by the different constants of the environment at stage 1. The components of input and coefficients vectors averaged across the modules within perceptive spot were used to describe the state of the process along with AL averaged across all modules. The results are in table 2

The table shows that at stage 3, the process was increasing CP and the relationship between the components of the vectors of coefficients gradually became similar to that between constants at stage 1. The opposite relationship between the components of input vectors results from formula 7, after inserting a constant as an input vector in it and taking the relationship between the components of CVs into account. Because the constants of the environment were equal at stage 3, the coefficient vectors of the system were influenced by these constants and, as a result, the relationship formed at stage 1 tended to disappear. This corresponds to the completion of the process owing to the influence of the situation. It is important to note that the action of the system cannot be explained by combination of the perseveratory activity of trained modules and the inactivity of untrained ones. The fact that at stage 3 the relationship between the components of the vectors of coefficients was already weakly present at 0,48, considerably beyond the area of modules changed at stage 1 means that a process including most modules indeed was formed at stage 2, while increasing the mean AL at stage 3 implies activity in modules untrained at stage 1.

 In another simulation, a process with a complex goal, including two constituents, was formed. In this simulation, stage 1 was divided in two phases. At the first phase, all constants were 20, the center of perceptive spot was at 0, 85, p3 was 4,2 while p2 and p11 were 0,2. After eight iterations this phase was completed, all constants became 80 and the center of perceptive spot was moved to 0,65 without changing p2, p3, and p11. This was the second phase of stage 1 and four iterations were performed. Stage 2 in this simulation was the same as in the previous one. At the beginning of the last stage all constants were 10, and the center was established at 0, 25 while p2 and p11 were 0, 2 again.

It was suggested that the process was to move the center of spot to the right and because there could be two groups of stable processors. The components of the vectors of coefficients within the spot could firstly increase and later decrease but the components of input vectors might change in the opposite direction following formula 7. To some extent, this can be considered as a very primitive form of multilevel activity.

Because at any moment all constants were identical, the components of input and coefficients vectors in modules could be averaged within each vector and across all modules in the spot. As a result, one number was sufficient to describe the state of input or

Intelligent Biosystems and the Idea of the Joint Synthesis of Goals and Means 339

criterion of functioning was used to construct and perform the processes and that the goals of the processes treated as a source of sustainability and its means were constructed

Stage 2 Stage 3

1 3 5 7 9 11 13 15 17 19 21 23 25 27 29 31 CV

Of course, it is easy to see certain shortcomings of the models. For example, the second process in model 2 was able to shift the center of the spot from one constituent to the other only because they were nearby. If the constituents would be established at opposite positions, such a shift would be impossible without changing the criterion of proximity. However, as is pointed out above the models simply are intended to demonstrate how the

With the relevance of the idea of joint synthesis to simulate goal-directed activity as shown by the models above, it seems useful to consider some theoretical perspectives of this approach to the construction of intelligent systems. First, it is of importance to note that the consideration of the joint synthesis as the basic class of human activity and intelligence does not mean the abandonment of other classes and approaches. Indeed, successful attempts to achieve natural functionality often are not based on the imitation of a natural architecture – for example, cars can be considered as analogous to horses but cars have no legs, etc.

Iterations

Fig. 4. The simulation of a goal-directed process with a two-constituent goal

idea of joint synthesis can be applied to simulate goal-directed activity.

 CP Al Constant IV

together.

Stage 1 Phase 1 Phase 2


**5. Future prospects** 

0

20

40

60

80

100

120


coefficients vectors within the spot at any iteration. In addition, AL averaged across all modules also was used as a characteristic of the process. The results of the simulation are in figure 2, where for convenience, CP and AL were multiplied by 100.

IV1, IV2, IV3 are the first, second, and third averaged components of input vectors; CV1, CV2, CV3 are the same components of the vectors of coefficients.

Table 3. The simulation of a goal-directed process with a simple goal

In my opinion, all of these simulations demonstrate that the processes constructed can be considered as goal-directed in the sense that there was a state (or states), which each process attempted to achieve using certain means. It is important to note that no innate

coefficients vectors within the spot at any iteration. In addition, AL averaged across all modules also was used as a characteristic of the process. The results of the simulation are in

Iteration CP AL Constant1 Constant2 Constant3 IV1 IV2 IV3 CV1 CV2 CV3

1 0,85 0,11 80 50 20 85,19 54,83 24,83 3,64 3,64 3,92 2 0,85 0,13 80 50 20 84,91 55,2 25,02 11,95 8,27 4,56 3 0,85 0,14 80 50 20 84,37 55,07 24,86 22,51 14,76 6,63 4 0,85 0,16 80 50 20 88,36 57,55 25,88 29,22 19,01 8,32 5 0,85 0,17 80 50 20 91,84 59,46 27,14 33,77 22,06 9,67 6 0,85 0,18 80 50 20 94,44 61,1 28,59 37,65 24,72 10,98 7 0,85 0,18 80 50 20 95,65 62,54 28,69 40,13 26,32 11,8 8 0,85 0,19 80 50 20 95,82 61,48 28,09 42,11 27,62 12,48 9 0,85 0,19 80 50 20 95,86 61,5 27,57 43,29 28,36 12,92 10 0,85 0,21 80 50 20 95,51 61,68 28,58 44,13 28,86 13,19

11 0,5 0,27 50 50 50 55,13 55,24 55,43 15,9 10,55 5,24 12 0,5 0,31 50 50 50 55,02 55,03 55,36 18,53 13,46 8,5 13 0,5 0,35 50 50 50 54,98 55,26 55,23 22,69 18,16 13,69 14 0,5 0,39 50 50 50 55,16 55,16 55,37 25,41 21,35 17,43

15 0,25 0,4 30 30 30 35,15 35,25 35,34 19,74 19,79 19,89 16 0,33 0,4 30 30 30 32,36 33,76 35,65 19,42 19,55 19,73 17 0,39 0,41 30 30 30 32,49 33,77 35,07 18,99 19,19 19,23 18 0,43 0,41 30 30 30 33,1 34,59 35,79 17,86 17,96 18,18 19 0,48 0,42 30 30 30 33,38 34,52 35,28 17,75 17,44 17,12 20 0,51 0,42 30 30 30 34,03 34,19 36,29 17,85 17,1 16,35 21 0,55 0,42 30 30 30 34,42 35,37 36,28 18,04 16,88 15,55 22 0,57 0,43 30 30 30 35,72 36,51 38,02 17,82 16,49 14,83 23 0,6 0,43 30 30 30 36,07 37,1 38,73 18,05 16,49 14,58 24 0,61 0,43 30 30 30 35,94 37,92 38,61 18,34 16,68 14,82 25 0,62 0,43 30 30 30 37,52 38,04 39,53 18,4 16,77 14,98 26 0,63 0,44 30 30 30 37,32 38,14 39,41 18,66 16,96 15,26 27 0,64 0,44 30 30 30 38,42 39,43 40,34 18,65 17,11 15,68 28 0,64 0,44 30 30 30 38,17 39,83 40,62 18,8 17,32 16,08 29 0,65 0,44 30 30 30 38,59 39,65 40,52 19 17,66 16,55 30 0,65 0,45 30 30 30 39,41 40,04 40,87 19,28 17,99 16,95 IV1, IV2, IV3 are the first, second, and third averaged components of input vectors; CV1, CV2, CV3 are the

figure 2, where for convenience, CP and AL were multiplied by 100.

Stage 1

Stage 2

Stage 3

same components of the vectors of coefficients.

Table 3. The simulation of a goal-directed process with a simple goal

In my opinion, all of these simulations demonstrate that the processes constructed can be considered as goal-directed in the sense that there was a state (or states), which each process attempted to achieve using certain means. It is important to note that no innate criterion of functioning was used to construct and perform the processes and that the goals of the processes treated as a source of sustainability and its means were constructed together.

Fig. 4. The simulation of a goal-directed process with a two-constituent goal

Of course, it is easy to see certain shortcomings of the models. For example, the second process in model 2 was able to shift the center of the spot from one constituent to the other only because they were nearby. If the constituents would be established at opposite positions, such a shift would be impossible without changing the criterion of proximity. However, as is pointed out above the models simply are intended to demonstrate how the idea of joint synthesis can be applied to simulate goal-directed activity.

#### **5. Future prospects**

With the relevance of the idea of joint synthesis to simulate goal-directed activity as shown by the models above, it seems useful to consider some theoretical perspectives of this approach to the construction of intelligent systems. First, it is of importance to note that the consideration of the joint synthesis as the basic class of human activity and intelligence does not mean the abandonment of other classes and approaches. Indeed, successful attempts to achieve natural functionality often are not based on the imitation of a natural architecture – for example, cars can be considered as analogous to horses but cars have no legs, etc.

Intelligent Biosystems and the Idea of the Joint Synthesis of Goals and Means 341

people sometimes are unable to solve simple problems, though their knowledge and skills are sufficient to find the right solution. To avoid similar difficulties an AI system may use the goals and means constructed jointly as a seed point in some cases and afterwards

Since the advent of first computers, the idea of construction of artificial intelligence similar to that of human beings has driven the work of thousands of brilliant scientists and engineers. However, the result of their activity seems unsatisfactory as compared to, for example, advances in computer hardware. One of the most fundamental reasons for this situation may be that the mechanisms of human intelligence are unclear. Though the imitation of human intelligence is not a necessary characteristic of artificial intelligence, obviously a particular view on human intelligence is a very important heuristic. Therefore, an incorrect understanding of human intelligence can be a serious obstacle to construct intelligent systems. Intelligence is a characteristic of goal-directed systems and two classes of such systems can be easily derived from observations of animals and human beings. In my opinion, the classes that underlie most approaches to the construction of artificial intelligence are not sufficient to explain human activity. A broader classification of goal-directed activities suggests such processes can be described as a two-dimensional structure rather than a one-dimension one. In such structure there is a cell where in my opinion, humans can be located i.e., humans are goal-directed systems that synthesize arbitrary goals and means together. Though the idea of joint synthesis seems contradictory to some aspects of everyday experience, it is consistent with psychological evidence. In addition, there is neural evidence favoring this supposition. Simple computer models demonstrate that the idea of joint synthesis can be applied to simulate goal-directed activity. I suggest that the idea of joint synthesis can be a useful method

Asaad, W. F., Rainer, G. & Miller, E. K. (2000). Task-specific neural activity in the primate

Bäck, T. (1996). *Evolutionary Algorithms in Theory and Practice: Evolution Strategies,* 

Bertino, E, Piero, G & Zarria, B.C. (2001*). Intelligent Database Systems*. Addison-Wesley

Buller, D. J. (1998) DeFreuding evolutionary psychology: adaptation and human motivation,

in Hardcastle, V. G., Ed. *Psychology Meets Biology: Connections, Conjectures, Constraints.* MIT Press/Bradford Books. http://cogprints.soton.ac.uk/documents

*Evolutionary Programming, Genetic Algorithms.* Oxford Univ. Press.

searches for goals and methods that are more suitable.

to advance research in the construction of intelligent systems.

p1=1; p4=2; p5=0, 3; p6=0,98; p7=0,5; p8=0,3; p9=0,8; p10=0,8

/disk0/00/00/03/26/cog00000326-00/defreud.htm

Bertalanffy, L. von, (1968). *General Systems Theory*.

prefrontal cortex. *Journal of Neurophysiology*, 84, 451–459.

**6. Conclusion** 

**7. Appendix** 

**8. References** 

p2=3\*10-6 ; p3=0,99; p4=0,001

Professional.

Model 1

Model 2

Similarly, it is not necessary that the joint synthesis is the only way to create full-scale artificial intelligence.

Though the joint synthesis seems appropriate to construct local projects like the models above, I will concentrate on the construction of general artificial intelligence. In my position, the construction of general AI should be a consequence of gradual changes within a system associated with its temporal functioning and complication, i.e. a result of a shift from short – term processes with simple goals and means to long-term processes with multilevel goals and complex methods. It is suggested that, although such a system may have complex innate architecture, the main source of its development is interactions with the system's environment via feedback loops.

The mechanism of joint synthesis is suggested to be a basis for such development. Indeed, because no innate criterion for the construction of goals is used, there are no constraints for the complication of goals. An appropriate means can be constructed for any goal because goals and means are constructed together,. As a result, the system is, in principle, able to adapt to any condition of training. It is suggested that the system can use blackboard architecture with flexible elements, then the formation and/or changes in elements can be considered as learning. The key component of the models above is the functional proximity that determines the possibility of the interaction between modules. It can be supposed that, in the hypothetical system, functional proximity can be dynamically changed in regard with the complexity and diversity of the system's components. The increase in the duration of goal-directed process may result from alterations in something similar to AL in the models above.

It can be suggested that the system should be able to create and use something similar to symbols. The generation of symbols may be performed as follows: because means in the system are constructed along with goals the system should be able to describe input and/or output information caused by these means in the terms consistent with goals and to prescribe a label associated with a goal to a sequence of such descriptions, thus creating symbols. The advantage of this method is that such symbols are grounded in the ongoing activity and therefore they can be used to construct novel goals and means through the involvement of symbols in the system's blackboard. Of course, this mechanism of symbolization can be gradual like changes in the duration of goal-directed processes. That is, at the beginning, the system will prescribe labels for the short sequences of ongoing states and gradually to for the longer and more complex ones.

It seems that the gradual increase in complexity and duration of goal-directed activities is a mechanism underlying human maturation. Indeed, in babyhood, the goal-directed activities of infants can be described as very short-term with primitive means but the activities of adults are long-lasting often life-ranging processes with very complex, hierarchical means and developed language. Therefore, the imitation of gradual human growth may be an effective way to achieve the human complexity of goal-directed activity and intelligence.

Of course, emphasis on the joint synthesis does not imply that other methods cannot be used within the given approach. For example, the criterion of minimal construction costs permits the system to synthesize a goal and means in any situation but if the minimum of costs found by the system is too local then the goal and means synthesized may be inappropriate to the situation. This mechanism seems to be an explanation for the fact that people sometimes are unable to solve simple problems, though their knowledge and skills are sufficient to find the right solution. To avoid similar difficulties an AI system may use the goals and means constructed jointly as a seed point in some cases and afterwards searches for goals and methods that are more suitable.

### **6. Conclusion**

340 Intelligent Systems

Similarly, it is not necessary that the joint synthesis is the only way to create full-scale

Though the joint synthesis seems appropriate to construct local projects like the models above, I will concentrate on the construction of general artificial intelligence. In my position, the construction of general AI should be a consequence of gradual changes within a system associated with its temporal functioning and complication, i.e. a result of a shift from short – term processes with simple goals and means to long-term processes with multilevel goals and complex methods. It is suggested that, although such a system may have complex innate architecture, the main source of its development is interactions with the system's

The mechanism of joint synthesis is suggested to be a basis for such development. Indeed, because no innate criterion for the construction of goals is used, there are no constraints for the complication of goals. An appropriate means can be constructed for any goal because goals and means are constructed together,. As a result, the system is, in principle, able to adapt to any condition of training. It is suggested that the system can use blackboard architecture with flexible elements, then the formation and/or changes in elements can be considered as learning. The key component of the models above is the functional proximity that determines the possibility of the interaction between modules. It can be supposed that, in the hypothetical system, functional proximity can be dynamically changed in regard with the complexity and diversity of the system's components. The increase in the duration of goal-directed process may result from alterations in something similar to AL in the models

It can be suggested that the system should be able to create and use something similar to symbols. The generation of symbols may be performed as follows: because means in the system are constructed along with goals the system should be able to describe input and/or output information caused by these means in the terms consistent with goals and to prescribe a label associated with a goal to a sequence of such descriptions, thus creating symbols. The advantage of this method is that such symbols are grounded in the ongoing activity and therefore they can be used to construct novel goals and means through the involvement of symbols in the system's blackboard. Of course, this mechanism of symbolization can be gradual like changes in the duration of goal-directed processes. That is, at the beginning, the system will prescribe labels for the short sequences of ongoing states

It seems that the gradual increase in complexity and duration of goal-directed activities is a mechanism underlying human maturation. Indeed, in babyhood, the goal-directed activities of infants can be described as very short-term with primitive means but the activities of adults are long-lasting often life-ranging processes with very complex, hierarchical means and developed language. Therefore, the imitation of gradual human growth may be an effective way to achieve the human complexity of goal-directed activity and intelligence.

