New AI Tool Forecasts Crop Harvests Six Months in Advance

A groundbreaking AI tool developed by researchers at the National Robotarium, in collaboration with scientific partners in Chile and Spain, is transforming how farmers predict harvest sizes. Using images captured with standard smartphones, the system can accurately count flowers on fruit trees, significantly improving crop yield forecasts months in advance. This innovation promises to enhance efficiency, sustainability, and profitability in agriculture.

AI technology identifies patterns and features in the images, such as the edges and shapes of petals, even when they overlap or are partially hidden. When tested in peach orchards in Catalonia, Spain, the system achieved a 90% accuracy rate in predicting flower counts. This marks a substantial improvement over traditional manual methods, which often have error margins of 30-50%. 

Accurate yield forecasts up to six months before harvest allow farmers to optimize water usage, allocate resources more efficiently, and better plan harvesting and distribution logistics. Given that agriculture consumes 65% of the world’s freshwater—almost half of which is wasted—and that approximately 45% of fruits and vegetables are lost or wasted annually, this technology offers a promising solution to these significant issues.

Researchers from the National Robotarium, based in Edinburgh and part of the Data-Driven Innovation initiative, plan to validate the AI's predictions against the peach harvest in September 2024. If successful, the technology could be adapted for other key crops like apples, pears, and cherries, benefiting growers in the UK, Europe, and beyond.

Dr. Fernando Auat Cheein, Associate Professor in Robotics and Autonomous Systems at the National Robotarium, highlighted the global potential of this technology. "Farmers worldwide often rely on manual yield estimation methods with significant error margins. Our AI approach, leveraging off-the-shelf smartphone technology, integrates seamlessly with traditional farming practices, offering a simple yet effective solution."

The project involved close collaboration with peach farmers in Spain to tailor the system to their needs. Farmers praised the AI for its accuracy and ease of use, recognizing its potential to inform decisions on crop management, such as targeted pruning and herbicide application. This enables farmers to optimize resources, reduce their environmental impact, and maximize harvest quality and quantity.

National Robotarium, supported by £21 million from the UK Government and £1.4 million from the Scottish Government, is committed to making Edinburgh the data capital of Europe. As part of the £1.5 billion Edinburgh and South-East Scotland City Region Deal, this initiative aims to drive sustainable and productive agriculture through AI and robotics.

The collaborative research project includes contributions from the Advanced Center for Electrical and Electronic Engineering at Federico Santa Maria Technical University and Universidad Andres Bello in Chile.

National Robotarium continues to lead in agricultural innovation, developing cutting-edge solutions to enhance crop yields, improve quality, and promote sustainability.

(Source: The National Robotarium)