Surrogate Safety Data

Crash data is limited by reporting accuracy and randomness associated with human events. Crash data only represents the events that are reported and contains limited detail about what happened. In addition, crash data can take several months to be reported and a sample of several years is usually required to understand patterns. Surrogate safety measures attempt to overcome crash data limitations by introducing new data sources and analysis methods.

Connected vehicles are one source of surrogate safety data being investigated by UDOT. Vehicles millions of data points about the vehicle's environment, telemetry, location, and interactions. These data could significantly improve our understanding of safety and how to improve it.

Two pilot projects were recently completed regarding connected vehicle data. The goal of these pilots is to understand what data is available, how it is delivered and collected, and how it may correlate to crash data. The UDOT Traffic Management Division is engaging in separate efforts to understand how these data could be used for traffic operations.

General Motors (GM)

UDOT Traffic & Safety contracted with GM to evaluate a web-based beta product providing access to connected vehicle data. The pilot site was used for approximately six months from November 2021 to April 2022. The data include various event readings from GM branded vehicles including hard braking, speeding, seat belt use, and vehicle type. The web interface provides a map based portal with data aggregated to a total risk score for roadway segments. Seat belt use metrics are also included.

For the pilot project, GM data was integrated into various existing UDOT Traffic & Safety processes in order to test potential uses. Internal and consultant staff involved with traffic studies, project safety analyses, intersection control evaluations, and HSIP applications used the GM data to look for ways to enhance their analysis. An evaluation team met monthly to share experiences and ideas. Feedback and examples of use are currently being summarized in a lessons learned report to assist with future evaluation or data initiatives.

Figure 1: The GM beta tool included a map interface with risk scores, trends, and seat belt usage.

Michelin

Michelin contracted with the UDOT Materials and Pavements Division to provide sample data for evaluation. Michelin is delivering several datasets representing insights from connected vehicle data. These data are aggregated from various manufacturers, insurance applications, fleet vehicles, and other onboard data collection systems. The data is delivered in tabular files and requires some manipulation before use.

Initial evaluations were performed to compare the Michelin data to crash data for the purpose of analyzing possible correlations. If a statistically significant relationship between crashes and near-misses can be established the data could be used to supplement crash data analysis. Similar to GM, the results of the analysis will be organized in a report detailing lessons learned and outcomes. Several universities are currently evaluating these same relationships, results of those studies will be carefully reviewed by UDOT

Future Work

Both pilot projects use speed and hard braking data. In the future Traffic & Safety would like to evaluate lane departure, rear-end collision avoidance, traction control, and other vehicle events that may correlate to hazardous conditions. In addition, many other suppliers of data have approached the department with similar products. The results of these two pilot projects will assist UDOT in making informed decisions in the future.