Methods For Yellow Light Dilemma Zone Boundary Modeling And Its Hazard Assessment At Signalized Intersections

Yellow light dilemma zone(called dilemma zone for short) is an area at the approach of high-speed signalized intersection, where the drivers face an indecisiveness of making stop or go decisions at yellow onset. As such, dilemma zone increases the probabilities of rear-end and side collision crashes at signalized intersections. The existing relevant literature have widely studied dilemma zone distribution, driver behaviors at yellow onset, dilemma zone hazard avoidance, and the like. Some dilemma zone hazard avoidance countermeasures have successively proposed. However, very few studies have explored dilemma zone hazard assessment. Furthermore, the effectiveness of the majority of conceived dilemma zone hazard avoidance measures is very limited, since most studies did not adequately consider drivers’ individual randomness and influence factor of traffic signal(such as signal countdown timer). Especially, a number of signalized intersections have been equipped with countdown timer in Chinese cites. This is bound to impact driving behavior at yellow onset, and then brings about the changes of dilemma zone distribution and corresponding hazard. Hence, based on observation data collected from signalized intersections with and without countdown timer, this research attempts to analyze the traveling parameters of first-to-stop and last-to-go vehicles during yellow interval, develop dilemma zone boundary model, propose approaches for assessing and avoiding dilemma zone hazard, and investigate the effects of countdown timer on dilemma zone distribution and hazard, in order to improve the safety level at signalized intersections.During yellow interval, first-to-stop and last-to-go vehicles on through lanes are regarded as observation objects, since these vheicles directly contribute to the formation of dilemma zone. Video observation scheme was designed and conducted at signalized intersections with and without countdown timer, so as to collect first-to-stop vehicles’ traveling parameters(i.e., yellow-onset distance to stop line, yellow-onset speed, perception reaction time, and deceleration rate), last-to-go vehicles’ traveling parameters(i.e., yellow-onset distance to stop line, yellow-onset speed, entry time, and acceleration rate), and road traffic related data.Based on observation data, descriptive statistics and fitting distribution of traveling parameters for first-to-stop and last-to-go vehicles under with and without countdown timer conditions were comparatively analyzed. Statistical testing results indicate that countdown timer significantly affects speeds and perception reaction times of first-to-stop vehicles, but does not significantly influence their distances to stop line and deceleration rates. Similarly, there are significant differences in distances to stop line and entry times for last-to-go vehicles under the with and without countdown timer conditions. However, no significant differences in their speeds and acceleration rates are found under such situatio ns. Kolmogorov-Smirnov testing outcomes suggest that those best fitting distributions for traveling parameters of first-to-stop and last-to-go vehicles under the two situations are different. Such findings lay the foundation for dilemma zone hazard evaluation.On the basis of observation data, yellow-onset speed was determined as the key influence factor for minimum perception reaction time, maximum deceleration rate, and maximum acceleration rate. The relationship models between yellow-onset speed and minimum perception reaction time, maximum deceleration rate, or maximum acceleration rate were developed, respectively. The accuracy of these models was validated through case analysis. After that minimum stop distance and maximum passing distance models were established, respectively. And then, impacts of countdown timer and yellow interval on type I dilemma zone distribution were explored. Comparative analysis results show that no matter whether the presence of countdown timer or not, it cannot form type I dilemma zone, when yellow duration is 4 second. If yellow duration is 3 second, countdown timer will lead to the formation of type I dilemma zone, but its range is very small.According to observation data, yellow-onset driving behaviors were analyzed to establish the binary logistic choice model for yellow-onset driving behavior. From the process of this model development, it is found that yellow-onset speed, yellow-onset distance to stop line, pedestrian, and countdown timer significantly influences drivers’ stop or go decisions making at yellow onset. In light of such choice model, yellow-onset stopping probability was deducted to develop type II dilemma zone boundary model. And then, effects of countdown timer and pedestrian on type II dilemma zone distribution were investigated. Comparative analysis results imply that no matter whether the presence of pedestrian or not, countdown timer does not change the length of type II dilemma zone, but moves it toward stop line.In view of type II dilemma zone covering type I dilemma zone(or option zone), without taking account of the collision between vehicle and pedestrian, study on type II dilemma zone hazard assessment was conducted. According to yellow-onset drivers’ driving behavior decision making, potential traffic conflicts within dilemma zone were analyzed, which includes rear-end and side collisions. After that estimation probabilities of dilemma zone rear-end and side collision risks occurrence were established and analyzed using Monte Carlo simulation technique. The analysis outcomes indicate that, the likelihood of dilemma zone side collision risk occurrence is much greater than that of rear-end collision risk. Moreover, the probability of rear-end collision risk occurrence is very small. Countdown timer obviously increases the probabilities of dilemma zone rear-end and side collision risks occurrence.Aiming at the issue of serious side collision risk within dilemma zone, the method for avoiding dilemma zone hazard based on dynamic all-red control was proposed in this study. The basic framework and technological process of dynamic all-red control were designed, including three steps. Firstly, vehicle s’ traveling parameters are collected by adopting Smart Sensor Advance. Secondly, yellow-light-running vehicles are distinguished in light of threshold deceleration rate and distance to stop line. Thirdly, the required all-red durations are determined based on yellow interval. Expected performance of dynamic all-red control was evaluated through repeating the flow chart for observation data obtained during the field study. Eventually, the pros and cons of dynamic all-red control were assessed using observation data. The results suggest that the safety benefit acquired by dynamic all-red control for avoiding dilemma zone side collision risk is much greater than the delay cost added on conflicting traffic.On one hand, the findings from current research enrich the theoretical system of dilemma zone boundary modeling and its hazard assessment study, provide a novel theoretical foundation for evaluating the effectiveness of dilemma zone hazard avoidance measures, and extend the method base for avoiding dilemma zone hazard. On the other hand, the established dilmma zone boundary models and the proposed dynamic all-red control approach can lay theoretical foundations for optimizing signal timing and enhancing traffic safety level at signalized intersections. In addition, the outcomes with respect to the influences of countdown timer on dilemma zone distribution and hazard can provide theoretical references for assessing safety benefits of countdown timer and developing regulations and guidelines of countdown timer installation.