Citation

Wang Y, Li Z, Liu P, Xu C, Chen K. Transp. Res. C Emerg. Technol. 2024; 161: e104543.

Copyright

(Copyright © 2024, Elsevier Publishing)

DOI

10.1016/j.trc.2024.104543

PMID

unavailable

Abstract

Traffic oscillations, also known as stop-and-go driving conditions, commonly emerge due to the alternative deceleration and acceleration of vehicles in congested traffic conditions. This study proposes a surrogate safety measure (SSM) specifically for estimating crash risks related to this phenomenon. The Crash Risk Indicator for Oscillation (CRIO) is established by integrating driving behavior during oscillations into the Time-to-Collision (TTC) framework. A year-long recording of a busy freeway on sunny weekdays at morning peak hours using unmanned aerial vehicles produced over 200 h of empirical trajectories of all vehicles, which encompassed 20 rear-end crashes that happened in the real world. The performance of the CRIO was assessed by comparing it with five commonly used SSMs. It was found that the oscillation amplitude induced by the leading vehicle and the behavior of the following vehicle during its reaction time significantly affected the likelihood of rear-end crashes. The CRIO diminished the false alarm rate from a range of 25% to 91% (by other SSMs) to 6% with a recall rate of 100%. The CRIO was further employed to predict real-time crash risk once the oscillation was formed. The results show that the CRIO can provide an early crash warning with an average time of 1.72 s in advance. This research contributes to the development of more robust proactive systems for early warning by enabling reliable and effective risk assessments for traffic oscillations.

Language: en

Keywords

Crash risk indicator; Driving behavior; Real-world crash trajectories; Traffic oscillation