基于多智能体强化学习的变外形飞行器协同 轨迹规划

Research on morphing flight vehicles predominantly addresses trajectory planning, threat zone avoidance, and morphing decisions for single flight vehicles, whereas cooperative trajectory planning for multiple morphing flight vehicles remains underexplored. A cooperative trajectory planning method based on multi-agent reinforcement learning is developed for multiple morphing hypersonic flight vehicles during reentry. Within the multi-agent proximal policy optimization (MAPPO) framework, the SAGRU-MAPPO algorithm is constructed by integrating gated recurrent unit (GRU) networks and self-attention mechanisms, significantly enhancing temporal information retention and high-dimensional state processing capabilities. Through unified design of trajectory planning commands and morphing decisions, coordinated threat avoidance and temporal synchronization are achieved for multiple morphing flight vehicles. Simulation results confirm that this method effectively addresses cooperative trajectory planning challenges for morphing hypersonic flight vehicles during reentry, offering technical support for multi-flight vehicle collaborative missions in complex scenarios.