A Hierarchical Optimization Approach to UAV Task Assignment in Large-scale Scenarios

This paper addresses the task allocation problem for unmanned aerial vehicle (UAV) swarms in disaster response scenarios involving large-scale mission demands. A hierarchical optimization framework is proposed, consisting of four stages: task clustering, UAV-to-task cluster matching, intra-cluster task allocation, and dynamic task adjustment. The model incorporates critical constraints, including limited communication range, task-specific requirements, UAV performance characteristics, and load balancing considerations. To solve each stage effectively, a suite of tailored optimization techniques is developed: mixed-integer programming for task clustering, a greedy-simulated annealing algorithm for swarm matching, a hybrid genetic-particle swarm optimization algorithm for task allocation, and an improved contract net protocol for dynamic adjustment. Extensive simulation experiments validate the proposed framework, demonstrating its superior performance in terms of convergence accuracy, robustness, and scalability in large-scale UAV task allocation scenarios.