Dhe idea sounds a bit like “Big Brother”. If the occupants stay in the bathroom too long, if they get out of bed at night but don’t return there, or if nothing happens in the apartment for an unusually long time, the system sounds an alarm. First it calls the residents themselves, and if they don’t respond, they call family, friends or the emergency call center. The home as protector. For Reiner Wichert, computer science professor at the Darmstadt University of Applied Sciences, a variant of Smart Home that is in the not too distant future, because that is exactly what he and his team are working on. As an old person in the home? It may not have to be the same in the future. The professor is convinced that digital care could be an alternative. Most of them prefer to stay in familiar surroundings anyway. Wichert’s research could help with that.
As is so often the case, the magic word is artificial intelligence (AI). With the help of software that uses self-learning systems, the scientists want to enable the elderly or disabled to live independently for as long as possible. Wichert has been dealing with the topics of smart home and personal health for many years. Safety plays a decisive role here, because a fall or accident is often the greatest risk for seniors living alone. There are emergency call buttons or bracelets, “but these are active alarms that have to be triggered by those affected,” says the researcher. Even passive systems such as smart watches on the wrist or fall detection on the floor do not always reliably register an emergency situation.
Neighbors, family and the emergency call center are alerted
The Darmstadt-based company’s idea starts elsewhere. Their help system is based on sensors that are installed in the living rooms – in the form of door contact switches, motion detectors or fall sensors. These should forward critical situations such as an accident, prolonged inactivity, smoke development or staying too long in the bathroom or living room to a controller that is also installed in the apartment. The computer evaluates this data with the help of artificial intelligence and analyzes whether there could be an emergency. In such a case, the device reports to the residents via loudspeaker. “They are told not to answer the phone if they need help,” explains the professor. In this way, an emergency call chain is triggered passively, alerting neighbors, family, friends, property management or the emergency call center by telephone. They can also use the controller to talk directly to those in need. If help arrives, the entrance door can even be unlocked digitally, depending on the house equipment.
Before moving to the university, Wichert did research on the subject at the Fraunhofer Institute for Graphic Data Processing in Darmstadt. The computer scientist helped develop a software platform for an EU-funded project to digitize care, which he now wants to further develop with his university team. The researchers are concerned with protecting privacy and at the same time developing a fast AI system with a high probability of detection. If necessary, help should arrive quickly. The AI is a self-learning system that Wichert describes as a cycle. Certain patterns, such as the behavior of residents, are recognized using huge amounts of data. The system thus learns about everyday life and people’s normal reactions – such as how long they are awake or asleep, when they are at home or when they leave the house. And it learns to recognize deviations from this with a high degree of probability.
26 apartments with sensors in the district of Görlitz
Whether this works is to be tested in a pilot project in Upper Lusatia. The study led by Wichert will start in January 2023 and will be funded by the Federal Ministry of Research with 1.14 million euros for four years. In addition to the Darmstadt University of Applied Sciences, the TU Dresden, the Zittau/Görlitz University of Applied Sciences, the Johanniter-Unfall-Hilfe and a housing association are all involved in the joint project. In the initial phase, 26 apartments in the Görlitz district will be equipped with sensors, with 50 more apartments to be added later. A total of 200 is possible.
The first application in Germany is being deliberately tested in an outdated and structurally weak region. Companies and institutions are to be specifically promoted in order to create jobs and make Upper Lusatia more attractive. Not only the hardware is produced by a local company. Electricians from the region are to install the sensors and attend training courses offered by the Zittau/Görlitz University of Applied Sciences. Local health services or health apps are also integrated. The software is flexibly programmed so that expansion is possible and providers can access the platform at any time, says Wichert. After a test phase in a laboratory environment, the system should go online in the apartments in mid-2024.
A basic set of the AI system is said to cost around 2,000 euros, plus monthly maintenance costs of around ten euros. “It’s more affordable than staying in a home,” says the professor. And in the future it should be usable for people of all ages, from small children to pensioners. Digital aids are conceivable, which detect fire or water damage in good time, act as babysitters or – most recently – help to save energy.