Impact Angle Control Guidance Law Considering Field-of-View Constraint Based on Deep Reinforcement Learning

A deep reinforcement learning-based multi-constraint guidance law is proposed in this study to satisfy field-of-view and impact angle constraints. Firstly, the relative motion relationship between the missile and the target is established, and the impact angle control optimal guidance law is derived based on optimal control theory. Secondly, the guidance parameters are extracted as agent actions, and a reward function that considers the field-of-view angle and impact angle accuracy is designed. Therefore, the multi-constraint guidance problem is reformulated as a Markov decision process. And then, a novel prioritized experience replay method is introduced in the deep deterministic policy gradient (DDPG) algorithm to enhance training efficiency. Meanwhile, the reward sparsity problem is effectively addressed. Finally, numerical simulation results validate the effectiveness of the intellignet guidance law.