Model Predictive Control-Based Lane Keeping Assist: Design, Validation, and Comparative Analysis for ADAS and Autonomous Driving

Abstract

Precise lateral control is a key challenge in the development of Advanced Driver-Assistance Systems (ADAS), particularly for Lane Keeping Assist (LKA) functionalities. This paper presents a Model Predictive Control (MPC) strategy tailored for LKA, designed to maintain the vehicle within its lane while ensuring stability and responsiveness. The controller leverages a dynamic vehicle model that integrates longitudinal, lateral, and yaw dynamics to realistically capture tire–road interactions and vehicle behavior. Through real-time optimization, the MPC predicts future vehicle states and computes an optimal control sequence that minimizes lateral deviation and heading error, while adhering to physical constraints such as steering limits and actuator saturation. The closed-loop stability of the system is theoretically validated using a Lyapunov-based approach. The proposed MPC controller is intended for integration within a modular ADAS architecture, alongside perception, lane detection, and longitudinal control modules. Simulation results using real-world trajectory data demonstrate the controller's effectiveness in delivering precise and robust lane keeping performance under varying driving conditions.