Analysis and Modeling of Drivers' Responses at Urban Signalized Intersections

Abstract

This study provided baseline data on how subjects approached intersections to support a follow-on study of augmented reality crash warnings. Twenty-four licensed drivers (12 ages 18-30, 12 greater than age 65) drove through 71 urban, signalized intersections in a fixed-base driving simulator while following a lead vehicle. Five types of conflicts were simulated at 10 intersections, straight crossing path (SCP), left turn across path – opposite direction (LTAP/OD), left turn across path – lateral direction (LTAP/LD), left turn into path (LTIP), and right turn into path (RTIP). The probability of stopping at yellow light was = ex/(1+ex), where x = -5.84 - 0.01*gap to stop line + 1.49*yellow light timing - 2.93*pedal position + 0.29*age + 0.15*gender, accounting for 19% of the variance because of individual differences. Subject vehicle approach speed to the intersection varied by conflict type and traffic light color and timing. When approaching a green light, 21% of subjects slowed down for LTAP/OD conflicts and 27% slowed down for RTIP. For yellow lights, when subjects chose to run light, they never slowed down for LTAP/OD. The approach speed to an intersection was = 13.70 - 0.17*gap to stop line - 5.38*run or not + 0.07*conflict or not + 3.21*yellow light timing 2.8 s - 0.18*yellow light timing 3.5 s - 1.58*age, with an R2 of 0.74.