Estimation of traffic conflicts using precise lateral position and width of vehicles for safety assessment

• Published in: Accident analysis and prevention 2019-11, Vol.132, p.105264-105264, Article 105264
• Main Authors: Charly, Anna, Mathew, Tom V.
• Format: Article
• Language: English
• Subjects: Crashes; Non-lane-based traffic; Safety evaluation; Surrogate measures of safety; Traffic conflicts; Traffic safety
• ISSN: 0001-4575; 1879-2057
• DOI: 10.1016/j.aap.2019.105264

•A generic methodology for traffic conflict estimation is proposed.•Considers lane-based, non-lane-based and multi-class traffic conditions.•Estimated conflicts show temporal and spatial correlation with real crashes.•Conflict estimation using Modified TTC is not effective for night traffic.•Results show suitability of proposed methodology for diverse traffic conditions. Surrogate measures of safety (SMoS) aims at road safety evaluation without depending on historical crash data. Existing studies have evaluated SMoS in traffic conditions having good lane discipline. However, in several traffic conditions vehicles do not follow good lane discipline resulting in high crash rates. Moreover, existing studies do not consider type of the vehicle explicitly while estimating conflicts. This study aims to address these gaps by proposing a generic methodology for safety evaluation applicable also in non-lane-based multi-class traffic conditions. It utilizes precise position of the vehicles and their widths to identify critical interactions between all types of vehicles. Conflicts are then estimated from these critical interactions using the modified time-to-collision (MTTC), an existing SMoS. The proposed methodology is evaluated in both lane-based as well as non-lane based traffic conditions. The former uses NGSIM trajectory data and compares the estimated conflicts with the literature. The latter, on the other hand, uses simulated vehicle trajectories from an expressway but compares the estimated conflicts with historical crash data. The results show that estimated conflicts exhibit significant temporal and spatial correlation with real crashes. It also shows the suitability of the methodology for diverse traffic conditions.