**3.1. Pattern recognition technique**

Please replace Fig.6 from given figure below. In the previous figure the spelling for the

Noise rejection/filtering

Mean absolute value (MAV) Root mean square (RMS) Wave form lenght (WL) Zero crossing (ZC) Slope sign changes (SSC)

Auto regressive coefficients(ARC) Frequency median (FMD)

Modified frequency median (MFMD)

Short time fourier transform (STFT)

Wavelet packet transform (WPT)

Frequency mean (FMN)

Wavelet transform (WT)

Frequency ratio (FR)

Amplifications

Invasive method: use of needle electrode

Non invasive method: use of surface electrodes

Disjoint segmentation

Overlapping segmentation

Signal conditioning

Time domain

> Frequency domain

Time frequency domain

Neural networks (NN) Bayesian classifier (BC) Fuzzy logic (FL)

Neuro-Fuzzy

Linear discreminant analysis (LDA) Support vector machine (SVM) Hidden markov models (HMM)

EMG processing in pattern recognition based control [5] and non-pattern recognition based

'Amplification' is wrong.

EMG data acquisition

242 Electrodiagnosis in New Frontiers of Clinical Research

EMG data segmentation

Feature extraction

Classification

**Figure 6.** Steps of pattern recognition based EMG processing

control are further illustrated in the next subsections.

3

The different steps coming under pattern recognition such as data acquisition, data segmen‐ tation, feature extraction and classification can be further broken down into sub areas consid‐ ering available options as shown in Fig. 6. Next subsections present further categorization of data segmentation, feature selection and classification respectively coming under the pattern recognition based EMG based control method.

**Data segmentation**: The EMG signal has two states: transient and steady. In the transient state, the muscle goes from rest to a voluntary contraction level [5]. Constant contraction of the muscle can be seen under the steady state. In addition the EMG signal in the transient state shows a large deviation of error compared to the steady state level. Therefore, in many cases, the steady state signal is used for the analysis of EMG. For the better result of data segmenta‐ tion, the selected time slot should be equal to or less than 300ms. This includes segment length and processing time to generate the control command. In addition bias and variance of features can be minimized by selecting adequately a large time slot and it contributes to better classi‐ fication performance [5, 12, 19]. Data segmentation is carried out with two major techniques: overlapping segmentation [5, 25], and disjoint segmentation which uses segments with predetermine length for feature extraction. Also processing time is a small portion of segment length and thus processor is idle for remaining time of the segment. The new segment slides over the current segment and has small incremental time for overlapping segmentation technique. According to [25], overlapping segment method increases processing time and hence better for the data segmentation [5].

**Feature selection**: Due to the large number of inputs and randomness of the signal, it is impractical to feed the EMG signal directly to the classifier [6, 18]. Therefore, it is necessary to create the feature vector, where sequence is mapped into a smaller dimension vector. Success of any pattern recognition problem depends almost entirely on the selection and extraction of features. According to the literature, features fall into one of three categories: time domain, frequency domain and time scale domain [25]. Many assistive robots use time domain analysis for feature extraction and in most cases, RMS calculation is adapted for feature extraction [6, 35]. Assistive robots based on frequency domain and time frequency domain were scarce.

**Classification**: Extracted features need to be classified into distinctive classes for the recogni‐ tion of the desired motion pattern [5]. Several external factors, such as fatigue, electrode position, perspiration and posture of the limb may causes changes in the EMG pattern over time. Therefore, this leads to large variations in the value of a particular feature. The important feature of the classifier is the ability to identify the unique feature throughout the varying pattern due to other influences. The speed of the classifier is an important aspect for generating required output from the controller. Further, training of the classifier is another way to improve the response of the control system of the assistive robot. Depending on the performance of the subject, practice required for rehabilitation etc can be customized through a training of the classifier and it produces the expected rehabilitation training for the patient too. According to the literature, several methods are used for EMG classifications. Some of them are Neural network [5, 35], Fuzzy logic [25], Neuro-fuzzy logic [5, 25], Probabilistic approach etc.

'Amplification' is wrong.

Lot of errors can be found in the reference section which were not in the final manuscript **Figure 7.** Steps of non-pattern recognition based EMG processing

#### **3.2. Non Pattern recognition technique**

Oct; 2(4):275–94.

