Abstract: Abstract: In space mission scenarios, rapid and accurate recognition of the intentions of space non-cooperative targets is a critical prerequisite for spacecraft to seize the initiative on the battlefield and ensure the reliability of subsequent decision-making. Meanwhile, after obtaining target intentions, spacecraft must possess efficient orbital game capabilities to formulate effective evasion, defense and even countermeasures against the intentions of space non-cooperative targets, so as to practically safeguard their own on-orbit safety. This paper systematically discusses the intention recognition and orbital game of space non-cooperative targets. Firstly, the relevant concepts of target intention recognition are introduced, and the process of intention recognition for space non-cooperative targets is clarified. Typical intention recognition methods including template matching, D-S evidence theory, Bayesian network and deep learning-based methods are emphatically reviewed, with their advantages and limitations objectively analyzed. Secondly, the relevant concepts of space orbital game are sorted out, orbital games are classified from different perspectives, and the methods for solving space orbital game strategies are systematically summarized. Finally, the existing key problems in the field of intention recognition and orbital game of space non-cooperative targets are deeply analyzed, and the future research directions are prospected combined with the development trend of technology, so as to provide a reference for the follow-up exploration in related fields.

Key words: space non-cooperative targets, intention recognition, deep learning, orbital game, space security