Joint Clustering and Hierarchical Reinforcement Learning for Complex Task Allocation in Heterogeneous Swarm Systems
Jiaqi Li, Jintong Yu, Yongzhao Hua, Xiwang Dong, Yang Chen, Changchun Zhao, Hao Xie · 2025
In complex missions involving heterogeneous swarm systems, effective task allocation is essential for the coordinated execution of dynamic tasks. Traditional methods, such as combinatorial optimization and heuristic algorithms, face challenges like high computational cost and slow convergence. This paper proposes a joint clustering hierarchical reinforcement learning (HRL) algorithm to address these issues. By combining option-critic HRL with joint clustering, the algorithm optimizes decisionmaking and accelerates convergence. Additionally, integrating an A* algorithm helps alleviate reward sparsity. Simulation results in dynamic environments show significant improvements in task allocation efficiency and adaptability compared to existing methods.