Date of Award


Document Type


Degree Name

Master of Science (MS)


Computer Engineering and Sciences

First Advisor

William Arrasmith

Second Advisor

L. Daniel Otero

Third Advisor

Barry Webster

Fourth Advisor

Philip J. Bernhard


Highway-rail grade crossing accidents are the second leading cause of fatalities within the railway transportation industry. Highway-rail grade crossing fatality statistics have failed to improve over the last decade, and in some years have increased. Because of these trends, research has found that considerable Federal funding initiatives exist to support and develop contemporary risk reduction methodologies involving machine learning and artificial intelligence. In accord with these initiatives, this thesis investigates the highway-rail grade crossing safety problem using machine learning to implement an ensemble decision tree-based classification application. The resulting classifier has been validated to achieve a less than three percent false omission rate and less than fifteen percent false negative rate for fatal accident classification. Classifiers such as the type developed within this research effort have strong potential to serve as risk reduction tools for existing infrastructure throughout the United States and can be used to prioritize funding for the most at-risk crossings. The classifier would also be beneficial to new railway systems engineering efforts by using a risk -based approach for assessing potential designs.


Copyright held by author