This paper investigates a dynamic confrontation problem where a swarm of agents with weak capabilities confronts multiple targets with strong capabilities. In such hostile situations, the agents typically strive to survive by cooperating to offend the targets and defend against potential attacks from the targets. To enhance the capability of such multi-agent systems,
it is necessary to develop an efficient mechanism in which each agent autonomously makes decisions and cooperatively confronts targets. In this paper, a cooperation and decision-making framework is proposed for a multi-agent system confrontation, which is comprised of target allocation, tactical decision-making, and swarm motion control algorithms. First, to address the exponential growth problem of possible behavioral interactions with increasing numbers of agents, a hedonic coalition formation game algorithm is designed for the agents forming
disjoint coalitions corresponding to different targets, i.e., the agents cooperating to confront the targets in the form of coalitions. Then, during the stage of attacking or defending the targets, each agent interacts with coalition members through information exchange. To implement effective cooperative behavior and explainable autonomous decision-making, a fuzzy cognitive map is designed to fuse situational information and obtain decision reference information for each agent’s tactical decision-making. Moreover, to design effective attack/defense tactics and strategies for each agent, the tactical pursuit point method is utilized to develop a tactical
pursuit point for each agent based on the decision reference information. Finally, a swarm motion control algorithm, including decision-oriented and swarm behavior rules, is designed to drive each agent towards the assigned target. Simulation results show the effectiveness of the designed framework and algorithms. In the confrontation, each agent adjusts its strategies according to different situations, occupies advantageous positions, and accomplishes cooperative attack/defense strategies to reduce casualties.