**6. Conclusion and perspectives**

In this paper, a new method based on the sensors' information to combine reactivity and anticipation in order to predict conflict situations between robots has been proposed. We have used fuzzy logic to develop a control system that enables the robot to navigate at real time and to minimise its interaction with other agents. Basing on the simulation experiments, we can conclude that our method operates in different cases of dynamic environments. The robot can easily predict the nature of obstacles in order to anticipate conflict situations. It can also predict the velocity of each obstacle and adjust its linear velocity in order to avoid collision. As perspectives for this work, we attend to improve the robot's behaviour by introducing both the ability to learn and the concept of communication between robots in order to explore the environment and to share knowledge.
