Drones, Machine Learning, and Urban Planning: A Parking Solution
This patent introduces an innovative methodology for analyzing urban parking behavior through automated aerial surveillance and computer vision. The system employs drone-based photogrammetry to generate high-resolution orthographic maps at regular intervals, creating a comprehensive spatiotemporal dataset of parking activity.
The patented process integrates several key technological innovations:
- Automated drone-based collection of aerial imagery
- Advanced photogrammetric processing for creating superresolution orthographic maps
- Machine learning models trained to detect vehicle presence and changes
- Colorimetric analysis for enhanced vehicle detection accuracy
- Integration with urban planning frameworks for practical application
Developed in collaboration with Michigan city planners, this system enables the automated collection and analysis of parking utilization data at scale. The technology processes sequential aerial imagery to identify vehicles in parking spots and detect changes over time, providing valuable insights for urban planning and parking management decisions.
The system’s particular innovation lies in its ability to create consistent, time-series datasets through automated mapping and analysis, offering a more efficient and comprehensive alternative to traditional parking studies. This approach has proven especially valuable for urban planners and municipalities seeking data-driven solutions for parking management and urban development.
Patent developed while serving as co-founder and technical consultant at Quantifly LLC.
US20200272837A1-1