As of October 1, 2016, the SWUTC concluded its 28 years of operation and is no longer an active center of the Texas A&M Transportation Institute. The archived SWUTC website remains available here.

167762-1 Report Abstract

Testing and Evaluation of Pedestrian Sensors

Shawn Turner, Dan Middleton, Ryan Longmire, Marcus Brewer, and Ryan Eurek, Texas A&M University, September 2007, 42 pp. (167762-1)

The foundation for several pedestrian safety measures is reliable and accurate detection of pedestrians.  The main objective of this study was to evaluate sensors for use in a pedestrian safety test bed in College Station, TX.  The following sensors were installed for use in the pedestrian test bed: 1) MS SEDCO SmartWalk 1400 (curbside detection) and SmartWalk 1800 (crosswalk detection); and 2) ASIM IR 201 (curbside detection) and IR 207 (crosswalk detection).  Several other sensors designed for counting pedestrians and bicyclist on trails also were evaluated: 1) Jamar Scanner, 2) TrafX Infrared Trail Counter, and 3) Diamond Traffic TTC-4420.

The ASIM and MS SEDCO intersection sensors provided fair to mediocre results, with error rates ranging from 9 to 39 percent.  The typical error rates were in the 20 to 30 percent range, which may not be sufficient accuracy for most pedestrian detection applications.  The accuracy of the sensors appeared to be very location-specific, in that pedestrian detection can be more effective in certain situations in which the pedestrian travel area is constrained.  The three trail sensors were able to accurately detect a single pedestrian at typical walking speed or a bicyclist at slow speed (5 to 10 mph).  The Jamar sensor had difficulty counting bicyclists at typical bicycling speed.  All three trail sensors did consistently undercount the actual ground truth counts, with the undercounting being more severe in situations with more pedestrian groups.  This undercounting presents a problem on busy shared-use trails.

Keywords: Pedestrian Detector, Pedestrian Sensor, Trail Counter, Automated Detection

ENTIRE REPORT (Adobe Acrobat File – 2.1 MB)