CenoBots S5: A Market Overview of the Industrial Sweeper

First the framing that decides how you should read everything below. CenoBots lists five models: S5, L3, L4, L50 and SP50. Three of them have been in service with us for a while — the L3 and the L4 as compact scrubber-dryers, the L50 as the industrial machine with 55 litres of fresh water. All three are wet machines. The S5 is the dry industrial machine of the line — the fifth model, the SP50, is a small indoor vacuum — and the S5 is the one that has only just joined our programme. What proximity to the brand gives us is the ability to read and place its data sheets. What it does not yet give us is a single measurement of our own on the S5. Where the text below says "per the manufacturer", that is not a formula — it is the boundary of what we know.

The S5 is an autonomous dry sweeping robot for industrial floors. Side brushes move material inwards, a suction motor pulls the dust through HEPA filtration into a 50-litre debris hopper, and an automatic filter shaker keeps the filter clear. Water and chemistry appear nowhere in this machine. The floor behind it is dry and immediately walkable and drivable — no film, no drying time, nothing that has to be kept clear of a forklift route. That is the decisive difference from the three L machines we run: those clean with water and leave a thin, quickly drying film. The S5 sweeps.

The interesting part is what CenoBots says about the debris range, because that is where the machine gets specific: wood chips, packaging film, plastic bottles, cardboard and screws are named explicitly in the brochure, and the product page adds fine dust, small metal rubbish and plastic bags. Bulky light debris is the category vacuum-type machines fail on — a sheet of film lies down in front of the intake instead of passing through it — and among the six sweeping robots whose documentation we read for our sweeper comparison, CenoBots is by some distance the most explicit about it. It is not alone in mentioning it: Pudu's official MT1 brochure names debris „like leaves and bottles" in passing. But naming a category in passing and specifying it item by item are different things. That is not proof the others cannot handle it. It is an indication of what this machine was designed for.

The official product page gives six values plus the computing platform: around 2,500 m²/h of cleaning productivity (27,000 ft²/h), 820 mm sweeping width, a 50-litre debris hopper, 2 hours of fast charging, 1,030 × 720 × 990 mm at 130 kg — the 720 mm being the width without side brushes — plus an NVIDIA chip with 100 TOPS and a 32-beam 3D LiDAR with up to 150 m of detection range. That is all of it. There is no specification table on cenobots.com, and the European mirror cenobots.eu shows exactly the same values without adding a single one.

Everything else — speed, battery, runtime, climb, passable width — exists only in the manufacturer brochure. And that brochure cannot be downloaded from cenobots.com: we pulled it from the server of an Australian dealer, and the German dealer Zwetko mirrors the same table one to one. From it: 1.4 m/s top speed, a maximum climb angle of 8°, an LFP battery with 75 Ah, 5–7 hours of runtime, 810 mm minimum passable width and 0 cm minimum distance from the wall. Those figures are solid, but they hang on a single document. Anyone procuring an S5 should have the official data sheet issued directly by CenoBots Europe B.V. rather than relying on a PDF from a dealer server.

Two of those numbers deserve a second look. The machine fits through 810 mm but sweeps 820 mm wide — not a contradiction but the consequence of the flexible side brushes, which project beyond the 720 mm body and give way on contact. In practice: standard doorways are not a knock-out criterion, but measure your narrowest real point including door frame, racking foot and parked pallet, not just the clear door width. And the 0 cm distance from the wall is a design statement, not a measurement: the side brushes pull debris out of the edge into the suction path. How clean that edge ends up is decided by your skirting, not by the data sheet.

