The agricultural use case involved the cultivation of vineyards, with an end user in France that produces agricultural machinery.
The agricultural use case demonstrated how AI can enable an enterprise not only to enhance its business-related activities, but also to ameliorate its environmental impact. Pesticides are a necessary factor in viniculture, but awareness has grown over the years of the damage they can inflict on the environment. Prior to AI SPRINT, the best that could be done was to calculate dosages based on the size of an area. But AI SPRINT made it possible to refine dosages down to the level of the foliage and its characteristics.
Using edge and smart sensors to optimise phytosanitary treatments in vineyards. As agriculture increasingly embraces the digital revolution, artificial intelligence will improve processes and production systems. AI-SPRINT will develop novel models for phytosanitary product optimisation.
These new AI models, which have never been used before in farming, will collect data from sensors deployed on grape harvesting machines and command the sprayer system of the machine to adapt the quantity of product to the actual need of the individual plant. Important impacts will come from optimising the use of chemical products and thereby lowering pollution levels and soil contamination, as well as reducing the overall quantity of phytosanitary products deployed. The technology developed by AI-SPRINT for this use case is expected to be applicable/adaptable also to other agricultural applications.
In terms of design and runtime, the underlying architecture is very similar to the AI-SPRINT reference architecture but without Federated Learning and encryption of the containers.
The deployable infrastructure consists of a Kubernetes cluster for training the AI-model and edge devices. The device will rely on OSCAR for FaaS functionality. At this early stage it is assumed that the edge device will run several containers orchestrated by Kubernetes and deployed using the Infrastructure Management.
The edge containers interacting with sensors or with tractor hardware (spraying system) will use containers running ROS. MinIO will be used to provide local storage services and synchronisation with the cloud.
Precision farming in viticulture allows optimised usage of phytosanitary products which brings financial benefits and - more importantly - has environmental benefits such as pollution reduction, increased good insects for the farm ecosystem , prevention of new species resistant to treatment, fewer chemical products, fewer products harmful to biodiversity.