Multi-Agent Mobile Robot Navigation through Genetic-Fuzzy Algorithm

The present paper deals with the navigation of multiple wheeled robots working in a common dynamic environment in a decentralized manner. Two different motion planning approaches have been proposed to solve the said problem. In Approach 1, automatic design of a Mamdani-type Fuzzy Logic Controller (FLC) using a binary-coded Genetic Algorithm (GA) has been considered. On the other hand, a potential field- based motion planner has been developed in the next approach. Since, the robots are agents and all the time competition is not good, a strategic approach has been proposed to solve the conflicts related to coordination among agents. Performance of the developed approaches have been tested through computer simulations. Computational complexity of both the approaches are also compared to see the feasibility of their on-line implementations. It has been observed that proposed coordination strategy along with the developed motion planners are found to generate movement of the robots in a human-like manner.