A guidance method for coplanar orbital interception based on reinforcement learning

This paper investigates the guidance method based on reinforcement learning (RL) for the coplanar orbital intercep-tion in a continuous low-thrust scenario. The problem is formu-lated into a Markov decision process (MDP) model, then a well-designed RL algorithm, experience based deep deterministic policy gradient (EBDDPG), is proposed to solve it. By taking the advantage of prior information generated through the optimal control model, the proposed algorithm not only resolves the con-vergence problem of the common RL algorithm, but also suc-cessfully trains an efficient deep neural network (DNN) controller for the chaser spacecraft to generate the control sequence. Nu-merical simulation results show that the proposed algorithm is feasible and the trained DNN controller significantly improves the efficiency over traditional optimization methods by roughly two orders of magnitude.