A Two-Stage Collaborative Hunting Algorithm for Multi-USV System with Dynamic Target Evasion

Aiming at the problem that the target escape strategy is unknown in the process of multi-USV cooperative hunting and the speed ratio between USVs and the target is uncertain, this paper establish the trajectory prediction model of the target for hunting pre-decision and proposes a multi-USV cooperative hunting algorithm based on double-layer switching strategy. The first layer of the hunting strategy optimize potential points allocation by Hungarian algorithm and plan collision-free encirclement path of USVs by APF algorithm to improve the efficiency of the USV to track the target. The second-layer hunting strategy considers the situation that USVs move as slow as the target, realizes dynamic formation control by establishing the behavior model to improve the hunting success rate. The first and second strategy can be converted to each other. The results of digital simulation shows that our algorithm shortens the capture time and improves the hunting success rate compared with the traditional potential point method and the equal-surround control method.