基于角色多智能体强化学习的异构无人机集群任务分配方法OA

Aiming to address the task assignment problem for heterogeneous UAV swarm collaboration in ground attack missions,this paper considers the characteristics of large scale,high dynamics,and strong adversarial conditions typically presented in real-world scenarios.This paper propose a role-based multi-agent reinforcement learning role-based multiple attention actor critic,(Role-MAAC)method driven by the attention mechanism.First,model the task allocation for heterogeneous UAV swarm collaboration in ground attack missions,describing both the task assumptions and adversarial interactions between the UAV swarm and ground targets.Second,to tackle the three key features of the task allocation problem,establish a network architecture that combines centralized training with distributed execution,mapping UAV functions to different roles.This paper propose a role-based architecture,where key states trigger new decision-making cycles in each round and optimize the construction of the sampling pool.Simulation results demonstrate that the proposed method effectively addresses the task allocation problem for heterogeneous UAV swarm collaboration in ground attack missions,achieving improvements in both model performance and convergence compared to baseline methods.