Three skylight cleaning machine architectures dominate 2026: frame-guided suction units for pitched glazed roofs, dock-based units for large flat atria, and rail-fixed systems for repeat cycles on one long elevation. Choose by roof geometry and access route. A machine that fits the wrong one wastes a season before anyone admits it.
Suppliers rarely present it this way. They sell a product, so everything looks like a fit. The geometry decides first, then the access, then the budget. Working the other way around is how buildings end up with a 60 kg unit and no safe place to dock it.
What is a frame-guided skylight cleaning machine?
A frame-guided unit uses the glazing bars as its track. It sits on the glass, climbs the pitch, and relies on the frame shoulders to keep alignment. Weight is 12-30 kg, so one person can lift it onto the roof. These are the machines that suit the classical monopitch lights over factory floors and the small vaulted roofs over shop entrances.
Limits: they are slow on large uninterrupted panes because the frame is what keeps them honest, and they need bars at 1.5-2.5 m centres. A big glass roof with 3 m wide unsupported spans gives a frame-guided unit nothing to guide against, and drift starts within the first hour.
What does a dock-based machine do differently?
Dock-based systems work on area, not on lines. The unit returns to a base station for water and charge, so runtime stops being the constraint. On a 2,000 m² flat atrium, this is the only architecture that makes a full cycle practical within one working day. Weight climbs to 45-70 kg, so you need roof access that takes a trolley and two people.
| Architecture | Best roof | Weight | Cycle limit |
|---|---|---|---|
| Frame-guided | pitched 10-25°, bars at 1.5-2.5 m | 12-30 kg | runtime per charge |
| Dock-based | flat 0-8°, 800 m² and up | 45-70 kg | water and filter, not power |
| Rail-fixed | one long elevation, 60 m and up | fixed structure | rack travel length |
When is a rail-fixed machine the right answer?
When the same glass gets cleaned 12 or more times a year and the roof is a long strip above a public concourse. Rail removes the alignment problem entirely: the carriage knows exactly where it is. It also removes the edge risk, because the carriage cannot leave the rail. The cost is the fabricating work, the bracket loads on the mullions, and the fact that a rail is now a permanent feature of your roof.
This is where a structural engineer earns their fee. Bracket loads look trivial until you multiply by a carriage that stops abruptly. Ask for a design that keeps the rail behind the mullion line so it does not show in photographs, and check the drainage path for the rail foot before drilling.
Who should choose which?
A factory owner with eight monopitch lights and a 20-year-old roof should buy frame-guided and keep two spare brushes. A mall with a 2,400 m² atrium needs dock-based, plus a trained operator, plus a service contract. A transport interchange with a 220 m glass canopy over a concourse should not buy a mobile unit at all; it should commission a rail.
The one group that should buy nothing yet: buildings with glass at more than four distinct angles. No single architecture covers that, and a mixed roof usually costs more to automate than it saves in the first three years. For those, the cheaper answer is a rope crew for the awkward sections and a mobile unit for the flat ones.
If your glass is mostly near-horizontal, start with the Lingkong K3 review for the dock-based class, and check the full skylight robot review archive before shortlisting. The comparison there covers what each class does under real water consumption, which decides more cycles than power ever does.
What roof access does each architecture require?
Access decides as much as geometry and costs more than most buyers expect. Frame-guided units need one person and a hatch: 30-45 minutes to set up, and the same to pack down. Dock-based machines need a trolley route to the roof, a level spot for the dock with drainage, and two people for the initial lift. On buildings where the only roof access is a 600 mm hatch with a fixed ladder, a 60 kg machine simply does not go up, and no amount of specification changes that.
Rail systems remove the lift problem entirely because they stay in place. The install week is the expensive part: a hire platform at 900-2,600 USD, plus a crew, plus the structural engineer on the bracket detail. After that, a single technician cleans the whole elevation from a controller at the parapet.
Consider the fall protection question too. Anyone working on a glass roof within 2 m of an unprotected edge needs either a guardrail, a harness and an anchor, or a documented safe system of work. A robot does not remove that obligation for the person who places and retrieves it. Budget for a guardrail on the access side, and put the dock behind it.
How does water supply change the choice?
A frame-guided unit carries 0.8-1.5 litres on board. A dock-based unit draws from a 20-40 litre tank that a technician refills. A rail system can be plumbed into a permanent supply if the building has a roof tap, which removes refills entirely and is the single biggest productivity gain in the whole category.
Water quality matters more than volume. Hard water above 180 mg/l calcium leaves mineral spotting as it dries, and on dark glazing those spots are visible from the floor. Every architecture needs filtration, and the filter is a consumable, not an accessory. Budget replacement every 3-4 months on hard water and check total dissolved solids before you sign the contract.
Key Takeaways
- Frame-guided: 12-30 kg, pitched 10-25° roofs with glazing bars at 1.5-2.5 m.
- Dock-based: 45-70 kg, flat roofs from about 800 m², limited by water and filters rather than battery.
- Rail-fixed: right choice above roughly 12 cycles a year on one long elevation; needs mullion bracket engineering.
- Weight dictates roof access: anything above 45 kg needs a trolley route and two people.
- Glass at more than four distinct angles is usually cheaper to clean by crew than to automate.