Ageing 1950-2050, www.un.org/esa/population/ publication ns/worldageing19502050/ (accessed 20 April 2012). [2] Population Division, DESA, United Nations, Demographic Determinants of Population Ageing, World Population Ageing 1950-2050, www.un.org/esa/ population/ publications/worldageing19502050/ (accessed 20 April 2012). [3] Mosher RS. Handyman to Hardiman [Internet]. Warrendale, PA: SAE International; 1967 Feb. Report No.: 670088. Available from: http://papers.sae.org/670088/ [4] Cloud W. Man Amplifiers: Machines that Let You Carry a Ton. Popular Science, vol. 187, no. 5, p70–73, 1965. [5] Oskoei MA, Hu H. Myoelectric control systems—A survey. Biomedical Signal Processing and Control. 2007 In the non-pattern recognition based method accuracy is not as high as the pattern recognition based method. The two main steps coming under non-pattern recognition based method can further be broken down into sub areas considering available options as shown in the Fig. 7. Typically, non-pattern recognition based method includes proportional control, threshold control etc. This is a simple structure and in most cases it uses ON/OFF control [5]. During the review, control methods of assistive robots based on non-pattern recognition based control were hardly found. This may be due to poor accuracy, low level of response, etc.

submitted. Therefore, please include the references as given below.

[1] Population Division, DESA, United Nations, Magnitude and Speed of Population Ageing, World Population

[6] Gopura RARC, Kiguchi K, Li Y. SUEFUL-7: A 7DOF upper-limb exoskeleton robot with muscle-modeloriented EMG-based control. IEEE/RSJ International Conference on Intelligent Robots and Systems, 2009.

[7] Kiguchi K, Hayashi Y, Asami T. An upper-limb power-assist robot with tremor suppression control. IEEE Int

[13] Schultz AE, Kuiken TA. Neural interfaces for control of upper limb prostheses: the state of the art and future

[14] Silva J, Heim W, Chau T. A Self-Contained, Mechanomyography-Driven Externally Powered Prosthesis.

[16] Fukuda O, Tsuji T, Ohtsuka A, Kaneko M. EMG-based human-robot interface for rehabilitation aid. IEEE International Conference on Robotics and Automation, 1998. Proceedings. May. page 3492–3497 vol.4.

#### IROS 2009. Oct. page 1126–31. **4. Review of EMG based control method of assistive robots**

Conf Rehabil Robot. 2011; 2011:5975390.

possibilities. PM R. 2011 Jan; 3(1):55–67.

[8] Merrill DR, Lockhart J, Troyk PR, Weir RF, Hankin DL. Development of an Implantable Myoelectric Sensor for Advanced Prosthesis Control. Artificial Organs. 2011;35(3):249–52. [9] Anon, Robots - The Big Picture - Boston.com, http://www.boston.com /bigpicture/ 2009/03/robots.html (accessed September 28, 2012). [10] Yanco HA. A Robotic Wheelchair System: Indoor Navigation and User Interface. Lecture notes in Artificial Intelligence: Assistive Technology and Artificial Intelligence. Springer-Verlag; 1998. page 256–68. [11] Hudgins B, Parker P, Scott RN. A new strategy for multifunction myoelectric control. IEEE Transactions on Biomedical Engineering. Jan.; 40(1):82–94. [12] Farina D, Merletti R. Comparison of algorithms for estimation of EMG variables during voluntary isometric This section presents a review of EMG based control methods of upper-limb exoskeleton robots and prostheses. For this review, several databases including IEEE explorer, Science direct and Google scholar were used. In total, more than forty five numbers of conferences and journal papers are included and reviewed. Further, EMG based control methods of upper-limb exoskeleton robots and prostheses are compared considering their country of origin, input signals, structure of the controller and special features. Table 1 shows the comparison of EMG based control method of assistive robots.

contractions. Journal of Electromyography and Kinesiology. 2000 Oct; 10(5):337–49.

[15] Christoph Guger WH. Prosthetic Control by an EEG-based Brain- Computer Interface (BCI): 2–7.

Archives of Physical Medicine and Rehabilitation. 2005 Oct; 86(10):2066–70.

