Kemaro K900 Gen II: A Market Overview of the Swiss Industrial Sweeper

One more time, plainly, because it decides how you should read everything below: every number here comes from KEMARO AG's own documentation, which we read ourselves rather than taking from a dealer page. Where the manufacturer makes a claim we cannot check, we mark it as a manufacturer claim and leave it there.

The K900 Gen II is an autonomous industrial dry sweeping robot from KEMARO AG in Eschlikon, Switzerland. Mechanically it is straightforward and, in the best sense, unglamorous: brushes move debris into a front opening while a blower puts the housing under negative pressure, drawing dust through a roughly 2 m² air filter into a 35-litre hopper. Kemaro states class M for that filter, which is the relevant class for industrial dust rather than household dust, and up to 2.6 m³ of air per minute passing through it. There is no water and no chemistry anywhere in this machine, and Kemaro's FAQ states clearly that it does not sweep liquids. The floor behind it is dry and immediately drivable — there is no film, no drying time, and nothing that needs to be kept off a forklift route.

Below the K900 sits the K700 Compact, and the difference is access rather than capability class. The K700 does 750 m²/h at 70 cm sweeping width with an 11-litre hopper, and it is 25 cm tall against the K900's 35 cm — built to get under low racking and between machines. Speed, runtime, charging time, battery and the 12 % maximum ramp are identical on both factsheets. Note the lifecycle difference, though: at the time of our research in August 2026 the K700 was in market launch and we found no German dealer listing it, while the K900 Gen II is the shipping flagship. Anyone considering the K700 should have Kemaro confirm delivery capability for Germany in writing.

From the official Gen II factsheet, which we extracted and read in full: 1,000 m²/h cleaning capacity, 90 cm sweeping width, 0.7 m/s cleaning speed, 35-litre hopper, up to 4 hours of runtime, 90 minutes charging time, lithium-ion at 36 V, 79 × 72 × 35 cm, 32 kg including batteries, maximum ramp slope 12 %. The product page adds IP54 protection, a docking station accepting 85–240 V, coverage of up to 20,000 m² per docking station and scaling to 100,000 m² with multiple stations, and the ECO, SMART and TOP tiers for areas below 2,500, 10,000 and 20,000 m² respectively, plus a K-Integrated variant for fleet-management integration. Perception is a 2D LiDAR "crown" covering 270 degrees plus a 3D camera for the lower blind spots; new in Gen II are the self-cleaning Automatic Filter Cleaner, the BlueSpot safety light that projects the robot's movement for forklift drivers, and Connected Rooms for autonomous movement between rooms.

Now the contradiction, and it is Kemaro's own. The factsheet, the product page and the manufacturer's November 2025 press release all state up to 4 hours of runtime. Kemaro's own Knowledge Hub states around 5 hours. The archived official site from 2023 shows why: the first-generation K900 was advertised with "a battery life of up to five hours", so the 5-hour figure is a Gen I value that survived in places Kemaro did not update. We publish 4 hours because the current factsheet is the most recent official document — but be aware that the same number is worth roughly 25 % of a night shift, so have it confirmed for the tier you are actually quoted.

Just as important is what Kemaro does not publish. There is no noise figure anywhere — not in the factsheet, not on the product pages, not in the FAQ, for neither the K900 nor the K700. A single US dealer quotes a decibel range; we will not reproduce it, because an unsourced dealer number is exactly the kind of figure that ends up in a tender and cannot be defended later. There is no pascal figure for suction either, and the Gen II battery is specified only as lithium-ion 36 V without a capacity. Treat all three as open questions for the vendor, not as gaps you fill from a comparison portal.

Two more things belong in the honest column. First, the hard limits: Kemaro's FAQ explicitly forbids operation in ATEX zones and forbids sweeping flammable or explosive substances — which matters because the machine is marketed for hazardous-materials storage, meaning storage areas people should stay out of, not explosion-protected zones. Ramps above 12 % and liquids are equally out. Second, the cost-savings marketing: the factsheet says up to 50 % of cleaning costs saved per year, the FAQ on the same website says up to 70 %. When a manufacturer's own two pages differ by twenty points, the figure is not a measurement, and neither the percentages nor the absolute euro savings quoted alongside them belong in a business case.

The only genuine overlap with our fleet is dry sweeping of large halls, which for us means the MT1 and the MT1 Max. On paper the gap looks dramatic: the MT1 is rated at up to 1,800 m²/h with up to 8 hours of runtime, the MT1 Max at up to 2,200 m²/h with up to 10 hours, both at 700 mm working width with a 35-litre hopper — against 1,000 m²/h and 4 hours for the K900. Anyone putting those numbers in one table without a footnote is misleading you, including in our favour.

Here is the footnote. The K900 runs at 0.7 m/s across a 90 cm sweeping width. Multiply it out and the theoretical maximum is around 2,270 m²/h, yet Kemaro publishes 1,000 — meaning Kemaro is quoting a realistic coverage figure with turning, overlap and obstacles already deducted. The Pudu figures sit much closer to the theoretical ceiling. So the honest statement is not "the MT1 Max is more than twice as fast". It is: on paper the Pudu machines lead on throughput and runtime, the real gap is smaller than the numbers suggest, and nobody can size it properly without measuring both on the same floor. If you want the reasoning behind area-performance figures in general, the sweeping-robot comparison goes through it, and the MT1 versus MT1 Max piece shows what the difference between two closely related machines really consists of.

