Pedestrian and Bicyclist Safety at Highway-Rail Grade Crossings – Assessment of Crash Predictions (Year 2 Report/Phase II Report)

Details

• Creators:
  Khattak, Aemal J. ; Aman, M. Naveed ; Bukhari, Moomal
• Corporate Creators:
  University of Nebraska-Lincoln. University Transportation Center for Railway Safety ; Mid-America Transportation Center
• Corporate Contributors:
  United States. Department of Transportation. University Transportation Centers (UTC) Program ; United States. Department of Transportation. Office of the Assistant Secretary for Research and Technology
• Subject/TRT Terms:
  Artificial Intelligence Bicycle Crossings Crash Analysis Crosswalks Data Science Mathematical Prediction Object Detection Pedestrian Safety Railroad Grade Crossings Safety
• Publication/ Report Number:
  UTCRS-UNL-O4CY24
• Resource Type:
  Tech Report
• Geographical Coverage:
  United States ; Kansas ; Missouri ; Nebraska ; Iowa
• Edition:
  June 1, 2024 – January 31, 2026
• Corporate Publisher:
  University of Nebraska-Lincoln. University Transportation Center for Railway Safety
• Abstract:
  This report presents results of a multi-state effort to improve pedestrian and bicyclist safety at highway-rail grade crossings (HRGCs) by explicitly incorporating non-motorist traffic exposure into crash prediction models. Building on an AI-based video pipeline from previous research, empirical non-motorist counts from 18 rail crossings in Lincoln, Nebraska, were used to calibrate a regression model that predicted daily pedestrian and bicycle traffic volumes at 13,672 public rail crossings in Nebraska, Kansas, Missouri, and Iowa. These traffic volume predictions were integrated into a modified Accident Prediction and Severity (APS) framework using a Zero-Inflated Negative Binomial (ZINB) model, showing an average of 5.5% increase in predicted crashes and much larger increases at crossings with substantial non-motorist traffic. A two-stage AI approach, Random Forest classification followed by Random Forest regression, further improved predictive performance relative to the ZINB and provided a practical tool for screening and prioritizing high-risk rail crossings. The findings demonstrate that non-motorist traffic is a critical but previously underrepresented factor in rail crossing safety assessment and support a scalable, data-driven framework for multimodal safety management.
• Format:
  PDF
• Funding:
  69A3552348340
• Collection(s):
  University Transportation Centers
• Main Document Checksum:
  urn:sha-512:d0d6191215005c6bf0c79cebc57d4aa07905126e408a5e44a48c34b9c78c58f436fa8a3030ae284576fa50254637e5b301199298e6757bd1efed1fbc1fd12757
• Download URL:
  https://rosap.ntl.bts.gov/view/dot/89550/dot_89550_DS1.pdf
• File Type:
  [PDF - 2.06 MB ]