The Smart Parking Insight effort under RM3P’s Commuter Parking Information Services (CPIS) program element aims to determine whether crowdsourced smartphone data can reliably estimate real-time parking occupancy at commuter lots. The technology works by using passive data from users running the ParkZen parking app to infer when vehicles enter or exit a lot, producing continuous updates. Through previous testing efforts, the project team concluded that with sufficient user adoption of the app and the commuter parking lots, ParkZen has the potential to furnish parking occupancy data that meets VDOT’s requirements. Before deploying the technology regionwide, VDOT initiated a proof-of-concept (POC) phase to validate whether employing targeted marketing strategies at selected parking lots and making adjustments to the app can aid in generating sufficient user adoption, which consequently will lead to meeting accuracy and timeliness requirements.
The first proof-of-concept phase, POC-1, tested the technology across six commuter lots over six months. However, the solution requires at least 15% penetration rate of lot users to actively run the app to generate accurate estimates, a threshold that was not reached due to low adoption. As a result, VDOT could not meaningfully evaluate the technology’s performance in POC-1. Alternative data sources, INRIX modeling data and Virginia Railway Express’ (VRE) in-ground counters, were also examined, but neither proved viable due to accuracy and long-term maintenance and cost issues.
To isolate the testing of the crowdsourcing algorithm itself, VDOT launched POC-2 at the Northern Virginia District Office lot, where employee volunteers were recruited to achieve the needed active penetration rate. Ground-truth validation, including drone-based vehicle counts and review of entry/exit camera footage, showed that accuracy improved modestly over time but did not consistently meet required thresholds. Still, incremental progress between validation rounds, combined with rising parking activity and ParkZen’s ongoing algorithm enhancements, supported the CPIS project team and steering committee’s recommendation to move forward with an additional POC phase.
The third proof of concept (POC-3) will combine expanded participant recruitment, targeted marketing, third-party integrations, and continued algorithm refinements.
Key Elements for POC-3
- Duration & scope: A 12-month effort, covering 7 commuter lots, with a 30-day decision gate.
- Typical occupancy profiles: Building a library of typical usage patterns that can provide valuable information even if crowdsourced adoption remains insufficient.
- User adoption & data partners: Expanded recruitment, localized targeted marketing campaigns, and potential integrations with formal data partners to boost penetration.
- App enhancements: Improved user experience, new Virginia-specific landing page, and additional contextual data.
- Cost efficiency: Crowdsourcing remains less expensive than hardware-based counting technologies while offering comparable accuracy and allowing easy expansion to additional parking lots without installing equipment.
At the end of POC-3, VDOT will assess whether the technology is ready for broader deployment, including identifying which parking lots it can reliably serve (for example higher-activity lots), or whether the program should pivot to a Profiles-only approach that still supports traveler decision-making.
