The collision avoidance planning in multi-robot system by genetic fuzzy control algorithm

Abstract

A collision-avoidance planning method in multi-robot system based on genetic algorithm optimized by fuzzy logic control is designed, which include a simplified three-tier structure: avoid robot, avoid static obstacles and moving to the goal. These actions reason independently, and take information from different sensors as inputs; all the outputs are next anticipant movement of robot. Then, it synthesizes the outputs of three actions based on the priority and weight. Subsequently, GA is used to optimize the width and central value of membership functions. Through the off-line self-optimization of fuzzy controller, a group of optimum parameters is gotten. The simulation results show that genetic algorithm improves the navigation performance of robot.