Choose a robot vacuum based on the obstacle that actually interrupts its cleaning cycles. Testing the exact model can help you assess how well it avoids small objects, while a no-go zone may suit a group of chairs that regularly stops the robot. However, the excluded area won’t be cleaned automatically. For low furniture or narrow gaps, measure the available space and compare it with the model’s dimensions. UFC-Que Choisir points out that both the height clearance and the space between furniture legs matter.
If objects are sometimes left on the floor, check tests of the exact model rather than assuming its camera or LiDAR will recognise every obstacle. Les Numériques reports test results for the eufy S1 Pro, but the same performance isn’t guaranteed in every home.
Chair legs or dark furniture can cause collisions or leave the robot stuck. If you’re happy to clean part of the floor separately, check whether the robot lets you set no-go zones. Dreame Support’s advice about certain black furniture applies to a specific case and system.
To clean under furniture or between closely spaced legs, compare the available height and width in your home with the model’s dimensions. UFC-Que Choisir highlights the importance of clearance under furniture and between chair legs.
Consider this if: objects are often moved around. Look for a test of the exact model using obstacles similar to yours.
The trade-off: a detailed map doesn’t prove that small objects will be detected or that areas the robot avoids will be cleaned.
Consider this if: a group of chairs or another fixed obstacle regularly stops the robot, and the model lets you edit the map.
The trade-off: the robot won’t clean the area within the exclusion zone, so you’ll need to clean it another way.
Consider this if: getting under a sofa or bed, or between closely spaced legs, is your main concern.
The trade-off: a low profile doesn’t guarantee the robot will avoid a crossbar or clean every corner.
| What matters to you | What to check | Who this is for |
|---|---|---|
| Avoiding getting stuck | Look for a test of the exact model using objects and chair legs similar to yours. Distinguish between map-based navigation and detecting obstacles on the floor. | I want the cleaning cycle to keep going around my usual obstacles; around 33% of consumers prioritise preventing the robot from getting stuck. |
| Setting exclusion zones | Check that the model lets you edit zones on its map, and see which areas will be left out of the cleaning cycle. | If a fixed obstacle is always in the same place, I could set an exclusion zone and clean that area another way; around 27% of consumers prioritise managing these problem areas. |
| Getting under furniture | Measure the clearance height and usable width between the legs, then compare them with the model’s dimensions. | I want to reach spaces under furniture or between its legs without relying on a general navigation claim; around 20% of consumers prioritise access to low or narrow spaces. |
| Managing rugs and objects on the floor | Check tests using rugs with the exact model, and see whether your lightweight rugs move or obstruct the robot during its first few cycles. | Before starting an automatic cycle, I want to know which rugs or objects to move; around 20% of consumers prioritise preparing the floor and handling rugs. |
In its archived 2024 review, Les Numériques describes the Tapo RV30 Plus’s navigation as methodical and its maps as easy to customise, but reports that it doesn’t detect small obstacles.
Check tests of navigation and obstacle avoidance separately, and make sure they cover the exact model number.
If a cleaning cycle is interrupted, you may need to free the robot or restart the clean.
A user on the Dreame forum described an X50 Ultra Complete that tried to climb onto a U-shaped chair leg and got stuck. This individual case doesn’t tell us how often robots generally get stuck.
Compare the shape of your chair crossbars with the scenarios covered in tests, and consider setting an exclusion zone if the robot repeatedly gets stuck.
The Deebot T50 Pro Omni tested by Les Numériques and the T50 Pro Omni Gen3 featured by Ecovacs are different models, so their advertised features aren’t interchangeable.
Check the full name and generation on the product page before relying on a test result or choosing accessories.
A no-go zone stops the robot returning to a problem area by keeping it away from that part of the floor during the cleaning cycle.
Make the zone as small as possible and check the results of the cleaning cycle in the room concerned.
The S1 Pro is a tested example of a model designed to avoid small objects in a home that can sometimes be cluttered.
The Deebot T50 Pro Omni, as tested, combines an 8,1 cm profile with obstacle detection, making it worth considering if you’re interested in both under-furniture access and obstacle avoidance.
The X50 Ultra, as tested, combines small-object avoidance, a retractable LiDAR and a climbing mechanism. These features may be useful around low furniture and certain thresholds.
The French product page for the Q10 S5+ Set and the Roborock app feature mapping, avoidance and zone management functions.
The RV30 Plus is an example of a model where map-based navigation alone doesn’t establish how well it detects small obstacles.
These models feature advertised obstacle detection, zone management or mapping functions to compare with the furniture and spaces in your home.
Image unavailable
If you want to explore advertised detection and exclusion features around fixed obstacles, the product page describes 3D detection, virtual walls and adjustable cleaning zones.
The page lists LiDAR, structured light and a claimed suction power of 8 000 Pa. It also mentions a 45-day dustbin; check the conditions associated with this claim.
Image unavailable
Worth considering if you want to compare advertised object recognition with mapping features for planning routes around furniture.
The product page describes the AIVI 3D 3.0 system, TrueMapping 2.0 and multi-floor mapping. It lists a height of 9,8 cm and suction power of 16 600 Pa.
Image unavailable
For managing some recurring obstacles, the product page lists AI.See, along with customisable no-go zones and virtual boundaries on the map.
The page describes suction power of 8 000 Pa and AI.See recognition of more than 100 object types. It advises removing small objects and cables from the floor before cleaning; these claims don’t guarantee that every obstacle in your home will be recognised.
Image unavailable
If objects often move into the robot’s path, this model lets you compare its advertised imaging technology with independent test results for the S1 Pro.
The product page describes MatrixEye with binocular infrared imaging and an RGB camera, and lists 8 000 Pa. Les Numériques reports that the S1 Pro detected and avoided the obstacles in its test, but the same result isn’t guaranteed in every home.
Image unavailable
To plan a route around furniture and exclude certain areas, the product page lists LiDAR navigation, maps and no-go zones.
The title lists 5 000 Pa suction power, five maps and ten no-go zones. The product page instructions say to remove small objects and cables before cleaning.
It can stop the robot returning to an area where it gets stuck, if the model lets you edit the map. However, the excluded area won’t be covered during that cleaning cycle, so you’ll need to clean it another way.
Start by clearing the route and creating a map if the model requires one. Then watch how the robot behaves around chairs, objects on the floor and rugs. Adjust no-go zones after these first runs: tests of different models report varying results when it comes to small obstacles.
No. Compare the height and diameter of the exact model with the available clearance and narrowest gaps in your home. Fitting under the furniture doesn’t guarantee that the robot can manoeuvre or avoid a low crossbar.
Check that the bags, filters and brushes are compatible with the exact model. For a new, second-hand or refurbished product sold by a business to a consumer, the statutory guarantee of conformity applies under the relevant conditions. The Ministry of the Economy and Finance states that you have two years to invoke it. This is separate from an optional commercial warranty.