Ask what matters about a floor cleaning robot and most buyers start with brush width and tank size. Those numbers tell you how much floor the machine touches per pass. They do not tell you whether it will still find its way at 2 a.m. in a building where someone moved a pallet, or whether it will slow down for a person in a busy aisle instead of expecting the aisle to be empty.
That is what cleaning robot navigation technology decides: whether autonomy works on your site, or only on a clean demo floor. This page covers what the machine senses, how those readings become a route, what happens when your building changes, and what to check in any supplier’s answer before you buy.
What an autonomous floor robot actually senses
A cleaning robot carries several kinds of sensor, and each one covers a blind spot the others have. This is the part most spec sheets compress into a single word — “LiDAR” — which hides more than it explains.
- Laser scanning (LiDAR). Measures distance to walls, racking and fixed structures. It is precise over long ranges, which is why it carries the mapping. It is also the sensor that struggles most in open, feature-poor halls and against glass.
- Depth and RGB cameras. Read the room in three dimensions. Where LiDAR sees a surface, a camera can often tell what the surface is — a chair leg to go around rather than a wall to follow.
- Ultrasonic sensors. Catch low obstacles and transparent or reflective surfaces that optical sensors misread.
- Drop-off detection. Watches the floor ahead for stairs, dock edges and mezzanine openings — the one failure mode there is no recovering from.
- Contact bumpers. The last line. If something is too soft or too low to be seen, the machine stops on touch rather than driving through it.
Two of our machines can be described concretely rather than in adjectives. The SC80 runs multi-line and single-line LiDAR together with six ultrasonic sensors and three cameras. The Scrubber 50 pairs 2D LiDAR with a 3D depth camera, an RGB camera and collision sensors. A machine carrying only one of these is not a cheaper version of the same thing — it is a machine with a known blind spot, and the blind spot is usually the one your building has.
How cleaning robot navigation technology turns readings into a route
Four steps repeat continuously while the machine works. Understanding them makes it far easier to judge a supplier’s claims.
- Map. On its first pass the robot builds a map of the floor, using the laser scan as the structural reference. Charging stations, work areas, lift lobbies and no-go zones are defined once.
- Locate. From then on the robot compares what it sees now against that map to work out exactly where it is. This is where a weak map hurts: too few features to match against and the position drifts — and the cleaning route drifts with it.
- Plan. A global route is calculated for the task, then revised continuously for whatever is in the way. That is why a pallet left in an aisle slows the machine down instead of stopping the shift.
- Clean. Deck pressure, brush contact and water are controlled along that path, and the machine works the edges with a side brush so the last strip along a wall is not left for a human to finish.
Why maps go out of date — and why that should not be your problem
Real buildings are never static. Racking is re-slotted, seasonal displays move, exhibition stands are rebuilt overnight. A robot that needs a technician to re-map the floor after every layout change is not autonomous in any useful sense; it is a machine with a maintenance dependency nobody costed at purchase.
Our waterless models are built around this problem. On the SW55A, routes are re-planned when the layout changes rather than hard-coded to a fixed line — a moved display changes the path instead of breaking the task. The SW80A maps up to 40,000 m² in a single survey, which is what lets it cover a large site without a re-mapping exercise every time something is moved.
There is a physical counterpart to this. In halls that are large and feature-poor, or glazed and reflective, both laser and visual matching get thin. Fixed QR floor markers give the robot known references to lock onto — a low-cost way to make navigation dependable in exactly the spaces where it is hardest.
People in the same aisle: detection is not the same as avoidance
Most sites are not empty at night, and almost none want to close an area to run a machine. The difference between suppliers shows up in one question: does the robot slow down for people, or does it require the floor to itself?
A machine designed to work around people does three things. It detects a person at a distance and reduces speed. It chooses to go around rather than stop and wait. And it announces itself with light and sound so the person has a chance to respond. Machines designed for “clean areas only” tend to stop and wait for the aisle to clear — which is fine in a warehouse at 2 a.m. and useless in a supermarket at 11 a.m.
