Agrifood and Beverage
Agri-food and Beverage value chain vision: To develop technologies, procedures, and AI algorithms to make European agri-food and beverage industry globally renowned for its leadership, resilience, and contribution to climate neutrality. The value chain integrates the edge AI technological developments into four demonstrators.
VCD 3.1: Vineyard – Yield Prediction demonstrator (lead: URCA, VPHV, CV, STM-FR) – aims to develop a reliable edge AI framework to perform yield forecast with a below 15% error margin, using RGB cameras mounted in a robotic tractor (Vitibot Bakus®, owned by VPHV) to collect images from the plants and fruits. A yield estimate must also be assessed with full knowledge of the climatic data from previous years and the bud burst, the number of buds stripping being decisive for the future yield.
VCD 3.2: Vineyard – Environment Monitoring demonstrator (lead: STM-FR, VPHV, CV, URCA, TECX, HMU) – aims to develop multi-objective AI models that can predict several threats and harmful conditions using self-powering. The demonstrator combines the WSN architecture using a gateway with swarm to gather information from the nearby sensors.
VCD 3.3: Vineyard – Disease Identification demonstrator (lead: URCA, VPHV, CV, STM-FR) – aims to advance on an edge AI framework for in-field, real time disease detection. Recent techniques are applied to object detection architecture, successfully evolving the tomato disease deep learning-based model for in-field real time prediction, which would allow for efficient large-scale processing. Initially, meta-edge analysis may help produce a map of the affected zones, allowing champagne manufacturers and winegrowers to rationalise the application of phytosanitary products.
VCD 3.4: The “Prise de Mousse” demonstrator (lead: VPP, CV, URCA, STM-FR, STMR-FR, TECX) – aims to develop and test edge AI solutions to monitor and localise bottle explosions during the second fermentation/ageing phase of Champagne production. Experiments are conducted on the deployment of a mesh of micro-edge sensors in a harsh environment (underground cellars, high level of humidity, reduced communication, and power source capabilities)