Testing a 20 Minute Robot Setup Claim on Your Floor

Can a cleaning robot really go live in 20 minutes?

Possibly, but only under conditions that a vendor controls. The figure comes from a demonstration. According to Gausium's October 8, 2026 post, with its Smart Deployment a 3,000 m2 supermarket went from mapping to its first task in about 20 minutes, a process that previously took one to two days.

That is the claim worth testing. The clock, the building and the people on the cart matter as much as the robot. Gausium itself says actual time varies with site conditions, a caveat that also appears on its Mira product page as of October 2026.

What the 20 minutes actually starts and stops on

Gausium describes the sequence as power on, start mapping and switch on assisted pushing, with the robot building its map from 3D LiDAR as it moves. The post says only the main aisles need covering, and shelf zones fill in automatically. It also says the robot generates the outer virtual wall, marks carpets, glass and escalators, and blocks no-go areas without manual editing. Those are vendor statements, not independent measurements.

The workstation step is optional in that description. The robot can search for a workstation and create it after one tap, without repeated docking attempts. The post notes that sites which marked the dock during the survey can run a first task the same day, while other sites usually stop to resolve power or drainage. That detail matters, because it means the clock can stall on your building rather than on the software.

Which steps to time on your floor

Split the claim into segments so a slow result points at a cause. Time each segment with a plain stopwatch and write the start and stop time next to it. Use the same five segments on every robot you test, so the comparison stays honest.

• Assisted mapping walk, from power on to a closed and saved map
• Aisle and shelf zone fill-in, including your check of the result
• Automatic annotation review for carpets, glass, escalators and no-go areas
• Dock and optional workstation creation, with any power or drainage stop logged separately
• First test run, from start command to a completed pass with no manual correction

How do you set up the test so the number means something?

Pick a real operating area, not a demo bay. Use the cleaning staff who will run the robot day to day, not the vendor engineer, since a trained operator changes the result. Write down who is on the clock and what they had already learned before the test began.

Settle the dock before the timer starts, or time that stop on its own line. Power and drainage problems are common causes of delay that have nothing to do with mapping. Record the floor area in square meters as well, because the vendor's headline number is tied to a 3,000 m2 supermarket and a comparison needs a common basis. Gausium's technology page separately gives 30 minutes to map a 2,000 m2 site, which is a useful second reference point but still a vendor figure.

How to log results so the claim can be judged

Keep one log sheet per test. Include the floor area, the number of zones, the minutes spent in each segment, the number of interruptions and why they happened, the staff count, the shift and the lighting and clutter conditions. Note every manual correction the operator made to the map, because a fully automatic claim that needs three fixes is not the same claim.

Run the test at least twice, ideally on different shifts. Divide total minutes by floor area to get a rough minutes per 1,000 m2 figure you can compare with the vendor's number. One run tells you how the robot behaved once. Repeated runs tell you whether the result is a range you can plan around.

What a setup clock does not capture

Setup is a one time event, and most of the cost of ownership is not. Gausium's post says training recurs as staff change and that mapping recurs whenever the layout changes. The same vendor says Mira updates its map in real time, so there is no remapping and no engineer revisit. Test that claim too, by moving a shelf line on purpose and timing the recovery.

An independent review of the Gausium Phantas lists the deployment work a buyer should price: site survey, mapping, route design, cleaning mode setup, commissioning and acceptance testing. The same review says the robots still need setup, maintenance, consumables, water management, exception handling and human supervision. The reviewer also reports that AB Vassilopoulos cited a 20 percent reduction in manual floor cleaning time across a 25 robot fleet. That is the reviewer's summary of a case, not a figure checked against the case study itself, so treat it as a lead to verify. AB Vassilopoulos announced the 25 robot deployment in a May 21, 2026 press release.

Where an integrator fits the test

Service Robot Co. is a full service commercial robot integrator that is OEM neutral. We compare robots across manufacturers, then handle deployment and integration, go live support, training and service through a nationwide US engineer network. A site assessment includes mapping, so the same timing log can be run on whichever robot fits your floor rather than on a single brand's demonstration.

Our offer is a lease, not a sale. We lease robots on a 3 year lease with one fixed payment quoted after the site assessment, so the setup clock, the training plan and the service coverage are priced together. Ask any integrator, including us, to put the setup steps and the recurring work in writing before you compare numbers.