That leaves the figure everyone reads first. 820 mm of sweeping width times 1.4 m/s works out to roughly 4,130 m²/h — the area the machine would cover driving straight ahead, with no turns, no overlap and no obstacles. Published is around 2,500, so a good 60 % of that ceiling. For comparison: our MT1 Max is rated at 2,200 m²/h with 700 mm and 1.2 m/s against a geometric maximum of about 3,020 — roughly 73 %. CenoBots therefore publishes the more conservative number, which speaks for the manufacturer. It is still not a planning figure: the S5 page puts a tilde in front of it, and the same company labels the area performance of our L4 on its own product page as "Max. Theoretical Productivity". The German dealer table that copies the brochure values consequently calls it "maximale theoretische Produktivität". How such figures come about, and what to measure instead, is set out in our sweeping-robot comparison.

There is no manufacturer figure for noise. Not on the product page, not on the EU mirror, not in the brochure — for a machine with a suction motor and a mechanical filter shaker, that is the most conspicuous gap in the data. One German dealer publishes a sentence about the noise level of the CenoBots fleet that does not break the S5 out separately; we do not reproduce it, because a dealer figure for a machine we have no measurement of our own on is exactly the number that ends up in a tender and cannot be defended afterwards. What can be said: treat the S5 as a loud machine until somebody gives you a measured value in writing. For warehouses and production that has no consequences — those are closed industrial spaces where night operation is fine. For supermarkets it means outside opening hours.

The list continues. On suction, the brochure says "a powerful suction motor" and nothing else — no pascal value. The battery is specified as LFP with 75 Ah but without a voltage, so the energy in kilowatt-hours cannot be derived, and without that nobody can plan your charging infrastructure seriously. An IP protection rating appears in none of the documents we read — and for car parks that is the decisive point: the S5 crosses speed bumps and climbs 8°, but whether it tolerates damp air and spray is something the manufacturer has nowhere explained. Our MT1 Max is the machine for which Pudu does state exactly that, with IP54 and an operating window of −10 to 45 °C. For the S5, covered decks only until further notice.

And then the procurement question: for the S5 we found no EC declaration of conformity, no list of test standards and no certification mark — neither at the manufacturer nor at any dealer. That does not mean the machine lacks one; it means you request the declaration instead of assuming it. The addressee is CenoBots Europe B.V. in Amsterdam, the manufacturer's European entity. Making it harder, there is no German-language manufacturer material for the S5 at all: cenobots.com and cenobots.eu are English, and cenobots.de is a domain offered for sale that does not belong to the manufacturer. Every German sentence about this machine therefore comes from a dealer — and at least one of them is wrong: a German dealer page lists "Frischwassertank: 50l" and "Abwassertank: 50l" for the S5. This machine has no water tanks. The 50 litres are the debris hopper, and those lines were visibly copied from a scrubber-dryer page.

Three claims, finally, do appear in manufacturer documents but are marketing, and are marked as such here. Spot cleaning is said to be "up to 4× faster than traditional full-coverage cleaning" — with no statement of what it was measured against or how. The CenoVLA AI model with 5 billion parameters is said to be "100× larger than standard cleaning robots" — a figure that cannot be checked without the comparison list. And the 24/7 operation: 2 hours of charging against 5–7 hours of running means a single machine is unavailable for roughly a quarter of every cycle. Round-the-clock sweeping is a property of a fleet, not of a machine. Also open: the additional sensors — depth camera, bumper, IMU, cliff detection — named only by the German dealer table, and the question of which fleet software the S5 runs on; the manufacturer documents we read say nothing about it.

The only real overlap with our fleet is the MT1 Max, likewise a waterless sweeper. On paper these are two tiers. The S5 weighs 130 kg, holds 50 litres, filters through HEPA with an automatic shaker, sees 150 m and claims vehicle recognition with avoidance path planning for forklifts and AGVs. The MT1 Max runs 2,200 m²/h at 700 mm working width with a 35-litre hopper, lasts up to 10 hours, and is the machine with a stated protection rating. In short: the S5 aims at heavier industrial debris and more area per hour, the MT1 Max at longer endurance without a charging break and at semi-outdoor use. And anyone laying the two number columns side by side is also comparing two publishing conventions — see above.

