Robot vacuum and mop buying guide 2026: suction, mops, docks and navigation explained

This guide explains which robot vacuum specs matter and which ones makers inflate or leave out: suction in pascals, mop designs, self-cleaning docks, navigation, height, threshold climbing, noise and smart-home support. It draws on 98 robot vacuum and mop records in the RankedRobot database, from a sub-$200 LiDAR vacuum to $1,800 flagships, and gives real ranges, medians and price tiers instead of marketing claims.

Last checked 1–9 October 2026. Spec-based analysis from maker-published figures; we have not tested these robots hands-on.

Start with your floors, not the Pa number

Most 2026 product pages lead with suction. It is the easiest number to print and the hardest to compare. The things that decide whether a robot suits your home are less obvious. Can it get under your sofa? Can it get over the step into the bathroom? Will it drag a wet mop across your rug, or chew a phone cable? How much will you handle dirty water?

Work through the sections below in order. Then use the robot finder to filter our robot vacuum category by the limits you have measured. Use the comparison tool to put two or three shortlisted models side by side.

Suction (Pa): why the headline number doesn't compare

Suction is given in pascals (Pa), a unit of pressure. It is usually a peak figure for the fan in its strongest mode. Brands do not share a test method for it. The international performance standard for household cleaning robots, IEC/ASTM 62885-7, measures dust and debris pickup from hard floors and carpets, fibre removal and energy use. Suction pressure is not one of its headline measures. So a 30,000 Pa robot from one brand and a 20,000 Pa robot from another can't be ranked against each other on that number alone.

The footnotes show how loose these figures are. Dreame's page for the X60 Max Ultra Complete says its 35,000 Pa figure is "based on our in-house lab data". Roborock's Saros 20 Flow page labels 36,000 Pa as "Max Digital Motor Power", measured internally "under specific conditions". Ecovacs's T90S Pro Omni page claims up to 40,000 Pa and gives no test conditions. iRobot's US page for the Roomba Max 705 Combo gives no pascal figure at all. It says "175x more suction" than a Roomba 600-series robot in spot-clean mode.

What our records show:

• 93 of 98 robots carry a suction figure. 90 of those come from the maker. Three iRobot figures were reported by an Australian iRobot retail site, not by iRobot itself.
• The median stated figure is 19,500 Pa. The middle half sits between 10,000 and 28,500 Pa.
• The range runs from 3,000 Pa (SwitchBot K10+ Pro Combo) to 40,000 Pa (Ecovacs Deebot T90S Pro Omni).
• 23 robots claim 30,000 Pa or more. A year or two ago, figures like that were rare.

How to use it: treat suction as a rough guide within one brand's range. A brand's 36,000 Pa model will probably pull harder than the same brand's 10,000 Pa model. Across brands, carpet handling (automatic boost, mop lifting, pile-height limits) and brush design tell you more. Our database never ranks robots by Pa, and our methodology explains why.

Mops: rotating pads, rollers, vibrating plates and extending arms

Mopping is now standard. Only one robot in our set, the Dyson 360 Vis Nav, is a vacuum with no mop. The mop design shapes how the robot handles edges, carpets and its own dirty water.

Rollers are the trend to watch. Ecovacs accounts for 8 of the 23 roller robots, eufy and Narwal for 4 each, and iRobot for 3. Dreame's Aqua10 Roller and Roborock's Saros 20 Flow show the big two pad brands moving that way too.

Mop lift height matters if you have rugs. 58 robots publish it. The median is 10.5 mm. The highest are 28 mm (eufy Omni S2) and 22 mm (Roborock Saros 10R). If your rugs are thicker than the lift height, choose a robot that detaches or leaves its pads in the dock, or set no-mop zones in the app.

Docks: what each station feature actually does

The dock now does most of the work. 96 of our 98 records list dock features. Counts below use those 96 as the base. A feature missing from a maker's page is recorded as "not stated", not as absent.

