Algorithm for Predicting Inattentive Signal Violators in an Infrastructure-Based Intelligent System

摘要

In June 2003, the FHWA introduced the first intelligent intersection, capable of assisting drivers in avoiding right-angle collisions that typically occur at intersections. One of the concepts that intelligent collision-avoidance systems displayed was a violator warning system, consisting of an illuminated sign with strobe lights to alert distracted drivers that they may violate an approaching red signal. Proposed is a novel approach for creation of an algorithm to predict and alert likely inattentive signal violators in time for them to react to an infrastructure-based warning. The algorithm is based on comparing velocity and acceleration data of vehicles traveling at free-flow speeds through an intersection with data from alert motorists stopping for a red signal on the same approach. Kinematic data were obtained at four intersections for two sets of motorists, those who were stopping for a red light and those who were going through a green light. It was presumed that inattentive violators act identically to attentive drivers with a green signal. Comparison of speed and acceleration profiles revealed distinct differences that could be used to distinguish into which subset a given vehicle would fall, well in advance of the intersection. These comparisons yield a simple, economical, and site-specific algorithm for detecting inattentive motorists about to violate a signal.