A Cooperative Guidance Law for Multiple Missiles based on Reinforcement Learning

Abstract

The traditional proportional guidance law lacks the limitation of time and field of view. In order to realize the coordinated attack of multiple missiles on targets and improve the attack efficiency, a reinforcement learning cooperative guidance law based on deep deterministic policy gradient descent neural network is proposed. According to the particularity of guidance process, the reinforcement learning agent is obtained by constructing state space, action space and reward function training. The simulation results show that the enhanced learning guidance law can strike maneuvering targets simultaneously and satisfy the field of view constraint, which is superior to the traditional cooperative proportional guidance law.