Reinforcement learning-based asymptotic cooperative tracking of a class multi-agent dynamic systems using neural networks

In this paper, a novel reinforcement learning-based cooperative tracking control scheme is proposed for a class of multi-agent dynamic systems with disturbances and un-modeled dynamics on undirected graphs by using neural networks (NNs). For each agent, two NNs are employed, i.e., an actor NN which approximates the unknown nonlinearity and generates the control input, and a critic NN which evaluates the performance of the actor and updates the weights of actor NN. Further, a RISE technique is utilized in the design of the actor NN and the critic NN to compensate for the external disturbances and the NN approximation errors. Based on the Lyapunov theory, it is proved that the proposed control scheme can guarantee the tracking error of each agent to converge to zero asymptotically. Additionally, the proposed control scheme is distributed in the sense that the controller for each agent only uses the local neighbor information. Finally, two simulation examples are given to verify the effectiveness of the proposed control scheme.