ABSTRACT：Interactions between road users are affected by drivers’ individual differences and driving styles. However, most current models on driver classification and driving style recognition are based on vehicle movement characteristics, such as velocity and deceleration. Safety performance has seldom considered in these models, such as risk critical events and driver’s response to them. This study applies a safety-oriented approach to classify and recognize driving styles from 60 drivers participated in Shanghai Naturalistic Driving Study. Five surrogate safety indicators are calibrated for each driver, including Time to Collison (TTC), Time Exposed TTC (TET), Time Integrated TTC (TIT), Deceleration Rate to Avoid Crash (DRAC), Crash Potential Index (CPI), and Proportion of Stopping Distance (PSD). Herein, six driving styles with different risky levels are recognized by clustering based on safety indicators. According to comparison analysis between safety and movements-oriented driving style recognition approaches, it is found that movement-oriented driving style recognition may not able to reflect risk exposure when the driver is interacting with neighboring vehicles. Moreover, cautious movements such as more likely to maintain large gaps in car-following and less likely to filter through close gaps may not reduce risk in certain driving scenarios.
Chen Chai, Xuesong Wang*, Ziyao Zhou, Linjia He. Safety-Oriented Driving Style Recognition Using Surrogate Safety Indicators from Naturalistic Driving. Transportation Research Board 98th Annual Meeting, Washington D.C., USA, 2019. 1.13-17.