Agriculture and AI in the service of the climate
| Reference number | |
| Coordinator | HS Hushållningssällskapens Service AB |
| Funding from Vinnova | SEK 6 880 879 |
| Project duration | November 2020 - December 2022 |
| Status | Completed |
| Venture | AI - Leading and innovation |
| Call | AI in the service of climate |
Purpose and goal
The goal of the project was to create a climate-positive control file model as a decision tool for the next level of precision farming, as well as to expand and improve the Hushållningssällskapet´s web-based decision support system Markkartering.se, to support climate-friendly and efficient food production as much as possible. An open mathematical platform, Lodyn, has been created where we actively seek collaboration within and outside of Sweden. Lodyn is based solely on OpenSource components.
Expected results and effects
The project results indicate that the basic conditions (soil properties and topography) predict 60-80% of actual yield, with 10% accuracy. With fertilization data and satellite data, the prediction is improved by 10-20%. Based on this, new user interfaces are created in markkartering.se to define fertilizer application on operational surface (20-50 meters) based on AI models calculated from satellite data (Sentinel-2) and from topographical data. Future projects will create more user interfaces.
Planned approach and implementation
The project collected as much data as possible from Scanian farms. The evaluation data is the crop mapping, continuous measurement with IR sensor. The data collection has been very time-consuming due to the absence of standards and functioning technical methods in Swedish agriculture. The situation can be summarized that agricultural machines collect data but data is not used operationally. The number of farms is relatively few and located in approximately the same fertility class.