Coordinative behavior in evolutionary multi-agent system by genetic algorithm

A strategy for motion planning of multiple robots as a multi-agent system is presented. The system has a decentralized configuration. All the robots cannot communicate globally at a time, but some robots can communicate locally and coordinate to avoid competition for a public resource. In such a system, it is difficult for each robot to plan its motion effectively while considering other robots, because the robots cannot predict motions of other robots as an unknown environment. Therefore, each robot only determines its motion selfishly for itself while considering a known environment. In the proposed approach, each robot plans its motion while considering the known environment and using empirical knowledge. The robot considers its unknown environment including other robots in the empirical knowledge. The genetic algorithm is applied to optimization of motion planning of each robot. Through iterations, each robot acquires knowledge empirically, using fuzzy logic. Path planning of multiple mobile robots is discussed, and simulations are performed