Correlation Analysis of Drivers’ Natural Driving Behavior Based on Kernel Density Estimation

Abstract

To investigate the interplay between driver handling behaviors, this article collects data on vehicle kinematic parameters characterizing driver handling characteristics under natural driving, estimates the probability density curves of the parameters using the kernel density method, and fits the curve equations. On this basis, a percentile correlation analysis was performed between the parameters to obtain the influence relationship between the handling behaviors. The results show that longitudinal maneuvers are frequent and intense in the 0–10 km/h speed range, lateral maneuvers are more intense in the 10–30 km/h speed range, and the interaction between longitudinal and lateral maneuvers is more intense in the acceleration phase. This study enriches the natural driving dataset and illustrates the correlation of driving behavior under natural driving, providing a theoretical and data basis for the development of driver-oriented intelligent driving technologies.