According to the USDOT Federal Railroad Administration, highway-rail grade crossing and trespasser fatalities still account for 96 percent of all rail-related deaths. A review of accident causes reveals that one of the main accident factors is human driving behavior.
One of the potential approaches to improve drivers’ behavior at crossings is to systematically examine their actions and use the outcomes to develop alternative methods or approaches to decrease the probability of missing or ignoring warnings. The Naturalistic Driving Study (NDS) approach, using in-vehicle video and other sensors to directly observe drivers during normal driving activities, is a technique that shows promise toward improved understanding of driver actions in the crossings.
This two-year project is divided into two phases with an overall objective to investigate driver behavior at highway-rail grade crossings using two distinct, but complimentary techniques. Phase I will use data collected under the Strategic Highway Research Program (SHRP) Naturalistic Driving Study to look at how normal drivers react at crossings in every day driving situations. Phase II will use the understanding developed in the first phase to create scenarios that resemble environments similar to those found in the NDS for use in our driver simulator environment. The research will look for two basic results. First, we will develop and use the organized NDS crossing database to examine behavioral trends at the crossings. Second, we will compare driver behavior in the simulator with that found in the NDS data to determine the level of correlation between the two environments. Our hypothesis is that a strong correlation would allow us to outline how to use the simulator environment in the future to predict driver response to a variety of crossing parameters.
Sponsor: Michigan Tech Transportation Institute (MTTI)
PI: Dave Nelson
Michigan Tech researchers have been involved in developing a program to investigate driver behavior at highway-rail grade crossings over the last couple of years. After a long downward trend for grade crossing accidents and fatalities, the statistics for problems at crossings have plateaued over the last few years.
One of the ways to address the issue is by comparing actual driver behavior at crossing with simulated one. Starting in late 2013 results from the Naturalistic Driving Study (NDS), conducted under a major Strategic Highway Research Program (SHRP) 2 program have been available to researchers that will allow us to fill some of the gap, and begin the process of validating the ongoing simulator research. This initiative grant is requested to secure NDS data for early analysis, so a stronger case can be made to funding agencies for ongoing research. The Michigan Tech research team will acquire camera (forward and backward video) and vehicle performance data (braking, throttle, and other data) showing what drivers were doing during normal day-to-day driving situations at grade crossings. Our initial results seem to indicate that drivers are not looking for trains at crossings. This is a big issue, especially at passive crossings, as drivers do not stop for trains if they do not look for them. We will use the NDS data to confirm that finding, and then compare the NDS data with our simulation research findings to validate the simulator process. Information about the NDS project can be found at http://www.shrp2nds.us/index.html, a document from the NDS, The SHRP 2 Naturalistic Driving Study, is included in this package.
This project will analyze driver behaviors at highway-rail grade crossings. Reducing collisions between cars and trains at these crossings has long been a goal of the Federal Rail Administration, the Federal Highway Administration, and the Federal Transit Administration. Improved understanding of driver behaviors can lead to improved traffic control devices for rail crossings and help meet this goal. The results of this study will also be used to validate research Michigan Tech is working on with a driving simulator to investigate crossing behaviors in a less costly fashion.