Speeding Behavior when Using Automation: A Descriptive Analysis of Naturalistic Driving Data
Event Type
Surface Transportation
TimeWednesday, October 12th11:30am - 11:45am EDT
DescriptionSpeeding is a prevalent and risky behavior. This study leveraged the MIT Advanced Vehicle Technology naturalistic driving data to identify and characterize speeding behavior across automation levels including Manual Driving, Adaptive Cruise Control (ACC), and Pilot Assist (PA) that supports both longitudinal and lateral vehicle control. We analyzed 146 hours of motorway driving from 15 participants who each drove an instrumented vehicle for one month. Speeding prevalence, magnitude, and duration distributions were compared across automation levels using logistic mixed effect models. Findings indicated that although PA was the most prevalent driving state during free flow motorway driving, drivers were more likely to speed with ACC compared to during Manual Driving or with PA. Automation, ACC and PA, were associated with longer speeding durations and lower speeding magnitudes compared to driving manually. These findings can inform the development of automation systems that may help reduce speeding behaviors and promote safe driving.