**4.1 The perception model**

One of the ultimate goals of robotics is to realise autonomous robots that are able to organise their own internal structure and to achieve their goals through interactions with dynamically changing environments. The robot should take decisions on-line using only sensors providing limited information (without prior information about the position of obstacles and their trajectories) when it navigates autonomously in dynamic spaces.

Our main objective is to make the robot able to predict the nature and the velocity of obstacles (static or moving) in order to minimise conflict situations, to avoid collision and to achieve a specific goal. At this stage, we use a perception model composed of eight ultrasonic sensors. We choose this type of sensors because it has some attractive properties, e.g. cheapness, reliability and soon, which makes it widely used in mobile robots. This type of sensor calculates the distance between the robots.

According to this model of perception, we divide the robot's space into three subspaces controlled by three groups of sensors (see figure 1). The subspaces are: (1) the front area "F" controlled by the first group composed of sensors number 1,2,7,8, (2) the left area "L" controlled by the second group composed of sensors number 3, 4, and (3) the right area "R" controlled by the third group composed of sensors number 5, 6.

Fig. 1. The Robot's perception areas.
