W. Zhang, E. Kontou

Pages: 87-100

Abstract
Accurate eye tracking technologies are becoming increasingly affordable to the research community, attracting prolific use of them in studying various types of driver behaviors. This paper presents a refined eye tracking study aimed at isolating out the sources, both stationary and temporal, that cause drivers to turn their visual attentions away from the roadway ahead. The eye tracking datasets were generated by 35 drivers using a dash-mounted eye tracker. Each driver drove through two test routes, Route A and Route B. The routes consisted of different types of roadways (expressways, on/off-ramps, local roads, et cetera.) and each took about 25 minutes to drive through. Tests were done during off-peak periods, in the early afternoon and after dark hours. About 1.5 million image frames with valid fixations, proper speed values, and associated roadway types were analyzed. The samples were grouped by roadway type, driving speed, and visibility. Fixation patterns across the 50 × 50 pixel grid cells corresponding to different sample groups were analyzed to screen out events that have a high likelihood of involving visual distraction. The study is built upon the lead author’s previous work presented at the 3rd International Conference of Road Safety and Simulation. It includes significant refinements in grouping criteria by speed.

Keywords: visual distraction; eye tracking; naturalistic driving; fixation pattern; fixed distraction source and temporal distraction source