Driving Style Classification and Dynamic Recognition Considering Traffic State

Identifying aggressive driving styles and assisting in providing driving behavior recommendations is necessary to reduce traffic accidents and improve traffic safety. However, driving behaviours reflect driving intentions that vary with traffic states, resulting in uncertainty in the classification of driving styles. In this paper, the driving behavior is evaluated comprehensively from two dimensions of speed and headway, and then the driving intention is fully recognized to lay the foundation for solving the driving style ambiguity. The driving style is divided into five categories by using k-means clustering method. The results of variance test showed that 11 parameters are significantly different between groups. This shows that the driving style classification method proposed in this paper can distinguish different driving styles.