Infrastructure Safety Assessment in a Connected Vehicle Environment

The goal of the Infrastructure Safety Assessment in a Connected Vehicle (CV) Environment project was to develop a method to identify infrastructure safety “hot spots” using CV data. Using these basic safety messages to detect hot spots may allow for quicker discovery than traditional methods, such as police-reported crashes. The basic safety message may be able to detect events that police normally cannot obtain, including unreported crashes and near-crashes. The project successfully explored some models and algorithms to detect crashes and near-crashes and also designed a methodology to apply to hot spot identification. With the data available, conclusive results were not achieved; however, the models showed some potential. Three techniques were tested to predict crashes using vehicles’ kinematic data. To predict where a crash was occurring, multivariate adaptive regression splines, classification and regression trees, and a novel pattern matching approach were all tested. The models were able to identify the majority of 13 known crashes with different amounts of false positives. The pattern matching approach outperformed a simple acceleration threshold by identifying nearly 70% of crashes in a crash-only test set and 74% of near-crashes in a near-crash only test set. On the training set, it was able to identify more crashes than the thresholds without increasing the number of false positives observed. Based on the work described in this report, the CVI-UTC is fully prepared to apply the methodology to data collected on the field test bed.

  • Record URL:
  • Supplemental Notes:
    • This research was sponsored by the U.S. Department of Transportation, University Transportation Centers program.
  • Corporate Authors:

    Virginia Tech Transportation Institute

    Blacksburg, Virginia  United States 

    University of Virginia, Charlottesville

    Center for Transportation Studies
    P.O. Box 400742, Thornton Hall, D228
    Charlottesville, VA  United States  22903

    Morgan State University

    Baltimore, MD  United States  21251

    Research and Innovative Technology Administration

    Department of Transportation
    1200 New Jersey Avneue, SE
    Washington, DC  United States  20590
  • Authors:
    • Smith, Brian L
    • Kluger, Robert
    • Park, Hyungjun
  • Publication Date: 2015-12-15

Language

  • English

Media Info

  • Media Type: Digital/other
  • Edition: Final Report
  • Features: Figures; References; Tables;
  • Pagination: 36p

Subject/Index Terms

Filing Info

  • Accession Number: 01594313
  • Record Type: Publication
  • Files: UTC, NTL, TRIS, RITA, ATRI
  • Created Date: Mar 2 2016 12:00PM