Memetic Algorithms in Evolutionary Robotics on Example of Virtual Bots

Abstract

Many crucial developments in the fast-growing area of evolutionary robotics take place not for real but for virtual robots. It is substantially easier to implement effective evolutionary search on a computer than on real hardware level. In any case the evolution of virtual robots needs highly effective evolutionary algorithms. One of the reasonable choices is to use coevolutionary (or self-generating) algorithms. In this communication we demonstrate and discuss the efficacy of our approach to evolutionary computations called the Mobile Genetic Elements approach in comparison with other related techniques. We discuss efficacy of the approach for several known tests in the area of animats and virtual robots.