29 robots list the full set: auto-empty, mop washing, hot-water washing, hot-air drying and automatic refill.

Tank-based docks still need manual top-ups. Clean-water tanks run from 2.2 L to 5.5 L (median 4.0 L, 44 published). Dirty-water tanks run from 1.7 L to 4.0 L (median 3.5 L, 38 published). A bigger clean tank means fewer refills. A dirty tank smaller than the clean tank is common, and it usually sets how often you empty it.

The robot's own bin matters less once you have auto-empty, but bins still range from 99 mL (Ecovacs Deebot mini 2) to 800 mL (Lefant M320, which has no emptying dock). The median is 250 mL across 44 robots.

Navigation and obstacle avoidance

Navigation is how the robot maps the home and knows where it is. 87 of 98 records name a laser or time-of-flight system: a spinning LiDAR or "LDS" turret, dToF, Mini-ToF or a 3D ToF module. Four rely on cameras alone, for example the Dyson 360 Vis Nav, iRobot Roomba Combo j9+ and Narwal Freo 20. Camera-only systems can struggle in the dark unless they carry their own lights, so check for an LED. The budget Lefant M320 uses a gyroscope, which means no proper room map.

Obstacle avoidance is a separate job. It covers spotting socks, cables and pet mess in front of the robot. Three approaches show up:

• Structured light: a projected laser line or pattern that measures the shape of objects. It works in the dark but does not know what an object is.
• Camera recognition (RGB): recognises object types and can treat them differently, for example giving pet waste a wide berth. 54 records name a camera in obstacle avoidance. 31 pair it with structured light or 3D ToF.
• LiDAR-only or infrared sensors: common on cheaper models. Fine for walls and furniture, weak with small soft objects.

Object counts are marketing, not a benchmark. 15 records state one. They range from 62 types (Roborock Qrevo Edge) to "280+" (Dreame L60 Ultra, Dreame X60 Max Ultra Complete). Minimum object size is more useful where given. Roborock states 5 cm wide and 3 cm tall for the S8 MaxV Ultra. Dreame states 10 mm for the X60 Max Ultra Complete.

Height, thresholds and carpets

Height decides what furniture the robot can clean under. 92 robots publish it. The median is 97.3 mm, and the range runs from 78.7 mm (Roborock Qrevo Edge 2) to 120 mm (Dreame Aqua10 Roller). 13 robots are 80 mm or lower: 11 Roborocks and 2 Dreames. All of them retract the LiDAR turret or use flat 3D sensors to get there.

Read the footnotes. Dreame's X60 Max Ultra Complete page gives 79.5 mm with the sensor retracted and 103 mm with it raised. It also says the robot needs 81 mm of clearance to enter with the retracting feature on. Roborock's Saros 20 page lists a lower clearance figure that relies on "passive compression" of the front bumper and warns that scraping may occur. Measure your lowest furniture and allow a few millimetres of margin.

Threshold climbing decides whether the robot can cross door strips and get onto rugs. 65 robots publish it. The median is 24.9 mm. 27 claim 40 mm or more. The top figures need care. Both the Dreame X60 Max Ultra Complete and the Roborock Saros 20 Flow quote 88 mm, but that is a two-step obstacle under set test conditions. For a single step, Dreame states 45 mm, and Roborock's Saros 20 page gives 1.77 in (about 45 mm). If you have one high threshold rather than a stepped one, compare single-step figures.

Carpet: look for automatic suction boost, mop lifting or detaching, and a stated pile limit. Roborock's Saros 20 page gives 1.18 in (30 mm) of carpet pile, with the feature switched on in the app.

Noise, runtime and battery

Noise and runtime are the specs makers leave out most often. Only 38 of 98 robots publish a noise figure, and only 39 publish a runtime.

Stated noise runs from 45 dBA (SwitchBot K11+) to 74 dBA (Samsung Bespoke AI Jet Bot Combo), with a median of 60.1 dBA. Those figures rarely refer to the same mode. Roborock's Saros 10 page gives 52 dB(A) for its quiet mopping mode and 72 dB(A) for the dock while it self-empties. The dock is usually the loudest part, and few brands publish its figure.

