Swarm dispersion via leader election, potential fields, and counting hops.

Abstract

Robot swarm dispersion is an important area of research with many applications including the deployment of sensor networks for military, search and rescue, emergency response, intruder alert, and other applications. In this work, we examine how robots with only laser data and the ability to communicate within a certain radius may disperse in a complex unknown environment. We present a python framework for running and evaluating robot swarms using controllers within the Player/Stage framework. Finally, we demonstrate a modular state-based algorithm for robot dispersion which attempts to maximize coverage while minimizing loss of connectivity during deployment. We demonstrate using the Player/Stage environment that this can in fact reliably achieve over a 100% percent increase in coverage over a variety of different environments.