Reinforcement learning-based prescribed finite-time optimal tracking control for a vehicle system regardless of initial position

The prescribed finite-time optimal tracking control (OTC) problem is investigated based on reinforcement learning (RL) for a vehicle system in this paper. A novel nonlinear mapping function is designed to develop a new approach of prescribed performance control (PPC) which is irrelevant to the initial position in the system. Meanwhile, the RL method adopting neural network with the actor-critic architecture is used to optimize the control input of the system. By means of a new prescribed finite-time performance function (PFTPF), a prescribed finite-time optimal tracking controller is obtained regardless of initial position of the vehicle system. The designed controller ensures not only that the tracking error satisfies the prescribed performance requirement, but also that all closed-loop signals in the vehicle system are bounded. Finally, the simulation results verify the effectiveness of the proposed controller.