Recognizing Safety-critical Events from Naturalistic Driving Data
Paper i proceeding, 2012

New trends in research on traffic accidents comprehend Naturalistic Driving Studies (NDS). NDS are based on largescale data collection of driver, vehicle, and environment information in realtraffic. NDS datasets have proven to be extremely valuable for the analysis of safetycritical events such as crashes and nearcrashes. However finding safetycritical events in NDS data may be difficult and timeconsuming. Safetycritical events are currently individuated using kinematic triggers (e.g. searching for deceleration below a certain threshold signifying harsh braking). Due to the low sensitivity and specificity of this filtering procedure, manual review of video data is –to date– necessary to decide whether the events individuated by the triggers are actually safetycritical. Such reviewing procedure is based on subjective decisions, is time, and is often tedious for the analysts. This study tested the hypothesis that automatic processing of driver video information could increase the correct classification of safetycritical events from kinematic triggers in naturalistic driving data. Review of about 400 videos from the triggered events collected by 100 Volvo cars in the euroFOT project, suggested that driver's individual reaction may be the key to discriminate safetycritical events. In fact, whether an event is safetycritical or not often depends on the individual driver. A few algorithms, able to automatically classify driver reaction from video data, have been compared. The results presented in this paper show that the stateofart subjective reviewprocedures to individuate safetycritical events from NDS can benefit from automated objective videoanalysis. In addition, this paper discusses the major challenges in making such a videoanalysis viable for future NDS.


Marco Dozza

Vehicle and Traffic Safety Centre at Chalmers

Chalmers, Tillämpad mekanik, Fordonssäkerhet

Nieves Pañeda González

Vehicle and Traffic Safety Centre at Chalmers

Chalmers, Tillämpad mekanik, Fordonssäkerhet

Procedia - Social and Behavioral Sciences

Vol. 48 505-515




Data- och informationsvetenskap



Mer information