Enhancing Intersection Safety: A Fusion of Historical and Real-Time Data For Route Planning

In 2022, approximately 42,795 people died in the United States from motor vehicle crashes, according to the National Highway Traffic Safety Administration (NHTSA). More than 50% of motor vehicle deaths and injuries occur at or near intersections. In Michigan, over 29% of fatal vehicle crashes occurred at intersections in 2022. Sensor technological advancements have allowed vehicles on the road to estimate the distance from other vehicles and provide real-time warnings to drivers. While sensors allow features such as Automatic Emergency Braking to engage the vehicle’s brakes when it detects that a collision with another object is imminent—be it another car or an obstacle—they also have limitations due to range, field of view, weather conditions, calibration, interference, and many more. In this paper, we employ a data-driven strategy that combines historical intersection accident data collected from the Michigan Traffic Crash Facts (MTCF) database with real-time road traffic data to reduce the likelihood of intersection accidents along a specific vehicle route. This approach is used to solve an optimization problem that generates a route minimizing the the overall probability of intersection accidents. Our results show that the proposed approach can enhance vehicle safety at intersections.