Mechanical PSO aided by extremum seeking for swarm robots cooperative search

Abstract

This paper addresses the issue of swarm robots cooperative search. A swarm intelligence based algorithm, mechanical Particle Swarm Optimization (PSO), is first conducted which takes into account the robot mechanical properties and guiding the robots searching for a target. In order to avoid the robot localization and to avoid noise due to feedback and measurements, a new scheme which uses Extremum Seeking (ES) to aid mechanical PSO is designed. The ES based method is capable of driving robots to the purposed states generated by mechanical PSO without the necessity of robot localization. By this way, the whole robot swarm approaches the searched target cooperatively. This pilot study is verified by numerical experiments in which different robot sensors are mimicked. © 2013 Springer-Verlag Berlin Heidelberg.