Baseline Analysis of Driver Performance at Intersections for the Left-Turn Assist Application 2017-01-0033

This study is aimed at supporting left-turn assist (LTA) applications, which provide warnings to drivers making a left turn across the path of oncoming traffic (LTAP/OD scenarios). The primary goal was to provide much-needed information on typical or “baseline” driving in LTAP/OD scenarios that can be used to refine alert criteria to reduce false and nuisance alerts. A secondary goal was to provide performance data useful for informing practical test procedures, e.g., setting turning speed when evaluating LTA applications on a test track. To accomplish this, LTAP/OD events were identified in the databases of two large-scale naturalistic driving studies. For these events, we estimated the size of the gaps in oncoming traffic into which drivers chose to turn, what factors (environmental, demographic, etc.) affected the choice to turn into a gap of a given size, and the speed profiles throughout each turn. The factors with the largest effect on gap size were age and gender, followed by road wetness, whether or not the turning vehicle stopped before turning, and the number of lanes the turning vehicle had to cross. As a counterpoint to this analysis of safe, typical turning behavior, we also identified instances in a national crash database where LTAP/OD maneuvers led to serious collisions involving vehicles equipped with event data recorders, i.e., scenarios where an alert could have been useful. Speed profiles for turning vehicles did not differ strongly for collisions as compared to baseline driving, but estimated gap sizes were shorter.