Inside the CenoBots family the S5 has a documented role, and it is the most interesting statement in the brochure as far as we are concerned: TeamClean — the S5 sweeps, the brand's own autonomous scrubber-dryers scrub afterwards. That is exactly the sequence that works on industrial floors: coarse debris first, then the film. For us that pairing has an odd symmetry. The wet half is our L50 — 2,200 m²/h, 510 mm brush, 55 litres of fresh and 55 litres of waste water, up to 6 hours, 220 kg, 67 dB(A) — a machine we run daily. The dry half we know only from documents. What TeamClean requires technically — shared map, shared account, shared software — CenoBots does not publish. That would be our first question.

The computing and sensing platform of the S5, by contrast, is not unfamiliar to us. The 32-beam 3D LiDAR with 150 m of range is the same configuration CenoBots states on the product page of our L4, and the NVIDIA chip with 100 TOPS is what the manufacturer also names for our L3. What is new about the S5 is not the hardware class but the job it does. That is the kind of thing you only notice if you operate the brand's other machines.

What we take from running the three L machines is above all an expectation about documentation. CenoBots publishes little, publishes it in English, and the complete tables sit with dealers rather than with the manufacturer. With the S5 that is not different, only more pronounced. Anyone buying into this brand should request data sheets actively, check every German figure against the English original, and collect the answers in writing. That is not an argument against the machines — ours run. It is guidance on how to procure them.

The S5 belongs on the shortlist when your problem is coarse dry debris on hard floors across large areas. Warehouses and logistics sites are the obvious case: pallet abrasion, packaging film, cardboard, mixed forklift traffic — that is what the vehicle recognition is for, and that is the debris range the brochure names. Production halls are the second: swarf, screws, fine dust, with HEPA and the filter shaker as a dust architecture rather than a brochure line. Parking decks are named by the manufacturer as a target environment, and 8° of climb plus speed-bump crossing fit that — as long as they are covered, see the protection rating. Retail floors are on the manufacturer list too, but there only outside opening hours as long as no decibel value exists.

It does not belong on the shortlist wherever wet soiling is the problem. Ground-in films, sticky floors, tyre marks — no sweeper helps against those, only a scrubber-dryer; our overview of which cleaning robot fits which operation sorts that question by floor type and soiling. Carpet is out, and so are guest-facing areas and quiet zones — not because we measured the machine as loud, but because nobody publishes a value and the manufacturer positions it for halls. Incidentally, the S5 is already available in Germany: at least one specialist dealer carries the complete CenoBots line including the S5, for purchase and for rental, and through us it comes in the same full-service model as the rest of our programme.

For the demo, a list pays off, and it starts with paperwork. Have the official data sheet issued by the manufacturer, not the dealer PDF. Demand the EC declaration of conformity. Get the noise level in dB(A) in writing, with the measurement condition attached. Ask for the battery voltage, so that 75 Ah turns into a kilowatt-hour figure. Ask for the IP protection rating before the machine drives into a car park. Ask whether depth camera, bumper, IMU and cliff detection are standard equipment — so far only a dealer table names them. And clarify what TeamClean requires on your side if you already operate a scrubber-dryer of the same brand.

Then measure the things that appear on no data sheet. First: how often is the 50-litre hopper full with your debris? No area-per-filling figure exists for this machine, and that number sets the staffing requirement, because the hopper is pulled and emptied by hand. Second: does your debris actually go in? Bring the bulkiest material you produce — film, strapping band, cardboard offcuts — instead of watching a swept demo floor. Third: does the machine fit through your narrowest real point, not your planned one? Fourth: how does it behave in live forklift traffic rather than in a cleared aisle? And fifth: how much area does it cover per charge in your layout? 2,500 times 6 hours is arithmetic, not a specification. If you would like this list applied to your own floor, talk to us — we run enough of this brand to know which questions stay open.