4

**Reference Country of**

Hand exoskeleton

Neuro-fuzzy exoskeleton robot

Exoskeleton hand

[27]

[18]

[19]

Hand rehabilitation robot [24]

SUEFUL-7 Upper limb exoskeleton [6]

Proportional EMG control upper limb exoskeleton [27]

Finger Prosthesis

EMG based Robotic Hand [36]

[46]

**origin**

Japan EMG and

Japan EMG and

W-EXOS [35] Japan EMG and

Saga Arm [22] Japan EMG and

DEKA arm [21] USA EMG/

force signal

force sensor signal

force signal

Kinematic signal

pressure signals

**Table 1.** Comparison of EMG based control method of assistive robots

**Input signal Structure of controller Special features**

Developed for hand rehabilitation. Adding force sensors will increase the complexity of blind source separation. increased when combined force

http://dx.doi.org/10.5772/56174

245

Mean absolute value (MAV) of EMG signal used to controller. Force sensor signals are used to control robot at low

Suitable control algorithm is needed to provide accepted level of rehabilitation

Motor speed varies according to joint

Joint torques and weights of musclemodel matrix are changed according to

Exoskeleton provides extra torque when it works at high proportional gains. This provides better assistance according to level of EMG activity

Allows flexible and natural motion to user according to motion intention.

Transhumeral Prosthetic Arm developed for predefined daily

for the 3-bit input signal of EMG

An improved control algorithm has

joint angles can be calculated for

Modular based controlling could be

activities

being proposed

controlling purposes

realized

angle variation of fingers

posture changes.

sensors

Recent Trends in EMG-Based Control Methods for Assistive Robots

muscle activity

recover the original EMG signal

Multiple neuro-fuzzy controllers are used to take the effect of posture changes during motion

microcontroller via serial connection

Neuro-fuzzy controller with muscle model oriented control method. Impedance controller is applied to change the impedance parameters

Proportional control gains, Kbic and Ktric are used to determine the

EMG based fuzzy-neuro controller. Fuzzy rules are used to determine torque required according to motion

EMG based fuzzy controller and a kinematic based controller

Japan EMG Based on ANN classification Together with motion capturing data,

Uses TMR controlling with pressure

Manus Hand [23] Spain EMG Based on EMG signal of three levels A command language was developed

LIBSVM and kNN

signals from the foot

Germany EMG Uses blind source separation to

USA EMG Uses binary, variable and reaching

Italy EMG EMG based proportional controller.

output

intention

AUS EMG Post-processing Classification using

Italy EMG Input signals are fed to the

control algorithms


**Table 1.** Comparison of EMG based control method of assistive robots

4

Please replace Fig.7 from given figure below. In the previous figure the spelling for the

Noise rejection/filtering

Amplification

Invasive method: use of needle electrode

Non invasive method: use of surface electrodes

Disjoint segmentation

Overlapping segmentation

Lot of errors can be found in the reference section which were not in the final manuscript

[1] Population Division, DESA, United Nations, Magnitude and Speed of Population Ageing, World Population Ageing 1950-2050, www.un.org/esa/population/ publication ns/worldageing19502050/ (accessed 20 April

[2] Population Division, DESA, United Nations, Demographic Determinants of Population Ageing, World Population Ageing 1950-2050, www.un.org/esa/ population/ publications/worldageing19502050/

[3] Mosher RS. Handyman to Hardiman [Internet]. Warrendale, PA: SAE International; 1967 Feb. Report No.:

[4] Cloud W. Man Amplifiers: Machines that Let You Carry a Ton. Popular Science, vol. 187, no. 5, p70–73, 1965. [5] Oskoei MA, Hu H. Myoelectric control systems—A survey. Biomedical Signal Processing and Control. 2007

[6] Gopura RARC, Kiguchi K, Li Y. SUEFUL-7: A 7DOF upper-limb exoskeleton robot with muscle-modeloriented EMG-based control. IEEE/RSJ International Conference on Intelligent Robots and Systems, 2009.

