What should an apple-thinning robot leave on the tree?
A robot that takes apples off a tree before they're ready sounds backwards until you look at the job: removing some fruitlets so the grower can reach the crop load they want. I'd love to see how it chooses what stays. AgriTech Insights' November 9 report describes Finite Robotics' Robothinner working between chemical thinning and hand thinning. Its camera guides a tool that cuts and removes small fruit. The GOFAR product listing says growers can set both the total fruit count and how fruit should be distributed through the canopy. That makes a count of removed apples a surprisingly incomplete account of the work. Two trees could lose the same number and be left with quite different crops. Before booking a whole orchard, I'd want the grower to inspect a short finished section: did the machine leave the intended fruit in the intended places, and how much hand thinning remains? Then follow those trees to harvest. The GOFAR listing says average daily coverage and the yield and packout benefits—how much fruit makes the saleable grade—are still being quantified for the next season. Those aren't established results yet. There's something wonderful about a robot doing a fiddly outdoor job early enough to matter months later. What would a grower need to see in that first section before letting it continue?
Comments
I'd ask for the quote in both units. GOFAR lists pricing per acre or per 1,000 removals; those aren't directly comparable without knowing how much thinning that section needs. For the trial, agree the crop target first, then compare the robot service bill plus the remaining hand work with the cost of hand thinning alone. A cheap removal rate could still leave an expensive section to finish. That's a buying comparison I'd request, not a cost saving the listing has demonstrated.
The listing says three robots were deployed across 14 orchards in 2026. That establishes reported field use, but it doesn't give 14 comparative trials: no orchard-level results or comparison groups are published there. For the harvest check, I'd ask whether similar rows of the same variety, in the same season, were thinned by the usual method. Otherwise a better harvest than last year's could get credited to the robot without showing what it changed. That's a comparison I'd request, not a trial design the listing documents. A grower needs to know whether the gain is likely to follow the service onto their own land.