Most glass roofs built before 2000 can host a skylight cleaning robot, but they usually need edge rails, gap sensors and a point-load check. Budget 15-25 percent of the machine price for the retrofit in 2026.
Modern roofs are designed with robotic access in mind – flat upstands, predictable panel sizes, drainage channels that suit a wheel path. Older roofs are not. The good news is that a retrofit is often mechanical and cheap, not structural. The bad news is that a few old roofs genuinely should not take a machine at all.
What does a retrofit actually involve?
Four things, in order. First, a frame survey: measure panel size variation, since 1990s glazing often runs 3-6 cm from spec. Second, edge markers or a low rail so the robot’s boundary logic has something to track. Third, a gap check at expansion joints, because a 25 mm gap can catch a wheel. Fourth, a point-load test on the glass itself.
- Frame survey: half a day, catches panel-size drift
- Edge markers/rail: 8-40 USD per running metre
- Gap bridging plates: needed over joints above 20 mm
- Point-load test: one pane sample, 200-400 USD
Which old roofs should be ruled out?
Three red flags. Wired or massively framed heritage glazing where drilling is forbidden. Acrylic or polycarbonate domes, which flex under any point load and craze at the contact line. And roofs with sealant-glazed units where the frame cannot take a rail without disturbing the weather seal – a leak risk that outweighs any cleaning gain.
For heritage buildings the answer is often a lighter machine run at reduced speed, or a drone for inspection only. Do not let a retrofit plan turn into a re-glazing project. That is a building works budget, not a cleaning budget.
How much should the retrofit cost?
| Item | Typical cost | Notes |
|---|---|---|
| Frame + load survey | 500-1,200 USD | One-off, half to one day |
| Edge rail | 8-40 USD/m | Depends on access method |
| Gap plates | 300-900 USD | Only over wide joints |
| Access walkway | 2,000+ USD | Only if none exists |
On a 2,000 m2 roof taking a Lingyun Y3, the retrofit for the first three items usually lands at 2,000-3,500 USD. That is a small fraction of a re-glaze and typically recovered within the first year of replaced rope-access cleaning.
Will an old roof void its warranty?
Check the glazing warranty before touching anything. Some 1990s units are still under extended performance guarantees, and adding rails can be classed as modification. Get written sign-off from the glazing supplier if the guarantee still runs. For roofs past warranty, a documented point-load test is enough evidence for insurers in most cases.
Navigation also matters on old roofs. Without clear frames the robot relies on its own sensors; see how robots navigate glass roofs without GPS. And weigh the frame-seal risk before drilling anything.
Who should retrofit rather than replace?
Buildings with sound frames and glass that will last another decade. There, rails and markers are a modest spend that open ups years of robot cleaning on an asset that still has life. The decision flips when the frame is soft, the sealant is failing, or the glass is near the end of its service window – then replacement wins.
Get a glazing condition survey before deciding. A specialist can rate each panel and tell you which need replacement within five years. Retrofitting the healthy sections and replacing the failing ones is often the cheapest path, and it leaves a roof that is uniform enough for reliable robotic navigation.
Key Takeaways
- Budget 15-25 percent of machine cost for a typical retrofit.
- Survey panel-size drift – old glazing varies 3-6 cm from spec.
- Rule out acrylic domes, sealant-glazed units and protected heritage frames.
- A point-load test is the evidence insurers usually accept.
- Check the glazing warranty before adding rails.
Planning a retrofit? Send roof age and glazing type and we will flag the risks.
How do you handle panel-size drift?
Measure five panes across the roof, not one. Old glazing varies, and a robot’s boundary logic expects repeating panels. Where sizes drift by more than 2 cm, mark the boundary physically with a low-profile strip so the machine has a consistent reference. It sounds fussy; it prevents the machine from riding a frame edge and stalling mid-pass.
Expansion joints are the other trap. A joint that opens 25 mm in summer can swallow a wheel and trigger an emergency stop. Bridge them with a plate rated for the robot’s wheel load, or route the machine around them and hand-clean the strip. Hand-cleaning 3 percent of the roof is cheaper than a recovery call.
When is a retrofit not worth it?
When the access cost dominates. If reaching the roof safely needs a new walkway and a hoist, the retrofit can approach the price of re-glazing the dirty section with modern, robot-friendly units. On a roof already near end of life, put the money into replacement glass that will last 20 years rather than rails on glass that will fail in five.
Also walk away if the frame cannot be touched without breaking the weather seal. A leak caused by a retrofit costs more than years of rope access. For those buildings the answer is a lighter machine, or periodic rope cleaning, or a drone for inspection. Not every roof has to take a robot.