Use of real-world connected vehicle data in identifying high-risk locations based on a new surrogate safety measure

Abstract

Traditional methods for the identification of high-risk locations rely heavily on historical crash data. Rich information generated from connected vehicles could be used to obtain surrogate safety measures (SSMs) for risk identification. Conventional SSMs such as time to collision (TTC) neglect the potential risk of car-following scenarios in which the following vehicle’s speed is slightly less than or equal to the leading vehicle’s but the spacing between two vehicles is relatively small that a slight disturbance would yield collision risk. To address this limitation, this study proposes time to collision with disturbance (TTCD) for risk identification. By imposing a hypothetical disturbance, TTCD can capture rear-end conflict risks in various car following scenarios, even when the leading vehicle has a higher speed. Real-world connected vehicle pilot test data collected in Ann Arbor, Michigan is used in this study. A detailed procedure of cleaning and processing the connected vehicle data is presented. Results show that risk rate identified by TTCD can achieve a higher Pearson’s correlation coefficient with rear-end crash rate than other traditional SSMs. We show that high-risk locations identified by connected vehicle data from a relatively shorter time period are similar to the ones identified by using the historical crash data. The proposed method can substantially reduce the data collection time, compared with traditional safety analysis that generally requires more than three years to get sufficient crash data. The connected vehicle data has thus shown the potential to be used to develop proactive safety solutions and the risk factors can be eliminated in a timely manner.