Highway-rail grade crossing (crossing) collisions and fatalities have been in decline, but a recent ‘plateau’ has caused the Federal Railroad Administration (FRA) to concentrate on decreasing further casualties. The Michigan Tech Rail Transportation Program has been selected to perform a large-scale study that will utilize the SHRP2 Naturalistic Driving Study (NDS) data to analyze how various crossing warning devices affect driver behavior and whether there are clear differences between the effectiveness of the warning devices.
The main results of this study are the development of a coding scheme for a visual narrative, used to validate machine vision head tracking data, and an improved baseline for the head tracking data using bivariate probability density. Head tracking data from the NDS and its correlation with coded narratives are vital to analyze driver behavior as they traverse crossings. This paper also presents preliminary results for the comparative analysis of the head tracking data from an initial test sample. Future work will extend the analysis to a larger data set, and ensure that use of the head tracking data is a viable tool for the ongoing behavior analysis work. Based on preliminary results from testing of the first data set, it is expected there will be significant positive correlation in future samples and the machine vision head tracking will prove consistent enough for use in the large scale behavioral study.