RUI: Behavior-Based Stochastic Traffic Flow Modeling for Intersection Safety Improvement

Cal Poly Pomona Foundation, Inc., Pomona CA

Investigators

Abstract

This research will investigate driver behavior at signalized intersections and derive sophisticated traffic flow models aimed at intersection safety improvement. Intersection safety has long been a national concern, partly due to the lack of understanding of complicated driving behaviors, especially decision-making mechanisms present when drivers face signal phase changes. This research tackles this issue by conducting a comprehensive investigation of driver decision-making at signalized intersections and, using insights from these investigations, developing a stochastic traffic flow model for such interrupted flow situations. Through the integration of the developed traffic flow model with connected vehicle technologies (specifically, V2X), the developed traffic flow model can be used to quantify the safety performance of signalized intersections, identify emerging hazardous situations, and contribute to the development of driver-assistance and intersection-accident-avoidance technologies. Furthermore, the outcomes from this research will create opportunities for students, particularly those from underrepresented groups, through educational and research opportunities in transportation engineering. It will also provide opportunities for college students, traffic engineers, and K-12 faculty and students through curriculum development, educational modules, and creation of pathways between high school and college, as well as college and graduate school. The objectives of this research include: 1) investigating complicated driving behaviors and the inner mechanisms of drivers' decision-making at signalized intersections through the analysis of vehicular trajectory data extracted from video images and 2) developing a stochastic traffic flow model to describe complicated driving behaviors and predict potential traffic conflicts at signalized intersections by considering the stochastic nature of drivers decision-making when facing signal phase changes under varying circumstances. This research will contribute to the theoretical development of traffic flow models. These traffic flow models will be able to describe complicated driving behaviors and estimate traffic conflicts at signalized intersections, both of which are absent from most other traffic flow models. Furthermore, this research is expected to contribute significantly to the improvement of intersection safety. This research will build a foundation for the future development of dynamic systems for alerting drivers of emerging hazards and helping to avoid intersection accidents.