Identifying Potential Tailgaters Using Matched Case-Control Logistic Regression

Abstract

Road collisions arise from interactions involving human factors, the environment, and road layout. Driving simulators, widely applied in rear-end collision studies, provide a secure environment to explore human errors, which are not observable through microsimulation tools. These simulators also facilitate the examination of driving behaviour in the presence of connected vehicles. This study aims to identify driver-related factors contributing to rear-end collisions in a driving simulator and to detect potential tailgaters behind a connected vehicle with connected cruise control. Using case-control logistic regression, participants with the potential to be involved in rear-end collisions are considered potential tailgaters, while non-potential participants serve as controls. The results reveal statistically significant factors, such as headway time and maximum brake mean values, in relation to rear-end collisions. Furthermore, employing regression outputs, log relative risk and survival function, with predefined thresholds effectively identifies potential tailgaters, achieving accuracy rates of over 90% and 97%, respectively.