SAC-Based Impact Time Cooperative Guidance Law for Multiple Flight Vehicles

This paper presents a new SAC-based impact-time cooperative guidance law for multiple flight vehicles under the Multi-Agent Reinforcement Learning (MARL) architecture. The Soft Actor-Critic (SAC) algorithm under MARL architecture is firstly introduced, employing a central Q network to guide policy networks for different flight vehicles more efficiently. Then the state, action and reward of SAC algorithm are specifically designed in conjunction with the multiple flight vehicle cooperation application scenarios. Finally, the numerical demonstrations are carried out to demonstrate the superior of the proposed method.