Most roof cleaning robots fail not on glass but on the gap: any skylight frame, curb or raised glazing bar over 35 mm stops a standard chassis. Measure your highest frame before you measure your glass.
The term “roof cleaning robot” covers two very different machines. One clings to glass with suction cups and crawls across glazing. The other rolls on tracks across standing-seam metal or membrane roofs. Search for the first and you will keep landing on listings for the second. That mismatch is where most disappointing purchases start.
I have watched contractors fit both kinds on commercial atria and factory roofs across the last three years. The reliable pattern: the glass-climbing type wins when the roof is mostly transparent and flat-ish, and loses badly when the skylight grid is dense with deep kerbs. Here is how to tell which side of that line you sit on.
What does a skylight cleaning robot actually need from the roof?
Three measurements decide almost everything: pitch, frame height, and pane size.
- Pitch. Suction-based machines hold well to about 30 degrees on dry glass and start slipping past 45 degrees on wet glass. Anything above 60 degrees usually needs a tether plus a fixed winch point, which changes the whole installation.
- Frame height. A body that clears 20-30 mm of glazing bar is common. Higher-profile timber or aluminium kerbs will high-centre the chassis and break suction seal. This is the single most common return reason.
- Pane size. A 900 x 900 mm pane wastes a lot of cleaning cycle on repositioning. Anything under 600 mm across is rarely worth automating with a crawler — you spend more time resetting than cleaning.
Weight matters less than people assume, but it matters at the edges. A 3-5 kg crawler puts real load on frameless glazing adhesive. If the roof is a silicone-bonded panel system, ask the glazier before you put a machine on it.
How does a skylight cleaning machine handle water?
Badly, if you assume one tank is enough. On a pitched roof the water runs away from the brush faster than on a vertical facade, so coverage per litre drops by roughly a third. Two-motor feed systems — one pushing water to the brush head, one pulling dirty water into a recovery tank — are not a luxury on roof glass. They keep the damp film even and stop dirty runoff from streaking down the pane you just finished.
The skylight cleaning solution matters as much as the machine. On roof glass, mineral-heavy tap water leaves visible spotting within one dry cycle because the pane is tilted and the water pools at the frame edge while it evaporates. Use filtered or deionised water, or a low-residue concentrate. Avoid anything with a silicone additive on self-cleaning coated glass; it fills the coating pores and the roof stays dull for months.
If you are choosing a chemistry rather than a machine, our breakdown of water, soap and pH for skylight glass goes deeper on that trade-off.
Which skylight cleaning machine fits which building?
| Building type | Typical glazing | What works | What to avoid |
|---|---|---|---|
| Retail atrium / mall roof | Flat laminates, shallow kerbs | Suction crawler with 30-degree rating, 100-150 m²/h throughput | Tracked roof machine — it cannot climb the glass |
| Factory saw-tooth roof | Steep, small panes | Pole-fed brush on tether | Crawler — pane size too small |
| Residential sunroom | Bonded panels, 3-4 m high | Light cordless unit, 2-2.5 kg | Industrial crawler — 5 kg on bonded glass |
| Airport concourse | Large geometry, wind-exposed | Crawler with dual tether points | Single suction line in gusty areas |
One note on the airport row. Rooftop wind speeds run 20-30% higher than ground measurements, and a crawler with one tether anchor has no redundancy if the suction fails at the moment of a gust. Two anchors cost an extra fifteen minutes of setup. Take them.
Do robot roof cleaners work on self-cleaning glass?
Partly. Hydrophilic coatings break down dirt as water sheets across them — but they only work if the water sheets. A robot brush passes water in a narrow band and leaves wider dry strips than rain does, so on a panel that has never been touched, expect two passes the first time and one thereafter. Do not use abrasive pads, and do not let the brush run dry on coated glass; friction polishing is what kills the coating.
You also lose the coating’s benefit at the frames. Every kerb and mullion is a dirt trap that rain never reaches, so a quarterly manual detail at the edges is still part of the routine.
What actually goes wrong in year one?
Suction loss is the headline risk, and it is usually the operator’s fault rather than the machine’s. A slightly greasy suction cup, a scratched rim, or a pane edge with old silicone residue is enough. On a flat atrium roof with a safety net that is a scare. On a 14-metre sloped lantern roof it is an incident report.
The second failure is the fall-arrest tether being treated as optional. It is not. Even on a 6-metre roof, a 4 kg machine leaving the glass and landing on a car park costs more than the robot.
- Budget for new brush and cup sets at 200-300 operating hours, not 500.
- Keep a spare battery — a 60-minute runtime on a 400 m² roof means two battery swaps minimum.
- Log every suction alarm. Repeat alarms on one pane almost always mean a lip or seal repair, not a glass problem.
If you are still weighing machine families, the skylight and window robot reviews archive lists what each tested unit does well and where it broke. For a closer look at a ladder-free glass-roof machine aimed at mid-rise work, our Lingyun Y3 review covers the frame-clearing chassis in field conditions.
Who should not buy one?
If your glass roof is under 20 m², clean twice a year, and sits directly above a walkable ceiling, a telescopic pole and a squeegee still beat automation on cost. You would spend more time charging, filling and repositioning a robot than simply cleaning it. The maths flips somewhere around 150-200 m² of glass or four or more cleans a year — below that, rent before you buy.
Key Takeaways
- Measure frame height and pitch before anything else — a 35 mm kerb can rule out a whole machine class.
- Suction crawlers hold to roughly 30 degrees dry, less on wet glass; above 45 degrees expect a slip.
- Water consumption on pitched roofs runs about one third higher than on vertical facades.
- Two tether anchors, not one, on any wind-exposed roof.
- Below 150 m² of glass and two cleans a year, a pole is the honest answer.