Does connected vehicle information reduce beyond-visual-range crash risk in foggy freeway conditions? A study based on extreme value theory

Intelligent connected vehicle technology can provide drivers with connected vehicle information (CVI) to enhance traffic safety. However, the impact of CVI on safety in the case of foggy freeway beyond-visual-range conditions is currently unclear. As such, this paper introduces extreme value theory (EVT) to assess and quantify this impact. Specifically, this paper first adopted driving simulation technology to build a connected environment experimental platform. A typical foggy freeway beyond-visual-range scenario was developed, and driving simulation experiments were carried out to collect driving behavior data in both traditional environments (without CVI) and connected environments (with CVI). With this data, the peak over threshold method of EVT was used to establish generalized Pareto distribution fitting models for the four indicators of time to collision (TTC), modified time to collision (MTTC), post-encroachment time (PET), and deceleration rate to avoid a crash (DRAC), respectively, and model comparisons and selections were performed. The optimal models were then chosen for risk assessment and impact analysis, which includes both crash probability and crash damage dimensions. The results show that the DRAC-based EVT models have better data-fitting performance and higher reliability. Additionally, CVI is effective in reducing the crash risk of beyond-visual-range events on foggy freeways, and there is diversity in crash risk and the effectiveness of CVI application between different driving groups. The study in this paper further extends the EVT and also helps to better understand the action and influence mechanisms of CVI.