Skip to main content
ICTD 2025 Main banner
Final Program


Icon Legend
This session is not in your schedule.
This session is in your schedule. Click again to remove it.
Back

ICTD 2025

Rail Transport

Innovations at Highway-Rail Grade Crossings

Deep Learning Techniques for Evaluation of Hang-Up Susceptibility of Highway Railway Grade Crossings

Wednesday, June 11, 2025
10:15 AM – 10:30 AM MT
Abstract: Abstract
Steep-profile Highway Railway Grade Crossings (HRGCs) pose substantial safety risks by increasing the likelihood of vehicles getting caught or “hung up” on the crossing, a situation that can result in severe collisions with oncoming trains. Assessing the hang-up susceptibility of HRGCs is essential for reducing these risks and for identifying crossings that may need to be designated as “hump crossings” to alert drivers to their elevated hazard level. Direct assessment methods, however, are often challenging due to the high costs associated with obtaining HRGC profile data and the need to implement traffic control measures during assessment. This study introduces a novel deep learning methodology, specifically utilizing a Convolutional Neural Network (CNN), to streamline the evaluation of HRGC hang-up susceptibility. Data collection efforts included gathering GPS-based elevation and Inertial Measurement Unit (IMU) readings from a sample of 225 HRGCs along the Red Rock BNSF railway corridor in Oklahoma. Using these field-collected data inputs, the CNN model was trained to predict the level of hang-up risk across five distinct levels (L1 to L5), which were verified through direct field observations. This approach presents a cost-effective and efficient alternative for assessing HRGC hang-up risks, with significant potential to enhance safety measures at railway crossings.

Keyword: Hang-Up Susceptibility, Highway Railway Grade Crossing, Deep Learning

Learning Objectives:

Attendees can expect to learn the following from this session:
  • na
  • na
  • na
  • na
  • na