Validation of bicycle counts from pneumatic tube counters in mixed traffic flows

FHWA and other public agencies encourage monitoring of nonmotorized traffic volumes and are collaborating with researchers in projects to test monitoring technologies and develop nonmotorized monitoring programs. This paper reports findings from research in Minnesota to validate the use of commercially available pneumatic tube counters to count bicycles in mixed traffic flows on urban roadways. Pneumatic tube counters and video cameras were installed in different configurations at two locations on arterial roadways with different traffic volumes. Bicycle counts from the tube counters were compared with manual counts from video to determine both percentage error and absolute error rates. Researchers matched time stamps from the counters and video and inspected discrepancies to determine causes of error. Calibration equations to adjust for systematic counter error were estimated by regressing hourly manual counts on hourly automated counts. The tube counters generally undercounted bicycles: error rates for the observation periods ranged from 6% to 57% depending on the location, configuration of deployment, type of device, and classification algorithm. Absolute error rates were substantially higher. Inspection of video indicated most false negatives (i.e., undercounts) were caused by occlusion. Calibration equations generally had moderate to very good fit (R2 values from .82 to .92). Agencies can adapt standard, commercially available pneumatic tube counters to collect bicycle counts in mixed traffic flows; but depending on the site, traffic volumes, configuration, and deployment, error rates may be high. The practicability of using tube counters to count bicycles in mixed traffic flows depends on the applications and the relative need for accuracy in measurement.