Finding control knowledge for multiple mobile robots using fuzzy classifier systems

摘要

This paper details our attempts to find control knowledge using Fuzzy Classifier Systems (FCSs). Three mobile robots are equipped with the FCS controllers and each robot with an FCS moves avoiding collision with other robots. This paper also presents new payoffs and credits. The new methods make it possible to fully utilize a powerful feature of the genetic algorithm, namely the effectiveness of crossover operations. Simulations are done to show that each FCS can find fuzzy rules for collision avoidance in a complex, changing environment.