StEMLF Funding for Developing New Machine Learning Methods for the Generation of Weed Application Maps for Mechanical Weed Regulation With Robots

Together with the TFZ and the LfL we successfully attracted funding from the Bavarian State Ministry for Food, Agriculture and Forestry for our new project “EWIS2: “Development and evaluation of weed application maps for mechanical weed regulation with robots”.

Based on digital information from remote sensing, AI-supported maps for the detection of weeds in sorghum and corn are to be further developed and optimized. For this purpose, we make use of previous findings and already developed AI models from the first phase of the EWIS research project (grant number G2/N/19/13).
By means of remote sensing (drone), images of sorghum and maize stands in Bavaria are created. These aerial surveys will be optimized in terms of image quality and efficiency (area coverage). For further use of the data, the images will be annotated. In an automated process, machine learning will be used to generate maps for site-specific weed control. For this purpose, AI models developed in EWIS can be used and further developed. These maps can form the basis for site-specific mechanical or chemical-synthetic plant protection in order to reduce both the erosion potential, especially on slopes, and the use of chemical-synthetic plant protection products. In the project, the necessary steps for the implementation of a site-specific use of robotics for mechanical weed control on agricultural land are investigated. Thereby the integration of field maps into the field robotics is necessary. Furthermore, based on the spatial distribution pattern of the weed infestation, an economic evaluation for the two application cases site-specific mechanical and chemical-synthetic plant protection will be carried out.


The project is supported by funds of the Bavarian State Ministry for Food, Agriculture and Forestry.