An evolutionary optimization framework for intelligent agents

Evolutionary algorithms have become popular in the recent years as a general, simple, and robust technique that can be used when other optimization methods cannot be applied. Presently, there are a number of evolutionary/genetic algorithm libraries publicly available; however, they are not specifically designed for multiagent systems. The framework described in this paper addresses this problem and presents a general architecture for evolutionary optimization aimed to be used by agents in a multiagent system implemented in the JADE framework inspired from the island model of parallel genetic algorithms. The characteristics of evolutionary algorithms that are more appropriate for use in a distributed, cooperative multiagent environment are also identified.