A glass facade robot is built for vertical or near-vertical glass with a rail or edge anchor; a skylight robot works on near-horizontal glass and depends on the frame for guidance. Buy by glass angle and mount type, not by suction rating. Most 2026 buying mistakes come from treating the two as one product category.
They share vacuum modules, brushes and water systems. That is where the similarity ends. A machine tuned for a vertical facade runs a lower brush load and a faster pass because the solution would otherwise run off before it lifts anything. Put that same unit on a 15° skylight and it leaves a wet film that dries with streaks. Reverse the install and a skylight unit scrubs too hard on a vertical panel and drags.
The same machine family covers facade work, but the physics changes. Vertical glass sheds water fast, so solution contact time is short and brush speed needs different tuning. Horizontal glass holds a film of solution and gives the brush time to lift dirt. A machine that streaks on a vertical facade can be excellent on a skylight, and the reverse. Do not assume a facade unit is your skylight answer without a trial.
Rake and vertical face cleaning is where the 2026 engineering is moving. Guide-rail systems now sit behind the mullion line so nothing shows in the facade photograph. Rail cost dominates: 40,000-90,000 USD for 200 m of elevation, and it needs structural sign-off on mullion bracket loads. Owners who cannot take that route use descent-only tethered units, which cover about 70% of the visible area predictably.
Where the truth is less flattering: silicone joints. Glazing sealant attracts dust and turns grey, and no brush removes it. The robot cleans the glass and leaves the joint line untouched. On white or silver facades this is invisible. On dark glazing the joints read as stripes and the property manager asks why nothing changed. Budget hand joint cleaning every four to six years whatever machine you buy.
Insurance deserves its own line. Carriers increasingly ask for a documented wind-downgrade procedure and a verified second anchor for anything launched above 100 m. Liability for a dropped unit in a public plaza is not a service-contract talking point, and the answer is an anchor engineer and a load test, not a longer warranty.
Which glass angles actually work?
| Glass angle | Best machine type | Practical note |
|---|---|---|
| 0-10° (roof lights, atria) | Skylight unit | frame-guided, watch pooled water near edge |
| 10-25° (glass canopy) | Skylight unit | reduce speed above 20° |
| 25-60° | Hybrid, trial first | suction load rises on wet steep glass |
| 60-90° (curtain wall) | Facade unit | rail or edge anchor strongly preferred |
| Overhanging soffit | Neither | specify manual pole access |
What should you test on site before buying?
Run the machine on your own glass, at your own angle, in your own weather. Ask for three passes on the dirtiest pane you have, filmed from below against a plain sky. Streaks show up against sky like nowhere else. Then ask to see a suction-loss event staged deliberately: lift the unit half off the glass and watch what it does. Any supplier who will not stage that should not get the order.
Check the edge behaviour too. On a glass roof with 8-10 mm gaps between panels, the drive wheel finds the gap. A skylight machine should bridge or cross it without a stall. The Lingyun Y3 review covers the crossing behaviour in detail, and it is the test most demo videos quietly skip.
Finally, ask about the dock and the tether path. On a flat glass roof the dock needs a level spot that is not a drain, and the tether should not cross a footpath. Small details, but they are what makes the difference between a machine that gets used weekly and one that gets covered with a tarp after a month.
Who should not buy either machine?
Buildings with glass at more than four different angles, owners who cannot put an anchor point on the roof, and anyone expecting the machine to clean glazing sealant or painted frit. Those are hand jobs. For everything else, browse all skylight robot reviews and match the machine to the glass rather than to the brochure.
The same machine family covers facade work, but the physics changes. Vertical glass sheds water fast, so solution contact time is short and brush speed needs different tuning. Horizontal glass holds a film of solution and gives the brush time to lift dirt. A machine that streaks on a vertical facade can be excellent on a skylight, and the reverse. Do not assume a facade unit is your skylight answer without a trial.
Rake and vertical face cleaning is where the 2026 engineering is moving. Guide-rail systems now sit behind the mullion line so nothing shows in the facade photograph. Rail cost dominates: 40,000-90,000 USD for 200 m of elevation, and it needs structural sign-off on mullion bracket loads. Owners who cannot take that route use descent-only tethered units, which cover about 70% of the visible area predictably.
Where the truth is less flattering: silicone joints. Glazing sealant attracts dust and turns grey, and no brush removes it. The robot cleans the glass and leaves the joint line untouched. On white or silver facades this is invisible. On dark glazing the joints read as stripes and the property manager asks why nothing changed. Budget hand joint cleaning every four to six years whatever machine you buy.
Suction load is the other spec that gets misused. A 2,500 Pa holding force sounds impressive until you note that it was measured on dry glass at 20°C. Wet glass with a film of solution drops holding force by 20-35% depending on the seal profile, and a hot day makes the vacuum pump work harder for the same effect. Ask for the wet-glass figure and the ambient temperature it was taken at, not the dry peak.
Insurance deserves its own line. Carriers increasingly ask for a documented wind-downgrade procedure and a verified second anchor for anything launched above 100 m. Liability for a dropped unit in a public plaza is not a service-contract talking point, and the honest answer is an anchor engineer and a load test, not a longer warranty.
Key Takeaways
- Facade units suit 60-90° glass; skylight units suit 0-25°. Do not substitute one for the other.
- A skylight unit leaves streaks on vertical glass because the solution runs before the brush lifts dirt.
- Demand three passes on your dirtiest pane, filmed from below, plus a staged suction-loss test.
- Panel gaps of 8-10 mm are where cheap units stall; test the crossing on your own roof.
- No robot removes glazing sealant grime or restores frit; budget hand work every 4-6 years.