Of course, emphasis on the joint synthesis does not imply that other methods cannot be used within the given approach. For example, the criterion of minimal construction costs permits the system to synthesize a goal and means in any situation but if the minimum of costs found by the system is too local then the goal and means synthesized may be inappropriate to the situation. This mechanism seems to be an explanation for the fact that

and gradually to for the longer and more complex ones.

artificial intelligence.

above.

environment via feedback loops.

Since the advent of first computers, the idea of construction of artificial intelligence similar to that of human beings has driven the work of thousands of brilliant scientists and engineers. However, the result of their activity seems unsatisfactory as compared to, for example, advances in computer hardware. One of the most fundamental reasons for this situation may be that the mechanisms of human intelligence are unclear. Though the imitation of human intelligence is not a necessary characteristic of artificial intelligence, obviously a particular view on human intelligence is a very important heuristic. Therefore, an incorrect understanding of human intelligence can be a serious obstacle to construct intelligent systems. Intelligence is a characteristic of goal-directed systems and two classes of such systems can be easily derived from observations of animals and human beings. In my opinion, the classes that underlie most approaches to the construction of artificial intelligence are not sufficient to explain human activity. A broader classification of goal-directed activities suggests such processes can be described as a two-dimensional structure rather than a one-dimension one. In such structure there is a cell where in my opinion, humans can be located i.e., humans are goal-directed systems that synthesize arbitrary goals and means together. Though the idea of joint synthesis seems contradictory to some aspects of everyday experience, it is consistent with psychological evidence. In addition, there is neural evidence favoring this supposition. Simple computer models demonstrate that the idea of joint synthesis can be applied to simulate goal-directed activity. I suggest that the idea of joint synthesis can be a useful method to advance research in the construction of intelligent systems.

#### **7. Appendix**

Model 1 p2=3\*10-6 ; p3=0,99; p4=0,001 Model 2 p1=1; p4=2; p5=0, 3; p6=0,98; p7=0,5; p8=0,3; p9=0,8; p10=0,8

#### **8. References**


**16** 

**Innovative Intelligent Services for** 

**Older Adults and Their Caregivers** 

Anelia Mitseva1, Sofoklis Kyriazakos2, Antonis Litke3, Paolo Barone4, Alessandro Mamelli4, Nikolaos Papadakis3 and Neeli R. Prasad2

ISISEMD project started in March 2009 is a European pilot project with the main aim to design, implement and test in real-life conditions innovative intelligent services for older adults with mild cognitive impairments and their formal and family care-givers (ISISEMD, 2009). Their ultimate goal is to improve the Quality of Life of the elderly and to help them to live more independently. The technology service platform is integration of systems from Hewlett Packard (Italy), Alcatel-Lucent (Italy), Converge ICT Solutions (Greece), Eltronic A/S (Denmark) and Socrate Medical (Italy). The service platform has been tested, validated and evaluated in four different European pilot sites (Frederikshavn in Denmark, Lappeenranta in Finland, Trikala in Greece and Belfast in UK) for a duration of more than one year and based on the outcome from the real-life operations and user feedback, it has been optimised, improved and adapted for diverse regional conditions. This has been a very challenging process because the user and functional requirements are very high w.r.t. intuitive user interface with minimum user interaction, diverse needs of the main user groups, personalisation needs, requirements for stable operation despite of the fact that it is a pilot system, but working in real-life conditions and exposed to interruptions in

**1.2 Understanding the unmet needs of the end-user groups of ISISEMD services** 

According to the 2009 World Alzheimer Report, the number of people with dementia in Europe is conservatively anticipated to increase by 40% over the next 20 years mainly due to the increase in the ageing population (ALZ, 2009). This means that, due to limited resources,

**1. Introduction** 

**1.1 European pilot project ISISEMD** 

connectivity, hardware failures, human errors, etc.

**Supporting Cognitively Impaired** 

*1North Denmark EU-Office/Aalborg Municipality, 2Center for TeleInFrastructure/Aalborg University,* 

> *3Converge ICT Solutions, 4Hewlett Packard Italy,*

> > *1,2Denmark 3Greece 4Italy*


## **Innovative Intelligent Services for Supporting Cognitively Impaired Older Adults and Their Caregivers**

Anelia Mitseva1, Sofoklis Kyriazakos2, Antonis Litke3, Paolo Barone4, Alessandro Mamelli4, Nikolaos Papadakis3 and Neeli R. Prasad2 *1North Denmark EU-Office/Aalborg Municipality, 2Center for TeleInFrastructure/Aalborg University, 3Converge ICT Solutions, 4Hewlett Packard Italy, 1,2Denmark 3Greece 4Italy* 

#### **1. Introduction**

342 Intelligent Systems

Cooper, R. P., & Shallice, T. (1995). Soar and the case for unified theories of cognition.

Evans, J. St. B. T. (2003). In two minds: dual-process accounts of reasoning. *Trends in* 

Joseph, R. (1999). Frontal lobe psychopathology: mania, depression, confabulation,

Holland, J.H. (1975). *Adaptation in natural and artificial systems.* University of Michigan Press,

Leslie, K. P. , Littman, M.L & Moore, A.W. (1996). Reinforcement Learning: A Survey.

Luria, A.R. (1966). Higher cortical functions in man. Tavistock. Publications, Andover, Hants. Luria, A.R. (1973). *Foundations of neuropsychology*. Moscow State University Press (in Russian). Luria, A.R. (1982). Variants of the "frontal" syndrome. In Luria A.R, Homskaya E.D. (eds.), *Functions of the frontal lobes of the brain*, pp. 8-48, Nauka ( in Russian). Miller, B.T. & D'Esposito, M. (2005). Searching for ''the top'' in top-down control. *Neuron*,

Miller, E.K. & Cohen, J.D. (2001). An integrative theory of prefrontal cortex function. *Annual* 

Prudkov, P. N. & Rodina, O.N.(1999). Synthesis of Purposeful Processes. *PSYCOLOQUY*

Rodina, O.N. & Prudkov, P.N. (2005). The principle of joint synthesis in purposeful

Russell, S. J. & Norvig, P. (2003). *Artificial Intelligence: A Modern Approach.* Upper Saddle

Stanovich, K.E. & West, R.F. (2000). Individual differences in reasoning: Implications for the

Tooby, J. & Cosmides L., (1992) The psychological foundations of culture, in J. H. Barkow, L.

Tooby, J., Cosmides, L. & Barrett, H. C. (2005). Resolving the debate on innate ideas:

Van der Velde, F. & de Kamps, M. (2003). A model of visual working memory in PFC.

Van der Velde, F. & de Kamps, M. (2006). Neural blackboard architectures of combinatorial

Wood J.N. & Grafman J. (2003). Human prefrontal cortex: processing and representational

Cosmides, and J. Tooby (eds.) *The Adapted Mind: Evolutionary Psychology and the* 

Learnability constraints and the evolved interpenetration of motivational and conceptual functions. In Carruthers, P., Laurence, S. & Stich, S. (Eds.), *The Innate* 

catatonia, perseveration, obsessive compulsions, and schizophrenia. *Psychiatry*,

*Cognition* , 55,2,:115-49.

*Cognitive Sciences* , 10, 454-459.

Jackson, P. (1998). *Introduction to Expert Systems*.

*Review Neuroscience*, 24,167–202

River, NJ: Prentice Hall.

*Neurocomputing*, 52-54, 419-424.

Craig, I. (1995). *Blackboard Systems*.

62(2), 138-72.

Ann Arbor.

48, 535–538.

Corkill, D.D. (1991). Blackboard Systems. *AI Expert*, 6(9),40-47.

Corsini, R.J. & Auerbach, A.J. (1998). *Concise encyclopedia of psychology*. Wiley.

Fuster, J.M. (1997). Network memory. *Trends Neuroscience*, 20, 451–459 Haykin, S (1998). *Neural Networks: A Comprehensive Foundation*. Prentice Hall

Heckhausen, H. (1980). *Motivation und Handeln.* Springer-Verlag

*Journal of Artificial Intelligence Research.* 4, 237–285.

Newell, A. (1990*). Unified theories of cognition*. Cambridge, MA: Harvard.

10(070). psyc.99.10.070.purposeful-processes.1.prudkov.

processes. *MGU Bulletin Psychology*, 2, 77-86 (in Russian).

rationality debate. *Behavioral and Brain Sciences*, 23, 645-726. Sutton, R. S. & Barto, A.G. (1998). *Reinforcement Learning*: *An Introduction*. MIT Press.

*Generation of Culture* pp. 19-136, NY Oxford University Press.

*Mind: Structure and Content.* NY: Oxford University Press.

structures in cognition. *Behavioral and Brain Sciences*, 29, 37-70

perspectives. *Nature Review Neuroscience*, 4, 139–147.

#### **1.1 European pilot project ISISEMD**

ISISEMD project started in March 2009 is a European pilot project with the main aim to design, implement and test in real-life conditions innovative intelligent services for older adults with mild cognitive impairments and their formal and family care-givers (ISISEMD, 2009). Their ultimate goal is to improve the Quality of Life of the elderly and to help them to live more independently. The technology service platform is integration of systems from Hewlett Packard (Italy), Alcatel-Lucent (Italy), Converge ICT Solutions (Greece), Eltronic A/S (Denmark) and Socrate Medical (Italy). The service platform has been tested, validated and evaluated in four different European pilot sites (Frederikshavn in Denmark, Lappeenranta in Finland, Trikala in Greece and Belfast in UK) for a duration of more than one year and based on the outcome from the real-life operations and user feedback, it has been optimised, improved and adapted for diverse regional conditions. This has been a very challenging process because the user and functional requirements are very high w.r.t. intuitive user interface with minimum user interaction, diverse needs of the main user groups, personalisation needs, requirements for stable operation despite of the fact that it is a pilot system, but working in real-life conditions and exposed to interruptions in connectivity, hardware failures, human errors, etc.

#### **1.2 Understanding the unmet needs of the end-user groups of ISISEMD services**

According to the 2009 World Alzheimer Report, the number of people with dementia in Europe is conservatively anticipated to increase by 40% over the next 20 years mainly due to the increase in the ageing population (ALZ, 2009). This means that, due to limited resources,

Innovative Intelligent Services

**2. ISISEMD service platform – a short overview** 

for Supporting Cognitively Impaired Older Adults and Their Caregivers 345

The overall architecture of ISISEMD is composed in such a way that on one hand it fulfils the service requirements, and on the other hand it addresses the reality to make the different systems (also identified as x-Servers) to be integrated into a whole with a web-portal functioning as a common entry point for the ISISEMD system. The portal is the actual single point of entrance into the ISISEMD system and is the component responsible for the management of the users as well as the association of users with services and other users. It contains various functional sub-components such as a User Management module, an Authentication and Authorisation module, a Logging module, and a Reporting module. At the user side, the central unit is a computer with a touch screen, which is called Carebox.

High level architecture of ISISEMD platform is presented on the figure below.

Fig. 1. High level architecture of the ISISEMD service platform

ISISEMD services for cognitively impaired adults include (Table 1): home and personal safety observations; reminders and prompts for basic daily activities – pre-defined reminders exist but there is also the possibility to set up and show personalised "free text reminders"; services for cognitive stimulation (Memory Lane shows a picture slide show on the Carebox and Brain Games that can be played on the Carebox); video communication service with care-givers; location service when the person is out of the home with the help of a simple GPS device called Lommy; emergency contact button in the home (Help button on

ISISEMD services for the formal (FCG) and informal care-givers (ICG) include: alerts, notifications, and alarm services which are distributed by mobile phone or email; an overview of daily activities shown on the portal; video-call service for communication with the elderly;

the Carebox touch screen) and outside the home (panic button on the GPS device).

there will be a significant challenge for the social care providers to meet their needs when the illness progresses. Currently, this group of citizens lives in the community and is being taken care of by their families which exposes them to care stress, social isolation, reduced employment and in many cases also leading to health deterioration. There are many types of dementia and for each person the disease develops individually. In general this group of older adults lacks structure of the day, their abstract thinking is drastically reduced, there are risks for home incidents from fire or a food forgotten on the cooker or they can get lost outside their home. All these risks prevent them to live independently and their family caregivers suffer a lot of stress and reduced quality of life. Their Quality of Life (QOL) can be maintained or increased and care stress can be reduced if intelligent technology services give the family care-givers a helping hand to notify about risks in the home or provide support information to the elderly person (EP) about the current day and time, upcoming appointments, etc. ISISEMD services have been initially designed for three main end-user groups – the older adults with mild dementia or mild cognitive impairments, their informal care-givers (partners, closest family, neighbours) and the formal carers.

The holistic approach of the ISISEMD services has a big potential for a positive impact but this requires "a smart system" with very high level of autonomous operation and intelligence of the services so they provide the exact type of home support needed for the specific dyad "elderly-family care-giver", with minimum interaction from the elderly and care-givers part. At the same time, the technology and the services must be "invisible" for the users and require very little or no user interaction at all.

The major contribution of ISISEMD project is that it aims at improving quality of life of fragile user groups by offering home support technology services in a holistic way, fulfilling most of their un-met needs. It involves all relevant end-user groups in the whole process of design, validation and assessment of the intelligent services in real-life conditions and in diverse regional settings. In this way it advances the developments one step closer in understanding the challenges that accompany the process of introducing Information and Communication Technology (ICT) services to older adults with mild cognitive impairments living in the community and their care-givers. Last but not least, it shares hands-on experiences and best practices.

#### **1.3 Chapter outline**

The challenges listed above were addressed by introducing intelligence in almost all of the services, in data and profile management, in the networking and in the integration and optimisation process. In this chapter we describe the final outcome of 30-month efforts, presented as follows: Section 2 presents a short overview of ISISEMD services. Section 3 focuses on highlighting the intelligent features in the service functionalities. Section 4 describes how ISISEMD project advances State-of-the-Art for intelligent systems and the advantages of ISISEMD system in comparison with other systems. Section 5 gives details how the services were piloted in real-life. Qualitative technical evaluation, service validation and user evaluation for satisfaction and acceptance were carried out and outcome from them is presented in Sections 6, 7 and 8 respectively. The positive impact from the use of the services is depicted with examples of "success stories" and users' statements in Section 9. The chapter is concluded in Section 10.

