Event-Triggered Image-Space Tracking Control of Space Manipulators Using Off-Policy Reinforcement Learning With Disturbance Observers

This paper aims to develop a methodology to address challenges in image-space tracking control for space manipulators, including external uncertainties, inherent dynamic uncertainties, and the balance between control performance and communication-computation resource utilization efficiency. Given the diverse types of external disturbances, this research enhances existing disturbance observers to accommodate a broader range of nonlinear disturbance forms. Additionally, an improved sample collection strategy is proposed for the model-free off-policy reinforcement learning method to enhance sampling efficiency and ensure the stability of optimal weight convergence. Moreover, a mixed-variable approach incorporating joint space and image space is employed in the control process to address the challenge of solving complex, coupled time-varying differential equations. Recognizing that different operational modes of space manipulators impose varying requirements on control performance and communication-computation resource utilization efficiency, this study introduces a generalized event-triggered scheme that integrates optimal control and maximized resource utilization while ensuring system stability. Finally, simulations validate the effectiveness of the developed algorithm, and comparative analyses with various control methods and disturbance observers demonstrate the advantages of the proposed approach.