The primary objective of this project is to create a flexible architecture to enable widescale deployment, monitoring, and maintenance of a non-intrusive computer vision truck parking availability system across diverse parking scenarios. Such scenarios involve facilities that allow either undisciplined parking or designated stall arrangements. Undisciplined parking facilities are actually relatively commonplace in many states in North America. The need for truck parking availability over contiguous regions is supported by the fact that driver perceptions of parking availability while traveling their routes are often incorrect. Our recent pilot study with Kansas--which adapted the multi-camera structure-from-motion (SfM) approach for undisciplined parking scenarios--has indicated over this year an overall detection accuracy at or above 90 percent, demonstrating feasibility of the approach to handle undisciplined parking. As a result, the state of Kansas instrumented 22 sites along two major interstate corridors that encompass both types of facilities, all of which will disseminate real-time parking availability through either roadside dynamic signs, or through an internet portal to third-party stakeholders and application developers. Eighteen of them went live at the beginning of 2019, with the remainder to go live before the end of 2019. Accordingly, monitoring tools must be tightly integrated with the vision processing and parking notification pipeline to query the health of different system components: 1) real-time notification and data retrieval, 2) computer vision processing, and 3) camera sensor management. Furthermore, other tools to enable remote recalibrations and real-time data visualization of the computer vision processing pipeline will also be developed. A technology transfer component that consists of system engineering documents for the software supporting the aforementioned components will be completed in the early phases of the work, with specific training to the Kansas Department of Transportation (KDOT) and other stakeholders. The architecture will then be evaluated by KDOT and University of Minnesota members to assess long-term systemwide performance.
- Project number: 2019067
- Start date: 01/2019
- Project status: Active
- Research area: Transportation Safety and Traffic Flow