#### **2. ISISEMD service platform – a short overview**

344 Intelligent Systems

there will be a significant challenge for the social care providers to meet their needs when the illness progresses. Currently, this group of citizens lives in the community and is being taken care of by their families which exposes them to care stress, social isolation, reduced employment and in many cases also leading to health deterioration. There are many types of dementia and for each person the disease develops individually. In general this group of older adults lacks structure of the day, their abstract thinking is drastically reduced, there are risks for home incidents from fire or a food forgotten on the cooker or they can get lost outside their home. All these risks prevent them to live independently and their family caregivers suffer a lot of stress and reduced quality of life. Their Quality of Life (QOL) can be maintained or increased and care stress can be reduced if intelligent technology services give the family care-givers a helping hand to notify about risks in the home or provide support information to the elderly person (EP) about the current day and time, upcoming appointments, etc. ISISEMD services have been initially designed for three main end-user groups – the older adults with mild dementia or mild cognitive impairments, their informal

The holistic approach of the ISISEMD services has a big potential for a positive impact but this requires "a smart system" with very high level of autonomous operation and intelligence of the services so they provide the exact type of home support needed for the specific dyad "elderly-family care-giver", with minimum interaction from the elderly and care-givers part. At the same time, the technology and the services must be "invisible" for

The major contribution of ISISEMD project is that it aims at improving quality of life of fragile user groups by offering home support technology services in a holistic way, fulfilling most of their un-met needs. It involves all relevant end-user groups in the whole process of design, validation and assessment of the intelligent services in real-life conditions and in diverse regional settings. In this way it advances the developments one step closer in understanding the challenges that accompany the process of introducing Information and Communication Technology (ICT) services to older adults with mild cognitive impairments living in the community and their care-givers. Last but not least, it shares hands-on

The challenges listed above were addressed by introducing intelligence in almost all of the services, in data and profile management, in the networking and in the integration and optimisation process. In this chapter we describe the final outcome of 30-month efforts, presented as follows: Section 2 presents a short overview of ISISEMD services. Section 3 focuses on highlighting the intelligent features in the service functionalities. Section 4 describes how ISISEMD project advances State-of-the-Art for intelligent systems and the advantages of ISISEMD system in comparison with other systems. Section 5 gives details how the services were piloted in real-life. Qualitative technical evaluation, service validation and user evaluation for satisfaction and acceptance were carried out and outcome from them is presented in Sections 6, 7 and 8 respectively. The positive impact from the use of the services is depicted with examples of "success stories" and users' statements in Section 9.

care-givers (partners, closest family, neighbours) and the formal carers.

the users and require very little or no user interaction at all.

experiences and best practices.

The chapter is concluded in Section 10.

**1.3 Chapter outline** 

The overall architecture of ISISEMD is composed in such a way that on one hand it fulfils the service requirements, and on the other hand it addresses the reality to make the different systems (also identified as x-Servers) to be integrated into a whole with a web-portal functioning as a common entry point for the ISISEMD system. The portal is the actual single point of entrance into the ISISEMD system and is the component responsible for the management of the users as well as the association of users with services and other users. It contains various functional sub-components such as a User Management module, an Authentication and Authorisation module, a Logging module, and a Reporting module. At the user side, the central unit is a computer with a touch screen, which is called Carebox. High level architecture of ISISEMD platform is presented on the figure below.

Fig. 1. High level architecture of the ISISEMD service platform

ISISEMD services for cognitively impaired adults include (Table 1): home and personal safety observations; reminders and prompts for basic daily activities – pre-defined reminders exist but there is also the possibility to set up and show personalised "free text reminders"; services for cognitive stimulation (Memory Lane shows a picture slide show on the Carebox and Brain Games that can be played on the Carebox); video communication service with care-givers; location service when the person is out of the home with the help of a simple GPS device called Lommy; emergency contact button in the home (Help button on the Carebox touch screen) and outside the home (panic button on the GPS device).

ISISEMD services for the formal (FCG) and informal care-givers (ICG) include: alerts, notifications, and alarm services which are distributed by mobile phone or email; an overview of daily activities shown on the portal; video-call service for communication with the elderly;

Innovative Intelligent Services

**3.1 Home safety services** 

and/or motion detectors.

captured by other sensors.

**3.1.1 Cooking monitor service** 

deliver the services.

focus on the intelligent features of the service platform.

**3. Intelligence in the service platform** 

for Supporting Cognitively Impaired Older Adults and Their Caregivers 347

The table above gives an overview of the ISISEMD services. In the following sections we

One of the advantages of the ISISEMD system is the hidden intelligence in the provided services. The basic idea is to abstract the services and make them transparent so the end users cannot actually realise the system complexity as they may not have much knowledge on ICT and this is not expected either. Drawing on this, the services act and pro-act in such a way that they are to define any critical circumstances that might occur before these actually happen and thus, they try to anticipate any consequences (e.g. the relatives of a person will receive a notification if the person is cooking for longer time than expected, anticipating thus a possible danger of having a fire in the home). This intelligence is possible to implement through a high-end home automation system identified as the Ecosystem.

Home safety services consist of sensors for monitoring the safety in the home environment. They function through sending email/sms to care-givers, posting alarms or notification messages on the web portal in cases of alarm events from the intelligent front door sensor, cooking monitor, fridge door sensor, fire/smoke sensor, flood detection sensor, bed sensors

The Ecosystem part of the Carebox technically consists of a virtual machine that runs a local instance of an Ecosystem Domotics server, which is a reduced set of services and processes tailored for the specific requirements. This server is responsible for monitoring and responding to events from the various input sensors in the patient's home. A special ISISEMD-specific service installed on each Domotics module is responsible for relaying those events up to the central X-Server hub process on the portal-side of the server to be evaluated and acted upon if required. Through this fact, a delegation of intelligence is possible to the first level of reaction so as to increase the efficiency of the whole system.

An example of domotic service operation is explained by the control of the home cooker the event is triggered by the fact that the person turns on the cooker. This activates an event of setting the appropriate timers that will monitor the duration of the cooker being on. Alert messages will be issued accordingly and the measures will be taken in such a way that will prohibit an accident and preventing fires, etc. Respective workflows are triggered by events

Elderly person is staying at home alone and the home is equipped with the aforementioned home automation services for preventing the events that have been mentioned. The services are pre-configured and running in the background without disturbing his daily life. Whenever the events that trigger the services are activated, the workflows will start-up to

The purpose of having installed a cooking monitor in the elderly peoples' homes is significant for the safety of receiving an alarm in case of having forgotten to turn off the

lifestyle pattern information over a period of time and remote doctor service. The added value for the informal care-givers relies on the fact of reducing the care stress towards an elder person and of being able to have the ability to receive information about potentially dangerous situations for their relative. This gives the feeling of safety for the beloved people and to be able to use the services without having any extensive knowledge on ICT - for instance a relative who takes care for the partner can be also a senior citizen. More details for the characteristics of the end-user groups and the added value for them from using the ICT services are provided in (Mitseva et al., 2009).


Table 1. List of ISISEMD services

Similarly, the ISISEMD platform adds value to the social care providers. They can unobtrusively monitor some activities that take place in homes and outdoors through general purpose, off-the-shelf devices. They have the ability to save travel time for performing unnecessary homes visits to the clients and to communicate remotely with the elderly clients using video-call service, and in this way, there is more time for care for more clients.

The table above gives an overview of the ISISEMD services. In the following sections we focus on the intelligent features of the service platform.

### **3. Intelligence in the service platform**

One of the advantages of the ISISEMD system is the hidden intelligence in the provided services. The basic idea is to abstract the services and make them transparent so the end users cannot actually realise the system complexity as they may not have much knowledge on ICT and this is not expected either. Drawing on this, the services act and pro-act in such a way that they are to define any critical circumstances that might occur before these actually happen and thus, they try to anticipate any consequences (e.g. the relatives of a person will receive a notification if the person is cooking for longer time than expected, anticipating thus a possible danger of having a fire in the home). This intelligence is possible to implement through a high-end home automation system identified as the Ecosystem.

#### **3.1 Home safety services**

346 Intelligent Systems

lifestyle pattern information over a period of time and remote doctor service. The added value for the informal care-givers relies on the fact of reducing the care stress towards an elder person and of being able to have the ability to receive information about potentially dangerous situations for their relative. This gives the feeling of safety for the beloved people and to be able to use the services without having any extensive knowledge on ICT - for instance a relative who takes care for the partner can be also a senior citizen. More details for the characteristics of the end-user groups and the added value for them from using the ICT

Service type Service name Validation status

Cooking monitor

Smoke/ Fire detection alarm Kitchen/Bathroom flood detection Fridge door alarm Leaving bed during night for long time Wake up sensor Intelligent front door

To Do List,

touch screen Cognitive stimulation Brain Games Validated during the

Medication Manager

Outdoor positioning

Panic button with outdoor position Fall alarm outdoor

Similarly, the ISISEMD platform adds value to the social care providers. They can unobtrusively monitor some activities that take place in homes and outdoors through general purpose, off-the-shelf devices. They have the ability to save travel time for performing unnecessary homes visits to the clients and to communicate remotely with the elderly clients

using video-call service, and in this way, there is more time for care for more clients.

givers Videophone

support Lifestyle Pattern

Calendar, Time/Date Validated during the Help/Contact request on the real-life pilot operation

Memory Lane real-life pilot operation

Remote Doctor Demo evaluation

Validated during the real-life pilot operation

Partially validated during the real-life pilot operation

Not validated during the real-life pilot operation

Validated during the real-life pilot operation

Partially validated during the real-life pilot operation

services are provided in (Mitseva et al., 2009).

Home safety

Structure of the day and contact to informal caregivers in case of emergency

Communication with care-

Communication with health and care professionals

Outdoor safety

Professional care-givers

Table 1. List of ISISEMD services

Home safety services consist of sensors for monitoring the safety in the home environment. They function through sending email/sms to care-givers, posting alarms or notification messages on the web portal in cases of alarm events from the intelligent front door sensor, cooking monitor, fridge door sensor, fire/smoke sensor, flood detection sensor, bed sensors and/or motion detectors.

The Ecosystem part of the Carebox technically consists of a virtual machine that runs a local instance of an Ecosystem Domotics server, which is a reduced set of services and processes tailored for the specific requirements. This server is responsible for monitoring and responding to events from the various input sensors in the patient's home. A special ISISEMD-specific service installed on each Domotics module is responsible for relaying those events up to the central X-Server hub process on the portal-side of the server to be evaluated and acted upon if required. Through this fact, a delegation of intelligence is possible to the first level of reaction so as to increase the efficiency of the whole system.

An example of domotic service operation is explained by the control of the home cooker the event is triggered by the fact that the person turns on the cooker. This activates an event of setting the appropriate timers that will monitor the duration of the cooker being on. Alert messages will be issued accordingly and the measures will be taken in such a way that will prohibit an accident and preventing fires, etc. Respective workflows are triggered by events captured by other sensors.

Elderly person is staying at home alone and the home is equipped with the aforementioned home automation services for preventing the events that have been mentioned. The services are pre-configured and running in the background without disturbing his daily life. Whenever the events that trigger the services are activated, the workflows will start-up to deliver the services.

#### **3.1.1 Cooking monitor service**

The purpose of having installed a cooking monitor in the elderly peoples' homes is significant for the safety of receiving an alarm in case of having forgotten to turn off the

Innovative Intelligent Services

Fig. 2. Workflow for service "cooking monitor"

**3.1.3 Fridge door service** 

SMS.

**3.1.4 Bed service** 

for Supporting Cognitively Impaired Older Adults and Their Caregivers 349

This service monitors the door of the refrigerator. Once the elderly person opens the refrigerator door, the system will wait for it to be closed after a certain predefined time

Letting a caregiver know if the door has been forgotten open for too long – by email or

 The following parameter can be adjusted: Time duration, since the fridge door was opened, after which to send an alert to a caregiver that the fridge door has probably

If there is an alarm situation, the system shows a blinking message on the top line of the Carebox and is "telling" to the elderly that the fridge door has been opened for too long and he/she needs to react. This notification is repeated periodically on the Carebox until the

Alarm is triggered if an elderly has left the bed for a long time during the night. Furthermore, this service allows a caregiver to be notified about disturbances or significant alterations in the patient's sleep pattern. From the service page, a caregiver can view whether the elderly person is in the bed or not, the last time the patient sat or slept on it, and

period. If it is not closed in the specified amount of time, an alert will be issued:

The open, close or alarm events will be posted on the portal

been forgotten open (specified in minutes).

event "fridge door closed" is detected by the system.

the last time the elderly person got up from the bed.

cooker. If an elderly has forgotten to turn off the cooker, a care-giver will be able to view the current status displayed on the portal when logging in. For instance, on the service page on the portal, it will be displayed if an elderly has "started cooking", "cooking on" or "not cooking" and the last time when this event took place will also be illustrated. If for instance the cooker has been on for a while or for too long, the system will send the information containing an alert message (email/sms) to the care-giver. At the same time, at elderly's home, a voice message will be played to the elderly person from the Carebox, and a blinking message will turn up on the top part of the Carebox screen to warn him/her about the forgotten activity. The duration for cooking before receiving an alert is to be adjusted depending on the "life style pattern" of the specific elderly person. For example, the time for the duration of a cooker being turned on for a while can be adjusted before it is to trigger an alarm to care-givers. The voice and text message for cooking for a while is a prompt for the elderly to react, defined as first level reaction. Then if the system detects that there is no response from the elderly, an alarm could be sent to care-giver for example that the cooker has not been turned off for more than a pre-defined period of time – this is characterised as the second level reaction. All these events, communicated as cooker on, cooker off, message for cooking for a while or for too long, are posted on the message board on the web portal for overview purpose of the daily activities. The workflow for this service is depicted on Figure 2 below.

Depending on the preferences of the regional care provider and the family care-givers, the cooking monitoring service can work with either one of the three installations - one temperature sensor, with two temperature sensors or with a power relay. In case of two temperature sensors, a threshold for temperature difference can be adjusted from the web portal.

#### **3.1.2 Smoke/fire detection service**

The purpose of the smoke/fire alarm is to send information for smoke or fire alarm to the web portal. Furthermore, the system notifies the elderly person on the Carebox about the dangerous situation. At the same time it notifies the caregivers by SMS or email. If caregiver accesses this service, a care-giver can see if a fire alarm is in progress, the last time when a fire alarm may have occurred, and if so, the care-giver can view the time when it ended (i.e. it was reset). All these events are posted on the web portal in the list with recent events so the care-giver can see them. If there is event "smoke/fire alarm ON", at the same time the system shows a blinking message on the top line of the Carebox and is "telling" to the elderly person that there is smoke/fire detected in the home and he/she needs to react. This notification is repeated periodically on the Carebox until the event "smoke/fire alarm OFF" is detected by the system. Improvement suggested during the real-life pilot operation for the smoke/fire alarm service was to send one additional SMS - initially the service was designed to send one SMS in case of "alarm on" event but there was a need for similar notification SMS for "alarm off" event and this was implemented. This was needed because of some cases with false alarms during the test period and also due to the fact that very often more than one relative receives SMS alarms but depending on their agreement, one is to react in case of incident. The second SMS was to inform all of them that there is no more danger.