Two practical layers sit around the technology itself:
- Drop-off detection must cover stairs, dock edges and open mezzanines. On a machine that weighs several hundred kilograms this is not an optional extra.
- Route and signage design does the rest. Marking where robots travel and where the station sits keeps the cleaning route predictable for your staff as well as the machine. Our robot in operation floor signs exist for that reason; the wider question of marking a site is covered in this guide to floor markings.
Where a formal standard applies, ask for it in writing. The Scrubber 50 can be supplied in a configuration built to IEC 63327, the standard covering automatically controlled floor treatment machines — worth having on record for hospitals, airports and any site that carries an audit trail.
Autonomy that survives a whole shift: docking, charging and reporting
Navigation only produces value if it runs unattended for a complete cycle. That depends on three unglamorous things: the machine returning to a station on its own, taking on charge and — on washing models — water, and reporting what it actually did.
On the SC80 an unattended overnight cycle is the design intent rather than an add-on: the machine returns to its station, recovers and resumes. The Scrubber 50 combines a single wash-and-dry pass with a workstation that handles charging and water. This is also where a small hardware detail turns into a reliability issue — the docking contacts on both machine and station are what every autonomous cycle depends on, and they are on the routine list in our cleaning robot maintenance schedule.
How the technology shows up across our machines
The same navigation stack is tuned differently for different floors. The table below is the short version; each machine page carries the full specification.
| Machine | Cleaning width | Rated efficiency | What the autonomy is built for |
|---|---|---|---|
| SC50 | 500 mm washing 700 mm drying | 1,800 m²/h | Narrow layouts. 900 mm minimum passage, so it works aisles and counters a large machine cannot enter — and can run during opening hours. |
| SC80 | 800 mm washing 1,160 mm sweeping | 4,000 m²/h | Unattended overnight cycles. Multi-line and single-line LiDAR, 6 ultrasonic sensors, 3 cameras; returns to its own station and resumes. |
| S100-N | 1,000 mm sweeping | 3,000 m²/h | Very wide indoor spans and outdoor yards, with a 1,700 mm minimum passage in autonomous mode. |
| SW55A | 550 mm, waterless | 1,200 m²/h | Mixed, frequently rearranged floor plans. Routes are re-planned when the layout changes rather than hard-coded. |
| SW80A | 850 mm, waterless | 1,600 m²/h | Large open floors, no plumbing. Single-survey mapping up to 40,000 m², 12% climb, fleet scheduling across one site. |
| Scrubber 50 | 460 mm (780 mm with side brush) | 500–1,300 m²/h actual | Wash and dry in one pass. 2D LiDAR plus 3D depth and RGB cameras, collision sensors, IEC 63327 configuration available. |
Five questions to ask any supplier about navigation
These apply to every brand, including ours. If a supplier cannot answer them in plain language, the specification is being used to avoid the question.
- Is navigation built on one sensor type or a fusion of several? If it is one, which situations is it weak in — glass, open halls, low obstacles?
- When our layout changes, who re-maps the floor? Your team, a portal you control, or a service visit you pay for?
- Does the machine slow down and route around people, or stop and wait? Ask to see it demonstrated in a busy area, not an empty one.
- What covers edges and drop-offs, and from which directions? Front-only detection is not enough near a dock or a mezzanine.
- What happens when the machine cannot work out where it is? It should stop safely and report — never keep driving on a stale position.
A buyer’s version of the same checklist, with the commercial criteria around it, is in how to choose a floor cleaning robot.
Where to go next
If you want the machine range, start from the machine overview. If the question is which machine fits which building, the solutions by industry pages take it site by site — warehouse, retail, hospital and hotel floors are already covered. And when the issue is keeping the sensing and drive hardware in condition, the spare parts and consumables range covers brushes, squeegees, filters, contacts and markers.
None of this is a substitute for testing a machine on your own floor. Tell us the floor area, the narrowest passage and when cleaning has to happen, and we can say what the navigation actually has to cope with on your site before anyone talks about price — speak to us here.