Fruit growers in Washington State and beyond are increasingly exploring artificial intelligence (AI) to improve harvest timing amid unpredictable weather conditions. Last year, during the apple harvest season, intense heat forced workers at Okanagan Specialty Fruits to halt picking early in the morning. Joel Carter, a representative from the company, believes that AI models forecasting optimal harvest windows by incorporating weather data could alleviate such challenges.
Okanagan Specialty Fruits manages over 1,250 acres dedicated to apple production. The company is implementing technology to enhance productivity, including genetically engineered apples that resist browning. Despite their efforts, scheduling harvests remains complex as market prices for fruit like strawberries and blueberries can shift significantly, affecting potential profits when timing is misjudged.
Advancing Technology for Harvest Projection
In an effort to perfect their harvest strategy, Okanagan Specialty Fruits is testing cameras developed by Vivid Machines, a Canadian company. These cameras, mounted on tractors, capture images of the apple orchards, identifying buds and fruits using AI analysis. Carter, however, notes that such AI systems rely heavily on robust historical data, illustrating that they cannot simply draw average yields from online sources but must be specifically tailored to each farm’s conditions.
While apples have a relatively generous three-week harvesting period, other fruits like strawberries have a much narrower window, making accurate predictions crucial. Raymond Martin from FruitCast, a UK-based company, underscores the urgency in harvesting berries to prevent disease and spoilage. FruitCast offers predictive services for various fruits, including strawberries, raspberries, blueberries, and tomatoes, with plans to expand to grapes next year.
Challenges and Developments in AI Forecasting
Martin explains that although experienced farmers might know their expected harvest timings, AI provides a comprehensive analysis across large or diverse farming areas, with their models boasting an accuracy of up to 90% a week in advance. However, the adoption of AI solutions, as noted by Angus Soft Fruits’ operations director Neill Finlayson, is ongoing and not yet fully integrated across the industry.
Research is also advancing in fruit inspection technologies. For instance, Yasaman Ghasempour from Princeton University is exploring millimetre wave technology to measure fruit ripeness without cutting into the produce. This system can detect internal qualities such as moisture and sugar content, potentially informing better harvest timing.
Adoption Challenges and Future Potential
Despite emerging technologies, many growers remain sceptical about the investment in such solutions. Jing Zhang from North Carolina State University highlights the importance of confidence in research for growers to adopt these technologies. Meanwhile, less costly systems like Kevin Wang’s drone-based crop-counting tool offer affordable alternatives for imaging crops.
While AI presents promising optimisation tools, Ben Palone from Western Growers emphasises the persistent value of human decision-making in critical harvest operations. Technology may complement but not yet replace the intuitive judgement of farmers.
Reporting based on coverage first published by bbc.co.uk.