The Impact of Highway Space Alignment on Autonomous Vehicles Trajectory Deviation

A stable and continuous driving trajectory is essential for road safety. The study examines the impact of highway geometric alignment space parameters on the driving characteristics of Autonomous Vehicles (AVs). AV implemented in this study was equipped with L4 driving automation. The study involved conducting driving simulation tests utilizing PreScan and MATLAB/Simulink software. The Gradient Boosting Decision Tree (GBDT) algorithm was employed to construct the driving trajectory deviation model for the tested AV. The SHAP value was utilized to examine and articulate the association between space alignment parameters and lane offset. The findings indicate that the driving trajectory of AVs is notably influenced by factors such as the curvatures of the space alignment, the lane in which the vehicle is driving, and the driving direction.