Cooperative Collision Avoidance Strategy for Multi-Agent Based on Dynamic Game
Guangri Li, Hao Wang, Jinghua Wang · 2023
Compared with a single agent, the multi-agent system composed of multiple agents in the network environment by means of communication has the characteristics of high intelligence, perfect function and strong robustness. It is more suitable for handling large and complex engineering tasks, and has great development potential in the fields of building construction, intelligent logistics and emergency disaster relief. Cooperative multi-agent not only has to face a complex static environment, but also has to coordinate the movement of each agent to avoid collision. Therefore, the research of cooperative collision avoidance strategy for multi-agent is an important research direction. In this paper, the idea of dynamic game is introduced into the cooperative motion collision avoidance problem of multi-agent. The collision risk assessment model is designed, and the dynamic collision avoidance game strategy based on collision risk is proposed for multi-agent. This method is suitable for collaborative motion of multi-agent under complex constraints to achieved the maximization of individual targets and system optimization.