Autonomous spacecraft operations in orbit are increasingly challenged by orbital threats such as debris collisions, close-range reconnaissance, and malicious interference. The avoidance process in such scenarios constitutes a dynamic game under uncertain adversarial behaviors. Influenced by environmental constraints, sensing limitations, communication disturbances, and stealth technologies, the information structure is both imperfect and incomplete, resulting in a dual dilemma for game-theoretic decision-making: significant uncertainty in adversarial actions and high sensitivity of strategies to situational dynamics. Additionally, the dimensionality of the strategy search space increases exponentially with the complexity of information, which, under the stringent constraints of onboard computational resources, often forces autonomous decision-making systems to adopt overly conservative strategies or compromises that may jeopardize spacecraft safety. To address these challenges, this study proposes a feedback-based strategy design framework featuring a vertically layered and horizontally adjustable architecture. By dynamically optimizing the game information structure in real time based on situational features, the method adaptively reduces the search space and improves strategy performance, thereby ensuring the security of space assets and satisfying the operational demands of on-orbit missions.
YUAN et al. (Fri,) studied this question.