Design and Implementation of Parameterized MPC Strategies for Lane Keeping Assist Systems in Real-Time HIL Architecture

The Lane Assist System (LAS) helps drivers stay within their road lanes. It uses cutting-edge technology (such as a camera system) to monitor marks for lane-following applications. The Lane Keeping Assist System (LKAS), a LAS with steering control, ensures appropriate lateral vehicle dynamics, avoiding unexpected steering operation. Despite LKAS solutions promoting comfort driving, robust steering control is lacking, requiring constant driver intervention, sometimes with premature LKAS deactivation. For this reason, this paper investigates parameterized Model-Predictive Control (MPC) solutions for real-time LKAS applications. The parameterized MPC calculates control actions for a constrained dynamic system, making it a potential candidate for robust LKAS. A sideslip and lateral speed bicycle model with three MPC formulations and various Quadratic Programming Solvers were considered. Trajectory tracking, lateral displacement and yaw angle regulation experimental tests were run through Hardware-In-the-Loop (HIL) Architecture. Experimental results, along with comparative performance and robustness analyses, demonstrate satisfactory results and provide valuable insights into the application of MPC for lane-keeping assist systems.