Distributed Pursuit-Evasion without Global Localization via Local Frontiers

This paper addresses a visibility-based pursuit-evasion problem in which a team of mobile robots with limited sensing and communication capabilities must coordinate to detect any evaders in an unknown, multiply-connected planar environment. Our distributed algorithm can guarantee evader detection and is built around maintaining complete coverage of the frontier between cleared and contaminated regions while expanding the cleared region. We detail a novel distributed method for storing and updating this frontier without building a global map or requiring global localization. We demonstrate the functionality of the algorithm through Player/Stage simulations in realistic environments and through hardware experiments. We also discuss the number of agents required to clear a certain class of environments, as well as compare Monte Carlo results for our algorithm to the theoretical optimum area cleared with a fixed number of robots.