TPFlabsBlueberry harvesting machine: types, costs and what actually works on fresh-market farms
A blueberry harvesting machine is not one product: it is three different tools for three different jobs. Over-the-row harvesters strip a bush in minutes but bruise fruit and take green berries with ripe ones. Walk-behind shakers suit processing fruit. Selective machines pick berry by berry and are the only category that keeps fresh-market grade. The choice depends on where your berries go after the field, not on how fast the machine drives.
Below is what each type actually does, what it costs per kilogram of picked fruit, and what to check before signing for one.
The three types of blueberry harvesting machines
Over-the-row harvesters. A tractor-mounted or self-propelled frame drives over the row and shakes the bush with beaters. Throughput is high, labour per hectare drops sharply, but the machine cannot tell a green berry from a ripe one. Fruit goes to freezing or processing, where a share of bruising and a few percent of loss are acceptable. For fresh-market packing, this is where most of the loss appears: about 5% of the crop stays in the field or leaves the premium grade.
Walk-behind and small-tractor shakers. Cheaper to buy and to run on small plots. Same logic as the big harvesters: whole-bush shaking, mixed ripeness, processing-grade fruit. On pick-your-own or farm-gate operations they can still make sense, but not for a punnet that has to travel to a supermarket.
Selective harvesting robots. A new and still small category. Instead of shaking the bush, a robot looks at each berry, decides whether it is ripe, and picks it with a soft gripper. The point is not raw speed — it is that the fruit never gets mixed, never gets bruised, and never waits for a picking crew to arrive.
What fresh-market blueberries require
Supermarket and export programmes pay for consistency: berry size within a narrow calibre band, no soft fruit, no bruising, and a cool chain that starts within hours of picking. Three machine characteristics decide whether you meet that:
- Time from bush to cooler. A machine that works 12-14 hours a day without a crew rotation compresses the harvest window instead of stretching it.
What a blueberry harvesting machine costs per kilogram
Purchase price is the number everyone asks about first, and it is the number that tells you the least.
- Robot-as-a-Service moves the cost to the season: service, software, updates and support included, paid per season rather than as capital expenditure.
For CALIBER+ the two commercial options look like this: RaaS at €26,000 per season, which works out to about €1.9 per kilogram of picked berries, or purchase — Lite (four arms) from €55,000, Flagship (six arms) from €80,000, with the prototype and pilot-series timeline running 2027-2028.
Five questions to ask before buying any harvesting machine
- What data do you get back? Ripeness distribution, yield per row and cost per kilogram are the numbers that make next season cheaper.
Where selective robots stand in 2026
Selective harvesting is the youngest category and the one that has just crossed from laboratory to field. CALIBER+, developed by TPF Labs, is an autonomous selective harvester for fresh-market blueberries: four SCARA arms on vertical carriages pick from two rows at once, a silicone-and-vacuum gripper holds each berry under 2 N of force, and an onboard AI — YOLOv9 with EfficientNet-B4 on an NVIDIA Jetson Orin NX, trained on a 50,000-berry dataset — decides ripeness with more than 97% accuracy and grades size within ±1 mm.
The platform runs on 48 V LiFePO4 with a 600 W solar roof and hot-swap batteries: 12-14 hours of daily autonomy, 17 kg/h across two rows, 450 kg on a tracked chassis built for soft soil. The gripper prototype is built and lab-tested; field prototypes are planned for 2027 with commercial launch in 2028.
The honest limitation is throughput. One robot today replaces part of a picking crew for a whole day, not the crew itself. The argument for it is quality and predictability: zero crop loss on the pass, no mixed-ripeness loads, and a harvest that does not stop when the seasonal workers do not arrive.
How to test a machine without buying it
The lowest-risk path is a pilot: a 30/60/90-day programme on your own field. Days 0-30 are a row survey and vision calibration for your varieties; days 30-60 the machine harvests one hectare while ripeness, calibre, loss and cost per kilogram are measured; days 60-90 you receive a full report and an option to lock a service tariff or buy for the next season. No financial obligation until a season contract.
Pilot slots for season 2027 are open. If you want to see the numbers on your own rows, write to drublev77@gmail.com with hectares, varieties and your harvest window.