Escape Planning via Single-Step Prediction for Impulse Propulsion Game Considering Optical Imaging Constraints

Aiming at high-orbit optical game mission, this paper proposes a single-step prediction-based evasion planning method. First, by comprehensively considering distance and solar angle constraints, and taking the effective optical imaging time as the minimum performance metric, an optimization model for the evasion trajectory of the escaping satellite is constructed. Then, a neural network-based predictive re-planning of counteractions for the imaging cluster is embedded to predict the total velocity increment consumed by the imaging cluster in the follow action. This predicted value is combined with the imaging time via weighted summation and incorporated into the performance metric. Thus, a dual-layer planning architecture is established, enabling the escaping satellite to complete its evasion while maximizing the expected maneuver cost of the imaging cluster. Simulation results demonstrate that the proposed strategy can not only achieve effective evasion of imaging but also significantly increase the total fuel consumption required for the imaging cluster to complete its mission.