Fig. 2. Workflow for service "cooking monitor"

#### **3.1.3 Fridge door service**

348 Intelligent Systems

cooker. If an elderly has forgotten to turn off the cooker, a care-giver will be able to view the current status displayed on the portal when logging in. For instance, on the service page on the portal, it will be displayed if an elderly has "started cooking", "cooking on" or "not cooking" and the last time when this event took place will also be illustrated. If for instance the cooker has been on for a while or for too long, the system will send the information containing an alert message (email/sms) to the care-giver. At the same time, at elderly's home, a voice message will be played to the elderly person from the Carebox, and a blinking message will turn up on the top part of the Carebox screen to warn him/her about the forgotten activity. The duration for cooking before receiving an alert is to be adjusted depending on the "life style pattern" of the specific elderly person. For example, the time for the duration of a cooker being turned on for a while can be adjusted before it is to trigger an alarm to care-givers. The voice and text message for cooking for a while is a prompt for the elderly to react, defined as first level reaction. Then if the system detects that there is no response from the elderly, an alarm could be sent to care-giver for example that the cooker has not been turned off for more than a pre-defined period of time – this is characterised as the second level reaction. All these events, communicated as cooker on, cooker off, message for cooking for a while or for too long, are posted on the message board on the web portal for overview purpose of the daily activities. The workflow for this service is depicted on

Depending on the preferences of the regional care provider and the family care-givers, the cooking monitoring service can work with either one of the three installations - one temperature sensor, with two temperature sensors or with a power relay. In case of two temperature sensors, a threshold for temperature difference can be adjusted from the web

The purpose of the smoke/fire alarm is to send information for smoke or fire alarm to the web portal. Furthermore, the system notifies the elderly person on the Carebox about the dangerous situation. At the same time it notifies the caregivers by SMS or email. If caregiver accesses this service, a care-giver can see if a fire alarm is in progress, the last time when a fire alarm may have occurred, and if so, the care-giver can view the time when it ended (i.e. it was reset). All these events are posted on the web portal in the list with recent events so the care-giver can see them. If there is event "smoke/fire alarm ON", at the same time the system shows a blinking message on the top line of the Carebox and is "telling" to the elderly person that there is smoke/fire detected in the home and he/she needs to react. This notification is repeated periodically on the Carebox until the event "smoke/fire alarm OFF" is detected by the system. Improvement suggested during the real-life pilot operation for the smoke/fire alarm service was to send one additional SMS - initially the service was designed to send one SMS in case of "alarm on" event but there was a need for similar notification SMS for "alarm off" event and this was implemented. This was needed because of some cases with false alarms during the test period and also due to the fact that very often more than one relative receives SMS alarms but depending on their agreement, one is to react in case of incident. The second SMS was to inform all of them that there is no more

Figure 2 below.

**3.1.2 Smoke/fire detection service** 

portal.

danger.

This service monitors the door of the refrigerator. Once the elderly person opens the refrigerator door, the system will wait for it to be closed after a certain predefined time period. If it is not closed in the specified amount of time, an alert will be issued:


If there is an alarm situation, the system shows a blinking message on the top line of the Carebox and is "telling" to the elderly that the fridge door has been opened for too long and he/she needs to react. This notification is repeated periodically on the Carebox until the event "fridge door closed" is detected by the system.

#### **3.1.4 Bed service**

Alarm is triggered if an elderly has left the bed for a long time during the night. Furthermore, this service allows a caregiver to be notified about disturbances or significant alterations in the patient's sleep pattern. From the service page, a caregiver can view whether the elderly person is in the bed or not, the last time the patient sat or slept on it, and the last time the elderly person got up from the bed.

Innovative Intelligent Services

out of the home (see above description).

inactive (see above description), specified in minutes.

to receive alerts if the patient leaves the home.

temporary solution was found for the special cases.

**3.2 Carebox for the end-user** 

kind of device are multiple:

to them

for Supporting Cognitively Impaired Older Adults and Their Caregivers 351

Time duration of inactivity that determines when the system decides that the patient is

Time duration after which the system will issue an alert if the elderly is considered

 The start and end times of the "night time" period. These two times tell the service what time is considered to be "night time", i.e. the period during which the care-giver wants

During the pilot operation, there were some cases when bed sensors could not be used – for example the elderly sleeps on a folding sofa and a bed sensor cannot be installed or if the elderly does not want to have a bed sensor installed due to some health problems. In this case, the current settings of this service cannot allow for optimal operation during the night period. Therefore, one of the suggestions for improvement was for the intelligent front door service to work without considering information from the bed sensor during the night. However, the final solution could not be implemented by the end of the project; instead a

The only visible interface toward the ISISEMD platform for the elderly persons is the touch screen-enabled device installed in their homes. The motivations behind the selection of such

1. First of all, it is very difficult for people with cognitive impairments to learn how to interact with new technologies: since they tend to forget what they already know, it would not be effective at all introducing in their life new elements that are not familiar

2. In addition, for the same reason described above, the need for direct interaction between them and the system must be reduced to the minimum, if not eliminated 3. Finally, in the cases where such interactions cannot be eliminated, they must be

The touch screen-enabled device selected for providing the ISISEMD services for the end

1. The shape, dimensions and aspect of the device are indeed identical to the ones of a television set, an appliance whose diffusion, in the last fifty years, has reached almost every home (at least in the developed countries). This means that it can be introduced in

2. The device can provide different interaction levels; the level for each end-user can be configured (at any time and remotely) by a care-giver based on the level of impairment.

3. The touch screen modality is currently the simplest and more intuitive technology available to interact with a computing system, which requires a very low level of

 Acts as the collector for all the messages that must be shown to the end-user. Running an intelligent software module, it makes decisions on when a message must be

In the simplest configuration, no interaction possibilities at all are provided

performed in the simplest and more intuitive modality possible

user can fulfil the above-mentioned requirements in the way that:

the elderly persons' homes in a natural manner

From a functional perspective, the device plays several roles:

training and cognitive capabilities

The service can send the following alerts:


The following three parameters can be adjusted:


Example of these parameters is when the system raises an alarm if the patient is in bed for longer than 10 hours, and also in case the patient left the bed between 10pm and 6am for longer than one hour (specified as 60 minutes). Since very often the elderly person takes a nap in the afternoon, it was suggested by the regional partners that additional time period of afternoon sleep is defined, similar to the night-time sleep period. However, this could not be implemented in the time frame of the project.

#### **3.1.5 Intelligent front door service**

This service monitors the status of the elderly's front door and monitors the presence of the elderly in the home with the help of motion detection (and bed sensor information if such a sensor is installed). On the main service page a care-giver can see the status of the front door, as well as the last times it was opened or closed and the last time a movement was recorded in the home.

Line for current status also shows whether the system has assumed that the elderly is present in the home or not at this moment. The system makes this conclusion by monitoring the various sensors, such as cooking sensor, door activity sensor, fridge door activity sensor, motion detection activity sensor. Each time when an activity is detected by one of these sensors it assumes that the elderly is active and in the home. The service also monitors the bed status (if bed sensor is installed), so that if there is no activity but the elderly is sleeping then this too will indicate an elderly's presence in the home. If there is no activity on any of these sensors for a certain amount of time (customisable from the service screen in the web portal) then the system will report to care-givers that the elderly is absent from the home and will make notification accordingly. Alerts are sent if the front door has been opened for more than a certain amount of minutes, when an elderly is absent from the house for certain hours and if an elderly is assumed to be absent from home for too long time.

The front door alarm is a service that can be customised depending on the elderly life pattern. For this purpose, four parameters can be adjusted according to:

 Time duration after which the front door will issue an alert if it stays opened, specified in minutes.


During the pilot operation, there were some cases when bed sensors could not be used – for example the elderly sleeps on a folding sofa and a bed sensor cannot be installed or if the elderly does not want to have a bed sensor installed due to some health problems. In this case, the current settings of this service cannot allow for optimal operation during the night period. Therefore, one of the suggestions for improvement was for the intelligent front door service to work without considering information from the bed sensor during the night. However, the final solution could not be implemented by the end of the project; instead a temporary solution was found for the special cases.

#### **3.2 Carebox for the end-user**

350 Intelligent Systems

If an elderly person has been on the bed too long time (a case in which the care-giver

If an elderly person has left the bed too long time during a certain period of time (such

Time duration after which an alarm is issued, if an elderly is still in bed (specified in

Time duration, during "night time" (see next parameter) after which a care-giver

 The start and end times of the 'night time' period. These two times tell the service what time is considered to be "night time", i.e. the period during which the care-giver wants to receive alerts if the elderly leaves the bed for a certain time (see previous parameter).

Example of these parameters is when the system raises an alarm if the patient is in bed for longer than 10 hours, and also in case the patient left the bed between 10pm and 6am for longer than one hour (specified as 60 minutes). Since very often the elderly person takes a nap in the afternoon, it was suggested by the regional partners that additional time period of afternoon sleep is defined, similar to the night-time sleep period. However, this could not

This service monitors the status of the elderly's front door and monitors the presence of the elderly in the home with the help of motion detection (and bed sensor information if such a sensor is installed). On the main service page a care-giver can see the status of the front door, as well as the last times it was opened or closed and the last time a movement was

Line for current status also shows whether the system has assumed that the elderly is present in the home or not at this moment. The system makes this conclusion by monitoring the various sensors, such as cooking sensor, door activity sensor, fridge door activity sensor, motion detection activity sensor. Each time when an activity is detected by one of these sensors it assumes that the elderly is active and in the home. The service also monitors the bed status (if bed sensor is installed), so that if there is no activity but the elderly is sleeping then this too will indicate an elderly's presence in the home. If there is no activity on any of these sensors for a certain amount of time (customisable from the service screen in the web portal) then the system will report to care-givers that the elderly is absent from the home and will make notification accordingly. Alerts are sent if the front door has been opened for more than a certain amount of minutes, when an elderly is absent from the house for certain

The front door alarm is a service that can be customised depending on the elderly life

Time duration after which the front door will issue an alert if it stays opened, specified

hours and if an elderly is assumed to be absent from home for too long time.

pattern. For this purpose, four parameters can be adjusted according to:

assumes that the elderly might be experiencing difficulty or health issues).

as night time) and which might mean that elderly person has fallen down.

The service can send the following alerts:

The following three parameters can be adjusted:

should be alerted if the elderly is out of bed.

be implemented in the time frame of the project.

**3.1.5 Intelligent front door service** 

recorded in the home.

in minutes.

minutes)

The only visible interface toward the ISISEMD platform for the elderly persons is the touch screen-enabled device installed in their homes. The motivations behind the selection of such kind of device are multiple:


The touch screen-enabled device selected for providing the ISISEMD services for the end user can fulfil the above-mentioned requirements in the way that:


From a functional perspective, the device plays several roles:

 Acts as the collector for all the messages that must be shown to the end-user. Running an intelligent software module, it makes decisions on when a message must be

Innovative Intelligent Services

safety

screen of an end user:

time)

of missing confirmation

which affects the set of messages to display.

**3.3 Knowledge and data management** 

notifications can be divided into two main categories:

user in maintaining the structure of the day

in the previous section on domotic-related services.

create custom ones, so called "free text reminders"

One-time only at a specified date and time

Recurring only in specified date ranges

notifications are sent to the caregiver by SMS and email

individuals or embedded in organisational processes or practice.

It can be selected the format for the notification: text, audio or both

disappearing from the screen and the interval among repetitions

for Supporting Cognitively Impaired Older Adults and Their Caregivers 353

services that are provisioned to all the end users (since they are strictly related to their safety) that require the delivery of notifications to them by means of the Carebox. Such

1. Messages coming from the domotic sensors, which are strictly related to the end-user

2. Messages configured by the care-givers on the web portal which should help the end

The configuration for messages of type 1 is made by the care-giver by accessing the "Domotic services" section of the web portal, and by filling the proper settings as described

The configuration for messages of type 2 is made by accessing the "To Do List" and "Calendar" section of the web portal. This section provides the caregivers with a huge set of possibilities for configuring and scheduling the prompt messages that must be shown on the

It is possible either to choose among a pre-defined set of commonly used messages or to

 It is possible to associate a sound to the message, to catch the end-user´s attention It can be defined how many times the message must be played and shown before

The system allows to configure every desired scheduling pattern for the message:

Recurring message (every day, once a week, once a month, once a year at the same

 It is possible for care-givers to specify, for each configured reminder, if a corresponding confirmation button must appear on the Carebox screen together with the message. If the message is not confirmed (e.g. the end user does not press the confirmation button),

 In case the reminder requires confirmation, the care-givers can define on the service page some rules for how and when they want to receive SMS or email messages in case

Every Carebox is constantly in communication with the server-side module handling the configuration of the reminders and updates in almost real time for every modification,

In general knowledge management comprises a range of strategies and practices used in an organisation to identify, create, represent, distribute, and enable adoption of insights and experiences. Such insights and experiences comprise knowledge, either embodied in

In ISISEMD platform knowledge is being used in order to assist the care-givers to take the appropriate decisions. Knowledge comprises the aggregated information gathered from

delivered and, based on several static and dynamic parameters - such as the user profile, the configuration made by the care-givers, the importance of the event to be notified, the surrounding context - what the proper formats for the delivery are (e.g. only text, text and audio message, text and sound, etc.)


The intelligent software module running on the device is constantly in communication with a central server module and is capable to enable/disable in real-time every single service provisioned, based on the configuration decided by a care-giver. Accordingly, it repaints the GUI elements displayed on the screen adding or removing them dynamically, for example by adding a new upcoming event in the list of "Next events". In addition to this, the module constantly considers the user context, leveraging the device equipment (e.g. the embedded web camera) or data coming from other devices (e.g. domotic sensors) and takes decisions based on the current situation.

Here follow some examples:


The layout of the GUI displayed on the Carebox and the related contents are strongly dependent on the settings made by the care-givers for each end-user. As already mentioned, the system is quite flexible, so that the care-givers can enable/disable at runtime a different set of services for each user depending on the level of cognitive impairments. Such configuration is made by accessing the web portal. Anyway, there is a core set of essential

 Allows the user to proactively use the platform by pressing a single "soft" button on the graphical user interface (GUI) to ask assistance from the care-givers. This can be accomplished by pressing with a finger on a help button which triggers the sending of notification messages in the form of SMS or email, or a button for receiving a video-

 Stimulates the cognitive capabilities of mild-dementia affected people by displaying information that helps them in maintaining the structure of the day (e.g. date and time) and the memories (e.g. slide show of personal pictures meaningful to them). Provides possibility to the elderly to confirm an activity by pressing a single "soft" button on

The intelligent software module running on the device is constantly in communication with a central server module and is capable to enable/disable in real-time every single service provisioned, based on the configuration decided by a care-giver. Accordingly, it repaints the GUI elements displayed on the screen adding or removing them dynamically, for example by adding a new upcoming event in the list of "Next events". In addition to this, the module constantly considers the user context, leveraging the device equipment (e.g. the embedded web camera) or data coming from other devices (e.g. domotic sensors) and takes decisions

 During the night time the system enters an energy saving mode by turning off the screen. Such modality is interrupted (the screen is displayed again) if movement is

 Multimedia alarm messages related to dangerous conditions in the home (e.g. a fire alarm or a cooker left on for a long time) are played on the device speakers and shown on the device screen on behalf of the domotic module. In this way, the elderly person can react to the situation (e.g. switching off the cooker) or, in case he/she does not within a pre-configured timeframe, alarm messages are sent to the care-givers and

 In case of external conditions that temporarily prevent the module to provide some of the functionalities (e.g. instable network connection), it changes the display layout showing proper notification messages which inform of the unavailability of one or more services. At the same time, a server side module is capable of detecting the error conditions and of alerting the proper actors (for example the service support team) to take action. As soon as the error conditions disappear, the module automatically

The layout of the GUI displayed on the Carebox and the related contents are strongly dependent on the settings made by the care-givers for each end-user. As already mentioned, the system is quite flexible, so that the care-givers can enable/disable at runtime a different set of services for each user depending on the level of cognitive impairments. Such configuration is made by accessing the web portal. Anyway, there is a core set of essential

returns in full-operation mode. Notifications are sent also to the care-givers.

detected in the surrounding by the context monitoring module

only text, text and audio message, text and sound, etc.)

