AGROTWIN: A drone-based agronomic DSS

#44, April 2024

PhD Niccolò Bartoloni

COO Agrobit s.r.l.

Agrobit is an Italian innovative agtech startup that develops terrestrial imaging solutions (iAgro, iTractor, iDrone) for the agriculture sector to support farmers/technicians in reducing/optimizing the use of chemical inputs and water. Agrobit uses different types of cameras in combination with computer vision and AI algorithms to assess crops status and relevant agronomic parameters.

Agrobit has been selected among the winning teams of Icaerus Push 1 Open Call; ICAERUS is an EU-funded project exploring the use of drones in agricultural production, forestry and rural communities.

The AgroTwin project aims to exploit 3D point clouds (big data) generated by consumer-grade RGB drones in order to develop a Decision Support System (DSS) based on innovative AI computer vision algorithms, that automatically analyse vineyards digital twin and assess for canopy biometrics and field parameters, to create vigor and prescription maps for optimized variable rate pesticide treatments. The proposed DSS will help farmers to reduce the use of pesticides, decreasing economic, social and environmental impacts in agriculture.

The challenges of the proposed project involve developing the computer vision AI algorithms, ensuring data accuracy and reliability and effectively reaching end-users in rural areas. These challenges will be addressed by testing and validating the solution in a real winery with real working conditions, also comparing results to field manual measurements, to minimize the risks, collect farmer/technician feedback and maximize the quality of the DSS output.

The technology employed in the solution includes the use of consumer-grade drones equipped with RGB cameras, open-source photogrammetry software to generate 3D point clouds, AI computer vision algorithms to analyse big data and assess canopy biometrics, as well as vegetative indices (LAI, LWA, TRV) and DSS models to assess the optimal pesticide and water doses for treatments.

The expected outcomes of the AgroTwin project will be:

  1. AI computer vision algorithms to automatically analyse the vineyard 3D point cloud and get biometric parameters with an error up to 10% with respect to manual measures.
  2. up to 90% increase in accuracy, repeatability and time-savings of plants biometrics data compared to the measurements carried out manually with a measuring tape.
  3. up to 60% reduction of pesticides and water compared to traditional treatments, based on farm history and without preventive monitoring of the intra-field spatial variability, guaranteeing at the same time an adequate coverage efficiency.
  4. up to 60% reduction of air/aquifers pollution, as a consequence of point 3;
  5. up to 60% reduction of pesticide residues in final products, as a consequence of point 3.

The starting TRL is 5 as the technology has been already validated in real environment and a research paper has been published ( Through an experimental project that will be conducted in a Demo farm, TRL 7 will be achieved with the production of a DSS prototype; the most easily achievable proposal now is to produce a DSS web app, which allows farmers to explore their fields through a Virtual Tour and operators to upload images from drones.


Figure 1 – Vineyard Digital Twin view