Conflict-based crash risk estimation of heterogeneous lane-changing traffic at the Panipat Toll Plaza (NH-44, India) using surrogate safety measures and UAV-based trajectory data

A gradual increase in the number of lanes and frequent lane-changing behaviour characterizes the approach of the toll plaza. These characteristics significantly increase the propensity for conflicts and collisions. This study aims to estimate the crash risk of heterogeneous lane-changing traffic at the approaching section of the toll plaza by analyzing vehicle trajectories. In this study, traffic data was collected from a toll plaza located on National Highway-44 in Haryana, India, using an Unmanned Aerial Vehicle (UAV). The vehicle trajectory data was retrieved using Data from Sky (DFS), a fully automated image processing software. The traffic crash risk was assessed using Extreme Value Theory (EVT) in conjunction with Lane Changing Time to Collision, a Surrogate Safety Measure (SSM) indicator. A comprehensive assessment of crash risk across vehicle categories indicates a negative relationship between vehicle size and conflict involvement, with larger vehicles such as trucks, buses, and Light Commercial Vehicles (LCVs) exhibiting a reduced likelihood of conflicts compared to two-wheelers and cars. Moreover, vehicle speed demonstrated a positive correlation with crash risk, indicating that higher average speeds are associated with an increased likelihood of crashes. The study is limited to a single morning peak-hour dataset and primarily covers motorized vehicles, as non-motorized traffic is prohibited on access-controlled highways. Additionally, the current video-classification technique could not differentiate between electric and conventional fuel-powered two-wheelers. These limitations should be considered while determining the scope and generalizability of the findings. The study findings are expected to assist engineers and toll plaza operators in selecting suitable traffic control measures to improve safety at the toll plaza. • Examining crash risk in toll plaza approach sections using the Peak Over Threshold (POT) method. • Lane Changing Time-to-Collision (LTTC) was chosen as a surrogate safety indicator. • A comprehensive assessment of crash risks related to different vehicle types is conducted.