On adaptive attitude tracking control of spacecraft: A reinforcement learning based gain tuning way with guaranteed performance

Abstract

This paper investigates the attitude tracking control problem of a rigid spacecraft subject to inertial uncertainties, uncertain space perturbations and actuator saturation. Firstly, a static inertial-matrix free attitude controller is designed to guarantee the appointed-time convergence and tracking accuracy indicators of the spacecraft. Then, two-layer critic-action NNs are used to tune and optimize the control gains adaptively to improve the robustness and tracking performance of the devised static attitude controller via exploring the reinforcement learning (RL) technique. Compared with the existing attitude control methods, the prominent advantage of our work is that the devised controller is inertial-matrix free with a simple RL-based adaptive parameter tuning scheme, which reduces the conservativeness of the traditional attitude controllers with fixed control gains and is also easy to be achieved. Finally, two groups of illustrative examples are organized to validate the effectiveness of the proposed attitude control method.