Application of Connected Vehicle Data to Assess Safety on Roadways

Abstract

Using surrogate safety measures is a common method to assess safety on roadways. Surrogate safety measures allow for proactive safety analysis; the analysis is performed prior to crashes occurring. This allows for safety improvements to be implemented proactively to prevent crashes and the associated injuries and property damage. Existing surrogate safety measures primarily rely on data generated by microsimulations, but the advent of connected vehicles has allowed for the incorporation of data from actual cars into safety analysis with surrogate safety measures. In this study, commercially available connected vehicle data are used to develop crash prediction models for crashes at intersections and segments in Salt Lake City, Utah. Harsh braking events are identified and counted within the influence areas of sixty study intersections and thirty segments and then used to develop crash prediction models. Other intersection characteristics are considered as regressor variables in the models, such as intersection geometric characteristics, connected vehicle volumes, and the presence of schools and bus stops in the vicinity. Statistically significant models are developed, and these models may be used as a surrogate safety measure to analyze intersection safety proactively. The findings are applicable to Salt Lake City, but similar research methods may be employed by researchers to determine whether these models are applicable in other cities and to determine how the effectiveness of this method endures through time.