Parallel Genetic network prgramming for self-sufficient trash collecting problem

Genetic Network Programming (GNP), one of the evolutionary computations, can generate behavior sequences of agents in order to solve agent-based problems. In this paper, a new method named parallel GNP has been proposed and applied to functionally distributed systems consisted of several tasks. In parallel GNP, GNPs corresponding to several tasks operate separately and independently, but concurrently dealing with the conflicts in task execution. Parallel GNP converges faster and has better fitness results than the conventional GNP, which was confirmed by simulations in dynamic problems.