Distributed Target Tracking in Cluttered Environments with Guaranteed Collision Avoidance

We propose a distributed, online algorithm for a group of pursuer agents to track a target through a cluttered environment. The pursuer agents must avoid collision with the obstacles at all times. We introduce the Obstacle-Aware Voronoi Cell (OAVC), a modified Voronoi tessellation that dynamically weights the boundaries between agents and obstacles such that an agent's OAVC is tangent but never intersecting the obstacle. The agents plan their control actions within their OAVC, guaranteeing collision avoidance among themselves and other agents. We demonstrate that by using tools from Voronoi-based coverage control, the pursuers successfully track a target given only an estimate of its position. Simulations conducted in Matlab demonstrate the performance of our algorithm.