Factors for Opportunity Left-turn Traffic Conflicts at Urban Signalized Intersections

抄録

Crashes at urban intersections account for more than 50% of the total traffic accidents in Taiwan. The most common type of crashes is side collision, with left-turn failure being the second most common cause. In this study, Post-Encroachment Time (PET) is used as a crash surrogate to quantify the proximity of left-turning vehicles to their oncoming counterparts at urban signalized intersections. The PET thresholds under different vehicle pairs are evaluated using the Peak over Threshold (POT) method in Extreme Value Theory (EVT). The correlation between traffic conflicts and crashes is first formulated. Second, count data and logistic regression models are developed to explore factors that might result in left-turn traffic conflicts both at the intersection and vehicle-pair levels, respectively. Using the vehicle trajectory data collected by Unmanned Aerial Vehicle (UAV), the empirical study results indicate that traffic conflict thresholds differ for different vehicle-pairing samples under a mixed traffic flow condition, and engineering-based countermeasures help reduce potential left-turn traffic conflicts at signalized intersections.