GUI if elderly is able to interact with the system

conference session

based on the current situation. Here follow some examples:

relatives in form of SMS and email

delivered and, based on several static and dynamic parameters - such as the user profile, the configuration made by the care-givers, the importance of the event to be notified, the surrounding context - what the proper formats for the delivery are (e.g. services that are provisioned to all the end users (since they are strictly related to their safety) that require the delivery of notifications to them by means of the Carebox. Such notifications can be divided into two main categories:


The configuration for messages of type 1 is made by the care-giver by accessing the "Domotic services" section of the web portal, and by filling the proper settings as described in the previous section on domotic-related services.

The configuration for messages of type 2 is made by accessing the "To Do List" and "Calendar" section of the web portal. This section provides the caregivers with a huge set of possibilities for configuring and scheduling the prompt messages that must be shown on the screen of an end user:

	- One-time only at a specified date and time
	- Recurring message (every day, once a week, once a month, once a year at the same time)
	- Recurring only in specified date ranges

Every Carebox is constantly in communication with the server-side module handling the configuration of the reminders and updates in almost real time for every modification, which affects the set of messages to display.

#### **3.3 Knowledge and data management**

In general knowledge management comprises a range of strategies and practices used in an organisation to identify, create, represent, distribute, and enable adoption of insights and experiences. Such insights and experiences comprise knowledge, either embodied in individuals or embedded in organisational processes or practice.

In ISISEMD platform knowledge is being used in order to assist the care-givers to take the appropriate decisions. Knowledge comprises the aggregated information gathered from

Innovative Intelligent Services

for Supporting Cognitively Impaired Older Adults and Their Caregivers 355

being promoted as a standard through OASIS (BPEL). The work of the workflow management coalition WfMC (WFMC) with the resulting XPDL standard (XPDL) provides a higher level of abstraction of workflow and aims to provide a format for process design. The focus of BPEL, and most business-oriented workflow languages, is control flow. Extensive research on workflow control patterns has shown that all languages have limitations in terms of what can be easily expressed (AALST, 2003). This insufficient expressivity and lack of rigorous semantics to allow automated checks on correctness and completeness mean that

The scientific community has also conducted considerable research into information and data processing applications that have similarities to some of the ISISEMD applications. However, these languages and tools do not support the constructs needed in ISISEMD and are not designed for Business to Business (B2B) applications that require distributed management, and trust and security based on the emerging industry Web Service standards. Beyond this, comes research into intelligent, autonomous, goal-oriented and knowledgebased service infrastructures. These infrastructures include facilities for: dynamic negotiation, adaptation and configuration; intelligent scheduling, resource and service selection; and optimised job execution and management. The infrastructure itself is able to decide on how to react in case of unexpected situations using self-organisation, selfmanagement, etc. Of particular relevance to ISISEMD is the use of semantic service descriptions to facilitate autonomic discovery, composition and use of services, for example

ISISEMD makes heavy use of emerging web service standards, including the areas of semantic service descriptions, SLAs and agreements, workflow and orchestration, management, trust and security, and transport and messaging. A significant research element of the work includes determining exactly how to apply and extend these standards in order to support ISISEMD applications. A particular challenge is balancing the use of industry standards, which are the key to meeting business to business needs and hence exploitation of the ISISEMD results. The standards that ISISEMD consider and engage with are manifold. Competing proposals from industry vendors may eventually converge through consolidation, but whilst this has the potential to improve interoperability, it often

In particular, ISISEMD adopts WS-Convergence where possible and evaluate/build upon/adapt the standards used for management, in particular to enable distributed management of networks of services that deliver real time applications in inter-organisation value-chains. Moreover, it uses Simple Knowledge Organisation System (SKOS) and Web Ontology Language (OWL), for formally describing SLA terms and their relationships, including QoS attributes of e-care applications. In particular, we address the problem of mapping high-level business objectives to low-level resource provisioning policies in a more

Summarising the key features and comparing them with the existing solutions available in

BPEL and related languages are unlikely to be a suitable foundation for ISISEMD.

within an organisation as an approach to resource management.

does so at the expense of compromising the specification.

**4.2 Potentials of ISISEMD services and limitations** 

the market, ISISEMD service platform has the following advantages:

automated, robust and verifiable way.

various sources (sensors, log data, events, etc) and is being considered in the service provisioning.

A dedicated service directly related with this is a Lifestyle pattern service. This service is not based on one workflow. It is a rather complex service that has to do with monitoring and logging the various events that happen in the domotics and Carebox environment of the elderly and in the sequel applying Business Intelligence in order to derive the patterns and the life style of the elderly. Based on this, important conclusions can be made for further optimisation of the home automations and care-giver strategies can be better adapted.

#### **3.4 System optimisation for real-life operation**

During the pilot operation, diverse interruptions of the normal system operation were observed due to some real-life factors, e.g. internet disconnection problems, high sensitivity of the touch screen, etc., all of them leading to generating false alarms or service unavailability. Therefore, additional layer of intelligence was introduced by fall back solutions and allowing for a higher tolerance to interruptions without jeopardising the safety of the end user. Also in case of planned technical maintenance operations, messages are sent to all care-givers and in case of portal or x- servers' unavailability, a notification is sent to the technical support team.

#### **4. Advancing state-of-the-art for intelligent systems**

#### **4.1 Service oriented architecture**

One of the major challenges for ISISEMD platform is the sustainability in the long-run, which can only be guaranteed if the technology can support a long term evolution, considering the user needs and the competition. Therefore, ISISEMD platform deploys a pure Service Oriented Architecture (SOA) that ensures the service integrity and the extensibility of the platform.

Service level agreement (SLA)-based web service infrastructures are the key for driving automation and system dynamics for ISISEMD platform. Web Services Distributed Management (WSDM) specifications and Web Services (WS)-Management only support a crude description of consumer-centric Quality of Service (QoS) properties not suitable to fully describe ISISEMD application requirements. However, the main challenge for ISISEMD is the specification of a QoS at the various levels in the value chain, including the ability to translate of high-level business objectives to low-level resource provisioning policies. Furthermore, well-defined metrics need to exist at all levels to allow monitoring and reporting of service usage against SLAs.

ISISEMD applications necessarily involve a network of services for their implementation. The application workflows involved, including both data and control flows, can be complex and require explicit support for e-care aspects and constraints.

The use of workflows for "programming in the large" to compose web services has led to significant interest in a standard workflow language within the web services stack. WS-BPEL (BPEL) has major industry support and was created through the agreed merge of their earlier web service composition languages WSFL and XLANG respectively. WS-BPEL and is

various sources (sensors, log data, events, etc) and is being considered in the service

A dedicated service directly related with this is a Lifestyle pattern service. This service is not based on one workflow. It is a rather complex service that has to do with monitoring and logging the various events that happen in the domotics and Carebox environment of the elderly and in the sequel applying Business Intelligence in order to derive the patterns and the life style of the elderly. Based on this, important conclusions can be made for further optimisation of the home automations and care-giver strategies can be better adapted.

During the pilot operation, diverse interruptions of the normal system operation were observed due to some real-life factors, e.g. internet disconnection problems, high sensitivity of the touch screen, etc., all of them leading to generating false alarms or service unavailability. Therefore, additional layer of intelligence was introduced by fall back solutions and allowing for a higher tolerance to interruptions without jeopardising the safety of the end user. Also in case of planned technical maintenance operations, messages are sent to all care-givers and in case of portal or x- servers' unavailability, a notification is

One of the major challenges for ISISEMD platform is the sustainability in the long-run, which can only be guaranteed if the technology can support a long term evolution, considering the user needs and the competition. Therefore, ISISEMD platform deploys a pure Service Oriented Architecture (SOA) that ensures the service integrity and the

Service level agreement (SLA)-based web service infrastructures are the key for driving automation and system dynamics for ISISEMD platform. Web Services Distributed Management (WSDM) specifications and Web Services (WS)-Management only support a crude description of consumer-centric Quality of Service (QoS) properties not suitable to fully describe ISISEMD application requirements. However, the main challenge for ISISEMD is the specification of a QoS at the various levels in the value chain, including the ability to translate of high-level business objectives to low-level resource provisioning policies. Furthermore, well-defined metrics need to exist at all levels to allow monitoring

ISISEMD applications necessarily involve a network of services for their implementation. The application workflows involved, including both data and control flows, can be complex

The use of workflows for "programming in the large" to compose web services has led to significant interest in a standard workflow language within the web services stack. WS-BPEL (BPEL) has major industry support and was created through the agreed merge of their earlier web service composition languages WSFL and XLANG respectively. WS-BPEL and is

provisioning.

**3.4 System optimisation for real-life operation** 

**4. Advancing state-of-the-art for intelligent systems** 

sent to the technical support team.

**4.1 Service oriented architecture** 

and reporting of service usage against SLAs.

and require explicit support for e-care aspects and constraints.

extensibility of the platform.

being promoted as a standard through OASIS (BPEL). The work of the workflow management coalition WfMC (WFMC) with the resulting XPDL standard (XPDL) provides a higher level of abstraction of workflow and aims to provide a format for process design. The focus of BPEL, and most business-oriented workflow languages, is control flow. Extensive research on workflow control patterns has shown that all languages have limitations in terms of what can be easily expressed (AALST, 2003). This insufficient expressivity and lack of rigorous semantics to allow automated checks on correctness and completeness mean that BPEL and related languages are unlikely to be a suitable foundation for ISISEMD.

The scientific community has also conducted considerable research into information and data processing applications that have similarities to some of the ISISEMD applications. However, these languages and tools do not support the constructs needed in ISISEMD and are not designed for Business to Business (B2B) applications that require distributed management, and trust and security based on the emerging industry Web Service standards.

Beyond this, comes research into intelligent, autonomous, goal-oriented and knowledgebased service infrastructures. These infrastructures include facilities for: dynamic negotiation, adaptation and configuration; intelligent scheduling, resource and service selection; and optimised job execution and management. The infrastructure itself is able to decide on how to react in case of unexpected situations using self-organisation, selfmanagement, etc. Of particular relevance to ISISEMD is the use of semantic service descriptions to facilitate autonomic discovery, composition and use of services, for example within an organisation as an approach to resource management.

ISISEMD makes heavy use of emerging web service standards, including the areas of semantic service descriptions, SLAs and agreements, workflow and orchestration, management, trust and security, and transport and messaging. A significant research element of the work includes determining exactly how to apply and extend these standards in order to support ISISEMD applications. A particular challenge is balancing the use of industry standards, which are the key to meeting business to business needs and hence exploitation of the ISISEMD results. The standards that ISISEMD consider and engage with are manifold. Competing proposals from industry vendors may eventually converge through consolidation, but whilst this has the potential to improve interoperability, it often does so at the expense of compromising the specification.

In particular, ISISEMD adopts WS-Convergence where possible and evaluate/build upon/adapt the standards used for management, in particular to enable distributed management of networks of services that deliver real time applications in inter-organisation value-chains. Moreover, it uses Simple Knowledge Organisation System (SKOS) and Web Ontology Language (OWL), for formally describing SLA terms and their relationships, including QoS attributes of e-care applications. In particular, we address the problem of mapping high-level business objectives to low-level resource provisioning policies in a more automated, robust and verifiable way.

#### **4.2 Potentials of ISISEMD services and limitations**

Summarising the key features and comparing them with the existing solutions available in the market, ISISEMD service platform has the following advantages:

Innovative Intelligent Services

the project (August 2011).

them.

deployment.

for Supporting Cognitively Impaired Older Adults and Their Caregivers 357

 Last but not least, the demo rooms were used as living labs because the improvements of the services were carried out in parallel with the pilot operation and in all cases the improvements were first applied and proofed in the demo machines, and then

In August 2010 the partners had a meeting to discuss experiences from the small scale pilot operation and how to make the transition to the full scale pilot operation more efficiently. In this way, the full pilot operation started in September 2010 and continued until the end of

To evaluate the effects of using the services in real-life, we followed overall assessment framework and carried out 15-month controlled study in the four European regions involving 71 elderly, 71 informal care-givers and around 15 formal care-givers, assessing baseline and final status of cognitive impairments, daily functioning and quality of life of the older adults. For the informal care-givers, quality of life and care stress were assessed. In the end of the controlled study, all groups were asked to evaluate their satisfaction and acceptability of the ICT services and their importance for care; the potential to promote independence and to increase feeling of safety. In three of regions, except in the region of

Significant results from the pilot operation were achieved even though the roll out of the pilots was more complicated than initially anticipated. There were a number of challenges to be addressed during the real-life pilot operation; some of those were not dependent on the consortium. The partners worked together and in many cases they were resolving technical issues in non-working hours/days based on good will and because of the eagerness to achieve a high impact. The whole process has been a positive learning experience for all of

The fact that most of the test users and their relatives are satisfied, feel safer and confident with ISISEMD services, is a significant result that is not measurable. Even somehow sceptical in the beginning, after giving them time to get used to the services, the elderly and their relatives accept the technology and can see the opportunities for positive impact. The relatives can better feel the difference by using the services in comparison with the elderly, because due to the disease, it is difficult for the elderly to think abstractly and understand it.

Technical evaluation was carried out to determine whether the system requirements were fulfilled by testing the services after the components of the system have been integrated into one system. The system evaluation had two aspects: *Functional evaluation* and *Non-functional evaluation*. The purpose of the *functional* tests were to evaluate all the functions that were enabled with the ISISEMD system. The *non-functional* evaluation included aspects such as e.g. scalability, reliability, flexibility. Non-functional requirements were validated based on qualitative and quantitative appropriate measures. More details about the overall assessment framework are presented in (Mitseva, 2010). In the next paragraphs we focus on the non-service specific evaluations which assess the readiness of the system for wider

**6. Outcome from technical evaluations - non-functional aspects** 

Frederikshavn, the trial participants were split in intervention and control groups.

transferred to the home installations of the test users


On the other hand studying the competition, following points have to be considered:


#### **5. Pilot operation and user evaluations**

The pilot operation activities started with final testing of the integrated platform and the beginning of the real-life pilot operation. During 15 months (May 2010-August 2011) the pilots were used by the test users under realistic conditions – older adults in their homes; the professional care-givers in their work tasks, performing their daily work to care for the elderly; the informal care-givers/family, also in their everyday activities to care for the seniors. The services were first tested in a smaller scale, with a few end-users at three of the pilot sites for a period of 3-4 months, in order to identify if any major problems exist before the large scale testing with all users during the rest of the testing period. Small scale pilots were carried out in all of the regions, except in the region of Trikala, with 2-3 home installations in each of the regions.

