Skip to content   Skip to footer navigation 

How we test robot vacuums

What happens behind the lab doors when we test robot vacs.

CHOICE tester Matthew Tung checking dust canister in a robot vacuum

If you think your stick vacuum is a convenient companion for a quick clean-up, you're going to be pretty impressed with the idea of a robot vacuum cleaner. Ideal for any time-poor person, just set up your robot vac and it'll get the job done for you – or so the ads say. 

Robot vacs are increasing in popularity but they still need to make improvements in some areas, so you may not want to part with your regular stick vac or barrel vacuum just yet.

Our expert testers have seen robot vacs evolve in functionality and capability over the years and with each test, more manufacturers jump on the bandwagon in the hope of producing the best robot vac yet. 

Here, we'll tell you how we put robot vacs through their paces to give you the results that help you choose what to buy.

Our expert testers

Our testers Matthew Tung and Adrian Lini are passionate about determining a product's technical performance. 

Matthew has worked at CHOICE since 2011, testing a wide range of products in the CHOICE labs from strollers to electric blankets, as well as stick vacuums. He also sits on Australian Standards committees. 

Adrian tests many small appliances including espresso machines, toasters and carpet shampooers.

How we choose what we test

With so many to choose from, what makes us choose one robot vacuum to test over another? As with most of our product testing, our aim is to test the most popular models on the market and what you're most likely to see in the retailers.

We survey manufacturers to find out about their range of models, we check market sales information and we also check for any member requests to test specific models. 

From this information we put together a final list that goes to our buyers. They then head out to the retailers and buy each product, just as a regular consumer would. We do this so we can be sure they're the same as any consumer would find them and not 'tweaked' in any way.

hard floor test_web

Robot vacuums may miss certain parts of the floor, even after repeated mapping.

pet hair test_web

We test how well the robot vacuums remove pet hair from carpet.

How we test robot vacuums

Dirt removal from hard floors

Our experts scatter a measured amount of sand evenly over a vinyl floor area and lets the robot run for 25 minutes, or until the vacuum completes its cleaning cycle, measuring the dirt pick-up at intervals throughout. 

They place objects (like chairs and a bench) in the room to make the set-up look similar to a home. With regular barrel vacuums, they pre-load the dust collector with dust, but robot vacs have a very small dirt receptacle which is normally emptied after a cleaning session, so these aren't pre-loaded.

Dirt removal from carpet

Sand is sprinkled evenly onto a section of carpet and the robot runs for a set period of time on the carpeted area. The sand collected is weighed and recorded.

Cleaning corners and edges

Sand is spread into a right-angled corner, and the robot vacuum cleaners are assessed to see how close to the corner they can get to pick up the dirt.

Pet hair removal

Cat hair is embedded into the carpet and the vacuum is allowed to run for six minutes. Our testers rate the vacuums based on how much cat hair is left behind after the test (this is a visual assessment).

Ease of use

While robot vacuums are generally very easy to use and operate, the increasing reliance on apps and connecting the machine to your home network means we started factoring this into our scores in 2023. The ease of use assessment includes:

  • ease of emptying the dirt receptacle (models with self-emptying functions tend to score better for this)
  • whether the app is easy to use, with clearly labelled functions
  • ease of connecting the robot and app to your home network.
mopping test_web

Robot vacuums can mop as well, so we now test how well they can remove stains and avoid carpet.

Mopping test

About half of the robot vacuums that come through the CHOICE lab now have mopping capability, so we've introduced a new mopping score (for models tested after August 2023).

Our experts assess how easy the mops are to program and prepare, and how easy the mopping pads are to wash.

When testing performance, they spread several common stains around laminate floorboards including coffee, red wine, a mud mixture, soy sauce and jam.

They then give the robot the best possible chance to do its job by selecting the most intense cleaning and water flow settings and rating how well the stains are removed.

Another part of this test is to see how well the robot avoids carpet (as mopping carpet can cause stains). Some robot vacuums don't have automatic carpet avoidance capability while mopping, and in those cases we follow the manufacturer instructions on how to set a defined cleaning area that will avoid the carpet. 

Some newer models can even lift their mopping pads up if they encounter carpet so you don't need to specify a 'no-go' area for the robot. If the mop makes things dirtier by getting on carpet and spreading soil around, a penalty may be applied to the overall mopping score.

Test criteria explained

For models tested prior to 2023, the CHOICE Expert Rating (our overall score) is made up of the following performance scores:

  • hard floor dirt pick-up (40%)
  • dirt removal from carpet (30%)
  • cleaning corners and edges (15%)
  • pet hair removal (15%).

From 2023 onwards, we include ease of use as 10% of the CHOICE Expert Rating, with the performance scores making up the remaining 90%.

Mopping scores (where applicable) don't form part of the CHOICE Expert Rating.

Our test lab

Our testing laboratory is up to date with the latest reference machines and calibrated measurement tools for our testers to bring you the right results.

We care about accuracy. See something that's not quite right in this article? Let us know or read more about fact-checking at CHOICE.

Stock images: Getty, unless otherwise stated.