Assessing motorcyclist safety at unsignalized intersection using automated trajectory data analysis.

Objective: In a developing country like India, where the share of motorcyclists is increasing exponentially, their road crashes are also rising at an alarming rate. The majority of these road crashes occur at unsignalized intersections. Therefore, the present study aims to analyze the safety of motorcyclists at unsignalized three-arm intersections under a heterogeneous traffic environment using a fully automated trajectory data extraction tool.

Methods: The study first analyses the most frequent types of interactions that occur between motorcyclists and other road users at unsignalized intersections. Then, the study examines the interactions between motorcyclists and other vehicles by analyzing the speed of both vehicles involved in these interactions. Lastly, the study employs a supervised classification technique, Support Vector Machine (SVM), to categorize these interactions into critical, mild, and safe based on surrogate safety indicators (for the proximity of interaction) and the maximum speed (for the severity of an interaction) at which the vehicles interact.

Result: The results indicate that rear-end conflict was the most frequently observed conflict at the unsignalized intersections. Further, the study emphasizes the crucial role of speed during interactions, particularly at higher speeds, where elevated threshold values of PET and TTC significantly influence the severity of the interaction.

Conclusion: Overall, the research provides an essential insight into motorcyclists' safety in terms of critical conflicts at an unsignalized three-arm intersection. The findings of the research demonstrate the remarkable potential of fully automated trajectory data analysis software in evaluating safety at unsignalized intersections.