Creating a real-time road safety evaluation of collision risks based on traffic state estimation

Summary

Vehicle-to-vehicle collisions are a major cause of traffic fatalities, and addressing them is needed for Minnesota’s Toward Zero Deaths. Such collisions are more likely to occur during certain traffic conditions. For example, "turbulence" in freeway traffic flow (variations in speed and flow among different lanes) is correlated with higher crash rates. If we could detect traffic conditions with high collision risk, we could take mitigating actions to reduce collisions or fatalities from collisions. Such actions might include reducing incident response times by focusing monitoring at traffic management centers on high-risk areas or preemptively relocating emergency response teams. In the long-term, road sections with patterns of high collision risks could be modified with signage, pavement markings, or other design changes to alert drivers or influence their behavior. This project will enable such mitigation by creating real-time estimates of crash risk for traffic engineers. To estimate collision risks, researchers start by estimating traffic conditions. They process detector counts with a traffic flow model such as the Payne-Whitham model, based on second-order kinematic wave theory, to interpolate traffic conditions between detectors. This approach has high potential for Minnesota freeways and roads, which typically have detectors at periodic intervals and at intersections. Using published literature on correlations between traffic conditions and collision risks, researchers will convert the estimated traffic conditions into predicted collision risks. This process will be implemented in software. To make it real-time, the research team needs real-time access to detector counts. For freeways, the team seeks to implement this system within MnDOT's open-source Intelligent Roadway Information System (IRIS) software, which will also expose the real-time collision risks to traffic management centers using IRIS.