Predicting Driving Distraction Patterns in Different Road Classes Using a Support Vector Machine

This study investigates driving behavior under distraction on four different road classes – freeway, urban arterial, rural, and local road in a school zone – using a high-fidelity driving simulator. Some 92 younger participants from a reasonably diverse sociodemographic background drove a realistic midsize network in the Baltimore metropolitan area and were exposed to different distractions. A total of 1,952 simulation runs were conducted. An ANOVA and Tukey Post Hoc analysis showed that distracted driving behavior demonstrates different patterns on various roads. This research developed a support vector machine model that achieved distraction prediction ability among different routes with an accuracy of 94.24%, which to the best of the authors' knowledge, is the best for such a task. The results indicate that driver distraction prediction models probably would be more accurate if developed separately for each road class.


  • English

Media Info

Subject/Index Terms

Filing Info

  • Accession Number: 01765392
  • Record Type: Publication
  • Files: TRIS
  • Created Date: Jan 25 2021 4:20PM