Optimal Control Design for Personalized Lane-Change Operation of Inteligent Vehicle

According to the personalized lane-change habits of different drivers, this study presents an automatic-driving personalized lane-change control system that ensures the safety of automatic-driving lane changes and meets the personalized needs of drivers. Constrained Delaunay triangulation (CDT) was used to generate a safe collision-free lane-changing area to ensure the safety of the planned lane-change path. Using the HighD datasets, a lane-change driving behavior identification model was established based on the BP neural network, divided into careful, normal, and aggressive according to driving habits. Taking driving safety and ride comfort as performance indices, quadratic programming (QP) algorithm is applied to generate personalized optimal paths to accommodate three different driving lane-change styles. Then, a linear quadratic regulator (LQR) feedback controller integrated with feedforward control was designed for tracking control of personalized lane-change paths. Finally, co-simulation results by using PreScan, CarSim, and MATLAB/Simulink demonstrated that the proposed personalized lane-change path-planning and the tracking control system were effective and feasibility.