Fitness Switching: Evolving Complex Group Behaviors Using Genetic Programming

This paper considers the problem of transporting a large table using multiple robotic agents. The problem requires at least two group behaviors of homing and herding which are to be coordinated in proper sequence. Existing GP methods for multiagent learning are not practical enough to find an optimal solution in this domain. To evolve this kind of complex cooperative behavior we present a novel method called fitness switching. This method maintains a pool of basis fitness functions each of which corresponds to a primitive group behavior. The basis functions are then progressively combined into more complex fitness functions to coevolve more complex behaviors. The performance of the presented method is compared with that of two conventional methods. Experimental results show that coevolutionary fitness switching provides an effective mechanism for evolving complex emergent behaviors which may not be solved by simple genetic programming.