Transportable Low Cost Traffic Data Collection Device for Rapid Deployment for Intersections and Arterials

Project summary:

In spite of recent advances in technology, most traffic engineering studies at intersections and arterial streets are still performed manually. This is especially true for measuring turning volumes at intersections; this task must be performed at regular intervals for re-timing traffic signals in order to minimize delays, stops, excess energy consumption, and pollution levels, as well as to improve coordination and reduce congestion levels. Manual measurements are subject to errors and, due to time, logistics, and cost considerations, they are usually performed only when absolutely necessary (for example, as a result of reconstruction, excessive public complaints, congestion, unusually high accident rates, or other emergencies). The goal of this project was to develop and test a rapidly deployable turnkey, low-cost, non-intrusive, stand-alone, video data-collection and surveillance system. Researchers developed a prototype that uses a self-raising extensible mast and a custom-fabricated base to elevate the camera approximately 30 feet above the road surface. It is designed to be secure-clamped to sign, light, or traffic-signal poles. A simple interface quickly creates daily recording schedules and operations of other hardware to utilize battery power as efficiently as possible. Approximately 40 hours of traffic video can be stored with a single battery charge. An intersection lane-approach topology was also developed and used to facilitate the selection of the actual field site locations. The deployment included four intersections and one midblock site, and standard traffic measurements (speeds, counts, vehicle classification) were extracted from each. Each site was analyzed to determine the accuracy of standard measurements with respect to ground-truth data. In addition, pole height and camera positioning with respect to the intersection, signal pole locations, and field of view, were assessed, and the data were analyzed to determine the most suitable algorithm to use for automated traffic measurement extraction. The detailed results are presented in the final report of the project.