A

Aerial LiDAR

LiDAR data acquired from an airborne platform — fixed-wing aircraft, helicopter, or drone. Aerial LiDAR is used for topographic mapping, forestry analysis, corridor surveying, and large-area infrastructure monitoring. Sensor payloads range from lightweight drone-mounted units to high-altitude survey-grade systems producing hundreds of points per square metre.

Aerial Photogrammetry

The process of creating 3D models, maps, or measurements from overlapping aerial photographs captured by drones or manned aircraft. Software reconstructs the geometry of the scene by identifying matching features across multiple images and computing camera positions. Aerial photogrammetry is often used alongside LiDAR for colourised point clouds and textured mesh outputs.

Angle of Incidence

The angle at which a laser beam strikes a surface relative to the surface normal. A perpendicular hit (zero angle of incidence) produces the strongest return signal. As the angle increases, the reflected energy decreases and measurement noise rises — a critical factor in scanning building facades, terrain slopes, and moving objects.

APD (Avalanche Photodiode)

A semiconductor photodetector that uses internal gain through impact ionisation to amplify the signal from incoming photons. APDs are widely used in LiDAR receivers because they offer higher sensitivity than standard PIN photodiodes while remaining below the gain levels of Geiger-mode devices. Indium gallium arsenide (InGaAs) APDs are the standard for 1550nm LiDAR systems.

ASIC (Application-Specific Integrated Circuit)

A custom-designed chip built for a specific function. In LiDAR systems, ASICs handle time-to-digital conversion (TDC), signal processing, and detector readout. Purpose-built ASICs enable smaller, faster, and more power-efficient sensor modules compared to systems built on general-purpose processors or FPGAs.

Autonomous Perception Stack

The complete software and hardware pipeline that allows an autonomous vehicle or robot to understand its environment. Typically includes LiDAR, cameras, radar, ultrasonic sensors, sensor fusion algorithms, object detection models, tracking, and prediction. The perception stack is the core intelligence layer that converts raw sensor data into actionable decisions.

B

Bathymetric LiDAR

A specialised LiDAR system that uses green-wavelength (532nm) laser pulses to penetrate water and map underwater topography. The green wavelength is used because it passes through water more effectively than the near-infrared wavelengths used in standard LiDAR. Bathymetric LiDAR is used for coastal survey, riverbed mapping, and shallow-water habitat assessment.

Beam Divergence

The angle at which a laser beam spreads as it travels away from the source. Lower divergence means a tighter beam, which translates to higher angular resolution and better performance at long range. Beam divergence is determined by the laser source design and the transmit optics — it is one of the key specifications that separates survey-grade LiDAR from consumer-grade units.

Beam Steering

The method used to direct the laser beam across the field of view. Mechanical systems use spinning mirrors or prisms. MEMS mirrors use micro-electromechanical oscillating elements. Optical phased arrays (OPA) steer the beam electronically with no moving parts. Flash systems illuminate the entire scene simultaneously and require no beam steering at all. The steering mechanism is one of the most consequential architectural choices in LiDAR design.

BIM (Building Information Modelling)

A digital process for managing building design, construction, and operations data throughout the lifecycle of a structure. LiDAR point clouds are frequently used as the as-built reference layer for BIM workflows — a process called “Scan-to-BIM” — enabling architects and engineers to work from millimetre-accurate representations of existing conditions.

C

Classification (Point Cloud)

The process of labelling each point in a point cloud with a category — ground, vegetation, building, water, powerline, vehicle, etc. Classification can be manual, rule-based, or AI-driven. Classified point clouds are essential for extracting useful information from raw LiDAR data, enabling applications like terrain modelling, feature extraction, and autonomous navigation.

Cloud Computing (for LiDAR)

The use of remote servers and GPU clusters to process, store, and analyse large LiDAR datasets. Point cloud processing, AI model training for object detection, HD map generation, and simulation environments for autonomous vehicles all require significant compute resources — typically hosted on platforms like AWS, Azure, or Google Cloud. Cloud processing enables scalability that local hardware cannot match.

Coherent Detection

A detection technique where the returned laser signal is mixed with a local oscillator (a reference beam from the same laser source) to measure both amplitude and phase. Coherent detection is the basis of FMCW LiDAR and enables instantaneous velocity measurement alongside range — a significant advantage over direct-detection (ToF) systems for moving-target applications.

Control Point

A physical ground marker with precisely known coordinates (typically surveyed with RTK GNSS) used to georectify and validate the accuracy of LiDAR datasets. Control points serve as the anchor between a point cloud and the real-world coordinate system.

D

Dark Count Rate (DCR)

The rate at which a single-photon detector (such as a SPAD or GmAPD) registers false events in the absence of light. Dark counts are caused by thermal excitation of charge carriers in the detector material. Lower DCR means cleaner data and longer usable range — it is one of the most important performance metrics for Geiger-mode LiDAR detectors.

DEM (Digital Elevation Model)

A 3D representation of terrain elevation derived from LiDAR or photogrammetry data. A DEM represents the bare earth surface with vegetation and buildings removed. DEMs are fundamental to hydrology, infrastructure planning, flood modelling, and terrain analysis.

