Autonomous Swarms of Unmanned Vehicles: Software Control System and Ground Vehicle Testing

Abstract

Unmanned vehicles are being developed for a wide variety of scientific, and military uses. Using a coordinated team allows for greater robustness, flexibility or ease of use. While there are several unmanned vehicle systems in use for a variety of applications, work on swarming has been largely theoretical. This thesis presents Woodstock, a software system implementing a swarming control law. The control law moves the swarm through waypoints specified by GPS coordinates. It was tested using a set of ground vehicles, which are also described. It is shown to be robust to noise, and superior to a previous implementation. The previous implementation was client server, making it vulnerable to a single vehicle loss, while the new system is peer to peer and thus more robust. The system is flexible enough to be portable to other vehicle systems easily.