Skylight cleaning robots cover roughly 20 to 40 m2/h in compact models and 60 to 100 m2/h in larger tracked units in 2026. Real daily output is usually 40 to 60 percent below the headline figure once refills, repositioning and battery changes are counted.
How many square metres per hour can a robot really clean?
The headline number assumes clean glass, a flat lane, a full tank and unlimited time. None of those holds on a real roof. A compact atrium unit such as the Lingkong K3 typically manages 20 to 40 m2/h across a shift. A larger tracked machine like the Lingfeng S1 can push 60 to 100 m2/h on open flat panes. That is a five-fold spread, so the machine class matters more than the brand.
| Machine class | Headline m2/h | Realistic m2/h | Main limiter |
|---|---|---|---|
| Compact suction unit | 30-40 | 20-30 | Tank refills, tight access |
| Mid tracked unit | 60-80 | 40-60 | Repositioning around obstacles |
| Large tracked unit | 90-100 | 60-75 | Battery swaps, water hauling |
| Interior ceiling unit | 25-35 | 15-25 | Public-area scheduling |
How does battery runtime cap a cleaning day?
Runtime lands between 3 and 5 hours per charge for most machines. That is the ceiling on any single session. If your roof needs six hours of cleaning, you either carry a second battery or split the job across two days. A second battery costs a few hundred dollars and doubles usable time, which is usually the cheapest capacity upgrade available.
Charging adds its own delay. A full charge can take 2 to 4 hours, so back-to-back work without a spare battery means downtime you cannot bill. Plan the day around charge cycles, not around the machine’s best-case speed.
Water capacity is the other hard stop.
Tanks range from about 10 to 40 litres depending on machine size. A small tank finishes roughly 120 to 200 m2 before a refill. If there is no water point on the roof, every refill means carrying containers up, and that labour eats the time the robot saved. Sites with a rooftop tap simply get more done per shift than sites hauling water by hand.
Why is real output so much lower than the brochure?
- Refills. Water and battery stops interrupt the run, and every stop costs setup time.
- Repositioning. Vents, frames and panels break the lanes, so the machine lifts and re-drops more than the spec imagines.
- Glass condition. Heavily soiled or streaked panes need a slower pass or a second one.
- Weather. Wind or light rain pauses the job, and a paused machine cleans nothing.
None of this means the machines are overhyped. It means the number to plan with is the realistic column, not the headline. Budget 50 to 60 percent of the brochure figure and your schedules will hold.
Should you size the fleet on throughput or on cycle time?
Cycle time wins. If a 5,000 m2 roof cleans at 50 m2/h realistically, that is 100 working hours, or roughly 15 short shifts. Whether you run one machine or two depends on how fast you need the job done, not on a single headline figure. Two machines halve the calendar time but double the capex and the crew. For most facilities, one machine on a regular cycle beats two sitting idle between jobs.
For a deeper cost view, see our earlier throughput analysis, and talk through your roof size via the contact page.
How do you measure real throughput on your own roof?
Pick one representative area, run the machine for a full shift, and count the square metres cleaned to a finished standard. Include the refills and the battery stop in the time. That number, divided by hours, is the only figure worth planning with. Vendor numbers are a starting point; your roof decides the truth.
Do the test twice, once in dry weather and once after rain, because wet glass changes both speed and traction. Sites that skip this step routinely under-budget cleaning cycles by half.
What should you do with the number once you have it?
Turn it into a shift plan. If your roof is 3,000 m2 and your real rate is 45 m2/h, one full shift covers about 2,700 m2 with allowance for setup, refills and a slower pass on the dirtiest panes. That means a two-shift clean, not a one-shift clean, and a schedule built on the honest number holds. A schedule built on the brochure number does not, and it fails in the first dirty week when the glass is worst.
Review the number each season. Glass soil changes with weather, nearby construction and pollen cycles, so a rate measured in spring may be optimistic by summer. Tracking it keeps the plan honest instead of hopeful.
Write the number on the machine or in the shift log. When a new operator takes over, the first thing they should see is the real rate, not the marketing figure. It sets expectations correctly and stops the team from promising a client a schedule the roof cannot deliver. Revisit it after any change to the machine, the glass or the cleaning chemicals.
Key Takeaways
- Headline throughput is 20-40 m2/h compact and 60-100 m2/h large tracked.
- Plan with 40-60 percent of the brochure figure for real scheduling.
- Battery runtime of 3-5 hours caps any single session.
- A second battery is usually the cheapest way to extend a day.
- Size the fleet on cycle time, not on a single speed number.