Multi-Agent Soft Actor-Critic-Based Task Allocation for Heterogeneous Multi-UAVs

To address the challenges of capability matching, strong task coupling, and environmental uncertainty in dynamic task allocation for heterogeneous multi-UAV cooperative reconnaissance, this paper proposes a Multi-Agent Soft ActorCritic (MASAC)-based dynamic task allocation method. Unlike traditional approaches, the proposed framework leverages a spatiotemporal decoupling mechanism to separately optimize spatial assignment and temporal scheduling, thereby reducing computational complexity. Simulation results demonstrate that the proposed method outperforms mainstream multi-agent reinforcement learning algorithms in terms of task completion rate, allocation fairness, and system stability, indicating strong potential for real-world applications.