An optimization approach to the multi-player pursuit-evasion problem

Abstract: In this paper a scenario of one evader being chased by multiple pursuers in two specific simulation environments is studied. The simulation environments are divided into an open area without obstacles and a closed area with obstacles. In the open area a fairly accurate system of dynamics are implemented for both pursuers and evader. The Virtual Vehicle Approach is used to provide a reference trajectory for the pursuers to follow in order to catch the evader. The main purpose of this thesis is to find a decentralized robust control method for the dynamics of the pursuers. In the closed area, the line of sight and field of view are introduced and the solution to the Minimum time UGV surveillance problem and the Centroidal Voronoi partitions. Different capturing strategies, encirclement and one-on-one chase, are both studied and compared. The numerical implementation and the resulting simulation are presented and analyzed. Conclusion on the optimal formation for the multiple pursuers is made.