Since the launch of the services, in three of the regions demo-rooms have been installed and in the fourth region the system was installed in the home of one of the formal care-givers. The demo-rooms existed in addition to the home installations and had the following goals:


 Different levels of interaction and easy interactive features, allowing the elderly to ask for contact&help, confirm actions, play cognitive stimulation games, receive video calls Highly customized and targeting all user groups involved in care provision for a person

Integrated system with open architecture that can be easily extended with new features

 Focuses on mild dementia persons in a holistic way, looking at all their needs (home and person safety, promote independence, prevent social isolation, increase quality of

 Maintain or increase quality of life not only to persons with mild dementia, but also for the relatives who suffer care stress and are also socially isolated because of caring for

 An end-to-end service approach should be addressed, so that this solution can be easily deployed. That includes: helpdesk, formal and informal care-givers, technicians, all

The "breakeven" requires a large number of installations in order to have affordable

The pilot operation activities started with final testing of the integrated platform and the beginning of the real-life pilot operation. During 15 months (May 2010-August 2011) the pilots were used by the test users under realistic conditions – older adults in their homes; the professional care-givers in their work tasks, performing their daily work to care for the elderly; the informal care-givers/family, also in their everyday activities to care for the seniors. The services were first tested in a smaller scale, with a few end-users at three of the pilot sites for a period of 3-4 months, in order to identify if any major problems exist before the large scale testing with all users during the rest of the testing period. Small scale pilots were carried out in all of the regions, except in the region of Trikala, with 2-3 home

Since the launch of the services, in three of the regions demo-rooms have been installed and in the fourth region the system was installed in the home of one of the formal care-givers. The demo-rooms existed in addition to the home installations and had the following goals: Demonstration of the services to potential test participants and their relatives – they were able to see, and experience the system before deciding to join the controlled study

and were convinced that it is very user friendly and also aesthetically acceptable In these demo rooms, the formal care-givers were able to try in reality the system and provide final feedback to technical partners for usability and functionality and

 Potential end-users were introduced to the services and training was provided to them Personnel from care-giver organisations gained hands-on experience with the services and became more confident in usage of the system and learnt how to report technical

On the other hand studying the competition, following points have to be considered:

with mild dementia or cognitive impairments and the person himself

Able to offer additional services with disease progression

in the future

the dementia relative

accessible through a single entity

**5. Pilot operation and user evaluations** 

installations in each of the regions.

suggestions for improvement

problems

life, etc)

service cost

 Last but not least, the demo rooms were used as living labs because the improvements of the services were carried out in parallel with the pilot operation and in all cases the improvements were first applied and proofed in the demo machines, and then transferred to the home installations of the test users

In August 2010 the partners had a meeting to discuss experiences from the small scale pilot operation and how to make the transition to the full scale pilot operation more efficiently. In this way, the full pilot operation started in September 2010 and continued until the end of the project (August 2011).

To evaluate the effects of using the services in real-life, we followed overall assessment framework and carried out 15-month controlled study in the four European regions involving 71 elderly, 71 informal care-givers and around 15 formal care-givers, assessing baseline and final status of cognitive impairments, daily functioning and quality of life of the older adults. For the informal care-givers, quality of life and care stress were assessed. In the end of the controlled study, all groups were asked to evaluate their satisfaction and acceptability of the ICT services and their importance for care; the potential to promote independence and to increase feeling of safety. In three of regions, except in the region of Frederikshavn, the trial participants were split in intervention and control groups.

Significant results from the pilot operation were achieved even though the roll out of the pilots was more complicated than initially anticipated. There were a number of challenges to be addressed during the real-life pilot operation; some of those were not dependent on the consortium. The partners worked together and in many cases they were resolving technical issues in non-working hours/days based on good will and because of the eagerness to achieve a high impact. The whole process has been a positive learning experience for all of them.

The fact that most of the test users and their relatives are satisfied, feel safer and confident with ISISEMD services, is a significant result that is not measurable. Even somehow sceptical in the beginning, after giving them time to get used to the services, the elderly and their relatives accept the technology and can see the opportunities for positive impact. The relatives can better feel the difference by using the services in comparison with the elderly, because due to the disease, it is difficult for the elderly to think abstractly and understand it.

#### **6. Outcome from technical evaluations - non-functional aspects**

Technical evaluation was carried out to determine whether the system requirements were fulfilled by testing the services after the components of the system have been integrated into one system. The system evaluation had two aspects: *Functional evaluation* and *Non-functional evaluation*. The purpose of the *functional* tests were to evaluate all the functions that were enabled with the ISISEMD system. The *non-functional* evaluation included aspects such as e.g. scalability, reliability, flexibility. Non-functional requirements were validated based on qualitative and quantitative appropriate measures. More details about the overall assessment framework are presented in (Mitseva, 2010). In the next paragraphs we focus on the non-service specific evaluations which assess the readiness of the system for wider deployment.

Innovative Intelligent Services

**6.5 Integration/openness** 

service/software developer.

**6.6 Manageability/flexibility** 

users in general.

**of ISISEMD platform** 

placement.

**6.4 Response time** 

for Supporting Cognitively Impaired Older Adults and Their Caregivers 359

During the pilot operation, response time metric is highly related to the services that activate alarms, especially alarms through SMS which is very much dependent on the network operator and SMS service provider. In the beginning of the pilot operation, the SMS service was assigned to normal level but since we observed delays in the delivery, the SMS service level was changed. The problem related to SMS service has been solved by subscribing to highest (and more expensive) service level of the SMS service provider.

Integration of new hardware (especially sensors), requires some testing period in order to get the correct information reading as expected, e.g. by configuration and correct

Integration of new software modules both on server and client side have been easily done in ISISEMD system since the platform has been designed in such a way that the services provided to end-user can evolve by providing API (Application Programming Interface) to

The assessment of ISISEMD system for these two parameters show that the platform is very easy and flexible to manage, i.e. in terms of adding/deleting new users, new services, new home installations, and new regions/sub-regions, etc., since it has been designed to support all these requirements. Furthermore, a user manual book written in four different languages (English, Finnish, Danish and Greek) has been provided as a guideline to regions and end

We would like to argue for having carried out a good piece of work, overcoming a number of challenges, and that the final services meet the user expectations and acceptances at a high level supported by the outcome of the user evaluations. Overall, the care-givers were satisfied with the services and consider them important for providing care. Most of the users now state their wish for continuing to use the services after the project end. As suggested by the user feedback, there is even stronger need for service personalisation so they can

Based on the continuous user feedback during the pilot operation, the services were improved where possible. But due to the limited project lifetime and resources, not all additional features suggested by the end-users as improvements could be implemented. However, they can be used for further development of the service platform as a commercial product. Subsequently, we emphasize on some of the suggestions that can be considered as **"wish list for the commercial system"** by the regional partners and by the test persons. We are of the interpretation that the following suggestions could improve the intelligence of the

**7. Outcome from service validation of the common functionalities** 

function in a more intelligent and autonomous way.

service platform in the next version of the system.

#### **6.1 Personalisation/customisation**

Overall, personalisation/customisation features of the services were highly appreciated by the regional care providers and the end-users because they give possibility for them to map precisely the services to the exact needs of the dyad "elderly-informal care-giver" and fitting the support services in the care-giving and coping strategies.

#### **6.2 Robustness of the equipments and connectivity**

Carebox: we experienced some issues with the Touch-screen computer due to high touch sensitivity and they have been resolved by context-aware solution. Some few cases related to hardware failures that required replacement of the computer were mainly due to the fact that such device has been designed in "consumer electronic" segment that is not intended to run 24x7 in highly reliable environment. Unstable internet connections were reasons for false alarms and freezing of Carebox screen but they were alleviated by applying more sophisticated filters on bypassing false alarm triggering conditions.

GPS device: we used a simple GPS device called Lommy that is in general very robust and has been already in use in different contexts. The only few issues were the connector for charging the Lommy as it was difficult for use by elderly people; LED indicators, blinking with different colours, were confusing for the elderly and their care-givers; in two of the regions, that had hilly and mountain landscape, problems with the coverage were experienced.

Sensors and RAck MOnitoring System (RAMOS): The type used was Ramos Mini C. It is a very robust device. It is meant to be used both in home and industrial environments. The only one issue was Ramos and Ethernet failures during the installation and then fixed in the installation process. Overall, the problems with RAMOS and sensor devices were related to setup, configuration, placement, and installation, which is more about practical issue than the robustness.

Overall, the equipments to be used when offering ICT services for home support to elderly must be robust and "invisible" i.e. placed in those areas of the home where they are difficult to reach. This will prevent un-wanted disconnections or damages due to some every day activities such as cleaning, dusting, etc. It is best if the wires, RAMOS, routers are hidden in or behind cupboards or in a box. On the contrary, it is best if the Carebox (the Touch screen for the end users) is placed in a kitchen or living room or close to a TV, to make it easer for the elderly to see it and refer to it. It is even more important the equipments and the integrated services to be extensively tested in conditions as close to the real-life operation as possible before installing it in the homes. In this way, specific local issues can be discovered and eliminated. For example we experienced some hardware problems in only one region that were not common to the other three regions.

#### **6.3 Scalability**

During the ISISEMD pilot operation, it has been shown that the current architecture of ISISEMD portal and X-Server have been able to handle between 30 and 40 end users concurrently. In the future, it would be good to study and make some prediction on ISISEMD's users growth and traffic volume, such that network dimensioning and simulation can be done, and finally to decide the best network architecture.

#### **6.4 Response time**

358 Intelligent Systems

Overall, personalisation/customisation features of the services were highly appreciated by the regional care providers and the end-users because they give possibility for them to map precisely the services to the exact needs of the dyad "elderly-informal care-giver" and fitting

Carebox: we experienced some issues with the Touch-screen computer due to high touch sensitivity and they have been resolved by context-aware solution. Some few cases related to hardware failures that required replacement of the computer were mainly due to the fact that such device has been designed in "consumer electronic" segment that is not intended to run 24x7 in highly reliable environment. Unstable internet connections were reasons for false alarms and freezing of Carebox screen but they were alleviated by applying more

GPS device: we used a simple GPS device called Lommy that is in general very robust and has been already in use in different contexts. The only few issues were the connector for charging the Lommy as it was difficult for use by elderly people; LED indicators, blinking with different colours, were confusing for the elderly and their care-givers; in two of the regions, that had hilly and mountain landscape, problems with the coverage were

Sensors and RAck MOnitoring System (RAMOS): The type used was Ramos Mini C. It is a very robust device. It is meant to be used both in home and industrial environments. The only one issue was Ramos and Ethernet failures during the installation and then fixed in the installation process. Overall, the problems with RAMOS and sensor devices were related to setup, configuration, placement, and installation, which is more about practical issue than

Overall, the equipments to be used when offering ICT services for home support to elderly must be robust and "invisible" i.e. placed in those areas of the home where they are difficult to reach. This will prevent un-wanted disconnections or damages due to some every day activities such as cleaning, dusting, etc. It is best if the wires, RAMOS, routers are hidden in or behind cupboards or in a box. On the contrary, it is best if the Carebox (the Touch screen for the end users) is placed in a kitchen or living room or close to a TV, to make it easer for the elderly to see it and refer to it. It is even more important the equipments and the integrated services to be extensively tested in conditions as close to the real-life operation as possible before installing it in the homes. In this way, specific local issues can be discovered and eliminated. For example we experienced some hardware problems in only one region

During the ISISEMD pilot operation, it has been shown that the current architecture of ISISEMD portal and X-Server have been able to handle between 30 and 40 end users concurrently. In the future, it would be good to study and make some prediction on ISISEMD's users growth and traffic volume, such that network dimensioning and

simulation can be done, and finally to decide the best network architecture.

**6.1 Personalisation/customisation** 

experienced.

the robustness.

**6.3 Scalability** 

the support services in the care-giving and coping strategies.

sophisticated filters on bypassing false alarm triggering conditions.

**6.2 Robustness of the equipments and connectivity** 

that were not common to the other three regions.

During the pilot operation, response time metric is highly related to the services that activate alarms, especially alarms through SMS which is very much dependent on the network operator and SMS service provider. In the beginning of the pilot operation, the SMS service was assigned to normal level but since we observed delays in the delivery, the SMS service level was changed. The problem related to SMS service has been solved by subscribing to highest (and more expensive) service level of the SMS service provider.

#### **6.5 Integration/openness**

Integration of new hardware (especially sensors), requires some testing period in order to get the correct information reading as expected, e.g. by configuration and correct placement.

Integration of new software modules both on server and client side have been easily done in ISISEMD system since the platform has been designed in such a way that the services provided to end-user can evolve by providing API (Application Programming Interface) to service/software developer.

#### **6.6 Manageability/flexibility**

The assessment of ISISEMD system for these two parameters show that the platform is very easy and flexible to manage, i.e. in terms of adding/deleting new users, new services, new home installations, and new regions/sub-regions, etc., since it has been designed to support all these requirements. Furthermore, a user manual book written in four different languages (English, Finnish, Danish and Greek) has been provided as a guideline to regions and end users in general.

#### **7. Outcome from service validation of the common functionalities of ISISEMD platform**

We would like to argue for having carried out a good piece of work, overcoming a number of challenges, and that the final services meet the user expectations and acceptances at a high level supported by the outcome of the user evaluations. Overall, the care-givers were satisfied with the services and consider them important for providing care. Most of the users now state their wish for continuing to use the services after the project end. As suggested by the user feedback, there is even stronger need for service personalisation so they can function in a more intelligent and autonomous way.

Based on the continuous user feedback during the pilot operation, the services were improved where possible. But due to the limited project lifetime and resources, not all additional features suggested by the end-users as improvements could be implemented. However, they can be used for further development of the service platform as a commercial product. Subsequently, we emphasize on some of the suggestions that can be considered as **"wish list for the commercial system"** by the regional partners and by the test persons. We are of the interpretation that the following suggestions could improve the intelligence of the service platform in the next version of the system.

Innovative Intelligent Services

technological solutions.

formal care-providers and saving them time.

**8. Results from the controlled study** 

**8.1 Important aspects of the piloting process** 

**7.2 Future work** 

for Supporting Cognitively Impaired Older Adults and Their Caregivers 361

interaction of the user with the system. This is mainly due to the fact that they understand, on the one hand, that elderly is not able to be familiar with the new technology and on the other hand, they are willing to increase the level of independence of the elderly but without the cost of adding more stress to their daily responsibilities. In addition, elderly cannot handle a lot of interactions with the system, since they are not familiar with the new technology and have somewhat of an aversion to learning. We have also seen that this may discourage use of the system, as the elderly may worry that they will break the technology, so they would rather not use it at all than be the cause of expensive repairs. The users want the system to be as automatic as possible; however, they also want inexpensive

Future work on the service platform can go in several directions. With respect to the technical side, it comprises mainly on implementing some of the suggested by the end users enhancements of the user interface of the Carebox and the portal. With respect to functionality, it can involve adding new services and enhancing the functionality of the Lifestyle Pattern service to automatically generate (graphical) reports for deviation from normal user behaviour. This service has also potential for improving quality of care for the

With respect to the services support, efforts can be dedicated to offering the services as one complete home-support service including acquiring the equipments, making the home installations, ensuring help-desk support, training the end-users, maintaining the services, etc. For user support, uploading short tutorial video-clips on the web portal for re-enforcing

For optimising the installation process, a direction to go is preparing installation software package with all needed settings and defining "a standard service package" in each region with the most desired services that will be initially offered to all elderly who subscribe to the

Taken as a whole, wide scale testing with higher number of test participants, more European regions and for a duration of a couple of years would show delay in

Developing and piloting ICT services for older adults with cognitive impairments or mild dementia and their family care-givers has been a challenging process. A system like the ISISEMD platform is complementary to the daily support provided by the family care-giver and the relative itself is considered a user of the system. In addition, spouses/partners and closest family played a key role in the level of independence of old people with cognitive impairments or mild dementia and also in ISISEMD pilot. In the process of piloting the ICT services, diverse aspects played an important role – such as trust towards technology, complexity of installations and the platform due to the fact that it included high number of services. It is natural for this elderly user group to feel scepticism towards the technology

the training process for service setting was a suggestion by the regional partners.

home-support services and further re-assessed and updated if needed.

institutionalisation and provide more real-life assessments of the services.

