Reinforcement Learning-Based Distributed Robust Bipartite Consensus Control for Multispacecraft Systems With Dynamic Uncertainties

In this article, the reinforcement learning-based distributed robust bipartite consensus control of multispacecraft systems with dynamic uncertainties is investigated. The developed control structure includes two parts, i.e., integral sliding mode control and distributed optimal bipartite consensus control. In the first step, an integral sliding mode controller is designed for each following spacecraft to address matched uncertainties such that the dynamics of nominal spacecraft is obtained. In the second step, a novel performance index function, which contains consensus errors and their derivatives, is designed for each nominal spacecraft. As a result, the system assumption of zero equilibrium and the discount factor in performance index function are not required, which simplifies the controller design process and improves the practicability of the developed control method. Moreover, in order to solve the coupled Hamilton–Jacobi–Bellman equation of each following spacecraft, a novel policy iteration algorithm is designed and its properties are analyzed. Finally, a group of spacecraft is employed to verify the effectiveness of the present control scheme.