Why this question matters now
On 26 October 2026, GXO and Agility Robotics announced plans to extend their humanoid robot partnership to two more US warehouse sites. They reported that Digit had already moved more than 100,000 totes at their existing site. The additional deployments are expected in early 2027. Agility Robotics and GXO
For cleaning companies and facilities managers, that makes Tony Sibley’s first question worth examining closely: how much of a cleaning shift could a humanoid reliably do?
The warehouse example shows a defined handling task in a particular setting. It gives us a reason to investigate cleaning capability, rather than an answer about it.
What the evidence shows
Cleaning robots already operate in UK buildings. Heathrow reported a fleet of 24 autonomous cleaning cobots in April 2026. These are specialist floor-cleaning machines. Their use helps us understand one part of the service, but cannot establish what a humanoid could do across a whole shift. Heathrow
The International Federation of Robotics makes a useful distinction in its 2026 overview. Floor cleaning is the main application for professional cleaning robots. Humanoid applications remain specialised and often involve a person controlling the machine remotely. The federation also identifies training, maintenance, safety standards and the business case as barriers to wider use. International Federation of Robotics
We therefore need to ask about the work being demonstrated. Following a mapped floor route, handling a tool and completing several different cleaning tasks each require their own evidence. A successful result in one of those tasks cannot settle the others.
Where a practitioner would start
Karl Haynes, a contributing FM practitioner, identifies the tasks he sees as the clearest fit for cleaning robots:
“Anything big, flat and repetitive. Floor scrubbing, vacuuming long corridors, overnight runs in malls, airports and car parks.”
He also points to doors, lifts, thresholds, changing furniture and unreliable Wi-Fi as practical obstacles. These are his observations from FM experience. He is discussing cleaning robots generally, rather than reporting a humanoid trial.
Karl Haynes, Facilities Management professional and Founder of Inside FM and FMHub Global.
What would make the claim convincing?
Tony’s original question asks manufacturers to explain what works reliably during a real cleaning shift, what still needs human help and what it costs. We would turn that into a short evidence request:
- Name the task. Which surfaces, rooms and duties were included? What was left for people?
- Show repeated operation. How long did the trial run, under what building conditions, and how often did someone intervene or control the robot remotely?
- Check the cleaning result. What agreed standard was used, and how was the result checked? A route-completion record describes activity.
- Count the support work. Include setup, training, charging, maintenance, recovery and supervision when describing the service and its cost.
For soft facilities management teams, this turns a broad technology claim into a service question that cleaners, contractors and customers can discuss together.
Our reading, and what remains open
The sources reviewed show commercial progress in bounded humanoid work and established use of specialist cleaning machines. They do not establish that a general-purpose humanoid can deliver an entire commercial cleaning shift at an agreed standard and lower total cost.
This is a review of published evidence and practitioner input. We have not conducted a robot trial. A documented cleaning deployment could change the assessment, particularly if it reports interventions, cleaning results and support costs.
The next question follows directly: when a machine takes over a task, what happens to the people, paid hours and work around it?
Sources
- GXO and Agility Robotics expand their humanoid partnership
Agility Robotics and GXO · 26 October 2026
Company announcement about warehouse handling. Performance figures are reported by the companies.
- World Robotics 2026: service robots and humanoids
International Federation of Robotics · 2026 edition
Industry-body overview. Cleaning robots and humanoids are discussed as distinct categories.
- Heathrow’s autonomous cleaning cobots
Heathrow · 2 April 2026
Operator announcement, rather than an independent evaluation of cleaning quality or cost.
Contribute to the inquiry
Have you run or assessed a humanoid cleaning trial? Share the task, operating conditions, human support required and how the cleaning result was checked. Please use information you can share publicly.
Join the discussion on the HI company page (opens in a new tab)