#### **7.1 "Wish list" for enhancements for the commercial system**

They were related mainly to higher intelligence&personalisation, improved usability and additional services.

#### **7.1.1 Services enhancement for higher intelligence&personalisation**

For the purpose of better personalisation, the free text reminders on the Carebox are suggested to be enhanced so they can contain longer text and, with advantage, to make it possible for the elderly him/her self to create their own text reminders, if the elderly is able to do it. It would also be an advantage for the users if the system is able to atomically create voice files for these free text reminders through speech synthesis. The reminders functionality can also be enhanced with more pre-defined reminders and advanced by having additional services such as shopping list reminders for the elderly people.

Other suggestions for enhancing the services and their usability concern implementing even more intelligence into the platform for automatic synchronisation of the calendar service events with the service for home presence. And for the cooker service, it would be an advantage to make separate time settings for the cooker plates and for the oven seeing that users have individual patterns and the oven shows to be used for a longer time than the cooker plates are. Users also suggest for the functionality of the intelligent front door service to be able to work without the bed sensor because there are cases when bed sensor cannot be used.

Further enhancement of the services is to create the possibility of having a small message board on the Carebox that can show to the elderly a SMS message sent by their care-givers mobile devices. In this way, the relatives will be able to send a SMS to the Carebox so the SMS text can be shown on the screen to the elderly person.

#### **7.1.2 Usability enhancement**

Some of the first suggestions focused on enhancing usability of the services, the portal and the Carebox - for instance, improving the visual aspect of the portal and the Carebox for enhancing the user interface to be usable by visual, hearing or cognitive impaired users. This will also create the possibility for other people with disabilities (blind/ deaf) to interact with the system.

#### **7.1.3 "Entertainment" features**

For the purpose of enhancing satisfaction of the elderly people and their family caregivers, it was suggested to include "entertainment" features - plug-ins for facilitating distribution of relevant information such as local news, the weather forecast, favourite music, etc. These entertainment features are to be accessible and shown on the Carebox screen by assigning for such a service.

#### **7.1.4 Reflections**

The automation degree of the services, together with their availability, was seen as a key feature for the family care-givers. Informal caregivers are willing to have very few degree of interaction of the user with the system. This is mainly due to the fact that they understand, on the one hand, that elderly is not able to be familiar with the new technology and on the other hand, they are willing to increase the level of independence of the elderly but without the cost of adding more stress to their daily responsibilities. In addition, elderly cannot handle a lot of interactions with the system, since they are not familiar with the new technology and have somewhat of an aversion to learning. We have also seen that this may discourage use of the system, as the elderly may worry that they will break the technology, so they would rather not use it at all than be the cause of expensive repairs. The users want the system to be as automatic as possible; however, they also want inexpensive technological solutions.

#### **7.2 Future work**

360 Intelligent Systems

They were related mainly to higher intelligence&personalisation, improved usability and

For the purpose of better personalisation, the free text reminders on the Carebox are suggested to be enhanced so they can contain longer text and, with advantage, to make it possible for the elderly him/her self to create their own text reminders, if the elderly is able to do it. It would also be an advantage for the users if the system is able to atomically create voice files for these free text reminders through speech synthesis. The reminders functionality can also be enhanced with more pre-defined reminders and advanced by

Other suggestions for enhancing the services and their usability concern implementing even more intelligence into the platform for automatic synchronisation of the calendar service events with the service for home presence. And for the cooker service, it would be an advantage to make separate time settings for the cooker plates and for the oven seeing that users have individual patterns and the oven shows to be used for a longer time than the cooker plates are. Users also suggest for the functionality of the intelligent front door service to be able to work without the bed sensor because there are cases when bed sensor cannot be

Further enhancement of the services is to create the possibility of having a small message board on the Carebox that can show to the elderly a SMS message sent by their care-givers mobile devices. In this way, the relatives will be able to send a SMS to the Carebox so the

Some of the first suggestions focused on enhancing usability of the services, the portal and the Carebox - for instance, improving the visual aspect of the portal and the Carebox for enhancing the user interface to be usable by visual, hearing or cognitive impaired users. This will also create the possibility for other people with disabilities (blind/ deaf) to interact with

For the purpose of enhancing satisfaction of the elderly people and their family caregivers, it was suggested to include "entertainment" features - plug-ins for facilitating distribution of relevant information such as local news, the weather forecast, favourite music, etc. These entertainment features are to be accessible and shown on the Carebox screen by assigning

The automation degree of the services, together with their availability, was seen as a key feature for the family care-givers. Informal caregivers are willing to have very few degree of

**7.1 "Wish list" for enhancements for the commercial system** 

SMS text can be shown on the screen to the elderly person.

**7.1.2 Usability enhancement** 

**7.1.3 "Entertainment" features** 

**7.1.1 Services enhancement for higher intelligence&personalisation** 

having additional services such as shopping list reminders for the elderly people.

additional services.

used.

the system.

for such a service.

**7.1.4 Reflections** 

Future work on the service platform can go in several directions. With respect to the technical side, it comprises mainly on implementing some of the suggested by the end users enhancements of the user interface of the Carebox and the portal. With respect to functionality, it can involve adding new services and enhancing the functionality of the Lifestyle Pattern service to automatically generate (graphical) reports for deviation from normal user behaviour. This service has also potential for improving quality of care for the formal care-providers and saving them time.

With respect to the services support, efforts can be dedicated to offering the services as one complete home-support service including acquiring the equipments, making the home installations, ensuring help-desk support, training the end-users, maintaining the services, etc. For user support, uploading short tutorial video-clips on the web portal for re-enforcing the training process for service setting was a suggestion by the regional partners.

For optimising the installation process, a direction to go is preparing installation software package with all needed settings and defining "a standard service package" in each region with the most desired services that will be initially offered to all elderly who subscribe to the home-support services and further re-assessed and updated if needed.

Taken as a whole, wide scale testing with higher number of test participants, more European regions and for a duration of a couple of years would show delay in institutionalisation and provide more real-life assessments of the services.

#### **8. Results from the controlled study**

#### **8.1 Important aspects of the piloting process**

Developing and piloting ICT services for older adults with cognitive impairments or mild dementia and their family care-givers has been a challenging process. A system like the ISISEMD platform is complementary to the daily support provided by the family care-giver and the relative itself is considered a user of the system. In addition, spouses/partners and closest family played a key role in the level of independence of old people with cognitive impairments or mild dementia and also in ISISEMD pilot. In the process of piloting the ICT services, diverse aspects played an important role – such as trust towards technology, complexity of installations and the platform due to the fact that it included high number of services. It is natural for this elderly user group to feel scepticism towards the technology

Innovative Intelligent Services

for Supporting Cognitively Impaired Older Adults and Their Caregivers 363

According to the test group's final evaluation, there was a significant decrease of the level of care burden among the informal care-givers after using the ISISEMD services whereas in the control group the informal care-givers burden indicated an increase after the evaluation period. Overall for the test group, 80% of the elderly maintained their basic Activities of Daily Living (ADL) and 40% of the elderly improved their Instrumental Activities of Daily Living (IADL). For the Quality of Life, 70% of the elderly had an improvement in their Quality of Life and 80% of the informal care-givers similarly had an improvement in their Quality of Life. 80% of the informal care-givers had a reduction of care-related stress (care-

From formal care-givers point of view, at least two thirds of formal care-givers rated the services to be easy to use. In extension, the majority of formal care-givers rated the services to be very important for care. They felt that services were easy to integrate into their existing care routines and were easy to personalise. Three quarters of formal care-givers also

From the qualitative evaluations, we can draw the conclusions that parameters that were important along the way for piloting the services were service maturity, flexibility and personalisation. Furthermore, the training of the end-users and the intelligence of the services played an important role for the overall satisfaction too, platform stability and

We would like to thank to our test users whom we accept as an equal partner of the consortium. They played a very important role in the process of bringing the services to a mature level and improving them in all aspects in order to meet their needs as best as possible. We can confirm that it is of high importance that the primary user and care-givers are to be motivated towards the usage of aiding technologies in their homes. For the acceptance of the services by the elderly, a key role was played by the family care-giver and the process was much more rapid and easier if the care-givers had previous experience with

The positive impacts from the use of the services are described below with the words of the real users as success stories and the overall interpretation of these, points towards an improvement in feeling of independence and safety and in communication and social relation between the elderly people and the relatives. Thus, an improved Quality of Life is

**ISISEMD services form a basis for good communication between the elderly and their relatives.** Hence, the services help the elderly people with mild dementia in the North Region of Denmark to create quick and easy communication with relatives. The Carebox equipped with a help button, is helpful for the elderly in the sense that it is easier to get in touch with relatives whenever it is needed, emergency or not. One case showing the

provided high ratings for services, in terms of their importance for care-giving.

service availability and offering the services at earliest stage of the disease.

**8.4 Acknowledgements for the test users in the pilot** 

**9. "Success stories" from the pilot** 

being observed in reality.

giver burden) and 0% had an increase in care-related stress.

**8.3 The view of the formal care-givers** 

technology.

and we took this into consideration in the process. There was a thin line and a trade off between testing the platform with real users while the services were still under improvement and adaptation for the real-life conditions. Finding many volunteers who fit the inclusion criteria and are willing to evaluate the services was another challenge for us. It also took us time to build understandable communication in cross-disciplinary teams because of the different paradigms for the regional and technical partners. By far, the most crucial factors were proved to be the maturity of the services; the thorough testing before installing them in the homes; matching the services to those clients who have the most benefit from them and openness for new technology.

During the trial period, all ethical rights of the citizens were respected and the trials were carried out according to high ethical standards and the national regulations and the privacy of the trial participants and all data related to this were ensured. All applicable national and international laws and acts were respected too. The trial participants were recruited only after approvals from Ethical Committees were granted for each region where it was required and consent forms were signed by all trial participants.

At baseline, for the cohorts combined (intervention and control groups), the female participants were 67.4%, the average age was 78.69 years, having mild to moderate decline in cognition. The assessment on basic daily functioning showed full or high dependency, while instrumental activities of daily living mean score indicated mid-range of dependency. The test group of informal care-givers showed severe effect on their quality of life, but the mean value for cohorts combined was on the border line of severe effect on their quality of life. The care burden indicated that the care-givers were in the mid-range of caregiverburden effects. Highest percentage of family care-givers was children.

#### **8.2 Findings from the final assessments**

The controlled study assessed in the end of the 15-month trial period cognitive decline and daily functioning for the elderly persons. For the informal care-givers – care stress. For both groups the impact of the services on the quality of life was investigated. We looked further into domains of which we expected the technology services to make a positive impact. Elderly persons and relatives were also asked for their willingness to use the services after the end of the project and their willingness to pay for the services in general.

User acceptance and satisfaction with the services were evaluated upon with the three main end-user groups and for instance, care-givers were asked about the importance for care and ability for independent living and, overall satisfaction. We observed a difference among the views from the four regions for the services showing a minimum/maximum satisfaction and acceptance among the elderly test users and a difference in the lowest/highest rating among informal care-givers of the services that are important for care giving. This also reflected cultural and care-model differences among the four European regions.

Regarding the feeling of safety, 70% of the elderly felt safer when using the ISISEMD system. Another 20% reported feeling significantly safer. 40% of the informal care-givers report feeling safer, with another 50% of them reporting feeling significantly safer. More than half of the test users reported independent living increases - 51.61% of elderly and 67.74% of informal care-givers. Also 3.23% of the informal care-givers reported it increases more than they thought.

According to the test group's final evaluation, there was a significant decrease of the level of care burden among the informal care-givers after using the ISISEMD services whereas in the control group the informal care-givers burden indicated an increase after the evaluation period. Overall for the test group, 80% of the elderly maintained their basic Activities of Daily Living (ADL) and 40% of the elderly improved their Instrumental Activities of Daily Living (IADL). For the Quality of Life, 70% of the elderly had an improvement in their Quality of Life and 80% of the informal care-givers similarly had an improvement in their Quality of Life. 80% of the informal care-givers had a reduction of care-related stress (caregiver burden) and 0% had an increase in care-related stress.

#### **8.3 The view of the formal care-givers**

362 Intelligent Systems

and we took this into consideration in the process. There was a thin line and a trade off between testing the platform with real users while the services were still under improvement and adaptation for the real-life conditions. Finding many volunteers who fit the inclusion criteria and are willing to evaluate the services was another challenge for us. It also took us time to build understandable communication in cross-disciplinary teams because of the different paradigms for the regional and technical partners. By far, the most crucial factors were proved to be the maturity of the services; the thorough testing before installing them in the homes; matching the services to those clients who have the most

During the trial period, all ethical rights of the citizens were respected and the trials were carried out according to high ethical standards and the national regulations and the privacy of the trial participants and all data related to this were ensured. All applicable national and international laws and acts were respected too. The trial participants were recruited only after approvals from Ethical Committees were granted for each region where it was

At baseline, for the cohorts combined (intervention and control groups), the female participants were 67.4%, the average age was 78.69 years, having mild to moderate decline in cognition. The assessment on basic daily functioning showed full or high dependency, while instrumental activities of daily living mean score indicated mid-range of dependency. The test group of informal care-givers showed severe effect on their quality of life, but the mean value for cohorts combined was on the border line of severe effect on their quality of life. The care burden indicated that the care-givers were in the mid-range of caregiver-

The controlled study assessed in the end of the 15-month trial period cognitive decline and daily functioning for the elderly persons. For the informal care-givers – care stress. For both groups the impact of the services on the quality of life was investigated. We looked further into domains of which we expected the technology services to make a positive impact. Elderly persons and relatives were also asked for their willingness to use the services after

User acceptance and satisfaction with the services were evaluated upon with the three main end-user groups and for instance, care-givers were asked about the importance for care and ability for independent living and, overall satisfaction. We observed a difference among the views from the four regions for the services showing a minimum/maximum satisfaction and acceptance among the elderly test users and a difference in the lowest/highest rating among informal care-givers of the services that are important for care giving. This also

Regarding the feeling of safety, 70% of the elderly felt safer when using the ISISEMD system. Another 20% reported feeling significantly safer. 40% of the informal care-givers report feeling safer, with another 50% of them reporting feeling significantly safer. More than half of the test users reported independent living increases - 51.61% of elderly and 67.74% of informal care-givers. Also 3.23% of the informal care-givers reported it increases

benefit from them and openness for new technology.

required and consent forms were signed by all trial participants.

burden effects. Highest percentage of family care-givers was children.

the end of the project and their willingness to pay for the services in general.

reflected cultural and care-model differences among the four European regions.

**8.2 Findings from the final assessments** 

more than they thought.

From formal care-givers point of view, at least two thirds of formal care-givers rated the services to be easy to use. In extension, the majority of formal care-givers rated the services to be very important for care. They felt that services were easy to integrate into their existing care routines and were easy to personalise. Three quarters of formal care-givers also provided high ratings for services, in terms of their importance for care-giving.

From the qualitative evaluations, we can draw the conclusions that parameters that were important along the way for piloting the services were service maturity, flexibility and personalisation. Furthermore, the training of the end-users and the intelligence of the services played an important role for the overall satisfaction too, platform stability and service availability and offering the services at earliest stage of the disease.

