Dilemma Zone Identification and Mitigation Approach at Signalized Intersections under Mixed Traffic Conditions Using UAV Data

Traffic safety at signalized intersections is significantly affected by the driver’s behavior, psychology, and the design of traffic signals, especially in mixed traffic conditions. At signalized intersections, the traffic signal design influences the traffic behavior during the yellow change interval. The indecisiveness or the dilemma in decision-making during the yellow change interval often results in hard deceleration or red-light running, which concerns the safety of the road users in the form of rear-end and side-angled conflicts. Thus, it is essential to understand how traffic behavior and signal design changes affect safety. The present work investigates traffic behavior at signalized intersections operating under mixed traffic conditions using high-quality vehicular trajectory data obtained using unmanned aerial vehicles (UAVs). The analyzed traffic behavior is related to Type I and Type II dilemma zones at intersections with two types of traffic flow conditions: oversaturated and undersaturated. However, it is found that the position of these dilemma zones is near the stop line for an oversaturated intersection and away from the stop line for an undersaturated intersection, as it depends on the approach speeds of vehicles. The driver’s decision-making behavior, which depends on the driver’s psychology, is also analyzed, and a binary logistic regression model is developed for determining the Type II dilemma zone by considering various parameters at the onset of the yellow change interval. Further, the variation in the length of the Type I dilemma zone is studied by performing a sensitivity analysis and proposing measures for mitigating its negative effects. The findings underscore the critical importance of considering both traffic behavior and signal design in the safety analysis of signalized intersections, particularly in mixed traffic environments where driver behavior, psychology, and vehicle interactions are more unpredictable.