Laser scanners and LiDAR: terrestrial, mobile, handheld and airborne

A laser scanner is a measuring instrument in the plain sense: it measures a distance, directly, hundreds of thousands of times a second, and every one is a point in space. It is the other way to make a point cloud, and the family of scanners, from a tripod to a drone to a diver’s sled, covers everything photogrammetry cannot. This page is how they work and what each kind is for.

How a scanner measures

Light travels about 30 cm in a nanosecond. A time-of-flight scanner fires a short pulse, starts a clock, and stops it when the reflection returns; the distance is half the round trip. A nanosecond of timing is 15 cm of range, so the electronics time to fractions of a nanosecond for millimetre results. Phase-shift scanners send a continuous beam modulated at several frequencies and compare the returned waveform with the sent one, which is faster and more precise at short range. Either way, the direction the beam left in is known from the mirror and the head, and direction plus distance is a point.

Two things set a scanner apart from a camera. It carries its own light, so it works at night and in shade. And one pulse can return more than once: a first return off the top of a canopy, more off the branches, and a last return off the ground. Keep the last returns and you have the terrain under the trees, which no photograph contains.

30 cm

How far light travels in a nanosecond

2.2 million

Pulses a second from a current survey-grade terrestrial scanner

3 mm

Its 3D accuracy at 50 m; 5 mm at 100 m

< 30 s

A full-dome scan at 6 mm point spacing, about 30 million points

The kinds of scanner

• Terrestrial laser scannerOn a tripod. Sweeps a full dome from each station; stations are joined by targets or by the overlap. Millimetres to hundreds of metres. Buildings, plants, tunnels, quarry faces, heritage.1–1,000 m · 1–6 mm · $20,000–120,000
• Handheld and backpack SLAMA spinning LiDAR carried through a space, positioning itself as it goes. A floor of a building in a walk, a mine drive, a forest plot. Centimetres, no GNSS needed.to 120–300 m · 5–15 mm · $20,000–60,000
• Mobile mappingScanners and cameras on a vehicle, with GNSS and an inertial unit, mapping a road and its verges at driving speed. Highways, rail, city streets.to 300 m · 1–3 cm · vehicle systems
• Drone and airborne LiDARA scanner on a drone or aircraft with its own GNSS and IMU, sweeping the ground below. Terrain under vegetation, corridors with wires, large areas.40–300 m AGL · 2–5 cm · $10,000–150,000+ for drone sensors

Terrestrial laser scanners

The animation at the top is a terrestrial scanner at work: set on a tripod, it rotates its head through 360° while a mirror sweeps the beam vertically, and every surface in line of sight becomes points. A current survey-grade instrument reaches a kilometre on a bright target, fires over two million pulses a second, and scans a full dome at 6 mm spacing in under half a minute, with a camera inside to colour the points. Accuracy is a few millimetres: 3 mm in 3D at 50 m, 5 mm at 100 m.

One station sees only what it can see, so a building takes dozens and a plant hundreds, joined (registered) by targets placed in the overlaps or by matching the overlapping scans themselves, and tied to control by surveyed targets. The point cloud that results is the standard record of as-built geometry: what a BIM model is checked against, what a tunnel’s profile is read from, what a quarry face is measured on when the drone cannot see it from above.

Strength

Millimetres, at distance, on any surface that reflects; interiors; vertical faces; night

Cost of it

Carrying it to every station, and registering the stations afterwards

Fails on

Glass, mirrors, water; very dark matte surfaces at range; anything that moved between stations

Output

A registered point cloud with intensity and colour, in control coordinates; meshes and sections cut from it

Handheld and mobile SLAM

A SLAM scanner solves the registration problem by never standing still: it works out its own position and orientation continuously from the geometry it is scanning (simultaneous localisation and mapping), so a person walks through a building and the floor is mapped behind them. Current handheld units run at 640,000 points a second or more, reach 120–300 m, and are specified at 5–15 mm; they tie to real coordinates through control points, GNSS where there is sky, or an RTK receiver on the unit.

They trade the tripod scanner’s millimetres for speed: a floor in minutes rather than a morning, a mine drive in a walk, a forest plot without a line of sight to the sky. Drift over a long walk is the thing to manage, by closing loops (returning to a scanned place) and by control points along the way.

Drone and airborne LiDAR

An airborne scanner has no fixed station; it is moving the whole time, so every pulse’s origin and direction come from a GNSS receiver and an inertial measurement unit on the sensor, and the quality of the point cloud is the quality of that trajectory. Drone sensors range from entry-level units around $10,000–15,000 through $30,000–50,000 mid-range to survey-grade systems over $100,000, plus the aircraft to carry them; a current mid-range unit returns 240,000 points a second with three returns per pulse and maps around 2 km² a flight at 2–5 cm vertical accuracy with good positioning and control.

The point of flying a laser is almost always the ground under vegetation, and after that the thin things (wires, fences, rails) that photographs lose. The LiDAR vs photogrammetry page weighs it against the camera case by case.

Bathymetric LiDAR

A green laser (532 nm) penetrates clear water, and a bathymetric scanner records a return from the surface and another from the bed, mapping rivers, lakes and shallow coasts to a few tens of metres’ depth where the water is clear enough. It is the only way to survey a riverbed from the air, and it fails in turbid water for the same reason a photograph does: the light does not get through.

Against photogrammetry

They are not rivals so much as neighbours on the range-against-accuracy chart, and the best surveys of complicated sites use both, merged in the same coordinates.