AI Pedestrian Traffic Safety System

According to the U.S. Department of Transportation (USDOT) and the National Highway Traffic Safety Administration (NHTSA), the number of traffic fatalities in the United States was 36,096 in 2019, which is more than 11.17 deaths per 100,000 population. In 2019, more than 240 people died in traffic accidents in the city of Los Angeles alone, with pedestrians accounting for more than half (55%) of these deaths.  These numbers underscore the criticality of improving national pedestrian and bicyclist safety.

Until recently, there was no effective automated system for monitoring the movement of these vulnerable groups, making it difficult to determine which safety interventions were most effective. Understanding the need for clear measurement, researchers from the Mineta Transportation Institute  developed and implemented a new automated system using artificial  intelligence (AI) and computer vision to monitor, track, and  count the movement of pedestrians in major urban areas using existing traffic cameras. Led by Dr. Mohammad Pourhomayoun, the team created the AI Pedestrian Traffic Safety System, which collects and processes this data in real time to enable dynamic traffic control. The system is already being implemented and tested in downtown Los Angeles, and because it requires no new equipment, it can be cost-effectively scaled over time to improve safety in LA and beyond.