Efficient vehicle lateral safety analysis based on Multi-Kriging metamodels: Autonomous trucks under different...

Different lateral control modes for autonomous trucks have been proposed in the past to improve road sustainability. To evaluate the lateral reliability and risk of autonomous trucks under different lateral control modes, especially in windy conditions and during overtaking, this study proposes a high-precision, efficient method based on Multi-Kriging (MK) metamodeling. Compared to Support Vector Regression (SVR), widely used in vehicle safety studies, the MK method builds Kriging metamodels for all potential deviation crashes of autonomous trucks, significantly reducing computational costs while ensuring accuracy. Additionally, instead of using the maximum lateral deviation as a dynamic risk indicator, this study innovatively suggests using the distance between the autonomous truck and the lane line after the maximum lateral deviation occurs. Considering the vehicle and environmental characteristics, the MK method is used to calculate the probability of a crash and the associated risk at various lateral positions of the autonomous truck, quantifying the crash probability and risk values under different lateral control modes. Furthermore, the study investigates the influence of the standard deviation of the normal distribution on the crash probability and risk values of autonomous trucks, providing valuable insights for future research on road sustainability and safety under autonomous driving conditions.