Comprehensive Assessment Framework for Collision Risk in Wide-Range Expressway Diverging Area Using Two-Dimensional Vehicle Trajectory Data

Abstract Expressway diverging areas exhibit a higher vehicle collision risk than basic segments due to frequent weaving and diverging traffic. Existing studies rely on 1D indicators and focus on the exit ramp and adjacent downstream sections, underestimating collision risks in wide-range diverging areas. This study proposes a two-step framework for comprehensive collision risk assessment in such areas. First, global vehicle trajectories with 2D kinematic parameters are extracted from videos recorded by multiple roadside cameras. Then, anticipated collision time (ACT), a 2D surrogate safety indicator, is integrated with the spatiotemporal overlap interaction filter to detect conflicts using the extracted trajectories. A case study is conducted on a wide-range expressway diverging area, divided into three sections for risk comparison. Results show that most detected conflicts involve only passenger cars, while those involving heavy vehicles are more frequently associated with unsafe scenarios. The spatial distribution of the conflict frequency revealed significant variations across sections. The midstream section exhibits the highest frequency, followed by the downstream section, while the upstream section has the lowest. Conflict-prone zones expand longitudinally and laterally across all sections as the ACT threshold is relaxed, and their spatial distribution demonstrates heterogeneity. The proposed framework supports specific traffic safety improvements in wide-range expressway diverging areas.