[7] Kiguchi K, Hayashi Y, Asami T. An upper-limb power-assist robot with tremor suppression control. IEEE Int

[8] Merrill DR, Lockhart J, Troyk PR, Weir RF, Hankin DL. Development of an Implantable Myoelectric Sensor

[9] Anon, Robots - The Big Picture - Boston.com, http://www.boston.com /bigpicture/ 2009/03/robots.html

[10] Yanco HA. A Robotic Wheelchair System: Indoor Navigation and User Interface. Lecture notes in Artificial Intelligence: Assistive Technology and Artificial Intelligence. Springer-Verlag; 1998. page 256–68. [11] Hudgins B, Parker P, Scott RN. A new strategy for multifunction myoelectric control. IEEE Transactions on

[12] Farina D, Merletti R. Comparison of algorithms for estimation of EMG variables during voluntary isometric

[13] Schultz AE, Kuiken TA. Neural interfaces for control of upper limb prostheses: the state of the art and future

[14] Silva J, Heim W, Chau T. A Self-Contained, Mechanomyography-Driven Externally Powered Prosthesis.

[16] Fukuda O, Tsuji T, Ohtsuka A, Kaneko M. EMG-based human-robot interface for rehabilitation aid. IEEE International Conference on Robotics and Automation, 1998. Proceedings. May. page 3492–3497 vol.4.

Signal conditioning

submitted. Therefore, please include the references as given below.

In the non-pattern recognition based method accuracy is not as high as the pattern recognition based method. The two main steps coming under non-pattern recognition based method can further be broken down into sub areas considering available options as shown in the Fig. 7. Typically, non-pattern recognition based method includes proportional control, threshold control etc. This is a simple structure and in most cases it uses ON/OFF control [5]. During the review, control methods of assistive robots based on non-pattern recognition based control

for Advanced Prosthesis Control. Artificial Organs. 2011;35(3):249–52.

This section presents a review of EMG based control methods of upper-limb exoskeleton robots and prostheses. For this review, several databases including IEEE explorer, Science direct and Google scholar were used. In total, more than forty five numbers of conferences and journal papers are included and reviewed. Further, EMG based control methods of upper-limb exoskeleton robots and prostheses are compared considering their country of origin, input signals, structure of the controller and special features. Table 1 shows the comparison of EMG

contractions. Journal of Electromyography and Kinesiology. 2000 Oct; 10(5):337–49.

[15] Christoph Guger WH. Prosthetic Control by an EEG-based Brain- Computer Interface (BCI): 2–7.

Archives of Physical Medicine and Rehabilitation. 2005 Oct; 86(10):2066–70.

670088. Available from: http://papers.sae.org/670088/

**4. Review of EMG based control method of assistive robots**

were hardly found. This may be due to poor accuracy, low level of response, etc.

'Amplification' is wrong.

EMG data segmentation

Non Pattern Recognition

**Figure 7.** Steps of non-pattern recognition based EMG processing

(accessed 20 April 2012).

IROS 2009. Oct. page 1126–31.

(accessed September 28, 2012).

Conf Rehabil Robot. 2011; 2011:5975390.

Biomedical Engineering. Jan.; 40(1):82–94.

possibilities. PM R. 2011 Jan; 3(1):55–67.

Oct; 2(4):275–94.

based control method of assistive robots.

**3.2. Non Pattern recognition technique**

2012).

EMG data acquisition

244 Electrodiagnosis in New Frontiers of Clinical Research

A considerable numbers of literature of EMG based control methods show the usage of different methods for EMG processing. This may give a number of options to researchers to conduct experiments for EMG processing in different angles. Most of EMG based assistive robots use surface EMG electrodes for EMG signal detection and few robots use needle electrodes [8, 13]. It was found that all methods of EMG processing belong to one of three main categories: time domain, frequency domain and time-frequency domain [5]. Another impor‐ tant aspect of an EMG based control method is signal classification. Generally, accuracy of EMG based control method highly depends on method of classification and which helps to identify muscles to generate the required output from the EMG based control method [18]. Different robots use different techniques for signal classification and many of them are based on neuro-fuzzy, fuzzy logic and neural network. All assistive robots considered for this review used an EMG signal as its main input signal and the architecture of the controller varies from one type to other. Some of them are based on proportional control and others use advanced control methods such as PID control. In another way, controllers can be classified based on its model as dynamic model [25], muscle model or other method. EMG based control methods of upper-limb exoskeleton robots and prostheses are respectively presented and reviewed in the next subsections. The authors make every possible effort to include all recent EMG based control methods of assistive robots in the next section. The logic for selecting particular control method is its key features and novelty.
