Gas-Mimetic Swarms for Surveillance and Obstacle Avoidance

Abstract

In this chapter, we explore a swarm algorithm for a challenging coverage task in which there is no prior knowledge of the environment and there are too few robots to monitor the environment continuously without movement. We propose a physicomimetic solution to the problem inspired by fluid physics. We show that by modeling each robot in the swarm as a particle in a kinetic theory simulation we are able to accurately predict the aggregate behavior of the swarm and empirically demonstrate that the swarm performs well on the surveillance task, providing coverage comparable to the best biomimetic approach while requiring only one-third the time.