Lidar

Lidar (light detection and ranging) measures the distance to a target by sending out laser light and analyzing what comes back. In the most common form, pulsed time of flight, a short pulse leaves the transmitter, reflects or scatters from the target, and a fast detector records its arrival; the delay $t$ gives the range. Light covers about 0.30 m per nanosecond, so a target 150 m away returns its echo after about 1.0 µs, and nanosecond timing gives decimetre-scale distances. Repeating the measurement in many directions builds a point cloud: a three-dimensional map of the scene. Typical sources are pulsed semiconductor lasers or fiber lasers near 905 nm or 1550 nm with pulse durations of a few nanoseconds.

Range and resolution

Because the light travels out and back, the range is half the round-trip path:

$$R = \frac{c\,t}{2}.$$

For $t = 1.0$ µs, $R = (3.00 \times 10^8 \text{ m/s})(1.0 \times 10^{-6}\text{ s})/2 \approx 150$ m.

Two targets along the same line of sight can be separated only if their echoes do not overlap, which sets the range resolution for a pulse of duration $\tau$:

$$\Delta R = \frac{c\,\tau}{2}.$$

A 5 ns pulse gives $\Delta R \approx 0.75$ m and a 1 ns pulse about 0.15 m. Resolution should be distinguished from precision: the arrival time of a single, well-separated echo can be located to a small fraction of the pulse duration by fitting its shape or timing its centroid, so centimeter precision is possible with nanosecond pulses. The pulse repetition rate also limits range: an echo must return before the next pulse leaves, giving a maximum unambiguous range $c/(2f_\text{rep})$, which is about 150 m at 1 MHz.

Sources: 905 nm versus 1550 nm

Near-infrared lidars around 905 nm use inexpensive pulsed laser diodes and silicon detectors, which are mature and cheap. Systems at 1550 nm use InGaAs detectors and often erbium fiber lasers, which cost more. The usual argument for 1550 nm is eye safety: light at that wavelength is absorbed in the front of the eye and does not reach the retina, so eye-safety regulations permit considerably more pulse energy there than at 905 nm, and the extra energy can be traded for range. Against this, water absorption is stronger at 1550 nm, which affects performance in rain and fog.

Scanning and flash architectures

A scanning lidar steers one beam, or a small fan of beams, across the scene with rotating mirrors, polygons, galvanometer scanners, MEMS mirrors or, in research systems, optical phased arrays. A flash lidar instead illuminates the whole field of view at once and images the return onto a detector array, measuring time of flight in every pixel in parallel. Flash systems have no moving parts but spread the pulse energy over the full scene, which limits their range; scanning systems concentrate the energy but need a reliable beam-steering mechanism.

Detectors

The returned signal is weak: for a diffuse target that fills the beam, received power falls as $1/R^2$, so moving from 10 m to 100 m costs a factor of 100, and the target reflectivity can vary from a few percent for dark surfaces to near unity for retroreflectors. Linear-mode avalanche photodiodes are the classic detector. Arrays of single-photon avalanche diodes and silicon photomultipliers detect individual photons and build a histogram of arrival times over many pulses, a form of time-correlated single-photon counting; they are well suited to flash and solid-state designs but must contend with ambient light triggering counts.

Coherent lidar

Instead of timing pulses, a coherent lidar mixes the return with a local copy of the transmitted light. In FMCW lidar the laser frequency is chirped and the range appears as a beat frequency, while the Doppler shift gives the radial velocity of each point directly. The trade-offs between the two approaches are compared in the article time of flight versus FMCW lidar.

Applications

Automotive and robotic lidars map the surroundings of vehicles and machines out to roughly 100–300 m. Airborne lidar surveys terrain and forest canopy, since some pulses pass through gaps in vegetation and return from the ground. Atmospheric lidars measure aerosol and cloud profiles from backscatter, wind from Doppler shifts, and gas concentrations by differential absorption (DIAL), where two wavelengths, one on and one off an absorption line, are compared.

Common questions

What does lidar stand for?

Light detection and ranging. The term is formed by analogy with radar (radio detection and ranging); it is written lidar, LiDAR or LIDAR.

How is lidar different from radar?

Both measure range from echoes, but lidar uses optical wavelengths several thousand times shorter than the 3.9 mm waves of 77 GHz automotive radar. The short wavelength allows a narrow beam and fine angular resolution, at the cost of stronger attenuation in fog, rain and dust.

Is lidar safe for the eyes?

Commercial lidars are designed and certified to meet laser eye-safety standards at their operating power. The permitted exposure depends strongly on wavelength, pulse duration and repetition rate, which is one reason 1550 nm systems can emit more energy per pulse than 905 nm systems under the same standard.

References: C. Weitkamp (ed.), Lidar: Range-Resolved Optical Remote Sensing of the Atmosphere (Springer, 2005); B. E. A. Saleh, M. C. Teich, Fundamentals of Photonics, 3rd ed. (Wiley, 2019).