Employ sensors that can account for a range of parking lot vehicle movements.
Experience from the smart parking field test at the Rockridge, Oakland BART station.
Made Public Date

Smart Parking Linked to Transit: Lessons Learned from the San Francisco Bay Area Field Test


The Bay Area Rapid Transit (BART) District, the rail agency serving the San Francisco Bay area, includes a total of 43 stations, with approximately 46,000 parking spaces at 31 stations. Due to the Bay area's high share of transit commuters, parking at the stations is in high demand. Many of the BART stations have a parking shortage, especially during peak commute hours, and it is difficult to secure land and funding for additional spaces. In 2002, BART implemented a monthly reserved parking program to guarantee commuters a space during peak hours. However, when monthly subscribers do not take transit everyday, reserved spaces are underutilized. From 2004 to 2006, researchers implemented a smart parking field test at the Rockridge, Oakland BART station to complement the monthly reserved program by providing daily flexibility during the morning commute to those who do not use transit everyday.

The project included in-ground sensors in the BART parking lot to determine available parking spaces, two changeable message signs (CMSs) located on the highway that display dynamically updated parking availability information for motorists, and a computer reservation system accessible via the Internet and a telephone Interactive Voice Response (IVR) system. This paper examines the institutional, user perspective, and operational lessons learned from the smart parking field test.

Lessons Learned

Evaluation of the BART smart parking field test revealed that the greatest technical challenges arose from the in-ground parking sensor system. The sensor system utilized had previously been used on roadways with traffic moving in one direction only. The system was modified to detect two-way vehicle movements and testing at an off-site location indicated that the sensors worked well. However, when implemented at the BART station, the sensors ability to accurately count vehicles moving at parking lot speeds was unreliable. It was later determined that the inaccuracy may have been a result of the magnetic field at the BART station, since the sensors work by detecting the changing magnetic fields from vehicles passing over the sensors. Also, sensors had difficulty accounting for atypical vehicle movements, such as cars driving into or out of the lot the wrong way.

Lessons to be learned from this experience are:

  • Ensure that the sensors for the system can account for a range of parking lot vehicle movements
  • Test the sensor systems on site to be sure that they conform properly to actual site conditions.

Eventually, the in-ground parking sensor system was replaced with an aboveground system. While the system was more accurate, the integration of the new sensors with the wireless counting system resulted in communication protocol problems. Researchers ultimately maintained count accuracy by using a proprietary algorithm that corrected the sensor problems and accounted for instances when vehicles queued above the sensors. In the end, the parking project managers determined that aboveground sensors were superior in providing an accurate vehicle count.

The experience of the BART field test demonstrates the unique issues associated with counting vehicles moving at parking lot speeds. In order to obtain accurate information agencies should employ sensors that can account for a range of parking lot vehicle movements. In addition, agencies should test the parking sensor systems onsite to ensure that they operate correctly under the specific site conditions.