#### **8.4 Acknowledgements for the test users in the pilot**

We would like to thank to our test users whom we accept as an equal partner of the consortium. They played a very important role in the process of bringing the services to a mature level and improving them in all aspects in order to meet their needs as best as possible. We can confirm that it is of high importance that the primary user and care-givers are to be motivated towards the usage of aiding technologies in their homes. For the acceptance of the services by the elderly, a key role was played by the family care-giver and the process was much more rapid and easier if the care-givers had previous experience with technology.

### **9. "Success stories" from the pilot**

The positive impacts from the use of the services are described below with the words of the real users as success stories and the overall interpretation of these, points towards an improvement in feeling of independence and safety and in communication and social relation between the elderly people and the relatives. Thus, an improved Quality of Life is being observed in reality.

**ISISEMD services form a basis for good communication between the elderly and their relatives.** Hence, the services help the elderly people with mild dementia in the North Region of Denmark to create quick and easy communication with relatives. The Carebox equipped with a help button, is helpful for the elderly in the sense that it is easier to get in touch with relatives whenever it is needed, emergency or not. One case showing the

Innovative Intelligent Services

**10. Conclusion** 

platform system.

*and plan their days, independently. "* 

use them after the project ends.

learn to refer to the services.

them time and money.

months.

for Supporting Cognitively Impaired Older Adults and Their Caregivers 365

*relative have the ISISEMD services installed and our relative can keep on doing everyday activities* 

ISISEMD platform with innovative intelligent services is scalable, open-system, validated in real-live environment. The competitive advantage of ISISEMD systems is that it is based on a modular scalable open platform that advances the State-Of-the-Art in the areas of systems for Ambient Assisted Living and can be integrated with other Health or Care Systems. Additional services can be included anytime, resulting in a very powerful

Last but not least, ISISEMD services have been extensively tested and validated in a pilot operation with 142 end-users (71 elderly and 71 relatives) for 15 months in four regions - in Denmark, Finland, Greece and UK. The final user evaluations, carried out in June 2011, showed high level of user acceptance and satisfaction with the services and willingness to

We knew that sceptical users are stoppers against introduction of new technologies. But our experience shows that the elderly and their relatives accept the technology and can see the opportunities for positive impact and added value from the use of the services in their everyday life even when the older adults with mild dementia and their family care-givers were sceptical in the beginning, after giving them time to get used to the technology. It can be expected, that after about one month, the elderly and their family caregivers can get used to the services. The most successful adoption of the services can happen when they are offered as early as possible in the disease. In this way, the technology services can be integrated in the coping and care strategies in the family and elderly has highest chances to

The benefits from using ISISEMD services were depicted in the user acceptance surveys, contributing to the Quality of Life of the end-users. The services improved the elderly ability for self-care by support for their basic daily activities in way that prevents health risks in their homes and promotes independence. They strengthened the daily interaction with their social sphere - partners and relatives, giving them the feeling of safety and improving their relationships. For the family care-givers, the services increased their quality of life and the feeling of safety; reduced their care burden and gave them a piece of mind while saving

Potential for direct savings from the usage of the ICT services is foreseen for both the careprovider organisations and informal care-givers. Costs can be saved from time and travel expenses for delivery of warm meals to elderly, performing on-line sessions with use of video-call service instead of physical home-safety visits, visits for administering medications, automatically registering information for such services, etc. The informal caregivers can save time and money for making telephone calls to elderly to check about status or to drive to elderly place or to cook meals for them. The highest potential for savings is at a society level - with delaying the admittance in nursing or dementia care institutions where savings at European level can range from approx. 1300-4000 euro for one person for 12

advantages of the ISISEMD Carebox comes into play due to the fact that one elderly with dementia has difficulties to make phone calls to relatives. Instead, by pressing the Carebox help button, the elderly person can easily get into contact with relatives. Contact is simply generated automatically by the system with the help of an SMS that is sent to relatives when the elderly presses the help button.

**The outdoor safety service helps the elderly people with mild dementia to stay healthy** by giving them the possibility to be active outdoors in their everyday lives. One case with an older man from North Region of Denmark reveals the advantage brought by the GPS device. The older man with diabetes who needs to keep his blood pressure down used to be afraid to go outside, as he could get lost. Today, with the GPS device the older man is no longer afraid to go for a walk. He explains, *"I feel more safe because I know that the Lommy will help me to get in contact with my family and then, they can find me if I get lost".* Neighbours and relatives reveal that they have even observed that the older man is going regularly for walks and that they can see his position on a map if he needs help or gets lost.

**Intelligent door alarms prevent the elderly people from leaving their homes unnoticeably in the winter time -** In Finland, a relative explains the significance of ISISEMD Intelligent door alarm and the impact it has done on both her mother and her self. The relative explains that now she can prevent her mother from going out of the house and getting lost in the cold winter in Finland. The Intelligent front door service sends an alarm message to a relative informing that the door opens and thus, the older person might be leaving the house. According to this, a relative has stated: *"In Finland, this service is very important so that we can be informed if our relatives get lost. It is so cold in the winter and this can be dangerous if the elderly goes outside for a longer period of time".* 

**Bed sensor helps the elderly to alert their families automatically when help is needed during the night -** In Finland, a relative expresses the importance of having a bed sensor installed. The relative explains that her mother has had several incidents of falling down from her bed during the night. The bed sensor sends a signal when an elderly is out of bed for too long, thus a relative or a caregiver will receive an SMS with the alert. In this particular case, the relative explained… "*When I received this message, I went to my mother's house and found her on the floor. I am very happy to have received this SMS so that I could help my mother, even if it was in the middle of the night".* 

**ISISEMD Calendar and To Do List, shown on the ISISEMD Carebox, helps the elderly with mild dementia to live more independently -** With the ISISEMD Carebox the elderly can keep track of the day and time. For instance, in North Ireland, a relative living with her mother states: *"Even if we only have had the Carebox for 10 days, both my mother and I already feel a difference. Normally my mother always calls me several times at work every day being anxious about not knowing the day and time. Now, with the Carebox, my mother is always informed about the day and time and this has helped her to become more confident*. "

**ISISEMD services are beneficial for enhancing the quality of life of the elderly and their relatives -** In Greece relatives taking care of the elderly explain that the services help them to feel safer knowing that their parents have the ISISEMD services installed in their homes. Hence, relatives are aware if the elderly leaves their houses or if they are out of bed for a long time. One relative explains … *"Before having the ISISEMD services we had to travel 30 minutes outside of town every day to check on our relative. Now we feel safer knowing that our*  *relative have the ISISEMD services installed and our relative can keep on doing everyday activities and plan their days, independently. "* 

#### **10. Conclusion**

364 Intelligent Systems

advantages of the ISISEMD Carebox comes into play due to the fact that one elderly with dementia has difficulties to make phone calls to relatives. Instead, by pressing the Carebox help button, the elderly person can easily get into contact with relatives. Contact is simply generated automatically by the system with the help of an SMS that is sent to relatives when

**The outdoor safety service helps the elderly people with mild dementia to stay healthy** by giving them the possibility to be active outdoors in their everyday lives. One case with an older man from North Region of Denmark reveals the advantage brought by the GPS device. The older man with diabetes who needs to keep his blood pressure down used to be afraid to go outside, as he could get lost. Today, with the GPS device the older man is no longer afraid to go for a walk. He explains, *"I feel more safe because I know that the Lommy will help me to get in contact with my family and then, they can find me if I get lost".* Neighbours and relatives reveal that they have even observed that the older man is going regularly for walks

**Intelligent door alarms prevent the elderly people from leaving their homes unnoticeably in the winter time -** In Finland, a relative explains the significance of ISISEMD Intelligent door alarm and the impact it has done on both her mother and her self. The relative explains that now she can prevent her mother from going out of the house and getting lost in the cold winter in Finland. The Intelligent front door service sends an alarm message to a relative informing that the door opens and thus, the older person might be leaving the house. According to this, a relative has stated: *"In Finland, this service is very important so that we can be informed if our relatives get lost. It is so cold in the winter and this can be dangerous if the elderly* 

**Bed sensor helps the elderly to alert their families automatically when help is needed during the night -** In Finland, a relative expresses the importance of having a bed sensor installed. The relative explains that her mother has had several incidents of falling down from her bed during the night. The bed sensor sends a signal when an elderly is out of bed for too long, thus a relative or a caregiver will receive an SMS with the alert. In this particular case, the relative explained… "*When I received this message, I went to my mother's house and found her on the floor. I am very happy to have received this SMS so that I could help my* 

**ISISEMD Calendar and To Do List, shown on the ISISEMD Carebox, helps the elderly with mild dementia to live more independently -** With the ISISEMD Carebox the elderly can keep track of the day and time. For instance, in North Ireland, a relative living with her mother states: *"Even if we only have had the Carebox for 10 days, both my mother and I already feel a difference. Normally my mother always calls me several times at work every day being anxious about not knowing the day and time. Now, with the Carebox, my mother is always informed about the* 

**ISISEMD services are beneficial for enhancing the quality of life of the elderly and their relatives -** In Greece relatives taking care of the elderly explain that the services help them to feel safer knowing that their parents have the ISISEMD services installed in their homes. Hence, relatives are aware if the elderly leaves their houses or if they are out of bed for a long time. One relative explains … *"Before having the ISISEMD services we had to travel 30 minutes outside of town every day to check on our relative. Now we feel safer knowing that our* 

and that they can see his position on a map if he needs help or gets lost.

the elderly presses the help button.

*goes outside for a longer period of time".* 

*mother, even if it was in the middle of the night".* 

*day and time and this has helped her to become more confident*. "

ISISEMD platform with innovative intelligent services is scalable, open-system, validated in real-live environment. The competitive advantage of ISISEMD systems is that it is based on a modular scalable open platform that advances the State-Of-the-Art in the areas of systems for Ambient Assisted Living and can be integrated with other Health or Care Systems. Additional services can be included anytime, resulting in a very powerful platform system.

Last but not least, ISISEMD services have been extensively tested and validated in a pilot operation with 142 end-users (71 elderly and 71 relatives) for 15 months in four regions - in Denmark, Finland, Greece and UK. The final user evaluations, carried out in June 2011, showed high level of user acceptance and satisfaction with the services and willingness to use them after the project ends.

We knew that sceptical users are stoppers against introduction of new technologies. But our experience shows that the elderly and their relatives accept the technology and can see the opportunities for positive impact and added value from the use of the services in their everyday life even when the older adults with mild dementia and their family care-givers were sceptical in the beginning, after giving them time to get used to the technology. It can be expected, that after about one month, the elderly and their family caregivers can get used to the services. The most successful adoption of the services can happen when they are offered as early as possible in the disease. In this way, the technology services can be integrated in the coping and care strategies in the family and elderly has highest chances to learn to refer to the services.

The benefits from using ISISEMD services were depicted in the user acceptance surveys, contributing to the Quality of Life of the end-users. The services improved the elderly ability for self-care by support for their basic daily activities in way that prevents health risks in their homes and promotes independence. They strengthened the daily interaction with their social sphere - partners and relatives, giving them the feeling of safety and improving their relationships. For the family care-givers, the services increased their quality of life and the feeling of safety; reduced their care burden and gave them a piece of mind while saving them time and money.

Potential for direct savings from the usage of the ICT services is foreseen for both the careprovider organisations and informal care-givers. Costs can be saved from time and travel expenses for delivery of warm meals to elderly, performing on-line sessions with use of video-call service instead of physical home-safety visits, visits for administering medications, automatically registering information for such services, etc. The informal caregivers can save time and money for making telephone calls to elderly to check about status or to drive to elderly place or to cook meals for them. The highest potential for savings is at a society level - with delaying the admittance in nursing or dementia care institutions where savings at European level can range from approx. 1300-4000 euro for one person for 12 months.

#### **11. Acknowledgment**

The work presented in this paper is partially funded by the ICT PSP EU project "ISISEMD - Intelligent system for independent living and self-care of seniors with cognitive problems or mild dementia." Contract Number: ICT-PSP-2-238914, www.isisemd.eu. The authors would like to acknowledge also the contributions of the whole consortium.

#### **12. Disclaimer**

This paper reflects only the views of the authors and the European Commission is not liable for any use that might be made of the information contained therein.

#### **13. References**


ALZ http://www.alz.org/national/documents/summary\_alzfactsfigures2009.pdf

BPEL http://www.oasis-open.org/committees/tc\_home.php?wg\_abbrev=wsbpel WFMC http://www.wfmc.org/

XPDL http://www.wfmc.org/standards/xpdl.htm

The work presented in this paper is partially funded by the ICT PSP EU project "ISISEMD - Intelligent system for independent living and self-care of seniors with cognitive problems or mild dementia." Contract Number: ICT-PSP-2-238914, www.isisemd.eu. The authors would

This paper reflects only the views of the authors and the European Commission is not liable

Aalst, W.; Hofstede, A.; Kiepuszewski, B.; and Barros, A. (2003). "Workflow Patterns

ISISEMD Intelligent System for Independent Living and Self-Care of Seniors with Cognitive

Mitseva, A.; Kyriazakos, S.; Litke, A.; Papadakis, N.; Prasad, N. (2009). *ISISEMD: Intelligent* 

Mitseva, A.; Peterson, C.; Dafoulas, G.; Efthymiou, A.; Abildgaard, A.; Bellini, S. (2010).

ALZ http://www.alz.org/national/documents/summary\_alzfactsfigures2009.pdf BPEL http://www.oasis-open.org/committees/tc\_home.php?wg\_abbrev=wsbpel

Problems or Mild Dementia. (2009) http://www.isisemd.eu/. 2009-2012,

*System for Independent living and self-care of Seniors with mild cognitive impairment or Mild Dementia,* © The Journal on Information Technology in Healthcare, Volume 7

*ISISEMD evaluation framework for impact assessment of ICT pilot services for elderly with mild dementia, living in the community and their relatives.*, Proceedings of NAEC 2010, pp. 123-148, ISBN 9780-0-9820958-3-6, Riva del Garda, Italy, October 7-10, 2010

like to acknowledge also the contributions of the whole consortium.

for any use that might be made of the information contained therein.

Distributed and Parallel Databases, 14(1):551, 2003

Issue 6 2009, pp. 383–399, ISSN 1479-649X

XPDL http://www.wfmc.org/standards/xpdl.htm

**11. Acknowledgment** 

**12. Disclaimer** 

**13. References** 

www.isisemd.eu

WFMC http://www.wfmc.org/

### *Edited by Vladimir Mikhailovich Koleshko*

This book is dedicated to intelligent systems of broad-spectrum application, such as personal and social biosafety or use of intelligent sensory micro-nanosystems such as "e-nose", "e-tongue" and "e-eye". In addition to that, effective acquiring information, knowledge management and improved knowledge transfer in any media, as well as modeling its information content using meta-and hyper heuristics and semantic reasoning all benefit from the systems covered in this book. Intelligent systems can also be applied in education and generating the intelligent distributed eLearning architecture, as well as in a large number of technical fields, such as industrial design, manufacturing and utilization, e.g., in precision agriculture, cartography, electric power distribution systems, intelligent building management systems, drilling operations etc. Furthermore, decision making using fuzzy logic models, computational recognition of comprehension uncertainty and the joint synthesis of goals and means of intelligent behavior biosystems, as well as diagnostic and human support in the healthcare environment have also been made easier.

Intelligent Systems

Intelligent Systems

*Edited by Vladimir Mikhailovich Koleshko*

Photo by solarseven / iStock