Modular robot path planning using genetic algorithm based on gene pool

Abstract

As a new generation of robotics, a modular robot is flexible enough to achieve self-replication by attaching a new modular, or perform self-assembly by transferring into different shapes. However, the path planning for modular robots, the fundamental function is seldom studied until now. In this paper, we improve the path schedule method of Molecubes, by designing a gene pool, to speed the convergence and avoid the uncertain of the original genetic algorithm (GA). Experiments show that the gene-pool based GA outperforms the old one in both success rate and speed in planning the long path. © 2010 Springer-Verlag Berlin Heidelberg.