Game confrontation model of USV based on multi-agent reinforcement learning | IEEE Conference Publication | IEEE Xplore