Low-Cost Real-Time Threat Notifications and Surrogate Safety Metrics at Highway-Rail Grade Crossings

Abstract: The Federal Railroad Administration reports more than 203,000 highway-rail grade crossings (HRGCs) in the US. In 2023, 228 fatal crossing incidents occurred at these HRGCs, with 96.9% of all rail-related fatalities attributable to HRGCs or trespassing. The National Highway Traffic Safety Administration reports that 94 percent of serious crashes at HRGCs are because of human errors and choices, and almost all these crashes are preventable.
Advanced warning systems can potentially reduce, and possibly eliminate human errors to enhance HRGC safety. With the emergence of connected vehicle-to-everything (V2X) technologies, it is technically possible to provide accurate information for train engineers, drivers, and pedestrians so that they can make correct stop-or-go decisions at HRGCs.
This paper will describe a low-cost prototype advanced warning system for reporting risks and calculating safety surrogate metrics in real-time at HRGCs. It will illustrate the potential uses, including providing driver and pedestrian information such as train arrival times and expected delays or wait times.
Although train arrival warning signals at HRGCs have been studied in the past, the proposed system leverages developments in advanced edge computing and machine learning to improve the practicality and availability of a low-cost modular and scalable system independent of existing infrastructure.
The proposed methodology will potentially allow for transferable deployments anywhere in the U.S. Additionally, it will provide a safety monitoring system that is independent of in-situ connectivity, making it especially impactful for rural locations and short lines.