Driver Behavior at Highway-Rail Grade Crossings Using Naturalistic Driving Study Data

Project Description Traffic incidents at highway-rail grade crossings continue to be a danger to both highway users and railroads causing over 250 fatalities each year. Michigan Tech will use the NDS data from grade crossing segments to analyze driver behavior at crossings, and to evaluate the effectiveness of a variety of traffic control devices currently in use at crossings. They used standard data analysis techniques, utilizing a variety of software tools, including Excel, MATLAB, and available database software. Project results will be used to investigate driver reaction to the existing traffic control devices, and impacts of the crossing environment, including weather, time of day, driver familiarity with the crossing, surface roughness. We will also investigate impacts from driver demographics. Future work will include using the NDS research results to inform and validate ongoing driver simulator research efforts. Data Request Scope The motivation for creating the dataset was to enable Michigan Tech to conduct a pilot investigation of driver behavior around highway-rail grade crossings. The data contained transits at three specific highway-rail grade crossings the first of which included two crossings, one active and one passive (NAVTEQ nodes 845714796, 96351877, and 96284604). Michigan Tech received video of front and rear views, time series data (excluding radar), and demographic information: participant ID, gender, and age group, for a total of 300 events. The event duration was ten seconds prior and ten seconds after each traversal. Data was sampled at 10 Hz. Additional reduction was provided for twenty transit events, ten active and ten passive crossings. These detailed narrative described driver behavior during each transit event. Data Specification Video files were provided as .mp4 format. Time-series files were delivered in .csv format, and demographic information was in .xlsx format. Please see attached data dictionary for a description of variables.