Short answer: LiDAR earns its place on a robot when the job needs direct, lighting-independent distance measurement: navigation and obstacle avoidance on mobile robots, safety monitoring, and mapping. It measures range by timing light pulses, so it works in darkness, glare and on the textureless surfaces where camera-based depth struggles. It does not read signs, colours or fine surface detail — which is why most serious robots fuse LiDAR with cameras rather than choosing between them.
Spinning, oscillating and solid-state designs
The original robot LiDAR is a spinning unit: a laser and receiver rotating through 360 degrees, producing a flat or multi-line scan. It remains the reference for mapping and navigation on AMRs and outdoor machines. Oscillating (MEMS and galvanometer) designs sweep a smaller field with a moving mirror — fewer moving parts, a defined forward field. Fully solid-state designs, including flash LiDAR, have no scanning mechanism at all and trade field of view and range for robustness and price. The directory profile of Benewake represents the solid-state end of the market: compact units aimed at robots and vehicles rather than survey instruments. Choose the architecture by the motion your robot actually makes — a warehouse AMR and a forward-looking delivery robot buy different sensors.
How to read the specifications
Range is always quoted at a target reflectivity — a unit rated at 40 metres against a bright target may see a dark one at half that. Compare ranges at the same reflectivity figure. Field of view splits into horizontal and vertical; a wide horizontal scan with a thin vertical slice suits flat-floor navigation and misses overhanging obstacles. Points per second and angular resolution decide how much detail the perception stack receives, and frame rate decides how stale the picture is at speed. Finally, check the environmental ratings as carefully as the optics: IP rating, operating temperature and vibration tolerance are what separate a robot sensor from a laboratory one.
Navigation, mapping and safety
On mobile robots, LiDAR does three distinct jobs that are often confused in purchasing. For localisation and mapping (SLAM), a 2D or 3D scan is matched against a map — this is the use case behind most AMR navigation, in the segment our AMR vs AGV guide compares. For obstacle avoidance, what matters is the near field and the update rate, not maximum range. For safety, only certified safety scanners may be used as protective devices — a navigation LiDAR, however accurate, is not a safety component and must not be wired as one. Suppliers in the vision and sensing section of the directory can be filtered by exactly these roles.
One specification buyers new to the category often ask about is eye safety. Reputable robot LiDAR is designed to Class 1 eye-safe limits under the IEC 60825 laser standard — safe under normal operating conditions, including for bystanders in warehouses and public spaces. Verify the class on the datasheet and in the certification documents rather than assuming it; it is one of the first things a safety assessor will ask for when the robot works around people, and it should be a stated property of the exact model quoted, not of the product family.
Where LiDAR degrades: rain, dust and glass
Light pulses scatter. Heavy rain, fog, dust clouds and exhaust haze all shorten effective range and can add false returns; sites with airborne dust should budget margin and plan sensor cleaning. Glass and polished metal are the indoor equivalents — specular surfaces reflect the pulse away from the receiver, so a glass wall may appear as open space. Perception teams handle these cases with fusion and conservative speed near known problem surfaces; buyers should ask a supplier specifically how the unit behaves in their worst environment, and test there.
The fusion answer
The industry's direction is combination, not substitution. LiDAR gives metric geometry that cameras cannot match; cameras give semantics — signs, markings, object classes — that LiDAR cannot see. Automotive programmes drove LiDAR prices down the scale curve, and robot makers now inherit those units: Hesai, whose sensors were designed for vehicle programmes, is a visible example of automotive volume changing what a robot programme can afford. The purchasing conclusion: specify the perception system, not the sensor — and evaluate suppliers on the fused output your navigation stack will actually consume.
Frequently asked questions
Is LiDAR better than a camera for robot navigation?
They measure different things. LiDAR gives direct, accurate distance independent of light and texture, which makes it the safer primary sensor for navigation and obstacle detection. Cameras provide the semantic detail — signs, markings, object types — that LiDAR lacks. Most capable robots carry both.
What does solid-state mean in a LiDAR specification?
That the scanning is done without a conventional rotating assembly — by MEMS mirrors, optical phased arrays or flash illumination, depending on the maker. It generally improves robustness and price, usually at the cost of field of view or range compared with spinning units.
How far can a robot LiDAR see?
There is no single figure: range depends on the model and on target reflectivity, and honest datasheets state both. A warehouse AMR may need 20–30 metres of reliable detection; an outdoor machine may need far more. Compare models at the same reflectivity and in your actual environment.
Does LiDAR work in rain, fog and dust?
With degraded performance. Precipitation and airborne dust scatter the pulses, shortening range and adding noise, and glass or mirror-like surfaces can be missed entirely. Outdoor deployments design for the worst expected conditions and test in them, rather than trusting clear-weather specifications.