When extreme summer heat sent temperatures soaring to 38°C in Washington State last year, apple pickers at Okanagan Specialty Fruits had to stop working by 10 a.m. To better navigate unpredictable weather and fluctuating market prices, agricultural businesses are turning to artificial intelligence. New AI models analyze orchard imagery, weather forecasts, and historical farm data to help growers determine the precise timing and labor required for crop harvests.
At Okanagan Specialty Fruits, which manages more than 1,250 acres of apple orchards, tractors equipped with cameras from Canadian firm Vivid Machines take footage as they move through the fields. AI algorithms then identify tiny flower buds and fruit to estimate yield sizes and harvest dates. While apples have a forgiving three-week harvest window, other crops like strawberries must be picked within days to prevent disease. In the UK, technology firm FruitCast uses drones, farm vehicles, and smartphones to map fruit development across entire farms, helping growers monitor fields affected by heat and drought.
Researchers are also building more detailed diagnostic tools. At Princeton University, scientists have tested millimeter-wave sensors that use high-frequency radio waves to measure moisture and sugar levels inside fruit without cutting into it. Meanwhile, researchers at North Carolina State University and the University of Florida are developing low-cost crop-counting tools that run on smartphones or inexpensive $100 drones, making harvest forecasting more accessible to smaller farms.
Despite the promising technology, adoption faces hurdles. Growers express concerns about sharing commercially sensitive farming data, such as irrigation and fertilizer plans, with third-party software developers. Industry experts, including representatives from Western Growers, emphasize that while AI tools offer valuable optimization, farmers will continue to rely on human experience and judgment to make critical harvest decisions.