TPFlabsBlueberry picking robot: how selective harvesting works, and when it pays
A blueberry picking robot does what a picking crew does, berry by berry. A shaker harvester does something else: it takes the whole bush at once. That difference sets the rest of the economics. The robot keeps fresh-market grade but has to be fast enough to replace hands. The shaker is fast but turns premium fruit into processing fruit. This guide explains how a selective robot works, where the limits are and how to decide if one belongs on your farm.
What a blueberry picking robot actually does
Every selective robot repeats the same four steps for each berry, thousands of times an hour:
- See. Cameras scan the bush from several angles. A vision model separates ripe berries from green, red and damaged ones, including berries half hidden behind leaves.
- Decide. The software chooses which berries to pick on this pass and in what order, so the arms do not cross or knock fruit off the branch.
- Pick. An arm moves to the berry and a soft gripper detaches it with a gentle twist or pull, the way a picker's fingers do.
- Place. The berry goes into a padded, cooled container instead of falling through a catch frame.
Ripeness is the hard part. A blueberry changes from red to blue in days, and the bloom (the white wax on the skin) is one of the things buyers check. A robot that rubs the bloom off or picks reddish berries fails the fresh market even if it is fast.
How CALIBER+ is built
CALIBER+ is the selective harvester we are developing at TPF Labs. Its design targets show the main choices behind a picking robot:
- Two rows at once. The machine straddles the row. Two manipulators per side work from vertical carriages, so one pass covers both sides of two bush lines.
- SCARA arms. SCARA arms are fast and precise in the horizontal plane, which matches how berries hang on a bush. The vertical carriage handles height.
- AI ripeness vision. A detection model trained on a 50,000-image berry dataset finds ripe berries and skips green ones, pass after pass.
- Soft hybrid gripper. Silicone fingers with vacuum assist hold each berry with under 2 N of force, to keep the skin and bloom intact.
- Chilled hopper. A 60 L cooled container slows the loss of quality between the bush and the packhouse.
- Hot-swap battery. The design target is 12–14 hours of work a day, and a battery swap takes under a minute.
These are design targets for a machine in pre-production, not results from a commercial fleet. Pilot farms will measure them on their own rows in 2027.
Robot vs crew vs shaker: what changes on the farm
| Picking crew | Shaker harvester | Selective picking robot | |
|---|---|---|---|
| Fruit quality | Fresh market | Mostly processing | Fresh market (design goal) |
| Ripeness | Pickers choose | Ripe and green mixed | Ripe only, berry by berry |
| People needed | A full crew | Driver plus sorting | One operator per machine |
| Cost driver | Wages, housing, transport | Machine price, fuel, losses | Season fee or purchase |
| Main risk | Finding people on time | Downgraded fruit | New technology |
A robot does not remove people from the farm. It changes the job: fewer people picking, and some people running machines, checking quality and handling fruit.
When a picking robot pays off
A robot pays off when three conditions meet:
- You sell fresh-market fruit. If your berries go to processing, a shaker is cheaper per kilogram and the robot's gentle handling earns nothing extra.
- Labour is your bottleneck. If you lose fruit or grade because the crew arrives late or is too small, every hour a machine works in the peak is worth more than its average cost.
- The numbers per kilogram work. Manual picking costs about €2.4 per kilogram. A CALIBER+ season contract (Robot-as-a-Service) is €26,000 per season with service and software included, about €1.9 per kilogram of picked berries. Purchase starts at €80,000.
On 100 tonnes of fresh-market berries that is roughly €240,000 for manual picking against about €190,000 with RaaS. Our cost per kilogram guide walks through that example in detail.
What to ask any robot supplier
- Measured, not promised. Ask for kilograms per hour, loss rate and share of unripe fruit picked, measured on a farm like yours.
- Your varieties and trellis. Bush shape, row spacing and variety change everything. Ask how the vision model is recalibrated for your plants.
- Service in the peak. Ask how fast a technician reaches the field in harvest week, and who pays for downtime.
- Who owns the data. Ripeness maps and yield per row are valuable. Make sure you get them.
- How to exit. A good offer lets you test before you commit. See how a machine compares with the other options.
How to test a picking robot on your own rows
The CALIBER+ pilot runs 30/60/90 days. First a row survey and vision calibration for your varieties. Then autonomous harvesting on one hectare, measuring ripeness, calibre, losses and cost per kilogram. At the end you get a full report and the option to lock a tariff or buy for the next season. There is no financial obligation until you sign a season contract.
Pilot slots for season 2027 are open. Write to orders@tpf-technology.com with your acreage, varieties and harvest window, or read more on the CALIBER+ page.
FAQ
Can a robot really pick blueberries without damaging them?
That is the main design challenge. CALIBER+ uses a soft silicone gripper with vacuum assist and a target grip force under 2 N, so the skin and bloom stay intact. Pilot farms will measure damage rates on their own fruit in 2027.
How fast is a blueberry picking robot?
Speed depends on the machine, the bush and the yield. The CALIBER+ design target is 17 kg per hour per machine, over 12–14 hours of work a day.
Will a picking robot replace my whole crew?
No. It reduces how many people you need for picking at peak. You still need people to run the machines, check quality and handle fruit in the packhouse.
How much does a blueberry picking robot cost?
A CALIBER+ season contract is €26,000 per season including service and updates, about €1.9 per kilogram. Purchase starts at €80,000.