Vehicle path tracking control using a prescribed performance framework

Abstract

To balance the trade-offs between accuracy, robustness, and smoothness in vehicle path tracking, this work introduces a unified control framework that integrates prescribed performance control (PPC). In the upper layer, a prescribed performance function is employed to predefine the convergence rate, maximum overshoot, and steady-state error bounds. This confines tracking errors within dynamically narrowing safety envelopes via an error transformation, converting the process into an unconstrained stabilization problem. The lower layer utilizes Model Predictive Control (MPC) to handle multivariable constraints and multi-objective optimization, achieving smooth error stabilization through receding horizon optimization. A three-degree-of-freedom vehicle dynamics model is established, with path tracking errors defined within a vehicle-fixed coordinate system. The PPC error transformation and the integrated quadratic programming for PPC-MPC are derived in detail. For evaluation purposes, the MPC module is replaced with Sliding Mode Control (SMC) to establish a PPC-SMC benchmark for ablation studies. Co-simulations using CarSim and Simulink are conducted under steady-state, constant-curvature arc conditions, where tracking accuracy and control smoothness are compared. The results indicate that PPC-MPC clearly surpasses PPC-SMC in both lateral and heading Root Mean Square Error (RMSE). Additionally, the Root Mean Square (RMS) of the steering angle rate is significantly reduced, providing smooth, chattering-free control signals while strictly satisfying transient constraints. This research offers a comprehensive theoretical framework and practical guidance for controller selection in high-performance autonomous vehicle path tracking.