**3.2 Fuzzy logic technique for autonomous navigation**

Fuzzy Logic (FL) has been investigated by several researchers to treat the problem of uncertainty in perception. This technique represents uncertainty via fuzzy sets and membership functions. It gives robustness to the system with respect to inaccuracy in data acquisition and uncertainty, and makes definition behaviour and their interactions quite easy by using simple linguistic rules (Klir et al., 1997; Sajotti et al., 1995). It also allows implementing of human knowledge and experience without requiring a precise analytical model of the environment. Probably, the greatest strength of behaviour-based fuzzy approaches is that they operate on/and reason with uncertain perception-based information, which makes them suitable even for difficult environments.

We present in table 1 comparison results between the characteristics of the Bayesian Networks and the Fuzzy Logic. According to this comparison study, we can conclude that the fuzzy logic technique is faster, easier to implement and well applied to autonomous robot navigation.

that minimise the conflict (Petti & Fraichard, 2005). Nevertheless, building a movement planning requires an important computation time in order to find the most appropriate way. However, if the frequency of environment's changes is high, the use of a planning method becomes not only inefficient, but also useless (if the robot generates plans without using

So, in this situation it is interesting to introduce prediction in reactive schemas in order to consider the environment's evolution in the future. Hence, the main concern of this paper is to propose a model that combines reactivity and anticipation in order to resolve the problem of conflict without using a motion planning. Thus, the robot should anticipate the nature and velocity of the obstacles in order to minimise interactions. We use the multi-agent system to simulate the robot's behaviour because there is a mapping between robots, while

Autonomous navigation in an uncertain environment involves using systems for navigation control that must not be too computationally expensive, as this would result in a sluggish response. In this case, soft computing methods have played important roles in the design of the robot controllers. The commonly used soft computing methods are: Bayesian Networks

Bayesian Networks are models which capture uncertainties in terms of probabilities that can be used to perform reasoning under uncertainty. It epitomises probabilistic dependency models that represent random stochastic uncertainty via its nodes (Darwiche, 2009). The Bayesian inference has been widely used especially in the localisation problem (Thrun, 1998), in the mapping and in the learning mechanisms in mobile robotics. But this technique is very slow and complex (NP-Complex) because it uses a probabilistic reasoning that require much more time to choose a suitable decision. Also, this technique requires a causal model of reality that is not always given. So, its application in autonomous navigation in the

Fuzzy Logic (FL) has been investigated by several researchers to treat the problem of uncertainty in perception. This technique represents uncertainty via fuzzy sets and membership functions. It gives robustness to the system with respect to inaccuracy in data acquisition and uncertainty, and makes definition behaviour and their interactions quite easy by using simple linguistic rules (Klir et al., 1997; Sajotti et al., 1995). It also allows implementing of human knowledge and experience without requiring a precise analytical model of the environment. Probably, the greatest strength of behaviour-based fuzzy approaches is that they operate on/and reason with uncertain perception-based

We present in table 1 comparison results between the characteristics of the Bayesian Networks and the Fuzzy Logic. According to this comparison study, we can conclude that the fuzzy logic technique is faster, easier to implement and well applied to autonomous

navigating in dynamic environments, and multi-agent systems.

**3. Autonomous navigation in uncertain environment** 

case of dynamic environments can be unprofitable.

**3.2 Fuzzy logic technique for autonomous navigation** 

information, which makes them suitable even for difficult environments.

them).

and Fuzzy Logic.

robot navigation.

**3.1 Bayesian networks** 


Table 1. Comparison between fuzzy logic and Bayesian network.

However, the majority of existing fuzzy logic methods deal only with static environments, and only use the distance between the robot and the obstacle to avoid collision and to minimise conflict with other agents sharing the same space (Bonarini et al, 2003; Selekwa et al., 2008). This kind of information gives a limited knowledge about the state of the environment, so the robot cannot take intelligent decisions. For this reason, in our work, we adopt a fuzzy logic technique to deal with uncertainty. We propose a fuzzy model for autonomous navigation in a dynamic and uncertain environment based on the nature and the velocity of obstacles.
