Multi-Constraint Trajectory Tracking Control for Spacecraft Based on an Integrated Decision-Control Architecture

This study introduces a novel integrated decision-making and tracking control framework, which enables optimal autonomous obstacle avoidance for spacecraft, overcoming the limitations of traditional approaches that separate trajectory planning and tracking control. By merging trajectory selection with tracking into a single optimization control model, the framework allows spacecraft to dynamically generate and track multiple trajectory references in real-time, achieving adaptive decision-making and precise control execution. This approach aligns naturally with the actor-critic architecture in reinforcement learning, where the critic functions as the trajectory selector, evaluating the quality of the state, and the actor serves as the trajectory tracker, ensuring the dynamic feasibility of the trajectory. The simulation results demonstrate the real-time effectiveness of the proposed algorithm.