Pilot study counts pedestrians and bikes using existing infrastructure
Knowing how many travelers use a transportation system is important for a number of reasons. Engineers and planners need to be able to estimate travel demand, and to do so they typically count the vehicles. Annual average daily traffic (AADT) counts have been collected for decades in the United States.
In recent years the demand has increased for non-motorized counts. For a multi-modal transportation system it is just as essential to know the number of road users who aren’t in vehicles, but traveling by bicycle or on foot. This research offers DOTs an approach for counting these travelers that doesn’t require purchasing expensive new equipment.
"Design and implementation of pedestrian and bicycle-specific data collection methods in Oregon" was a research project funded by the Oregon Department of Transportation (SRP 754) as part of an effort to count multimodal road users. The project’s principal investigator was Prof. Miguel Figliozzi and the lead graduate research assistants were Pamela Johnson and Bryan Blanc, all of Portland State University. The final report can be accessed here.
As part of the project, the Portland State team investigated how feasible it would be to come up with reliable counts of pedestrians and cyclists using the existing hardware at traffic signals. Blanc and Johnson will present a paper with the research highlights at the 2015 annual meeting of the Transportation Research Board; the paper is entitled “Leveraging Signal Infrastructure for Non-Motorized Counts in a Statewide Program: A Pilot Study”.
The researchers set up the pilot study to test the accuracy of data collection using signal infrastructure already in place. For pedestrian counts, they made use of “walk” phases and to count bicyclists they took advantage of the inductive loops in bike lanes. The team used 2070 signal controllers to record pedestrian phases and bicycles and the data were retrieved utilizing Voyage controller software. Recorded 2070 data were compared to the counts using 24 hours of video footage of the intersection.
The signal controllers were found to provide a reasonable proxy for pedestrian activity; a total of 596 pedestrians used the intersection while 482 pedestrian phases were logged, resulting in an average of 1.24 pedestrians per phase logged. The number of pedestrians using this highly trafficked and congested suburban intersection was something that caught the attention of ODOT staff. There was no bicycle and pedestrian count data in this area before this study and counting over 500 pedestrians in a 24-hour period was surprising; prior estimates were significantly lower. This result highlights the importance of statewide counting stations, in as many locations as possible, that can provide a reasonable estimate of the level of pedestrian and bicycle activity.
The results were inconclusive on the feasibility of inductive loops for bicycle counting, but nonetheless revealed important lessons to be taken into account if inductive loops are to be used. The shape, size, and placement of the inductive loop can affect the count. The proper location of bicycle loops in relation to motorized traffic trajectories (for example, away from right turning vehicles) is essential. In all instances the loops were over-counting, detecting vehicles in the traffic lane in addition to bikes. To further complicate things, nearly half of the bicycle volume through the intersection was riding on the sidewalk and two bicycle loops (one for advance detection and the other at the stop bar) use the same 2070 input. These factors made it challenging to collect accurate bicycle counts. An ongoing Oregon DOT research project is continuing this line of research and evaluating the accuracy of different loop detectors and tube counters for bicycles.
Read more about NITC's TRB research