Mobile robot control using combined neural-fuzzy and neural network

This paper describes the concept of the navigation system for a mobile robot. The system is using a combination of two navigation algorithms: self-learning neural network, necessary to form a movement plan for a robot, and a collision-free control algorithm based on heuristic neuro-fuzzy approach. The basic task of neural network is to generate initial path. Heuristic base of rules for collision free algorithm are limited and does not cover all situations. Main contribution of proposed navigation is related to neural network property to supplements special cases that are not covered by present heuristic rules from the data base. Both algorithms are adapted and implemented to navigate real platform of a mobile robot equipped by two independent wheel drives, encoders and a set of short-range sonars. The combined reactive algorithm (using two control methods) is used in real time for obstacle navigation in robotized system. Navigation algorithms are placed into a PC, which is connected to mobile robot by wireless and wired links. Experiments have shown ability of collision-free navigation of mobile robot in real time.