Risk assessment of various car-following models of new generation vehicles in mixed traffic

Abstract

New generation vehicles, namely autonomous vehicles (AV) and connected autonomous vehicles (CAV), are advancing rapidly with goals of full automation on roads in the near future. The modelling of car-following (CF) behaviour has progressed significantly over the last few decades with many variants of car-following models appearing. However, very little research focuses on the safety and risks of car-following models. This paper aims to analyse and provide a comprehensive comparison of different car-following models based on their safety impact in a traffic network. The Pan Island Expressway (PIE) and Bukit Timah Road will be utilised as the two networks on which the simulations will run on. Simulations of Adaptive Cruise Control (ACC), Cooperative Adaptive Cruise Control (CACC) and SmartSK car-following models against the Intelligent Driver Model (IDM) car-following model will be conducted. In the simulations, market penetration rate (MPR) and traffic density are also taken into consideration. After each simulation process is completed, Surrogate Safety Assessment Model (SSAM) measures such as Time-to-Collision (TTC) and Post Encroachment Time (PET) will be produced to analyse the risks and occurrences of potential collisions. Results show that IDM and SmartSK car-following models fare better in terms of their safety performance in traffic networks as compared to ACC and CACC car-following models. Furthermore, the varying performance of car- following models on highway and intersection can also be seen through the results. At the intersection, vehicles tend to have greater critical potential collisions in comparison to the highway. Insights from this study will help deepen understanding on the safety performances of the respective car-following models used.