BERLIN — Dirk Siemers pulls his smartphone out of his pocket and opens an app.
“Here I have the values from my stables,” says the 43-year-old farmer who runs a hen laying operation in Diepholz, a town of less than 50,000 residents in Lower Saxony .
In 2019, he took over the business from his father and built a modern stable for free-range farming and now keeps 12,000 animals there.
Siemers can use the app to check temperature data, humidity levels, the fill level of his feed silos, or the fresh air requirement in the barn. Modern sensors also provide data on water consumption per animal or feed consumption.
“We try to capture as much data as possible here,” explains the farmer, who studied computer science and worked as an IT specialist for many years before taking over the business.
Data is the key to the use of artificial intelligence (AI) in agriculture, he says.
“Artificial intelligence learns from examples,” says AI expert Christian Lamping from Big Dutchman , a supplier of equipment for livestock buildings. He explains that there are fundamental differences between AI and conventional digital applications.
In conventional applications, individual features are defined. For example for a camera to recognize a chicken, there is a white area with a small red area.
The problem is that when it gets dark or the light changes, the system can no longer recognize a white area because everything looks reddish.
Systems using AI work differently. The system is fed data – in this case, many different images of chickens.
Several thousand different images are needed, which the AI stores as images of chickens. On this basis, the system can train itself independently. The AI-based system is therefore less prone to errors. What’s more, AI can recognize patterns that humans cannot.
Every AI that is trained is only as good as the data that is fed into it, says Lamping.
“You want an AI that can handle as much data as possible, that is robust, so you need enough data.” The challenge is not only to get data, but also to ensure its quality and preparedness.
Huge amounts of data are needed for AI systems to work. Any physical quantity can be recorded, explains Jörg Kleine-Klatte, who heads the digital business unit at Big Dutchman .
Not only water and feed consumption or weights, but also values for ammonia levels in the air in the barn or the concentration of carbon dioxide (CO2). All these factors influence the health and welfare of the animals.
Image processing has made great strides in the last five years. It is now possible for AI to determine the weight of animals using cameras. But it is also now possible to recognize movement patterns, explains Kleine-Klatte.
Conclusions can also be drawn from such data about animal health, for example. A well-known example in chickens is dust bathing, which involves chickens rolling around in dust or dry earth to remove parasites from their fur, feathers or skin.
“This is a clear sign that the animal is feeling well,” says Kleine-Klatte.
“The big advantage of AI is that it can monitor a large farm 24/7,” explains Lamping. Until now, farmers have only been able to get a snapshot of what is happening in the barn when they check on the animals during a tour.
Only then can they observe the animals’ behaviour or determine what the climate in the barn is like or how clean the litter material is. With AI, on the other hand, you are able to see what is happening in the barn every second of the day – and can react faster.
The first applications of AI are already being used in many areas of agriculture, says Joachim Hertzberg . The computer scientist taught at the University of Osnabrück until his retirement last summer and was managing director of the local branch of the German Research Center for Artificial Intelligence (DFKI).
Harvesters are equipped with sensor systems that ensure grain is threshed correctly.
According to Hertzberg, these sensors check if the grain is being processed at the right thickness, taking into account factors like grain size and moisture content.
Cameras installed in the machines can also monitor the harvesting of crops like beets and potatoes.
“This is unobtrusive AI, which is as it should be: you don’t even notice that it’s there,” says Hertzberg. It involves small software modules that are built into the machines’ information technology and function.
This does not make the machines completely autonomous, but rather they are assistance systems that support the drivers – similar to in a car.
Siemers expects that thanks to further developments in camera technology, there will soon be AI applications for his stalls as well.
“In the area of broiler fattening, there are already applications that evaluate the movement patterns of the animals.”
As a farmer, he has developed a sense of how the animals are doing when he walks through the stall. “You get a feeling for how the animals are behaving, how they look when they are doing well – that’s what experience teaches you.”
AI-based applications could be an important aid, says Siemers. They could far outperform even the best laying hen farmer because they can process much more data than a human.
“A computer can react much faster than we can, and a computer sees everything.”
©2025 dpa GmbH. Distributed by
Tribune Content Agency, LLC.
link