A multi-agent cooperative encirclement strategy integrating dynamic target-point allocation and PPO control

To address the challenge of cooperative encirclement of a maneuvering target by multiple agents in cluttered environments, this paper proposes a distributed control framework that integrates a dynamic target-point assignment mechanism with Proximal Policy Optimization (PPO). The method first constructs an ideal enclosing circle around the target and dynamically allocates encirclement points in real time according to factors such as relative positions and heading offsets between the target and the agents. A PPO-based reinforcement learning controller is then employed to generate control policies that guide the agents to their assigned positions while ensuring obstacle avoidance. Simulation results in multi-obstacle environments demonstrate that the proposed approach achieves efficient, robust, and stable encirclement across varying numbers of agents.