Physical-Informed Neural Network for MPC-Based Trajectory Tracking of Vehicles With Noise Considered
Long Jin, Longqi Liu, Xingxia Wang, Mingsheng Shang, Fei–Yue Wang
Abstract
The trajectory tracking plays a vital role in unmanned driving technology. Although traditional control schemes may yield satisfactory outcomes in dealing with simple linear tasks, they may fall short when handling dynamic systems with time-varying characteristics or lack of ability to complete a given task with the disturbance of noise. Therefore, a predictive control scheme under the framework of artificial systems, computational experiments, and parallel execution (ACP) is proposed. Within the ACP framework, the scheme integrates a model predictive control (MPC) controller and a physical-informed neural network (PINN) model to tackle intricate trajectory tracking tasks effectively with noise considered. Moreover, soft constraints that can enhance model robustness and improve solution efficiency are considered in the scheme. Then, theoretical analyses on the PINN model are provided with rigorous mathematical proofs. Finally, experiments and comparisons with existing works are conducted to illustrate the effectiveness and superiority of the constructed PINN model for MPC-based trajectory tracking of vehicles.