For the English version please click "English" in the top right corner menu.
Are you looking for a specific topic? Use the search function at the top right.

Sind Sie auf der Suche nach einem bestimmten Thema? Nutzen Sie die Suchfunktion oben rechts.

Wenn Sie unser Magazin kostenfrei abonnieren möchten, schicken Sie eine Mail mit dem Betreff FELD und Ihren Kontaktdaten an:
Sie möchten die neuen FELD-Ausgaben lieber online nachlesen?
Dann registrieren Sie sich für den Newsletter auf unserem Online-Blog zum Magazin:

vor 1 Jahr

FELD 01/2022

  • Text
  • Articifial intelligence
  • Precision farming
  • Intensification
  • Patchcrop
  • Biodiversity
  • Climate
  • Agriculture
  • Farmers
  • Ecosystem
  • Zalf
  • Researchers
  • Sustainable
  • Landscape
  • Soil
  • Agricultural
Small squares instead of large fields: Together with a real farm, a research team is testing an unusual cropping system in the patchCROP landscape laboratory. // ZALF researchers are developing agricultural strategies to explicitly promote valuable ecosystem services like fertile soils and clean drinking water. // Striving for a resource-efficient agriculture without yield losses, more and more farmers are implementing measures of sustainable intensification on their farms. // Using precision farming to detect pest outbreaks and predict climate change effects: artificial intelligence holds great potential for agriculture.

Artificial Intelligence

Artificial Intelligence Artificial Intelligence technology required for »precision farming« has been developed and steadily improved. The first self-sufficient machines are already in use, for example in weed control. However, precision farming is still not very common. »Like many other aspects of agriculture, the transition is associated with very high investments that have to be planned well in advance. In many places, the technology is also not yet economically viable. In order to train the machines for the diverse operational needs, we often lack the data; for example, on the applied quantities of pesticides«, Ryo explains. One problem: the necessary data is not easy to obtain, also for reasons of data protection. »In this regard, we need a trustful cooperation between researchers and farmers. It is essential that data is collected anonymously«, says Ryo, »and we also need a clearer political commitment to precision farming in the form of support on the one hand, and regulation on the other.« AI NEVER STOPS LEARNING Prof. Ryo and researchers from all over the world anticipate even more complex tasks for artificial intelligence in the future. For example, it is supposed to provide predictions in view of the challenges posed by climate change. Which weather conditions will become more frequent? How can I adapt my farm to these conditions? »When we talk about predictions in agricultural research, we are actually always talking about computer models«, says Ryo. So far, artificial intelligence has been used in so-called »data-driven models«. Currently, these models do not yet predict the future, but they help in understanding how agricultural yields, for example, have been affected by the climate in recent years. »In short, if you feed data-driven models with very large datasets, they reveal the most important relationships«, Ryo explains. All this happens at a speed and with an »understanding« of complexity that no human brain could ever achieve. »How well these models work can be tested by comparing the results with real data. However, just how exactly the models arrive at their results is hardly comprehensible for us beyond a certain point.« He thereby addresses a fairly significant problem: »Future artificial intelligence that makes specific predictions is likely at least as complex as our current models. For minor predictions and recommendations, many farms would probably soon rely on AI-driven machines. But for major decisions, it’s a different story«, Ryo points out. Suppose a model made a specific recommendation that a farm should completely change its entire planning for the following years and cultivate other crops. Even if the model were correct, the farmers would still have to understand how this recommendation came about. »If a decision is really important, one cannot just blindly trust the machine, because in the end the responsibility always lies with the human being«, says Ryo. In any case, artificial intelligence still has a lot to learn before it is capable of making such predictions. Prof. Ryo’s research group is currently working on a model that precisely predicts crop yields with the help of drone images and field measurements. Artificial intelligence is also helping them to spatially map the diversity of agricultural landscapes in Brandenburg. Based on images, a well-trained AI can recognise exactly which crops are growing where at any given time. ZALF has been promoting the networking idea for years, for example as the coordinator of projects such as DAKIS (Digital Agricultural Knowledge and Information System). With the new working group, ZALF is now also a partner in the network »Artificial Intelligence (AI) and Internet of Things (IoT) for Digital Agriculture«. All of these activities contribute to training artificial intelligence for its future role in agriculture. In any case, Prof. Ryo is optimistic about the advance of intelligent machines: »In contrast to many science fiction films, they will not take over the world. Instead, they will rather help to harvest our potatoes.« Text: Tom Baumeister THE RESEARCHER The engineer Prof. Masahiro Ryo is head of the working group »Artificial Intelligence« of ZALF’s Research Platform »Data Analysis and Simulation«. He is also a professor of environmental data science at the Brandenburg Technical University of Cottbus-Senftenberg. Video Clip 36 37

If you would like to subscribe to our magazine (free of charge), please send an e-mail with the subject FELD and your contact data:
Would you prefer to read the new FELD issues online?
Then register for the newsletter on our online blog: