Two robot vacuums can look almost identical on the shop shelf and still be worlds apart in your living room. One cleans in tidy rows, heads back to its dock when the battery runs low and picks up where it left off. The other wanders, misses the hallway and ends up stuck under the sofa. Suction gets the headlines, but navigation is usually what separates them: how the robot works out where it is, where it has been and where it still needs to go. Here is how the main systems work, in plain English, so you can tell what you are paying for.
What "SLAM" means
Most capable robot vacuums rely on SLAM, short for simultaneous localisation and mapping. The robot is doing two jobs at once: drawing a map of your home and working out its own position on that map. Robotics researchers have long described it as a chicken-and-egg problem, because a good map depends on knowing where you are, and knowing where you are depends on having a good map. The ideas date back decades in academic robotics; the clever part is squeezing them into a battery-powered disc that costs a few hundred pounds.
A robot that can do this cleans systematically, can be sent to a single room, respects no-go zones you draw in the app and can find its dock. Robots without it have to make do with much cruder methods.
The four navigation methods you'll see
1. Random bounce. The simplest robots have bump sensors and little else. They drive until they hit something, turn and carry on. Given enough time they will cover a small room, but with lots of overlap and no guarantee of reaching every corner. Early Roombas worked without maps, and iRobot still lists random-bounce navigation for at least one current model.
2. Gyroscope and wheel tracking. A step up, these robots estimate where they have been by measuring how far their wheels have turned and which way they are facing. That lets them clean in straighter lines, but small measurement errors build up over a run, so the robot's sense of position slowly drifts. You'll find this approach in budget and some mid-range models.
3. Camera-based navigation (vSLAM). Here a camera picks out distinctive features such as picture frames, door frames and windows, and uses them as landmarks to build a map and locate itself. iRobot introduced this to the Roomba with the 980 in 2015, and Dyson took it further with a 360-degree camera on its robots. Britain has a stake in the technology: Dyson's first robot, the 360 Eye, drew on visual SLAM research by Professor Andrew Davison's team at Imperial College London. The downside is light. Camera systems work less well in dim rooms or spaces with few visual features, so iRobot advises turning the lights on for mapping runs.
4. LiDAR (laser mapping). The little turret on top of many robots spins a laser and measures how long the light takes to bounce back, giving precise distances to walls and furniture all around. It builds an accurate map quickly and works in the dark. Roborock helped make LiDAR mainstream in robot vacuums from 2016, and the approach has become so dominant that iRobot moved its entire 2025 Roomba line-up to LiDAR. The weaknesses: the turret adds height, which can stop a robot fitting under low furniture, and mirrors and other shiny surfaces can confuse it.
Hybrids are now the norm
At the top of the market most robots combine methods. LiDAR handles the map; a front camera, often paired with structured light (a projected pattern that measures the shape of nearby objects), handles obstacles such as cables, socks and pet mess that a laser near the top of the robot can miss. iRobot pairs LiDAR with its PrecisionVision camera on some models, Roborock combines laser navigation with RGB cameras, and Dyson's newer Spot+Scrub Ai pairs LiDAR with an HD camera. Makers are also finding ways round the turret: Roborock's Saros 10, for example, can retract its laser sensor to bring the robot's height down to under 8cm.
Dyson's 360 Vis Nav shows the camera-led route at its most ambitious. It uses a fish-eye lens for a panoramic view plus 26 sensors, and Dyson says it can pinpoint its position to within 71mm.
Which method suits a UK home?
| Your home | Best fit | Why |
|---|---|---|
| Studio or small one-bed flat | Gyroscope or budget LiDAR | A simple layout doesn't need a sophisticated map. |
| Several rooms on one floor | LiDAR, or LiDAR plus camera | Accurate maps make room-by-room cleaning and no-go zones work. |
| Pets, children, cables and toys | Camera-based or hybrid | Object recognition copes with clutter better than a laser alone. |
| Dark rooms, cleaning at night | LiDAR | Lasers don't need light; cameras can struggle. |
| Two or more floors | Any mapping robot with multi-map support | Saved floor plans let it recognise each level after you carry it there. |
The UK angle: prices, Matter and privacy
Laser mapping is no longer a premium feature. In our guide to the best robot vacuums in the UK, the Roborock Q7 L5+ offers LiDAR navigation and a self-emptying dock for around £160 at the time of writing.
Smart home support is improving too. The Matter standard added robot vacuums as a device type in version 1.2 in 2023, and in March 2025 Apple's iOS 18.4 brought them to the Apple Home app, with Roborock, iRobot and Ecovacs issuing firmware updates. Matter control is still fairly basic, covering things such as starting and stopping a clean, so you'll keep using the maker's app for maps and zones. Our beginner's guide to Matter explains what you need.
If a camera in your home bothers you, check the privacy settings before you buy. iRobot, for instance, lets owners turn on or off a feature that photographs obstacles for review in its app. UK law now also requires connected devices to publish how long they will get security updates, which is worth checking for any robot that carries a camera.
The bottom line
If your home has more than one room, choose a robot with genuine SLAM. Pick LiDAR for accuracy and cleaning in the dark, a camera for spotting clutter, or a hybrid for both. When Which? tests robot vacuums it measures how much floor they cover and how they handle trailing wires and raised thresholds, which tells you more than any suction figure. Suction sells robot vacuums; navigation decides whether you keep using one.