An AutoML based trajectory optimization method for long-distance spacecraft pursuit-evasion game

Current successes in artificial intelligence domain have revitalized interest spacecraft pursuit-evasion game, which is an interception problem with a non-cooperative maneuvering target. The paper presents automated machine learning (AutoML) based method to generate optimal trajectories long-distance scenarios. Compared conventional deep neural network (DNN) methods, the proposed dramatically reduces reliance on manual intervention and expertise. Firstly, differential game theory costate normalization technique, trajectory optimization formulated under assumption of continuous thrust. Secondly, AutoML technique sequential model-based (SMBO) framework introduced automate DNN design process. If recommended architecture exists, tree-structured Parzen estimator (TPE) used, otherwise efficient search (NAS) morphism used. Thus, novel high computational efficiency achieved. Finally, numerical results demonstrate feasibility method.