Microsimulation Approach to Investigate the Impact of Incorrect Automated Pedestrian Detection on the Operation of Signalized Intersections

Abstract

Numerous studies have assessed video detection systems, but automated pedestrian video detection systems (APVDS) have not received as much attention. Although some researchers have evaluated the accuracy of APVDS, none quantified the consequential effects that false or missed pedestrian calls have on critical vehicular and pedestrian performance metrics. To address this gap, this study introduces a microsimulation-based approach to analyze the impact of missed and false pedestrian calls on signalized intersection operation. This method employs simulation inputs that mimic various detection scenarios from the field, allowing for greater flexibility in analysis. Additionally, the study compares the impacts of incorrect calls under different operational strategies, including APVDS with and without the pedestrian recycle feature of the controller, pedestrian recall operations, and common push-button operations with or without pedestrian recycle. Results show that the proposed microsimulation approach is a valuable tool for future studies, especially considering the idiosyncratic nature of specific sites where the percentage of false and missed calls depends on local conditions. The study reveals that neither of the two analyzed APVDS systems is mature enough to work as a standalone option. Microsimulation results show that even though the single-camera video detection system (VDS) appears to be a safer option than the double-camera VDS, it fails to provide any benefits in pedestrian and vehicular delay over the pedestrian recall operations. This research may be of practical significance as transportation agencies could utilize the proposed microsimulation approach to evaluate APVDSs effectively. Knowing the strengths and limits of APVDS can help agencies make informed implementations.