The turbine stands in the North Sea, off the Danish coast, and the damaged part is a strip of coating along the blade's leading edge — the line that meets the air first. The traditional fix is a technician on a rope, hanging high above the water in whatever weather the schedule allows, sanding by hand.
In a campaign whose results were published in July, the job went differently. A heavy-lift drone carried a suitcase-sized machine up to the blade, clamped it on, and an operator on deck drove it through the repair sequence from a screen. Sanding, cleaning, coating. Then the drone brought it back down.
Key takeaways
- During a commercial campaign at Ørsted's Horns Rev 2 wind farm, the Sparrow robot completed a leading-edge repair on a single blade in 27 minutes, with up to eight blades repaired in one working day.
- Ørsted and Clobotics (扩博智能) report the robotic method cuts repair cost by 50–75% and runs up to ten times faster than rope access.
- Two months earlier, the same team's best time was 38 minutes with six blades a day; in 2025, the first offshore robotic repair of this kind reportedly took one hour and 40 minutes.
- The robot was developed by a company founded in Shanghai in 2016, which bought two Danish wind-service firms in 2020 and now runs its wind business from Europe and Singapore.
Why the leading edge is the expensive part
A turbine blade is a composite wing that spends its life in a sandblaster. Rain, salt spray and airborne grit erode the leading edge, and once the protective layer is gone the underlying laminate starts to degrade. Left alone, the aerodynamic profile changes and the turbine produces less power for the same wind.
Industry estimates put the annual energy loss from leading-edge erosion at 3–20%, depending on site and severity. For an offshore farm, a few percent of production is a large number, and the repair itself is one of the least scalable costs in the business: a vessel, a crew, a weather window, and a technician suspended in the air.
That combination — high value, low scalability — is why blade maintenance became one of the first places where offshore wind was willing to pay for a robot.
What changed in 2026
Robotic inspection of blades is not new. Clobotics' IBIS system, a drone that photographs all four surfaces of a turbine's three blades in about 15 minutes and resolves defects down to the millimetre, has flown more than 180,000 blade inspections across 40-plus countries.
Repair is harder, because the machine has to do work, not just look. Sparrow is essentially a portable machine shop: it attaches to the blade, prepares the surface, and applies a new leading-edge protection layer while an operator runs it remotely. The point is not that it is faster at sanding. The point is that the technician is no longer the constraint on the schedule.
The progression in the published numbers is unusually legible. First offshore robotic repair of this kind, in 2025, reportedly took one hour and 40 minutes. May 2026: 38 minutes, six blades a day, 29 blades across six days. July 2026: 27 minutes, eight blades a day. Each step is a logistics gain as much as an engineering one — containerised storage on the vessel, battery handling, a dedicated operator station.
The part that actually saves money
Offshore maintenance economics are dominated by access, not labour. Getting people and equipment to a turbine consumes most of the budget, and every extra transfer adds weather risk and vessel time.
A repair that finishes in half an hour lets an operator compress more work into the same weather window, which is the scarce resource. That is why the reported 50–75% cost reduction is plausible even though the robot itself is expensive: the saving comes from cancelled vessel days and fewer rope-access crews, not from cheaper sandpaper.
It also changes what an operator is willing to schedule. If a repair is a 27-minute task rather than a two-hour ascent, minor damage gets fixed before it becomes structural, and the maintenance plan shifts from reactive to routine.
Where the Chinese part fits
Clobotics was founded in Shanghai in November 2016 by George Yan, a former Microsoft Greater China vice-president, and built its first business on computer vision — first for retail shelves, then for wind blades. Its 2020 acquisition of two Danish wind-service companies gave it the European field operation that Ørsted's campaign required.
The company now describes itself as Singapore-headquartered, with wind-service offices in Europe, the Americas and Asia-Pacific, and it showed Sparrow at China Wind Power 2026 in Beijing in October. The engineering is split across Shanghai, Seattle and Denmark — a supply chain that is increasingly normal in robotics and increasingly awkward for anyone trying to assign a nationality to a machine.
Honest limitations
The 27-minute figure, the eight-blade day and the 50–75% cost reduction are reported by Ørsted and Clobotics; neither has published an audited comparison against a matched manual campaign, and "suitable operating conditions" is doing real work in the daily-output claim. Sparrow repairs leading-edge erosion — it does not fix structural cracks, lightning damage or internal defects, which still need other tools. The 3–20% energy-loss range is an industry estimate, not a measurement from this site. The company's earlier records and inspection totals are self-reported. And a robot that needs a heavy-lift drone, a vessel and a trained operator is not cheap to deploy; the economics work at fleet scale, not for a single turbine.
What readers can do now
- If you invest in or buy power from offshore wind, ask operators how much of their maintenance budget is access cost. That is the line robotics is attacking.
- If you run a field-service business, study the Sparrow deployment model: the product is the campaign, not the robot.
- If you follow Chinese robotics, watch where the companies register and where the field crews sit. Headquarters tells you less than the service depot does.