Multi-missile Cooperative Guidance Law Based on Deep Reinforcement Learning

Considering that the traditional guidance laws need to carefully adjust the parameters of guidance commands according to specific combat scenarios. In this paper, deep reinforcement learning is used to design a guidance law, which has the autonomous decision-making and planning capabilities, to improve combat effectiveness in uncertain battlefield environment. Improved on the basis of leader-followers strategy: the leader uses the traditional guidance law and the followers adjust the guidance commands by TD3 algorithm. By designing the appropriate reward function, the relative range and leading angle of the following missiles tend to be the same as that of the leading missile. Finally, simulation experiments are carried out for maneuvering targets to verify the validity and correctness of the guidance law.