Systematic evaluation of lane-change risk for multi-vehicle types on freeways using high-resolution data

Traffic collision technology is widely used to identify lane change risk, which provides support for improving the safety of automatic driving assistance system. In most existing studies, micro-conflict indicators such as Time-to-Collision (TTC) and Stopping Distance Index (SDI) are used to determine the risk level during the lane change process. Therefore, this study focuses on the safety of the whole lane changing process, regards the lane changing vehicle and its four surrounding vehicles as a system, and proposes a method to quantify the comprehensive risk of lane changing from both the spatial and temporal dimensions. First, lane change samples are extracted from the real trajectory data set. Then, based on TTC and SDI, a macro conflict index, Risk Exposure Level (REL) and Risk Severity Level (RSL), is formulated to quantify the risk between lane changing vehicles and surrounding vehicles in the whole process of lane change. Finally, two macro conflict indicators are integrated to quantify the risk of the whole lane change system via Fault Tree Analysis method. The effectiveness and authenticity of the proposed method are verified by the HighD dataset. Dividing vehicles into different types and analyzing the factors that influence the risk levels of each vehicle type. The results are expected to provide a reference for intelligent vehicle risk warning and trajectory planning, and is valuable in the formulation of highway safety operation management measures.