Driver's Distraction Detection Function in Cooperation with Artificial Intelligence

Abstract

Reduction of traffic accidents may be one of urgent challenges to create sustainable mobility society. Currently preventive safety technologies of vehicles are expected to play important role to reduce the number of traffic accidents. This research proposes a concept of artificial intelligence based driver’s states monitoring function. From analysis of Internet survey, driver’s distraction state is one of dominant psychosomatic state which drivers often fall in inattentive driving. By means of using pattern recognition, this research established a method to detect driver’s cognitive distraction. Four types of classification feature were examined as alternative characteristics of distraction states, which are gaze direction of eyes, head orientation, pupil diameter and heart rate from electrocardiogram (ECG) waveform. Loss-based Error-Correcting Output Coding (LD-ECOC) was adopted as a classification algorithm. Driver’s states monitoring function was comprised of driver’s distraction detection function, which was part of an artificial intelligence unit in cooperation with autonomous driving function for reduction of the number of traffic accidents.