On the Feasibility of Mass-Spring-Relaxation for Simple Self-Deployment

Abstract

Self-deployment describes the task of spreading an autonomously moving swarm of mobile robots over a given area. All these robots have to move to locations such that the set of robot locations satisfies a desired property. In this work, we describe a fully distributed deployment algorithm executed locally at each robot. The approach requires only few local information per node: the distances and very coarse angular information to immediate neighbors. It has been developed for use on small robots with very restricted memory, communication, and processing capabilities. In this paper, we specify the algorithm and evaluate it in an empirical study. This includes both simulation studies and real test bed experiments. For the test bed, we consider two different platforms: ground moving robots and aerial robots. The results of our simulations show that our local deployment rules achieve almost globally optimal results. The test bed study supports and substantiates our simulation study and shows as a proof of concept that our algorithm works both with real ground based and aerial based robot swarms.