Incorporation of Driver Distraction in Car-following model based on Driver's Eye Glance Behavior

Abstract

This paper aims to incorporate driver distraction into car-following model to demonstrate hazardous situations such as vehicle collision. Many reports point out a driver distraction as one of the major causes of automobile collisions and driver's eye glance behavior is representative indicator for quantitatively evaluating distraction. To analyze distraction, a 100-car Naturalistic Driving Study data including eye glance data is used. This study classified several variables affecting the glance behavior into two different groups and analyzed in different ways. Based on the results of the analysis, decision tree analysis is conducted to derive driving scenarios according to the eye glance behavior and total of eleven scenarios are derived. Driver's glance behavior is modelled by scenarios and incorporated into the existing car-following model. As one of the existing car-following model, Oversaturated Freeway Flow Algorithm (OFFA) is extended to distracted OFFA. We show that distracted OFFA describes real-world driving more closely in terms of vehicle safety by comparing a distribution of time to collision (TTC) with existing car-following models.