Characterizing driver behavior using naturalistic driving data

Abstract

This study highlights the significance of understanding and categorizing driving styles to improve traffic safety and increase fuel efficiency. By analyzing a comprehensive dataset of naturalistic driving records from taxi drivers, it offers insight into driving behaviors in various environments. Utilizing deep clustering methodology, the research develops a novel framework for categorizing driving behaviors into Baseline Driving Characteristics (BDC), encompassing aspects such as turning, cruising, acceleration, and deceleration. These characteristics are instrumental in creating an abnormal driving index that serves as a quantitative measure for evaluating driving styles concerning traffic safety. Furthermore, the study elaborates on the utility of the abnormal driving index and its correlation with headway distances, enabling the formulation of personalized safety guidelines for drivers. This research contributes to the field of traffic safety by using the BDC to offer insight into driving behaviors. It lays the groundwork for future research aimed at enhancing driving behavior analysis through the integration of advanced driver assistance systems and exploration of linkages between the abnormal driving index and actual crash risk. The results of this study advance understanding of driving behaviors and their implications for traffic safety, paving the way for the development of broader and more effective safety measures in transportation.