How LiDAR mapping works
LiDAR (Light Detection and Ranging) uses a rotating sensor that emits laser pulses in all horizontal directions and measures the time it takes each pulse to return after bouncing off a surface. By triangulating thousands of these pulse return times per second, the robot builds a precise 2D map of its environment (walls, furniture, doorways) in real time.
In a robot vacuum, the LiDAR sensor sits in a spinning tower on top of the robot, rotating at 5 to 10 revolutions per second. Each rotation sweeps 360 degrees and adds another layer of precision to the map. Within 15 to 30 minutes of the first run, the robot has a usable floor plan. After the first complete run, the map is typically accurate enough for room segmentation and no-go zone placement.
The map is stored in the robot's memory and updated on subsequent runs. Moving furniture is automatically incorporated on the next run without requiring manual remapping.
How camera-based navigation works
Camera navigation (used by iRobot's Roomba line) builds maps visually. The robot uses optical sensors — typically downward-facing cameras and forward-facing cameras — along with accelerometers and gyroscopes to track its position relative to visual landmarks in the environment.
This approach is called vSLAM (visual simultaneous localization and mapping). The robot processes visual features in its camera feed (furniture legs, wall colors, floor patterns, door frames) and triangulates its position based on what it recognizes. Maps are built over multiple runs as the robot accumulates visual data from different angles.
On premium iRobot models (j7+, j8+, s9+), a forward-facing camera is also used for AI obstacle avoidance: the camera captures what is in front of the robot and an onboard AI model identifies specific objects (cables, socks, shoes, pet waste) to classify them and decide whether to avoid them.
Map accuracy and first-run speed
LiDAR: accurate from run 1
LiDAR produces a precise 2D floor plan within the first cleaning run. By the end of run 1, walls and doorways are accurately mapped. Room boundaries are typically identifiable within two to three runs. No-go zones can be placed accurately after the first run.
LiDAR maps are independent of lighting conditions, floor color, and wall texture. The laser measures distance, not visual appearance, so the map is equally precise on white walls and dark walls, in rooms with no visual landmarks.
Camera: improves over multiple runs
Camera-based maps stabilize over 5 to 10 runs. The first run produces a rough navigable map. Room boundaries and accurate room segmentation typically require 5 to 10 runs to reach the same precision a LiDAR unit achieves in 1 to 2.
For most users, this maps initialization difference is a one-time inconvenience during the first week of use. After that, both systems perform equally in day-to-day navigation.
Camera maps are affected by significant environmental changes: if a room is completely redecorated, the visual landmarks change and the robot may need to remap. LiDAR maps measure geometry, not appearance, and are more robust to cosmetic changes.
Low-light performance
This is one of LiDAR's clearest advantages. LiDAR uses laser pulses, which are not dependent on ambient light. A LiDAR robot navigates identically in a dark room or a bright room.
Camera-based navigation on iRobot units uses floor-facing sensors and downward cameras that also function in darkness through infrared sensing. Basic navigation in dark rooms is handled adequately. The forward-facing AI obstacle avoidance camera, however, requires some ambient light to identify objects. In complete darkness, the robot navigates safely but cannot classify what objects are in front of it — falling back to proximity detection without identification.
Practical impact: if you schedule your robot to clean at 3 AM in a home where lights are off, a LiDAR robot handles this without issue. An iRobot camera unit also handles it for navigation, but with reduced obstacle identification capability.
Robot height: LiDAR's trade-off
The spinning LiDAR tower adds height to the robot chassis. Roborock and Dreamebot LiDAR units are typically 3.7 to 4.2 inches (9.4 to 10.7 cm) tall. This prevents them from fitting under furniture with clearances below that height.
Camera-based iRobot models have no spinning tower and are significantly lower profile. The Roomba j7+ stands 3.4 inches (8.6 cm). This allows it to reach under sofas, beds, and furniture that stop LiDAR competitors.
Before buying: measure the clearance under key furniture pieces in your home. If your sofa has 3.5 inch clearance and you want the robot cleaning under it, a camera-based iRobot is the only option. If all furniture is well-elevated (4+ inches), height is not a practical concern.
| Robot | Navigation Type | Height |
|---|---|---|
| iRobot Roomba j7+ | Camera (vSLAM + AI) | 3.4 in (8.6 cm) |
| iRobot Roomba s9+ | Camera (vSLAM) | 3.5 in (8.9 cm) |
| Eufy RoboVac 11S | Bump-and-go | 2.85 in (7.2 cm) |
| Roborock S8 | LiDAR + AI camera | 3.85 in (9.8 cm) |
| Roborock Q5 | LiDAR | 3.8 in (9.65 cm) |
| Dreamebot L20 Ultra | LiDAR + 3D obstacle | 3.85 in (9.8 cm) |
Obstacle avoidance: where camera has a real advantage
LiDAR detects obstacles (it measures that something is there) but cannot identify what they are. A LiDAR robot sees a toy, a cable, and a shoe as the same shape: an object to navigate around or push out of the way.
Camera-based AI on premium iRobot units identifies specific object categories and decides how to respond to each:
- Pet waste: avoid completely, continue cleaning elsewhere
- Charging cables: avoid, flag in the app
- Socks: avoid, do not push them
- Shoes: navigate around
Some premium LiDAR units (Roborock S8 with Reactive AI 2.0, Dreamebot L20 Ultra) have added forward-facing cameras on top of LiDAR specifically for obstacle identification. These are hybrid systems that combine LiDAR mapping with camera obstacle classification. They perform well for general obstacles but do not include the pet waste avoidance guarantee that iRobot offers specifically for the j-series.
Which technology wins overall
For most buyers without specific under-furniture or pet waste requirements: LiDAR-based units offer more choice, better value, and faster map initialization. The difference in daily navigation quality is not perceptible once both types have completed their initial mapping runs.
Which brands use each technology
| Brand | Navigation Type | Notable Models |
|---|---|---|
| iRobot (Roomba) | Camera (vSLAM + AI on j-series) | j7+, j8+, i4+, s9+ |
| Roborock | LiDAR (+ hybrid camera on S8+) | S8, S8 Pro Ultra, Q5, Q7+ |
| Dreamebot | LiDAR (+ 3D obstacle on L-series) | L20 Ultra, D10 Plus |
| Eufy (premium) | LiDAR | L35, X8, L60 |
| Shark | AI navigation (camera-based, not vSLAM) | AV2502WD, Matrix series |
| Ecovacs | LiDAR + TrueDetect | Deebot T20, N8 |
FAQ
Does LiDAR or camera navigation have better battery efficiency?
LiDAR navigation is slightly less battery-efficient because the spinning sensor motor runs continuously during the cleaning session. In practice, the difference is small: LiDAR units' larger batteries more than compensate. Roborock S8 (LiDAR): 180-minute runtime. Roomba j7+ (camera): 75-minute runtime. The battery difference between models is due to battery capacity choices, not navigation type efficiency — LiDAR robots tend to use larger batteries.
Is LiDAR safe (laser exposure)?
Yes. Robot vacuum LiDAR sensors use Class 1 lasers, the safest category. Class 1 lasers are safe under all normal use conditions, including if you look at the spinning sensor. They are the same class as barcode scanners and DVD players. No eye protection is required; no special handling precautions apply.
Top LiDAR and camera robots to buy
- Best LiDAR robot: Roborock S8 on Amazon →
- Best camera-navigation robot: iRobot Roomba j7+ on Amazon →
- Best budget LiDAR: Roborock Q5 on Amazon →