Multi Robot System based on Boundary Value Problems

Abstract

In this paper, we introduce a modification in the Potential Field Method based on Boundary Value Problems to control multiple robots. In this new method, that we call Locally Oriented Potential Field (LOPF), the robots share a single map, but the field is adjusted locally for each robot using a bias vector. We show that the LOPF method is very appropriated to control robot soccer teams, in which each robot needs to adapt independently to the environment while aiming to perform the same task. Simulation results are presented to illustrate the richness of behaviors allowed by the technique.