Autonomous machines with potential, digitalization with obstacles
After five years, the Zukunftslabor Agrar (Future Lab Agriculture, ZLA) project is coming to an end. Research institutions from Lower Saxony, including the DFKI (German Research Center for Artificial Intelligence), studied aspects of the digitalized agriculture of the future.
One concept is Spot Farming, which views a field as a heterogeneous area. Different parts of the field are planted differently to preserve the environment and increase yields. New machines like robots are expected to be used for this type of farming. The ZLA project closely examined the technology. However, a current review reveals a major gap: there is no consistent digitalization across the industry. In some cases, agricultural businesses still submit mandatory documentation to authorities by mail.
Spot Farming challenges traditional thinking in agriculture. Crops like maize or potatoes are currently grown on large, uniform fields, optimized for the power of large machinery. Fertilizers and pesticides compensate for the disadvantages of specific plants due to soil or location, allowing for good yields.
The idea of Spot Farming
Professor Jens Wegener of the Julius Kühn Institute co-developed the idea of Spot Farming in 2017:
“In the future, resources for such a blanket approach will be limited. Climate change will also make weather a major challenge for crop production. The goal of Spot Farming is to grow different crops where they find optimal growing conditions. This will make them more resilient, increase yields, and help protect the environment.” Technology must follow suit, requiring smaller machines and robots for specific spots on the field.
So far, this approach exists only in theory. “In the ZLA, we have made significant progress toward realizing Spot Farming,” says Wegener. The Institute for Plant Protection Technology at the Julius Kühn Institute developed the agronomic method for identifying spots with different growth conditions on a field using publicly available geodata. Together with DFKI researchers, a tool was created to automatically generate field maps with spots that can be cultivated by robots. TU Braunschweig built a physical prototype of a universal seed drill, which plants seeds at ideal distances for plant growth and can be attached to a robot.
Text continues below picture
Robots as a foundation
For robots to later target and manage spots on the field, they must navigate their environment. DFKI’s research area of Plan-Based Robotic Control advanced this basic technology within the ZLA. Researchers built a semantic map using geo- and environmental data from a real farm. This map enabled a robot to perceive, understand, and navigate its surroundings, autonomously moving around the farm. In the future, information about how a field is divided for Spot Farming could also be integrated.
Work from Osnabrück University of Applied Sciences within the ZLA also contributes to Spot Farming. A GPS-controlled robot was used to automatically collect data on soil compaction at various spots in a field. In areas with high compaction, water is less available to plants, affecting yields. This data can be factored into the spot mapping process.
No AI and robotics without data
Data is always needed for intelligent digital technologies. However, such data isn’t always readily available. Benjamin Kisliuk, a researcher at DFKI, notes: “For a robot to work independently on a farm, you need a digital twin—a virtual representation of reality. If you want to know where field boundaries are, for example, you need the corresponding geodata. But in some states, this data isn’t easily accessible. In North Rhine-Westphalia, you can simply download it, but in Lower Saxony, you first have to send an email to the authorities.”
This observation aligns with findings from other ZLA studies. Agricultural producers must fulfill various reporting and record-keeping requirements. The Thünen Institute analyzed the data flows between livestock farms, government authorities, veterinary offices, and quality and certification bodies. They found that data collection on farms is often not fully automated through sensors but done manually via digital farm management systems. Data transmission to control authorities remains a hurdle, often done by storing the data for on-site inspections or sending it by mail or fax.
Joachim Hertzberg adds: “Unified digital interfaces for interaction between agricultural stakeholders don’t yet exist but are absolutely necessary. The government needs to make this possible, similar to how tax returns are handled today. Otherwise, farmers can’t use data from their digital processes effectively with authorities and face extra work.”
Text continues below picture
Transformation must involve people
To ensure the transition to data-driven AI tools and new concepts like Spot Farming succeeds, the social aspect must also be considered, says Professor Silke Hüttel of the Georg-August-University of Göttingen, Department of Agricultural Economics and Rural Development, in the ZLA project.
“We need digitalization in agriculture. Ideally, it will become the new normal to protect our climate and environment. However, some people remain skeptical and hold on to what they know. We need to provide evidence showing that digital technologies are ecologically, economically, and socially beneficial.”
Hüttel’s team interviewed sugar beet farmers in northern and western Germany to understand the factors that influence the acceptance and willingness to use autonomous robots. The results showed that farmers are more likely to adopt autonomous robots if they are effective, reliable, and don’t create additional work. Farmers also want to maintain control. Those who are satisfied with their current methods are more reluctant. Positive feedback from the public improves their willingness to embrace robotics.
Join 17,000+ subscribers
Subscribe to our newsletter to stay updated about all the need-to-know content in the agricultural sector, two times a week.
