€223,000 AgXeed W4 robot wins over dairy farmer — but not his wallet
A Dutch dairy farmer rate the AgXeed W4's integration five stars but walked away: €223,000 and 75 hp make it too costly and too weak for livestock work, says an NPPL researcher.
- Entered
- Plot
- 1.9 ha
- Record
- 774 words
- Walk-through
- 4 min
DrillSprayHarvestPlanning window

Agronomist’s notes
- The AgXeed W4 costs €223,000 and delivers 75 hp with a single PTO.
- Dairy farmer Joram Hofman tested the robot for one month, mowing with his 20-year-old Lely mower needing only one added sensor.
- A software bug limited the robot to 3 km/h instead of 10 km/h early in the trial.
- NPPL researcher Koen van Boheemen expects the W4 to find its niche in arable crop care, not livestock.
- Hofman predicts field robots will be as common as milking robots within about 20 years.
Dutch dairy farmer Joram Hofman spent a month testing the AgXeed W4 autonomous field robot and came away convinced the technology works — and equally convinced he will not buy this one. The price, more than €200,000 (€223,000 by Future Farming's assessment), is not financially viable for his farm.
"It's quite magical. You switch on the robot and it simply gets on with the job," Hofman said. He is no robotics novice: his farm already runs five robots, including milking and feeding units, so the learning curve was manageable. "I think that in around 20 years' time, field robots will be as indispensable to agriculture as milking robots are today," he concluded.
What did the robot actually do?
The W4 handled mowing as its main task, plus some tedding towards the end of the trial. Hofman mowed one field per day and baled the following day, a rhythm that suited autonomous operation. The robot mowed neatly, left no grass standing and gave an exact finish time, useful for planning.
Integration with existing machinery required almost nothing. The only modification to his 20-year-old Lely mower was a sensor monitoring rotational speed; if speed fell outside the permitted range the robot stopped and raised a fault. That stop never happened in practice.
Working width barely mattered. Despite running a single 2.80 m mower instead of his normal twin combination, autonomy freed time so effectively that the robot coped with the workload. Hofman believes it would also have suited overseeding grassland damaged by mice, but the robot had left by then.
Where did it struggle?
Headlands were the weak point in work. The robot reversed and manoeuvred at even small corners, wasting capacity. Instructing it not to reverse significantly raised output, at the cost of some uncut grass that Hofman finished with his own tractor.
A software bug early in the trial capped speed at 3 km/h instead of the programmed 10 km/h. The fault was already known from the tracked version and was quickly fixed on the wheeled one.
The safety sensors worked almost too well. Tall thistles — grown large because grazing cows left them standing and the delayed first cut — registered as obstacles, halting the robot repeatedly. The temporary fix was mounting the sensor as high as possible to see over the weeds.
Power was the structural limitation. With 75 hp and a single power take-off (PTO, the shaft driving implements), the robot struggled to condition the crop behind even one mower. Hofman would have preferred two mowers. "How was I supposed to know it would come with 710 mm tyres? A size smaller would have been a bit more practical," he added.
Setup demanded time on the first pass. Mapping fields, surveying and creating tasks through the AgXeed website took considerable effort initially, though repetition cut the workload. One field with a drainage ditch needed the ditch mapped as a boundary beside a watercourse. Hofman used TeamViewer to create tasks remotely from his phone, though he admits a desktop at the desk suits most people better. The joystick control felt too direct around the yard, even though it excelled at attaching implements.
How did it score?
The user review rated construction and hardware, integration and maintenance/service at five stars out of five. Autonomy, operating costs and overall value scored three, held back by thistle stops, the investment level and monitoring demands. Dealer Weevers Nieuwstad, backed by AgXeed, was reachable even at weekends; no maintenance was needed all month.
His verdict: no purchase. "I see potential in autonomous technology, but this robot is too expensive and too limited in power to be a practical addition to my farming operation."
What is the W4's real market?
Koen van Boheemen, technical researcher in precision agriculture at NPPL, calls the trial a technical success: "From a technical perspective, the robot is capable of carrying out a large proportion of the work on a dairy farm." But limited power relative to price, plus trouble on uneven ground and tall weeds, will hinder livestock adoption — especially since the robot was used for tasks it was not designed for.
Van Boheemen expects the W4 to find its niche in crop care on arable farms. For dairy farmers, he points to autonomous retrofit kits such as Iquus or the American Sabanto system: cheaper, and they leave the tractor free for manual feeding work. Hofman expects autonomy to beat manual work only in year two or three — the same curve his milking and feeding robots followed. Farmers watching this space should track retrofit kit pricing against purpose-built robots like the W4.
via Future Farming (Source)
More from James Calloway
Show full bio
Correspondent covering regulation, policy and their impact on production farming and field operations at Arable Wire.
201 articles
Nearby records
- HE 94AI on the farm is retraining workers, not replacing themNovember 14, 2026
- ZM 73Farming GT hoeing robot: 60 sold, but Dutch grower still tweaksOctober 19, 2026
- VP 61First ex-factory autonomous tractor leads ATH field-tech showcaseOctober 10, 2026
- LN 37Field robot or retrofit autonomy? Dutch grower tests bothNovember 2, 2026
- XY 53SwarmBot 5 logs 7,500 hours in Australian farm user reviewOctober 10, 2026