Research on Multi-Agent Region Coverage Search Based on Multi-Task Reinforcement Learning
Gaofeng Deng, Bo Wang, Xiao He · 2025
Cluster search and area coverage technologies are widely applied in both military and civilian fields. This paper divides the task area into grids and studies the multi-agent coverage search algorithm in unknown environments based on multi-task reinforcement learning. The main focus is on accelerating the training of existing schemes on the basis of the multi-agent area coverage search algorithm based on reinforcement learning. A multi-task reinforcement learning training method based on a hard parameter sharing network model is proposed. The area coverage search problem is decomposed into coverage, search, and obstacle avoidance problems, and corresponding network update strategies are designed. Through comparative experiments, it is verified that the multi-task reinforcement learning training method can effectively improve the convergence speed of the algorithm and increase the robustness of the model.