Q-Learning Based Optimal Tracking Control of Free-Flying Space Manipulators with Unknown Dynamics

Hongxu Zhu,Shufan Wu,Qiang Shen,Ran Sun
DOI: https://doi.org/10.23919/ccc55666.2022.9901557
2022-01-01
Abstract:This paper investigates a Q-learning based optimal tracking problem of a free-flying space manipulator with unknown dynamics. First, via a series of equivalent transformations, we obtain a simplified model-free solution form of the original optimal tracking problem. Then we construct a Q-learning based policy iteration mechanism with an actor/critic neural network structure to solve the optimal control policy. By following a Lyapunov analysis we prove the asymptotical stability of the closed-loop system. Finally, a numerical simulation of a 2-DOF space manipulator verifies the effectiveness of the proposed method.
What problem does this paper attempt to address?