Prediction of Red-Light Running on Basis of Inductive-Loop Detectors for Dynamic All-Red Extension

Abstract

Dynamic all-red extension (DARE) has recently been reported as a countermeasure to safety hazards caused by red-light running (RLR). DARE dynamically extends an all-red interval by a few seconds when detection has determined that an RLR hazard has a high probability of occurring. Previous studies of RLR prediction methods were usually based on high-quality sensors for vehicle detection. This study addresses the issues of (a) the feasibility of the use of existing configurations of inductive-loop detectors (ILDs) for RLR prediction and (b) the achievable performance of ILDs in this capacity. One challenging problem is the limited resolution and type of measurements that ILDs can provide. To overcome this problem, an innovative set of driver behavior parameters was developed, and two arrival time estimators with car-following characteristics were designed to realize the prediction of RLR. The empirical data from the field intersection were used to validate the performance of the RLR predictors. The results showed that a reasonable prediction of RLR could be achieved for an operational DARE from existing ILD vehicle detection systems that contain at least two ILDs: the advance loop closest to the stop bar and the leading presence loop.