Evolving mobile robot controllers using evolutionary algorithms

Abstract

Creating autonomous mobile robot control systems using evolutionary algorithms and neural networks is a new approach and has attracted increasing attention of researchers in recent years. It attempts to make robot controllers emerged from the interaction between robot and its environment. The developed controllers adapt to partially unknown or dynamic environments. In this paper, the main technologies used in evolving robot controllers are introduced. The developing modes are assorted based on whether simulation is employed. Research results obtained up to now are summarized. Some key challenges are discussed.