Intelligent vehicle path tracking control strategy considering data-driven dynamic stable region constraints

Abstract

The path tracking controller can easily reduce the tracking error, but often exceed the limitations of vehicle stability. In this paper, an intelligent vehicle path tracking control strategy considering data-driven dynamic stable region constraints is proposed. Firstly, based on the two-degree-of-freedom (DOF) vehicle model and nonlinear tire model, the vehicle sideslip angle-sideslip angular velocity ([Formula: see text]) phase plane is established. Then, the stable region dataset is made considering the influence of vehicle speed, adhesion coefficient, and front wheel angle. To get the vehicle driving stable region, a back propagation neural network (BP-NN) regression model is trained offline. Subsequently, a path tracking control strategy based on adaptive-model predictive control (MPC) is designed, which considers the vehicle dynamic stable region constraints with the BP-NN predicting online. Finally, model-in-the-loop (MIL) and driving simulator is designed to test the control strategy, which indicates that it has a better performance compared with the linear quadratic regulator (LQR) path tracking controller.