Multi-Stage Multi-Robot Cooperative Hunting with Dynamic Alliance

This paper examines the multi-robot cooperative hunting problem with dynamic alliances. Based on the contract network protocol, an efficient dynamic task allocation approach is used for optimal online allocation when new evaders arrive during the mission and resolve local conflicts. Then, a multi-stage cooperative hunting strategy based on the Apollonius circle is introduced, dividing the cooperative hunting problem into three stages, namely, a self-organized wolf-pack-particle-based strategy for pursuing and forming a preliminary encirclement at a long distance, an improved potential point strategy for smoother terminal encirclement at short distances, and an interception strategy based on the Apollonius circle after encirclement. By integrating the advantages of various strategies, the proposed multi-stage strategy enhances the efficiency of the hunting task. Both simulation and experiment results are provided to demon-strate the effectiveness of the proposed approach.