Localization Using Triangulation in Swarms of Autonomous Rescue Robots

A challenge for autonomous rescue robots is determining their position in an unstructured environment. Landmarks are rarely useful since any floor plan for a structure that predates a disaster scenario is likely to have been disrupted by the event that necessitated the rescue response. Navigational aids useful for outdoor navigation, like GPS, are not generally useful in a structure and aids like beacons require too much manual setup. To address this problem, we have been researching methods for localizing rescue robots in a robot swarm, using triangulation between cells of three robots. This paper describes the real-world problem of localization in unstructured environments and some of the approaches taken by other researchers to address the problem. The approach we have been researching is then described in some detail, together with simulation results. There have been three attempts to implement this approach on small rescue robots at Utah State University. These implementations and the lessons learned from them are described. Finally, an implementation that addresses the issues with the earlier implementations is proposed.