The K900 differentiates itself structurally rather than through throughput, and those differences are real. Class-M filtration with a self-cleaning filter is a dust-handling architecture, not a marketing feature — for wood, metal and mineral dust that is the relevant question, not square metres per hour. IP54 permits damp, splash-prone environments such as bottling plants, where the machine still cleans dry. The docking coverage of up to 20,000 m² per station with scaling to 100,000 m² is a genuinely different deployment model for very large sites. The three-year or 5,000-operating-hour warranty on main components including battery and motors — Kemaro's stated terms, wear parts excluded and conditional on maintenance by Kemaro or an authorised partner — is above the industry norm. Mechanical rather than optical drop protection is the more robust choice on a loading ramp, and the BlueSpot exists precisely because this machine is expected to work among forklifts.

What our machines have that the K900 does not is less exciting and more decisive: a published noise figure of below 65 dB for both MT1 variants, longer runtime per charge, and the fact that we run them, know their failure modes, and hold spare parts. That last point is not a spec-sheet argument. It is the difference between a documented machine and a known one, and this article can only give you the first for the Kemaro.

KEMARO AG has been developing and building in Eschlikon, Thurgau since 2016, and the company puts that front and centre: designed and produced in Switzerland, Swiss quality components on the factsheet, the Swiss-made mark on the homepage. For German industrial buyers this is more than a flag on a data sheet, but it is worth being precise about which parts of it are checkable.

The checkable part is documentation and language. There are full German product pages, German press releases, and a manufacturer whose home market speaks German — support and paperwork in German are a realistic expectation, not a hope. There is also a data-protection angle that increasingly decides tenders: Kemaro states on its own site that it complies with European and Swiss privacy regulations with servers located in Switzerland, and describes a "minimized sensors" approach. That is the manufacturer's own statement and a good starting position for a works council or a data-protection officer — but it is a statement, not a contract. Ask for the data processing agreement, the list of subprocessors and the actual server location in writing before anyone relies on it.

On service, Kemaro maintains its own sales structure for Germany and Austria with named regional contacts, alongside a dense dealer network mostly recruited from intralogistics and forklift dealers, and it offers purchase, temporary rental and service packages including cloud reporting, predictive maintenance, spare parts and remote support directly. That is a mature distribution setup, not a pilot.

And here is the caveat that matters more than the country of origin: geographic proximity is not a response time. A Swiss manufacturer with a German dealer network can be excellent or slow, and the factory address tells you nothing about which. Before signing, get a committed response time in writing, ask where the spare parts for your machine physically sit, ask who does the annual maintenance that the warranty depends on, and ask what happens in week two if the machine is down. Those four answers determine your uptime. "Swiss made" does not.

The K900 belongs on the shortlist when your problem is dust on hard industrial floors over large areas: production halls with wood, metal or mineral dust, warehouses and logistics sites with pallet debris and packaging residue, bottling plants where damp is a fact of life, car parks, and hazardous-materials storage in the sense of areas you want people out of. Mixed forklift traffic is an argument for rather than against it, lights-off operation in a dark hall is explicitly supported, and if your site is far larger than a single machine can cover, the docking-station scaling is a structural argument.

It does not belong on the shortlist for offices, retail during opening hours, or anywhere near guests and quiet zones — not because we measured it as loud, but because nobody publishes a decibel value and this is a blower-driven industrial machine that the manufacturer positions for halls. It is also out for ATEX zones, which Kemaro forbids outright, for liquids of any kind, for carpet, and for ramps steeper than 12 %. If your problem is adhesions, tyre marks or sticky floors rather than dust, you need a scrubber and not this machine; which cleaning robot fits which operation sorts that question by floor type and soiling.

For the demo, ignore the brochure and take a list. Get the noise level in writing with a measurement condition attached. Get the Gen II battery capacity in watt-hours, not just the voltage. Ask for the EC declaration of conformity — we did not sight it anywhere public. Ask what technically distinguishes the ECO, SMART and TOP tiers beyond the licensed area, and specifically whether the sensor set is identical. If you plan to integrate with an AGV fleet, ask what the advertised VDA5050 compatibility requires on your side. And clarify the runtime discrepancy for the exact configuration being offered.

Then measure the two things that decide the business case and appear on no data sheet. First: how often does the 35-litre hopper need emptying with your dust, not with a demo floor's dust — that number sets the staffing requirement. Second: how much area does the machine actually cover per charge in your layout, with your racking and your obstacles. Multiplying 1,000 m²/h by 4 hours is arithmetic, not a specification, and the difference between that arithmetic and a measured run is where every disappointed robot project we have ever taken over began. If you would like the same list applied to your own floor, talk to us — including in the case where the honest answer is that a Kemaro fits you better than anything we operate.