
Start outside the peak, on a route that has already been walked. Our pick for a route that passes 800 mm and can keep a workstation is the CleaniBot C5. Our pick for a hall the size of Frankfurt Terminal 1 stays the Tennant T16AMR, because that is what Tennant’s own case study says was chosen. Deployment is the order of work after that choice: survey, no-go walls, dock, first off-peak task, then the report. It is not a second shortlist.
Disclosure. OrionStar publishes this page and sells CleaniBot. Machine figures are from manuals and manufacturer specifications. Airport events are the publisher’s own text, accessed 5 November 2026. Which zone gets which robot is the airport page.
Airport deployments have to account for live terminals, passenger queues and limited cleaning windows. Examples from Frankfurt, Milan and George Bush Intercontinental show why the rollout order matters. Staff still do restrooms, high-touch surfaces and the queue.
| Step | Done when | The number to hold | Who signs | Stop the rollout if |
|---|---|---|---|---|
| Zone walk | Hard floor, carpet, queue and kerb are four lines | Terminal 1 in the Tennant case study is over 500,000 m² of stone | Operations | One “airport floor” is written for all four |
| Machine match | Ride-on hall or 800 mm route, not both | T16AMR pass 1.45 m, 190 L / 225 L. C5 pass 800 mm, 45 L / 45 L | The buyer, in writing | The C5 is offered as the Frankfurt machine |
| No-go walls | Queue, exit and accessible route are on the map before go-live | C5 turn 1,720 mm. Neo 2W turn about 3.05 m | Installer | The first task is aimed at the security lane |
| Dock live | Return, charge, and water only if that dock does water | C5 refills and drains. T16AMR charge is off-board | Building engineer | “All night” is promised with no water dock |
| First task off-peak | One closed or low-traffic window, then a report | C5 ≤70 dB(A), 3 h. T16AMR 71 dB(A). Neo 75 dB(A) | Named attendant for week 1 | The first autonomous run is at a departure peak |
| Staff list | Restrooms and high-touch stay with people | Houston’s newsroom says the four TASKI machines left those jobs to custodians | The contractor | The proposal deletes the custodial posts |
| Fleet claim verified | The vendor statement is identified as such | Gausium, 6 February 2026: Omnie at Malpensa and Linate. SoftBank, 27 August 2026: 100 robots, 15 locations | Nobody treats it as an audit | Vendor statements are presented as measured fact |
Walk one concourse that passes 800 mm, climbs within 2%, and has a room for the C5 workstation. Map it empty. Run the first scrub after the peak, for up to 3 h, and read the report before adding a second route. Do not size that night from 1,980 m²/h maximum theoretical. Cut it for kiosks and turns.
If the walk shows a ride-on hall, stop and use the T16AMR path: 910 mm, pass 1.45 m, 5,890 m²/h theoretical manual, no water dock published. Tennant’s 2025 case study says Fraport Facility Services, about 800 people, introduced robots in 2024 and preferred the T16AMR after trials because it also drives manually. The study does not say how many robots. Brain Corp BrainOS is that operating system. It does not publish a water dock.
Kärcher KIRA B 50, 2,365 m²/h theoretical autonomous and 69 dB(A), deploys on a mid-size concourse only if the optional full dock is on the order. Gausium Phantas, pass 650 mm and quiet mode ≤60 dB, is a gallery install, not the stone hall. LionsBot R3 Scrub Pro’s practical rate is 800–1,200 m²/h. Nilfisk Liberty SC50 belongs on the shortlist when the lease names CSA/ANSI C22.2 No. 336. Pudu MT1 Vac, max 1,400 m²/h, is a separate carpet deployment. Pudu BG1 at about 344 kg does not go in the queue.
No terminal price and no payback are printed here. The dock type is the cost line that changes the night. A vendor’s fleet sentence is not a substitute for that line.
The C5 station self-cleans the sewage tank. The weekly check stays on the day roster. The T16AMR crew still handles water, because no water dock is published. Consumable hours are per manual.
A robot in a queue is a blocked route. Where the terminal uses the accessible-route guide, keep it out (Access Board, chapter 4). Where it uses NFPA 101, do not park it in an exit. Wet stone is still a slip risk (HSE slips). In the EU the employer duty is the OSH Framework Directive. IEC 63327 and IEC 60335-2-72 cover the machine class. IEC 60529 defines IPX3. The C5 manual names no robotic-certificate number. The T16AMR spec row does. 47 CFR Part 15 on the C5 is the radio claim, not an airport approval.
Hold the walk and the docking station to these figures. The 500,000 m² terminal is the building, not a rate the C5 can finish.
| Model | Path | Rate to plan on | Tanks or bag | Runtime | Noise |
|---|---|---|---|---|---|
| CleaniBot C5 | 550 mm path, pass 800 mm, turn 1,720 mm | 1,980 m²/h maximum theoretical | 45 L / 45 L tanks | 3 h scrub | ≤70 dB(A) |
| Tennant T16AMR | 910 mm path, pass 1.45 m | 5,890 m²/h theoretical manual | 190 L / 225 L tanks | 4 h scrub, or 5.5 h lithium | 71 dB(A) |
| Avidbots Neo 2W | 790 mm path, turn about 3.05 m | 3,941 m²/h theoretical | 109 L / 135 L tanks | 4–6 h | 75 dB(A) |
| Kärcher KIRA B 50 | 550 mm path, pass 900–1,000 mm | 2,365 m²/h theoretical autonomous | 55 L / 55 L tanks | ~3.5 h scrub | 69 dB(A) |
| Gausium Phantas | pass 650 mm path | 350–700 m²/h practical | 11.5 L / 10.5 L tanks | 4.5 h scrub | quiet mode ≤60 dB |
| Nilfisk Liberty SC50 | 510 mm path | 1,895 m²/h theoretical autonomous | 57 L / 53 L tanks | up to 6 h scrub | 63 dB(A) |
| Pudu MT1 Vac | 550 mm vacuum path, pass 75 cm | 1,400 m²/h max | 14 L dust bags | 3–6.5 h | minimum 65 dB(A) |
Walk the zone, match the machine, draw the no-go walls, prove the dock, run one off-peak task, and leave restrooms to staff. The zone choice is the other page.
Frankfurt Terminal 1, in Tennant’s case study, is over 500,000 m². The C5 is a 45 L machine with an 800 mm pass. The case study says the T16AMR was preferred.
Gausium’s Milan note and SoftBank’s 100-robot note are company statements. Pudu deploys as a vacuum, or as the BG1 only on a wide closed hall. LionsBot R3 Scrub Pro provides the compact practical band. Kärcher deploys if the dock was ordered. Nilfisk is the named-certificate option. Brain Corp is the system on the T16AMR.
No. Noise, weight and queues are why week 1 is off-peak. A 75 dB(A) Neo or a 71 dB(A) T16AMR is a worse peak choice than waiting.
The practical rate when it exists. Otherwise the labelled ceiling, after the walk. Not a 500,000 m² terminal on a 45 L tank.