Modeling for Driver's Dilemma Behavior in Choosing Pass/Stop After Yellow Time with Mixed Logit Models (Case study of Qazvin)

At the onset of the yellow phase, drivers often come across a dilemma situation where they are unable to stop comfortably before the stop line or clear the intersection (without excessive acceleration) prior to the onset of the red signal phase. yellow time is designed to inform drivers about passing time and preventing extreme changes in cars' speed in timing of intersections with traffic lights. However, studies have confirmed that drivers face high level of uncertainty during yellow time. drivers visually sample their surroundings while driving so they are able to change their behavior based on other vehicles’ movements, the roadway environment and traffic signal data. This implies that drivers’ behaviors are affected by surrounding factors such as other vehicles’ headway or intersection conditions. In the dilemma zone, drivers’ decisions are influenced not only by their own condition (e.g., distance to the stop line, speed, red time)but also by the surrounding environment at an intersection. The primary goal of  the research described here was to develop a comprehensive knowledge of the stopping characteristics of dilemma zone drivers at signalized intersections. Physical, traffic, timing and phasing of intersections and weather conditions are assessed factors. The research performed here involved macroscopic evaluation of driver behavior; thus, characteristics of individual drivers were not investigated as it was not feasible to determine information such as age, experience, route familiarity, and sex of each driver. This study investigates actual data of traffic cameras and central smart program in four intersections in Qazvin in which traffic lights are set up. Peak, normal sunny and rainy conditions and drivers' behavior in yellow and red times are studied using binary and mixed logit model. A field study was performed using a video-based data collection system to record several attributes related to the behavior of the last vehicle to go through and the first vehicle to stop in each lane during each yellow interval. The researchers concluded that a driver’s decision to stop or go through when presented with a yellow indication is complex but can be predicted reasonably well based on several factors. Pedestrians in streets and headway are the most effective factors on drivers' pauses in yellow or red phase. High speed of cars and also waiting time (red phaseare the most influential factors on drivers motion in yellow or red phases. The results of model show binary logit model has a higher accuracy than the combined logit model for assessing driver behavior.