Робот-прореживатель из Онтарио удалил более тысячи яблок в час в коммерческих садах

An automated apple thinner developed in Ontario worked in 14 commercial high-density orchards during the 2026 season, using machine learning to identify fruit for removal. The system was also demonstrated at the University of Guelph-managed Simcoe Research Centre, where its six robotic arms reached for fruit but did not perform removals because the orchard had already been thinned. The robot makes up to 10 thinning decisions per second and directs its arms to "cut, suck, eject" fruit that does not meet its size criteria. It achieved a top rate of approximately 1,000 removals per hour this season. The development team aims to reach 500 removals per hour per robotic arm. The 2026 prototype was designed and built in Ontario. It uses a solar-powered cart and drive system, with a gas generator as backup. The unit operated for 10 to 13 hours per day without requiring the generator during the thinning season. For 2027, the plan is to increase the system to 20 robotic arms, with 10 on each side of the cart. At maximum capacity, the unit is projected to cover 18 acres per day, equivalent to about 7.3 hectares. Field testing is also being used to improve the robot's decision-making. Growers can provide orchard maps based on their experience, allowing the system to account for factors beyond fruit size. The current model can identify isolated fruit that should remain on the tree, but cannot yet identify deformities or insect damage when those fruit are the largest in a cluster. The technology could also allow thinning to begin earlier, when fruit is marble-sized or when spray residues prevent workers from handling it. "It's really hard to count 180 marble-sized fruit," said company president Matt Stevens. "But the robot can do it." The company is continuing to increase robotic capacity while using commercial orchard operations to collect data and refine machine-learning decisions.