Machine learning-based framework for reidentifying vehicles at a signalized intersection approach

Abstract: Signalized intersections are equipped with advance and stop-bar detectors that detect vehicles at discrete locations without linking or reidentifying them over the approach area. Accurate tracking and reidentification of vehicles between these detectors could provide valuable information on driver behavior, especially during the safety-critical yellow onset periods. However, reidentifying vehicles using non-visual high-resolution event data is challenging and not well-explored, with existing analytical models relying on a priori-calibrated parameters. To this end, we propose a machine learning (ML)-based reidentification framework for accurately tracking vehicles over the advance and stop bar loop detectors. The framework comprises two major components: ML models for accurately predicting the travel time between detectors and a novel optimization model that utilizes these predicted travel time and actuation events for reidentifying vehicles. Tests carried out on a major intersection approach in Phoenix, Arizona, showed that the XGBoost-based optimization framework reidentified vehicles even at congested conditions with 95.4% precision and 92% recall, outperforming other ML models and state-of-the-art analytical methods. The low false alarm rate and high recall of this reidentification framework opens avenues for obtaining valuable driver behavior data at the yellow onset to analyze stop/go behavior, dilemma zone entry/exit, red light running, and crossing conflicts at signalized intersections.