Choosing a skylight cleaning robot comes down to three site numbers: roof pitch, panel span and access width. If your glass sits above 25 degrees or your panels are older than 15 years, most robots are the wrong tool and you should stay with rope access or a boom lift.
What three numbers decide the right robot?
Start with pitch. Flat and low-slope glass between 0 and 15 degrees is the easy case, because the machine can hold with suction alone and move in straight lanes. Between 15 and 25 degrees you need a robot that pairs suction with a traction drive, so it grips as it climbs. Beyond 25 degrees, gravity starts winning. The Lingyun Y3 tops out near 25 degrees on dry glass, and once the surface is wet that safety margin shrinks fast.
Second number is panel span. Robot mass is not the only load you have to think about. A 28 kg unit such as the Lingfeng S1 spreads its weight across track contact points, but if you have 8 mm laminated panes on thin aluminium mullions from a 1990s atrium, the deflecting glass is the risk, not the machine. Ask the glazing supplier for a point-load rating before you buy anything. If they cannot produce that number, treat the roof as fragile.
Access width is the number people forget
A robot has to get onto the roof. Door widths, parapet heights and crane lift points decide whether you can actually deploy the machine you picked. Contractors who bought a large tracked unit and then discovered they could not lift it past a 900 mm service hatch are a common story this year. Measure the whole route, from truck to glass, before the spec sheet gets signed.
Robot or rope access: which suits which roof?
| Site condition | Better choice | Why |
|---|---|---|
| Flat glass, 0-10 degrees, large area | Skylight cleaning robot | No harness crew, repeatable coverage |
| Pitch over 25 degrees | Rope access crew | Suction margin drops in rain |
| Fragile 1990s glazing | Manual + boom lift | Point-load limits unknown |
| Curtain wall, vertical | Dedicated facade robot | Different machine class entirely |
| Mixed atrium with tight access | Lingkong K3 class | Compact footprint, narrow deck |
This is not a robot-versus-human argument. It is a matching problem. On a 4,000 m2 flat factory skylight, a robot wins on safety paperwork alone, because the crew never needs a rescue plan for the ordinary clean. On a narrow pitched church roof with leaded glass, no robot should be up there, whatever the brochure promises.
Where do buyers get the specification wrong?
- Buying for the steepest panel on site instead of the average pitch. One steep bay does not justify a machine that is weak everywhere else.
- Ignoring water. Wet glass cuts suction by a real margin, and most robots are rated on dry surfaces. A rain sensor is not a luxury add-on.
- Forgetting the tether and fall arrest. If the robot drops, someone below is at risk. Brochures rarely mention this.
- Underestimating panel gaps. Rails narrower than 40 mm stop some chassis cold at every seam, turning a one-hour job into a half-day.
There is a quieter mistake too: buying for the glass you see and not the dust you cannot. Sites near roads or construction pick up gritty film that wears brushes and cups faster. That changes the consumable budget, not just the machine choice.
Who should buy, and who should not?
Buy if you clean the same glass more than four times a year, the roof is above 8 metres, and your panels are rated for the load. The payback math on a Lingkong K3 in that scenario is usually under two years, and the safety case is easy to defend to a client.
Do not buy if your glass is fragile, your site has fewer than three cleaning cycles a year, or your access is so tight that deploying the machine takes longer than the job itself. Renting for a season first is the cheap way to find out which camp you are in, and it gives your operators real hours before you commit capital.
For deployment planning questions, the team at our contact page can review a roof plan before you commit budget, and this earlier note on pitch limits covers the steep-roof edge cases.
How much water and power does a roof robot need?
Cleaning glass is mostly moving water, so the tank and hose setup decides how long a run lasts. A compact machine carrying a 10 to 15 litre tank finishes roughly 120 to 200 m2 before a refill, depending on how greasy the glass is. Larger tracked units with 25 to 40 litre reservoirs roughly double that. If your roof has no water point, you are carrying containers up, and that labour cost belongs in the business case.
Power is the other constraint. Battery runtime typically lands between 3 and 5 hours per charge. That sounds generous until you add setup, refills and repositioning around obstacles. A realistic working window for a two-person team on a 4,000 m2 roof is one long morning, not a full day.
Spare parts lead time is a hidden project risk
A machine down for three weeks waiting on a suction cup is worse than a slightly slower machine with regional spares. Before buying, ask three questions: where are cups and brushes stocked, what is the typical lead time, and are the cups a standard size or a proprietary mould. Proprietary consumables give the vendor pricing power for the life of the machine, and that changes your five-year cost.
Key Takeaways
- Pitch, panel span and access width decide the machine, not brand preference.
- Above 25 degrees on wet glass, most robots lose traction margin.
- Check glazing point-load ratings before buying a 28 kg unit.
- Robots pay back fastest on large flat roofs cleaned four or more times a year.
- Measure the access route before the spec sheet.