A 12,000 m2 enclosed mall atrium cut glass-cleaning cost 47 percent over 18 months after moving to a skylight cleaning robot fleet in 2025, and reported zero trading-hour closures. The catch was a three-week navigation tuning period that the original quote did not mention.
This is a composite case drawn from mall-scale projects we have scoped, with the numbers rounded and anonymised. It is useful because malls combine everything that makes skylight cleaning hard: high glass, occupied trading floors below, brand-sensitive visibility and constant footfall.
What did the mall look like before robots?
Roughly 12,000 m2 of glass roof and curtain wall over three atria. Rope-access crews cleaned twice a year on night shifts, with gondola setups for the taller sections. Each visit cost a five-figure sum and required partial trading-hour precautions on the ground floor below. Between visits, dust and pollen film built up on the south-facing glazing within weeks, so the glass looked tired for most of the year.
How was the robot fleet set up?
Two tracked units were split across the atria, working 2-6am in rotation. A single operator ran both, moving between roofs via service stairs. The flat atrium sections ran at about 140 m2/h effective; the raked glass over the escalators ran slower at 85 m2/h because of repositioning and overlap. Battery swaps were timed to the operator’s coffee break, so no separate charging crew was needed.
- Glass area: 12,000 m2 across three atria.
- Fleet: 2 tracked skylight robots, one operator.
- Cycle: every 8 weeks instead of twice a year.
- Effective rate: 85-140 m2/h depending on slope.
- Cleaning window: 2am-6am, no trading impact.
What were the 18-month results?
Cleaning spend fell 47 percent against the prior rope-crew baseline once lease, labour and consumables were counted. Trading-hour closures dropped to zero. Tenant complaints about drips and water streaks fell sharply after the first two cycles, once the operator learned to pre-wipe humid interior glass dry before the wet pass. The mall also started cleaning during spring pollen season, which rope crews had simply skipped.
What went wrong?
Two things. First, the navigation tuning period. Escalator glass sits among handrails and signage, and the robot needed three weeks of waypoint mapping before it stopped clipping edges. Second, an early attempt to run during the noon lull caused noise complaints from a ground-floor restaurant, so the night-only rule stuck. Neither was fatal, but both should have been in the plan and budget.
Who should copy this approach?
Malls with 5,000 m2 or more of roof glass and a willing night operator. Smaller centres with under 2,000 m2 often do better with a single leased unit used monthly. And any mall where glass is a brand asset, not a maintenance afterthought, sees the fastest payback because the cleanliness gain is customer-visible. Rope access still has a role for the odd inaccessible pocket the robot cannot reach.
If you are scoping a mall project, send floor plans and glass areas through the contact page. Our airport, mall and factory guide covers the other large-venue types.
Key Takeaways
- A 12,000 m2 mall cut cleaning cost 47 percent over 18 months with two robots.
- Night-only runs (2-6am) eliminated trading-hour closures entirely.
- Budget a 2-3 week navigation tuning period for complex glazing.
- Effective rate was 85-140 m2/h depending on slope and obstacles.
- Fleet payback is best above roughly 5,000 m2 of roof glass.