Self-adaptive pattern formation with battery-powered robot swarms

Abstract

This paper presents a distributed, energy-aware algorithm for an autonomous deployment of battery-powered robots in a specified pattern. While each robot gradually discharges and leaves the formation to recharge, the algorithm presented in this paper assures that the formation pattern is preserved. This is achieved by defining the desired pattern as a point cloud where each point is occupied by a robot. The point cloud is transformed into a tree model that is shared among all robots. This model is used by each robot independently to govern its behavior, resulting in a self-adaptive network of robots which automatically generate paths for joining the formation and leaving it to recharge. Robots which leave the formation are replaced by neighbors to preserve the formation pattern. To demonstrate our algorithm we use a physics-based simulator and evaluate the persistence of the pattern formation formed by a robot swarm in an environment without global positioning, using only range and bearing.