Stated runtimes run from 65 minutes (Dyson 360 Vis Nav) to 350 minutes (Ecovacs Deebot T90S Pro Omni), with a median of 180. These are almost always lowest-power figures. For mopping robots runtime matters less than it used to, because the robot returns to wash its mop and recharges between rooms.

Consumables, warranty and running costs

A robot with a full dock uses dust bags, filters, side brushes, main brushes, mop pads or rollers and, if it doses detergent, cleaning solution. We do not record consumable prices, and most product pages don't put them next to the robot. Before you buy, look up the bag and pad prices on the maker's store, and check that third-party parts exist for your model.

Warranty is published for 53 of 98 robots: 29 at 12 months, 6 at 24 months and 16 at 36 months. 14 of those 16 are Dreame. Samsung's regional page for the Bespoke AI Jet Bot Combo states 60 months. Long warranties sometimes cover only the motor, so read the terms.

App, privacy and Matter

Privacy. A robot with a front camera drives through your home. Before buying, check whether images leave the device, whether you can turn the camera off or delete maps, and how long the brand promises security updates. ETSI EN 303 645 is the common baseline for consumer IoT security. It covers no universal default passwords, vulnerability reporting, software updates and easy deletion of user data. Two of our records list third-party security or privacy certifications. The Roborock Qrevo Curv 2 Flow lists a TÜV certificate against EN 303 645. The Dreame X60 Ultra lists a UL Solutions IoT security mark and a TÜV SÜD privacy certificate. In the UK, the product security regime has applied since 29 April 2024. Sellers of connected products must publish the minimum period for security updates, so look for that support period before you buy.

Matter. The Connectivity Standards Alliance added robot vacuums to Matter 1.2 in October 2023. That covered remote start, progress, vacuum versus mop modes, and status such as charging and errors. Matter 1.4.2 (August 2025) tidied up how robots handle commands sent one after another. Matter gives you basic control from Apple Home, Google Home, Alexa or SmartThings. Map editing and dock settings stay in the maker's app.

24 of our 98 robots list Matter, 13 of them Roborocks. The SwitchBot K10+ Pro Combo appears in the Alliance's certified product listing against Matter 1.4. Platform requirements apply. iRobot notes that the Roomba Max 705 Combo needs iOS or iPadOS 18.4 or later for Apple Home pairing, plus a home hub for automations.

What you get at each price

The table groups the 79 robots that show a US-dollar price on the maker's US store. Each robot is placed by the lowest price shown when we recorded it, which is the sale price where one was displayed. Prices move often.

The jump that matters is around $700. Below that, mop-washing docks are common but camera-based obstacle avoidance is rare: 4 of 29. From $700 up, most robots have both. Above $1,000 you mainly pay for a lower body, higher climbing, hotter washing and rollers.

Example robots compared

Figures are maker-stated unless noted. "—" means not published on the pages we used. Threshold is the single-step figure where the maker separates it.

Two things stand out. Suction tracks price only loosely: the $299 T30C claims twice the Pa of the $299.99 Q7 M5, and the Roomba has no maker figure at all. And the slimmest bodies (79.5–81 mm) appear at both $399.99 and $1,699.99.

How to choose by home

• Low furniture: filter for 80 mm or less, then check the "entry height" footnote, not just body height. Retracting-turret Roborocks and Dreames dominate here. The Ecovacs T50 Pro Omni at 81 mm is the cheapest near-slim option in our data.
• Pets, cables, children's toys: pay for camera recognition paired with structured light or 3D ToF. Most robots priced from $700 up have it.
• Mostly hard floors and kitchen spills: a roller mop with hot-water washing, or high-temperature pad washing. Check robot height, because rollers tend to be taller.
• Thick rugs: prioritise mop lift (20 mm or more) or mop detaching, a stated pile limit and automatic carpet boost.
• High door strips or a step: compare single-step climbing figures. The 60–88 mm claims are mostly two-step tests.
• Small flat or renting: a compact vacuum-first robot such as the 248 mm-wide SwitchBot K11+ avoids a big dock. Plumbed refill/drain needs a water connection you may not be allowed to add.

