Analysis of Destination Lane Choice at Urban Intersections

Abstract

In some states in the U.S., it is required by law that drivers use a designated destination lane at intersections, so as to avoid a potential collision with another concurrent turning movement; however, drivers do not always select the correct designated destination lane when turning and currently, there is no transportation model that can accurately predict which destination lane drivers will choose when turning at urban intersections. With the advent of connected and autonomous vehicles, there will be a period of time when some vehicles on the road are automated (either partially or fully), and others are still driven by humans. Therefore, there is a dire need for a model that can be used in conjunction with sensors in autonomous vehicles, to detect and predict if a human-driven vehicle will turn into the incorrect destination lane. This research makes use of two Next Generation Simulation (NGSIM) arterial street data sets in order to perform a comparative analysis of drivers’ destination lane choice behavior, specifically for left turns. The model accurately predicts when drivers choose lane 1 as their destination lane and performs poorly when predicting lanes 2 or 3 as their destination lane. This model (i) can be used in conjunction with sensors in automated vehicles to avoid potential collisions with human-driven vehicles; (ii) may be incorporated into microscopic traffic simulation tools in order to improve the safety of urban intersections; and (iii) illustrates the need for policy improvements at intersections, specifically as it relates to designated destination lanes.