Automated Bicycle Counts

Abstract

Quantifying bicycle use is fundamental to understanding bicycle travel. Methods of counting bicycles vary from limited-time, manual counts to permanent overhead imaging sensors. One common permanent counting method uses inductive loops embedded in the pavement to count cyclists on paths. Although inductive loop detectors have been found to be a highly accurate method of counting bicycles under ideal test conditions, their accuracy after years of use has not been systematically studied. This study focuses on bicycle counts collected by the City of Boulder, Colorado, since 1998, on multiuse paths with inductive loop detectors. To estimate the accuracy of the devices in use, two individuals manually counted path users at six locations. On average, the loop detectors counted 4% fewer bicycles than the manual counters at the same locations. Of the 22 detector channels with sufficient counts to judge their accuracy, roughly 68% were considered accurate. The most dramatic inaccuracies were caused primarily by detector settings and software-related problems. This study found that inductive loop detectors can provide accurate measures of bicycle use on a pathway, but only when detectors are properly installed, calibrated, maintained, and free of external interference.