For ranked picks see our best robot vacuums list. Our separate robot vacuum buying page covers retailers.

What the numbers say (Analysis)

• Pa is close to universal, but it isn't a benchmark. 90 of 98 robots carry a maker-stated suction figure, with a median of 19,500 Pa and a range of 3,000–40,000 Pa. None of the pages we checked links that figure to a published test method.
• Suction climbs with price, but loosely. Median stated suction rises from 10,000 Pa (10 robots under $400) to 30,000 Pa (9 robots at $1,400 and up). Within every tier, figures overlap heavily.
• Docks have converged. Of 96 robots with dock details, 85 auto-empty, 76 wash the mop and 76 hot-air dry it. Only 29 list the full set including automatic refill, and 17 of the 57 that publish a wash temperature claim 100 °C.
• Low bodies belong to two brands. 13 of 92 robots that publish a height are 80 mm or lower. 11 are Roborock and 2 are Dreame.
• Noise and runtime are the biggest gaps. Only 38 of 98 publish noise and 39 publish runtime. Ecovacs (14 of 14) and Roborock (12 of 32) supply most of the noise figures.
• Sale prices are deep and constant. 27 robots show both a list and a sale price on the maker's US store. The median markdown is 36%, and the range is 10% to 62%. Treat list prices as a ceiling, not a guide.

Frequently asked questions

Is 30,000 Pa better than 20,000 Pa?

Only within the same brand, and even then only roughly. There is no shared test method for Pa across makers. The standard performance test for these robots measures pickup, not pressure.

Do I need a hot-water mop wash?

It helps with greasy kitchen floors and smell. 58 of 96 docks list it. Stated maximums range from 55 °C to 100 °C. Hot-air drying (76 of 96) matters at least as much against odour.

Roller mop or rotating pads?

Rollers rinse themselves continuously and are spreading fast (23 robots, plus 3 hybrids). Pads are still the majority (45). They often extend into corners and suit lower bodies. Choose on robot height and edge cleaning, not mop type alone.

How low does a robot vacuum go in 2026?

The lowest in our data is 78.7 mm. Several robots need slightly more clearance than their body height when their sensor retracts, so read the footnote.

Can a robot vacuum climb a 4 cm threshold?

27 of 65 robots that publish a figure claim 40 mm or more. Check whether that figure is for a single step or a two-step obstacle.

Does Matter let me control everything from Apple Home or Google Home?

No. Matter covers start, stop, cleaning mode and status. Maps, no-go zones and dock settings stay in the maker's app. 24 of 98 robots list Matter.

Are camera robots a privacy risk?

They can be, if images or maps leave your home without controls. Look for a camera off-switch, local processing claims, a published security-update period and independent certification against standards such as ETSI EN 303 645.

Methodology

All robot figures come from the RankedRobot database: 98 robot vacuum and mop records that passed our quality checks, each linked above to its robot page, which lists the source for every field. Sources are maker product pages, spec sheets, user manuals and press releases. Where a figure came from elsewhere, it is marked "reported" (for example, three iRobot suction figures). We checked selected footnotes and prices directly on maker pages between 1 and 9 October 2026, plus the standards and regulator pages linked in the text.

Counts and medians are computed only over robots that publish the relevant spec, and denominators are stated. Price tiers use the lowest US-dollar price shown on each maker's US store when the record was taken. Prices change often and may differ from what you see today. Figures are as the maker states them, in the maker's units and currency. kPa is converted to Pa. We have not tested these robots hands-on. This is spec-based analysis. See our methodology page for field definitions.

Spotted an error or a newer figure? Email corrections@rankedrobot.com and we'll check it and log the correction.