„Ich drehe schon seit Stunden, hier so meine Runden, Es trommeln die Motoren, es dröhnt in meinen Ohren. Ich finde keinen Parkplatz, ich komm‘ zu spät zu dir, mein Schatz”, is German singer Herbert Grönemeyer stating in one of his songs, which roughly translates to “I’ve been driving around for hours, the engines are drumming, it’s thundering in my ears. I can’t find a parking spot, I’m late for you, my darling,”. In Holbaek, a Danish town with just under 30,000 inhabitants, he might be better off. There, a AI-supported traffic analysis software allows traffic planners to better handle stationary traffic.
Software for traffic cameras
Conventional parking management systems are error-prone and expensive, both during installation and operation. The system of the Austrian AI provider Swarm Analytics and the traffic technology corporation SWARCO is supposed to do better. It is based on AI-assisted evaluation of anonymous data from simple IP cameras. This should increase reliability and reduce deployment times, installation and maintenance costs. With little effort, municipalities should be able to significantly reduce their parking traffic, resource consumption, and accidents – not to mention a quick, successful search for a parking spot without frustration for drivers.
Blueprint for intelligent parking management
Many cities around the world can learn from the lighthouse project in Holbaek, Denmark, about traffic planning and management. A medium-sized city with just under 30 thousand inhabitants, well 60 kilometers west of Copenhagen. Postcard-colored wooden houses in yellow, red, blue. A picturesque harbor with many small sailboats on the Baltic Sea, nestled in the green hilly landscape of the island of Zealand. And eleven larger parking lots with the well-known, problematic parking traffic. A typical city with typical traffic problems.
In January 2022, the starting signal was given with an analysis phase to better control parking traffic in Holbaek. Since September 2022, Swarm Analytics’ project has been running with traffic technology partner Swarco Denmark. 75 cameras and 35 computer units were installed to monitor a total of 900 parking spaces. The deployment time, cost, and maintenance of a conventional parking management solution are not comparable to the AI-supported system.
Classic parking management systems are often too expensive
So far, monitored parking management in cities and municipalities has been expensive, maintenance-intensive, and time-consuming. Usually, underground sensors are used for each parking space, or entrance and exit induction loops.
This not only means a large initial outlay but also ongoing costs: loop-based solutions need to be recalibrated every two to four weeks, and individual space sensors often require batteries. Barriers as an alternative are also associated with high costs and are also error-prone. “That’s why many cities are hesitant to tackle parking management projects,” says Michael Bredehorn.
Flexible and cost-effective thanks to AI
The Swarm Analytics system relies on very simple, sometimes already existing, anonymous, and GDPR-compliant data from simple IP cameras. It consists of one or more cameras, a mini-PC, the Swarm Perception Box, and the AI-assisted analysis software, the Swarm Perception Platform.
The evaluation system of Swarm Analytics thus bypasses the classic municipal cost drivers: earthworks, expensive hardware, and permanent maintenance. Another advantage: if the systems are needed elsewhere, for example, because a parking lot has been moved, converted, or closed, they are simply dismantled and installed and put into operation at the desired new location.
How it works
A sensor station analyzes up to 50 parking spaces from a height of up to twelve meters with a range of 40 meters (consisting of a camera, perception box, and perception platform). However, individual data is not recorded. A grid with “virtual parking spaces” is overlaid on the real parking lot. The camera can recognize with this positioning where space for a car should be, without the need for real sensors to be installed and maintained. The AI evaluates whether this space is occupied or not.
Processing takes place in the perception boxes on-site in seconds without images or videos being saved. Only the status of a parking space, whether occupied or free, is transmitted via mobile or cable network to the “MyCity” processing platform of the technology partner Swarco, processed, and displayed at traffic information points. Of course, third-party systems can also be used to process traffic data via APIs. The system of the manufacturer Swarm Analytics, as well as any other software with the corresponding interface, can be used.
LED displays and dashboards now inform drivers in Holbaek about 900 parking spaces and whether they can park frustration-free and economically – or whether they should organize their trip differently. This benefits not only traffic and the environment but also the pulse of parking hunters.
This text was published online on Kommunalnet.de (German language).
