Collaborative search and tracking based on dynamic map and information fusion for UAV swarm

To improve the accuracy and stability of multi-target tracking tasks for UAV swarm, a UAV scheduling method based on multi-agent reinforcement learning is proposed. In the proposed method, a multi-agent actor-critic with attention (MAAC) algorithm is proposed, which is used to optimize scheduling strategies effectively. By incorporating dynamic mapping and information fusion mechanisms, the UAV swarm can continuously update target histories, environmental search statuses, and UAV positions, to better perceive and adapt to the environment. Simulation results demonstrate that the proposed method outperforms traditional methods in tracking accuracy, response delay, and collision rate, especially in dynamic and communication-limited environments.