Using Inductive Loops to Count Bicycles in Mixed Traffic

Abstract

Inductive loops are commonly used for bicycle detection both on- and off-street, but until recently, few such detectors were able to differentiate between bicycles and motor vehicles. For this reason, automated bicycle counting is usually confined to off-street locations. With bicycle use increasing around the world, particularly on shared roadway facilities such as bicycle boulevards, there is a growing need to detect bicycles in on-street traffic conditions. This study tests the accuracy of an off-the-shelf inductive-loop technology designed to count bicycles in mixed traffic conditions and compares this accuracy to similar inductive loop technology used for detection on separated bicycle facilities. The results show that the inductive loop technology is capable of differentiating bicycles from motor vehicles and counting bicycles in traffic with reasonable accuracy, but an individual bicycle may be undetected or counted more than once. Overall, there was a 3 percent undercount for the counter on the separated path and a 4 percent overcount for the counter on the shared roadway. The results show that refinements in inductive loop detector/counter software and setup have made it possible to distinguish bicycles from motor vehicles; however, care must be taken in installation, calibration, and maintenance to ensure that the counters are accurate.