Tinus Kanters is project manager of Living Lab Stratumseind Eindhoven, a nightlife area experimenting with sensors to improve the atmosphere on the street. He oversees an algorithm that can predict, based on data, where a fight will occur. Rathenau Institute took a look in his office.
On a drizzly Wednesday afternoon, Stratumseind does not look inviting with closed doors and few people on the streets. Not exactly an area where you would expect sensors to measure crowds. Yet the Stratumseind is one of the Netherlands' best-known research and measurement laboratories for safety and livability in public spaces. The municipality works with police, pub owners, property owners, breweries and local residents to improve the atmosphere and livability. To this end, many, at first glance invisible, sensors have been placed on the streets.
Tinus Kanters, project manager of the Living Lab Stratumseind, who keeps office above a café in the entertainment street, knows all about it. His office resembles a laboratory from a science fiction movie, with dozens of screens, sensors, cables and prototypes. He is asked whether the Stratumseind is a model for the future of the Dutch entertainment street.
Currently, there are about eight hundred incidents on the Stratumseind on an annual basis. To reduce this number, real-time measurements of the atmosphere on the street are being taken, and data are being collected from many different sources for this purpose. 'We measure how many people come and leave here, where they come from, the weather conditions, the amount of noise on the street, the light intensity and the amount of garbage.' In addition to these measurements, we look at social media messages, an event calendar, police statistics and the amount of beer delivered.
Measuring the number of visitors on the street is done with counting cameras, which automatically convert people into unrecognizable dots. "That's how people are registered anonymously. I am not the police. I'm not allowed to look out images." Based on the group behavior of the "dots" and sound analysis, the algorithm could predict two to three seconds in advance when a brawl will occur.
This does not yet work flawlessly: "There are still a lot of false positives in the sound analysis. All fights are included, but also a lot of bachelor parties. We hear from the police whether something was a brawl and we pass that back to the algorithm to learn."
Lamps and scent diffusers are used to try to influence the atmosphere: extensive research has shown that orange light and orange scent are the most atmospheric. It remains to be seen whether this also works in practice: "The scent will mix with stale beer and shoarma smell on the street, for example, so you get wrong on that in terms of effect," Kanters said.
Meanwhile, researchers and students are working to find connections between all those data streams, such as what's on the events calendar and where people come from. The data are basically public data, which can be found in the opendata portal of the Municipality of Eindhoven. In the experiment, these data are supplemented with purchased data from phone providers that accurately indicate how many visitors are in an area. For internal use, all data are displayed on a dashboard, which will soon also be shared with the police.
"It is not only the citizens who have to get used to this technology, but also the police themselves. New technology can appear threatening, but we think it makes an important contribution." For example, based on a notification message, a signal could be sent to police on the street in the event of an incident. That could save crucial seconds, according to Kanters. This makes everyone better off, police, victim and offender alike. The data can also show that a particular spot needs more police deployment, or less.
Technology and privacy are not always mutually exclusive, according to Kanters: more privacy can be created by deploying technology. Camera images in control rooms, for example, can "go black" until sensors detect that something is happening, according to Kanters. "In the absence of legislation in the Netherlands, we drafted our own data principles. If you want to build a house in the Netherlands, there are books full of rules before even one brick has been laid. For the use of data there is nothing. You can then reason, it is not forbidden so just do it. But that is the wrong starting point."
The principles focus, among other things, on privacy, citizen participation, and guarding control over technology. For example, the project in Eindhoven is modularly organized. That means that the basis of the system, the algorithm that converts data into useful information, is in-house. "If you let that be done by the big commercial players of this world, you lose all control," he said.
The advantage here is also that you are not dependent on one software vendor. This phenomenon is called vendor lock-in and, according to Kanters, it happens too often in this type of project. It then becomes incredibly expensive to make relatively easy changes to the software structure, for example. For the police, Kanters still has some advice to offer. "Don't make yourself dependent on one company, and set up an ethics committee."
The Living Lab on Stratumseind has been running for about four years now. Has the street become safer, according to Kanters? 'Research by TU Eindhoven shows that the number of fights has decreased compared to four years ago. And the nightlife public's perception of safety has gone up. In all fairness, I should note that the number of visitors to the street has declined, although the turnover of the hospitality industry is stabilizing. Probably the latter is a sign that there is now a target group that has a bit more to spend. In any case, Stratumseind has become a lot nicer in recent years.'
source: Rathenau Institute
