Safe to school thanks to Artificial Intelligence. In an interview, Wieland Alge from Swarm Analytics explains how video data analysis can be used to defuse particularly dangerous traffic situations such as crosswalks.
Mr. Alge, what makes protected zones such as crosswalks so dangerous, especially for the way to school?
Crosswalks work when they are well built. That is, with multiple signs, easy to see and well lit. But they often aren’t, and this creates a false sense of security that leads to more critical situations.
Do flashing signals help?
No, because unfortunately there is the so-called ‘routine effect’. If the system always warns, and five times there are no pedestrians, motorists like to ignore the warnings.
What are the ways to make a crosswalk safer?
It’s important to figure out where action makes sense. A simple survey with good data helps classify whether and how dangerous certain crosswalks are. Here, we rely on camera-based traffic analysis. A system consists of one or two cameras installed at neuralgic points. The video data is processed in real time on site using Artificial Intelligence and translated into information. The important thing here is that video data is not stored. It is only a matter of recognizing whether a person is in the crosswalk, but not who that person is. The system is controlled via a cloud-based dashboard with predefined monitoring routines and individual ‘if-then’ analytics rules. The analytics system detects rule violations. If there occur frequently, measures can be taken on the basis of reliable data and their success can also be monitored. For example, setting up speed cameras, tackling 30 km/h zones, turning a crosswalk into a traffic light system, or even installing an intelligent traffic signal that, thanks to ‘if-then’ monitoring routines, only comes on when vehicles and pedestrians could actually meet.
Would that also allow for some sort of ‘speed camera’ for crosswalk speeders?
Absolutely. The system is a bit more complex here, because legally binding evidence has to be recorded, such as photos of the license plates. But this effort pays off very quickly on busy roads. Simply approaching a crossing point at a ‘non-moderate speed’, as the fine catalog puts it, is punishable by 80€ and a point in Flensburg.
Are there already municipalities that use such systems?
In fact, the systems have so far been used primarily for traffic data collection and mobility management. The Austrian municipality of Lustenau uses camera-based traffic analysis to test traffic-calming measures such as 20km/h zones and to develop measures to keep the massive through traffic out of the city center and to increase the share of bicycles. Increasingly, pedestrian traffic and its needs are coming into focus, and municipalities of various sizes now want to not only count pedestrians, but specifically protect them.
What does such a system cost?
The costs for a simple analysis system are around 2,000€ one-time and 600€ annually. A ‘speed camera replacement’ is 10,000€ one-time and 2,000€ annually. No comparison to radar devices. The real benefit is increased safety, because word of monitored dangerous spots spreads rapidly and driving behavior is changed almost instantly.