Institut für Rebenzüchtung
'In viticulture quantity and quality of yield significantly affects the product quality and therefore the profitability. For vine grower yield forecasts are an important ecological factor, which can be highly influenced through weather conditions showing strong year-to-year fluctuations. Up to now yield forecasts are very error-prone leading to yearly surprises, positive or negative, in the companies. Simultaneous accurate yield forecasts offer an important base for operational management: (1) adjusting the desired yield level (quantity-quality- relation; quality management); (2) transport logistics (transport capacity from the vineyard to the cellar. reduction of waiting time at grape delivery) and (3) cellar logistic (capacity of grape press and fermentation tanks). Trans-technologically and application-oriented a machine learning method (section of Artificial Intelligence (AI)) will be used to develop a yield forecast model.KI-iREPro will be based on extensive vineyard yield experiences in combination with environmental data. In a second step KI-iREPro will be extended with sensor-based yield parameters acquired directly in the vineyard in the ongoing season. This vineyard specific information will be used to improve the accuracy of the yield forecast (KI-iREProext). Project goals are the approaches for an adjusted yield forecast on the one hand and methodological technical innovations for an AI driven, modular and low-cost realtime sensorsystem for parameter detection directly in the field.
Federal Ministry of Food and Agriculture