Skylight cleaning robots cut fall-from-height risk but do not remove liability for dropped objects or glass damage. In 2026 most commercial insurers want documented suction monitoring, a tether or retention system, and named trained operators before they price cover favourably.
Building owners often assume the robot makes the whole operation low risk. Underwriters disagree, for good reasons, and the paperwork now reflects that.
What do insurers actually ask about robotic glass cleaning?
Four things come up repeatedly. Does the machine stop automatically if vacuum drops? Is there a secondary retention system, such as a tether or safety line, in case suction is lost? Who is trained to operate it and how is that recorded? And what is the procedure if the machine stops on a panel above a public area?
The last one trips people up. A robot stopped at 14 m on a mall atrium roof is not an emergency, but it is a controlled operation that needs a plan. Insurers want to see that plan written down.
Does a robot lower your premium?
Not automatically. It changes the risk profile, and it is up to you to present that clearly. What usually improves terms: fewer people at height, no rope access contractor on site, a machine with logged vacuum data, and a maintenance record showing pads were checked.
What makes terms worse: an untethered machine working above occupied space, no operator training record, and a maintenance log that does not exist. Underwriters notice absence of records far more than they notice good ones.
What risks does a skylight robot create that manual cleaning does not?
Dropped objects, mainly. A 19 kg machine falling from a roof is a serious event, and pad failure on a sloped skylight is a real failure mode on wet glass. A tether, an anchor point and a designated drop zone below are the standard mitigations.
Second, glass damage. Suction pads can mark or crack a weak panel, and a mis-set pad on a 6 mm annealed pane is a genuine hazard. This is why the load note discussed in our weight limits article belongs in the insurance file, not just the engineering file.
What should your risk assessment cover?
- Maximum slope on each roof section and the vacuum threshold that triggers a halt.
- Tether anchor points and their rated load.
- Exclusion zone on the ground under the working area.
- Operator competency record and refresher interval.
- Procedure for a machine stopped on a live panel above public space.
- Water spillage over electrical equipment or occupied areas.
Keep it to two pages and make it specific to your building. A generic template with no panel schedule in it will not satisfy either an insurer or a health and safety inspector.
Who should reconsider the whole approach
If your building sits above a busy public concourse with no practical way to close the area below, a roof robot adds a dropped-object exposure you may not be able to mitigate cleanly. In that case, internal access platforms or cleaning from fixed walkways during closed hours is the better route.
Similarly, if the site has no one who will own the equipment, a robot on an unsupervised roof is worse than no robot. Deploy, inspect, log. If that loop breaks, stop using the machine.
For spec sheets, vacuum monitoring details and maintenance intervals that satisfy an underwriter, talk to us via the contact page. Our safety and CE compliance guide covers the certification side.
How do you handle a machine that stops mid-roof?
Treat it as a controlled recovery, not a crisis. Keep a written sequence: confirm the exclusion zone below, check whether the machine is on a live panel, decide between local reset and lowering, then log the event. Practise it once with a stopped machine on the ground. Teams that have rehearsed recover in 15 minutes; teams that have not improvise and take an hour.
Do you need to notify anyone before each cycle?
On a building with public areas below, yes. A short pre-work notice to the site manager, the security desk and any tenants under the working zone takes minutes and removes most of the friction. Sites that skip it end up being stopped mid-cycle by security answering a complaint about a machine on the roof.
Who owns the machine when contractors run it?
Be explicit in the contract. If a cleaning contractor operates your robot, ownership, maintenance responsibility and damage liability all need writing down. A machine returned with worn pads and a flat pack after a season of use is a dispute waiting to happen. One page of terms prevents it.
What training does an operator actually need?
Half a day is enough for basic operation, but competence comes from the third and fourth cycle. Train on the boundary rules, the vacuum alarm, the recovery procedure and the pad inspection, and record it. The single most valuable habit is the pre-cycle pad check, because a dusty or torn pad skirt is the most common cause of both sliding and scratching. Make it a signed checklist, not a habit in someone’s head.
What does a typical cleaning cycle look like?
A full cycle is more than the time the machine spends cleaning. It is the walk to the roof, the water fill, the boundary setup, the run itself, and the pack-down. On a well-mapped roof the setup is 20 minutes and the run dominates. On a difficult roof with restricted access it can be 45 minutes before the machine lays down its first pass. Budget the whole cycle, not the pass, when you plan a shift.
Key Takeaways
- Robots remove the person from height but create dropped-object and glass damage exposures.
- Insurers want documented suction monitoring, tethering and trained named operators.
- Premium improvements follow good records, not robot ownership in itself.
- A 2-page building-specific risk assessment beats any template.
- No owner for the equipment means no robot on that roof.