An Algorithm for Identifying Red Light Runners from Radar Trajectory Data

Drivers often experience uncertainty when traffic signals change from green to yellow, and they must quickly decide whether to stop at an intersection or continue driving. This situation is called the "dilemma zone," and a great deal of research has been performed to minimize the uncertainty in that zone. The goal is to have the minimum number of vehicles caught in the dilemma zone and reduce red light runners while maximizing the green light period. In order to determine how well dilemma zone protection systems perform, we need to measure the frequency of red light runners. This frequency is closely correlated with crashes, but is difficult to collect. Recent algorithms utilize radar for dilemma zone protection. However, verification of the number of red light runners requires additional video detection. In this study, we propose a new algorithm to predict red light runners and distinguish them from right turners on red. We used Canonical analysis to exclude right turners from red light runners. Our model has 96% accuracy.