Model‐Free Adaptive Finite‐Time Optimal Tracking Control for Second‐Order Multiagent Systems With Input and Full State Constraints

For a class of second‐order multiagent systems such as networked Euler‐Lagrange systems, while the input saturation and full state constraints are considered, a model‐free adaptive finite‐time control is proposed in this article. The mean value theorem and nonlinear mapping techniques are used to deal with the input saturation and state constraints, respectively, and then the transformed agent dynamics without constraints are constructed. The unknown dynamic functions in the transformed models are coped with neural networks. To achieve the optimal cooperative tracking control, the improved actor–critic architecture based on reinforcement learning is deployed to approximate the optimal solution of the Hamilton–Jacobi–Bellman equation. The novel nonlinear controller and adaptive laws based on the improved actor–critic architecture are proposed. According to the finite‐time Lyapunov stability theorem, it is proved that the cooperative tracking under input saturation and full state constraints can be achieved in the finite time with the near‐optimal performance guarantees. Finally, the simulation of networked robots is carried out to further show the effectiveness of the control method proposed in this article.