LiDAR

See also: Depth sensing, Time-of-flight camera and SLAM

LiDAR (an acronym for light detection and ranging, also written lidar) is a depth-sensing and ranging method that measures the distance to objects by illuminating them with laser light and timing or analysing the reflected return. A LiDAR sensor sweeps or floods a scene with laser pulses, records how long each pulse takes to return, and converts those times into distances, producing a set of 3D points known as a point cloud.

The technology originated for surveying, atmospheric science, and autonomous vehicles, but it became directly relevant to consumer augmented reality when Apple added a LiDAR Scanner to the iPad Pro in March 2020 and to the iPhone 12 Pro that October. In AR and mixed reality devices, LiDAR supplies a real-time depth map of the surroundings, which is used for spatial mapping of rooms, fast placement of virtual objects on real surfaces, and occlusion, where virtual content is correctly hidden behind nearer real objects. The same sensors feed SLAM (simultaneous localization and mapping) pipelines in robotics and positional tracking.

How it works

Most LiDAR systems measure distance by time of flight (ToF). The sensor emits a short laser pulse, and the distance D to the surface that reflects it follows from the round-trip time t as D = ½ ⋅ c ⋅ t, where c is the speed of light. Because light travels about 30 cm per nanosecond, accurate ranging needs very fine timing: for centimetre-level single-shot precision the timing circuit must resolve delays of well under 100 picoseconds. Pulses themselves typically last a few nanoseconds or less.

This pulsed approach is called direct time of flight (dToF). A second family, coherent or frequency-modulated continuous-wave (FMCW) LiDAR, sweeps the laser frequency in a chirp and mixes the return with the outgoing beam by optical heterodyne detection; it recovers range from the beat frequency and can also measure target velocity directly through the Doppler shift. FMCW is common in long-range automotive sensing but is not used in current consumer AR hardware.

LiDAR is distinct from structured-light depth sensing, which projects a known dot or stripe pattern and infers depth from how the pattern deforms over a surface. Apple's front-facing TrueDepth camera is a structured-light system, whereas its rear LiDAR Scanner is a time-of-flight system.

Beam steering: scanning vs solid-state

LiDAR designs differ mainly in how they aim the laser across a scene, which sets the size, cost, and durability of the sensor.

Approach	How the beam is directed	Notes
Mechanical scanning	A spinning housing or rotating mirror sweeps the beam, often through a full 360 degrees	Wide field of view and long range; bulky, with moving parts that wear and add cost and vibration
MEMS (semi-solid-state)	A small micro-electro-mechanical mirror oscillates to steer one or a few beams	More compact than spinning units; still has a moving micromirror
Optical phased array (solid-state)	Phase shifts across an array of on-chip emitters steer the wavefront electronically, with no moving parts	Fully solid-state; smaller, quieter, more robust, and cheaper to mass-produce
Flash (solid-state)	A single wide pulse floods the whole scene at once and a 2D detector array captures depth for every pixel in parallel	No moving parts; well suited to compact devices, with range limited by available laser power

Solid-state designs (flash and optical phased array) are generally smaller, quieter, lower cost, and less prone to failure than spinning mechanical units because they have no moving parts. Compact LiDAR sensors commonly pair a vertical-cavity surface-emitting laser (VCSEL) emitter array as the light source with a single-photon avalanche diode (SPAD) array as the detector; SPADs can register individual returning photons, which suits the very low light levels and fast timing that dToF requires.

Use in AR and VR

In augmented reality, a per-frame depth map of the surrounding space is the basis for placing and behaving correctly relative to real geometry. LiDAR provides that depth directly and quickly, rather than inferring it from camera motion alone.

On Apple devices, the LiDAR Scanner works with ARKit, Apple's AR framework. When the iPad Pro scanner launched, Apple said it lets existing ARKit apps place virtual objects automatically without the user first sweeping the camera around the room, improves motion capture and people occlusion, and speeds up the Measure app, while a new Scene Geometry API exposes a mesh of the environment to developers. The depth data combines points from the LiDAR Scanner with the cameras and motion sensors and is processed by computer vision algorithms on the device's chip.

LiDAR also reaches headset-class hardware. The Apple Vision Pro, Apple's mixed-reality headset, lists a LiDAR Scanner among its sensors alongside a TrueDepth camera, six world-facing tracking cameras, four eye-tracking cameras, and four inertial measurement units; the LiDAR contributes to the depth perception, occlusion, and scene understanding the headset uses to anchor content in the room. TechInsights, which teardown-analysed the headset, reported that the Vision Pro's LiDAR module is an adaptation of the established iPad Pro LiDAR, uses direct time of flight, and pairs a 940 nm VCSEL array (driven by a Texas Instruments part) with a Sony SPAD detector, flashing about twice per second.

Not every AR or VR headset uses LiDAR. Many devices derive depth and tracking from stereo cameras and visual-inertial SLAM instead, and LiDAR is one of several depth-sensing methods alongside structured light and passive stereo. Where it is present, its advantage is that it works in the dark and does not depend on visible surface texture, since it supplies its own light.

Beyond live AR rendering, the smartphone LiDAR scanners are widely used for 3D capture: scanning rooms, objects, and small landforms into meshes or point clouds that can be reused as assets or measurements. A 2021 Scientific Reports study led by Gregor Luetzenburg evaluated the iPhone 12 Pro LiDAR for geoscience surveying and found an absolute accuracy of about ± 1 cm on small objects and ± 10 cm on a coastal-cliff model, with point density falling sharply over distance (roughly 7,225 points/m² at 25 cm versus 150 points/m² at 250 cm) and reliable performance limited to within about 5 m and to surfaces larger than 10 cm. The authors described it as a low-cost alternative to established close-range survey methods rather than a replacement for professional instruments.

Camera and other on-device uses

On the iPhone 12 Pro and later Pro models, the LiDAR Scanner also assists the rear camera. Apple states it improves autofocus by 6x in low-light scenes, reduces capture time, and (with the A14 Bionic Neural Engine) enables Night mode portraits with a low-light bokeh effect.

Origins

LiDAR followed quickly from the invention of the laser. Theodore Maiman's team at Hughes Research Laboratories demonstrated the first working laser, a ruby laser, in 1960, and Hughes Aircraft Company built an early laser-ranging system soon after, around 1961 to 1962. The term was originally read as a blend of "light" and "radar" (laser radar) before settling on the "light detection and ranging" expansion. One early high-profile use was in space: a lunar laser altimeter built by RCA flew aboard Apollo 15 in 1971 to measure the spacecraft's distance to the Moon's surface. The technology later became central to topographic surveying, atmospheric remote sensing, and self-driving cars before its miniaturisation into phones and headsets.