The multi-agent cooperative search has garnered increasing attention in recent decades. Within the decentralized framework of multi-UAV cooperation, precise task allocation and assignment are paramount to balancing the workload of each UAV. To achieve reasonable task allocation, we employ spatial segmentation within the search domain, dividing it into distinct partitions. The aim is to allocate equivalent search workloads to individual UAVs across these partitions. To enhance algorithm efficiency, we utilize the Voronoi Diagram as the spatial segmentation generator and integrate deep reinforcement learning to refine the topological structure of these partitions. Finally, to assess the robustness of our proposed algorithm, we conducted experiments under various search scenarios. The results demonstrate significant improvements in the overall search efficiency of the swarm.
|