Search

Driving Performance Evaluation based on Driving Volatility Measures – A Case Study on Indian Drivers

Yarlagadda, Jahnavi / Pawar, Digvijay S.

Driving safety is largely dependent on the driving styles that people have formed over time. The majority of previous research on the classification of driving styles concentrated on identifying driving patterns using kinematic feature magnitudes. The variance in the instantaneous driving decisions termed as “driving volatility” is not explored in the context of performance assessment. This study proposes a methodology for categorizing driving behaviors based on the magnitudes and volatilities of instantaneous driving decisions. The driving profile data of 47 professional car drivers collected in real-time is segmented as per the driving regime, resulting into 9691 accelerations and 6778 braking maneuvers. The primitive driving features (speed, acceleration, and yaw rate) of each maneuver are extracted, and then the respective volatilities are computed defining 12 measures of volatility at event-level. Further, the driving patterns are grouped by means of K-means clustering performed at two-levels, using feature magnitudes and volatilities independently. Total four patterns of driving styles are identified under acceleration and braking regimes, ranging from normal to volatile driving behaviors. The study insights are useful in monitoring individuals for driving assistance and feedback provision.