A skylight cleaning machine in 2026 comes in three mechanical families: suction crawlers, rail systems and rotating brush poles. Pick by roof geometry and cleaning frequency, not by price; the cheapest family is the wrong one about a third of the time.
The word "machine" covers everything from a 2 kg crawler to a fixed gantry, and sellers rarely say which family they mean. That ambiguity is why buyers end up comparing a 1,400-dollar unit against a 40,000-dollar track and conclude the market is insane. It is not; they are different tools.
Which skylight cleaning machine family fits which roof?
Suction crawlers work on any reachable sloped glass plane and cost least. They need a start point, a tether and a clean frame edge, and they scale with labour: more area means more passes, not more throughput.
Rail and gantry systems put a fixed track along the roof and move a cleaning head along it. Setup cost is high, throughput is steady, and the geometry must be regular. They win on large, repeat-cycle, uninterrupted glazing, which in practice means long atria, airport halls and industrial roofs above about 1,200 m².
Rotating brush poles are the honest workhorse. They reach 6 to 12 m, need a person below the glass, and remove dirt without removing water; that limits them to rain-washed or lightly soiled roofs. For a single skylight over a corridor, a pole is often the correct answer and no one selling robots will tell you so.
What decides the choice: area, pitch or access?
Access decides first, then area, then pitch. If a technician can stand on a walkable roof next to the glazing, poles and crawlers both work. If the only access is from inside the atrium, a crawler with a tether from an interior anchor is the practical route. If nobody can get on the roof at all, you are designing a building system, not buying a machine.
Pitch matters mostly for crawlers. Above roughly 35°, tether and seal design start to dominate the purchase decision, and the range of viable products narrows sharply.
Family Typical cost Area fit Main risk Suction crawler USD 1,400–22,000 20–1,500 m² Seal failure, tether discipline Rail / gantry USD 35,000–60,000+ 1,200 m² and up Capital lock-in, geometry mismatch Rotating brush pole USD 400–1,800 Any, height-limited No dirty-water removal, reach limit
What do buyers get wrong about skylight cleaning machines?
Two things. First, they size the machine to the total glass area instead of to the largest uninterrupted pane. A 500 m² roof split into 40 framed bays behaves like forty small jobs, and the crawler that looked underpowered is suddenly fine while the rail system is impossible.
Second, they ignore water. Dirty water has to go somewhere. A machine that scrubs and spreads is worse than a pole, because it turns dust into a film that bakes on. Look for a dirty-water return or at least a squeegee-and-vacuum arrangement, and check where the machine discharges at the frame.
A third, subtler error: buying for the glass you have now. Retrofits and roof extensions are common, and a rail system bolted to a specific mullion grid does not forgive additions.
How do you run the pilot before committing?
Pick your worst roof, not your best. Dirty, sun-exposed, framed, with a vent or a step in the middle. Run one full cleaning cycle and time it end to end including setup, water fill and anchor rigging. Then multiply by your annual cycle count. If the number of technician-hours is not clearly below what you spend now, the machine is not the answer for that roof.
For a structured comparison of the crawler family, our breakdown of three machine types goes model by model, and the reviews archive holds nine tested units with measured coverage and edge performance. The Hutt DDC55 Pro review is a useful reference point for the higher-throughput end of the portable class.
Who should skip robots entirely in 2026?
Buildings with fewer than five skylights, cleaned twice a year, where a contracted rope or pole crew already visits. The training, storage and anchor inspection overhead eats the savings. Also anyone whose roof glazing is listed or heritage, because every family on the list above assumes you can put load on the frame.
Everyone else, run the pilot. One afternoon on the worst roof tells you more than a year of brochures.
Funding models change the arithmetic more than most buyers expect. Rail systems are often financed as building plant and depreciated over fifteen years, while a portable crawler is an operating expense. For a landlord with a capital plan, the rail can look cheaper on paper even when the total cash outlay is ten times higher. Ask your finance team which line it lands on before comparing quotes.
Training is the other hidden cost. A crawler needs one trained operator and a written method statement; a gantry system needs a maintenance regime and a control-system owner. The second is closer to a building services contract than a cleaning contract, and if nobody on your site owns that, the system will run well for a year and then quietly decay.
Finally, decide who is accountable for the result. If cleaning is contracted out, the contractor will choose the method and your machine specification becomes advice. Owners who want control over method usually keep the equipment in-house and contract only the labour. That choice, made early, determines whether any of the numbers above apply to you.
Key Takeaways
- Three families: suction crawlers, rail/gantry systems, rotating brush poles, each with a clear area fit.
- Access decides the family before price does.
- Size to the largest uninterrupted pane, not total glass area.
- Machines that scrub without removing dirty water leave a worse roof than a pole.
- Run the pilot on your dirtiest roof and count technician-hours, not cleaning minutes.