Controlling a Mobile Robot Herd: Theory and Practice

Abstract

This paper discusses the difficulties of analyzing and synthesizing intelligent collective behaviors for a collection of mobile robots. It reviews available approaches and presents a bottom-up experimental alternative consisting of a set of robust global behaviors based on simple local rules. These basic behaviors are designed to be sufficient and additive, in order to be combined into more complex collective task achieving behaviors. The paper also describes the role of interaction in a multi-agent environment, and ways of analyzing and predicting it. Finally, we present the experimental set up, consisting of twenty physical mobile robots, on which the interference studies were performed and the basic behavior repertoire was developed and tested.