Digital Twin

A virtual replica of a physical asset, facility, or environment — typically constructed from LiDAR scans, photogrammetry, IoT sensor data, and BIM models. Digital twins are used for simulation, condition monitoring, predictive maintenance, and planning. In infrastructure, a digital twin enables engineers to test scenarios virtually before making physical changes.

Direct Detection

A LiDAR detection method where the receiver measures the intensity and arrival time of returned photons without mixing them with a reference signal. Direct detection is simpler and cheaper than coherent detection but does not provide velocity information. Most time-of-flight and flash LiDAR systems use direct detection.

DSM (Digital Surface Model)

A 3D representation that includes all objects on the earth’s surface — buildings, trees, vehicles, and other structures — as captured by LiDAR. Unlike a DEM (which shows bare earth), a DSM represents the highest reflective surface at each point. DSMs are used for urban planning, telecommunications line-of-sight analysis, and solar potential assessment.

E

Edge Computing (LiDAR)

Processing LiDAR data locally on the sensor platform — inside the vehicle, drone, or robot — rather than transmitting raw data to a cloud server. Edge computing is essential for real-time autonomous navigation where latency must be measured in milliseconds. Modern LiDAR systems increasingly integrate on-chip or near-chip processing to classify objects at the sensor level.

Eye Safety Class

A regulatory classification (IEC 60825) that defines the maximum permissible laser exposure for a LiDAR system. Eye safety is determined by wavelength, pulse energy, and beam characteristics. 905nm systems are more restricted than 1550nm systems because the human retina absorbs 905nm light more readily. Eye safety constraints directly limit the range and power of automotive LiDAR systems.

F

False Alarm Rate

The frequency at which a LiDAR detector or perception system incorrectly identifies a signal or object that isn’t there. In single-photon detectors, false alarms come from dark counts and background noise. In perception stacks, false alarms come from misclassified point cloud clusters. Minimising false alarm rates is critical for safety in autonomous driving and defence targeting applications.

Flash LiDAR

A LiDAR architecture that illuminates the entire scene with a single broad laser pulse (like a camera flash) and captures the returns simultaneously across a 2D detector array. Flash LiDAR has no moving parts and can capture a complete 3D frame in a single shot. Trade-offs include limited range (the laser energy is spread across the full field of view) and detector array complexity.

FMCW (Frequency-Modulated Continuous Wave)

A LiDAR architecture that emits a continuous laser beam with a linearly varying frequency (chirp). By comparing the frequency of the returned signal to the transmitted signal, the system measures both range and instantaneous velocity simultaneously. FMCW is inherently immune to interference from other LiDAR sensors and can operate at very long ranges. It is considered the next-generation architecture for automotive LiDAR by several major sensor companies.

Focal Plane Array (FPA)

A 2D grid of photosensitive detector elements — analogous to the pixel array in a camera sensor, but designed for LiDAR wavelengths. In Geiger-mode and flash LiDAR systems, the FPA captures an entire 3D scene in parallel, with each pixel independently measuring time-of-flight. FPA resolution (e.g. 32×32, 128×128, 256×256) determines the angular resolution of the image.

FOV (Field of View)

The angular extent of the scene that a LiDAR sensor can observe — typically specified as horizontal FOV and vertical FOV in degrees. Automotive long-range LiDAR might have a narrow FOV (e.g. 25° × 10°) while a mapping sensor might cover 360° horizontally. The FOV determines what the sensor can see without physically reorienting.

G

GaAs (Gallium Arsenide)

A III-V semiconductor material used to manufacture VCSEL laser arrays for 905nm LiDAR systems. GaAs is mature, well-understood, and can be fabricated at wafer scale — making it the material of choice for high-volume automotive LiDAR illumination sources.

Gaussian Splat

A rendering technique that represents point cloud data as overlapping Gaussian distributions rather than discrete dots. Gaussian splatting produces visually smooth and photorealistic 3D reconstructions from sparse point cloud data. The technique has gained significant traction in real-time 3D visualisation and neural radiance field (NeRF) workflows.

GCP (Ground Control Point)

A physical marker placed in the survey area with coordinates measured to high accuracy (typically centimetre or sub-centimetre) using RTK or PPK GNSS. GCPs are used to georectify aerial LiDAR and photogrammetry datasets, validate absolute accuracy, and provide a traceable quality assurance chain.

Geiger-Mode APD (GmAPD)

An avalanche photodiode operated above its breakdown voltage, in the Geiger regime, where a single absorbed photon triggers a full-scale avalanche current. GmAPDs are the most sensitive class of LiDAR detector — capable of detecting individual photons at ranges exceeding 10 kilometres. Originally developed for defence reconnaissance, Geiger-mode arrays are now moving into commercial mapping and autonomous navigation applications.

Georeferencing

The process of assigning real-world coordinates to every point in a LiDAR dataset. Georeferencing combines GNSS position data, IMU orientation data, and ground control points to place the point cloud accurately on the Earth’s surface. Without georeferencing, a point cloud is a dimensionally accurate but locationally floating 3D object.

GNSS (Global Navigation Satellite System)

The umbrella term for satellite-based positioning systems used to determine location on Earth. Includes the US GPS, European Galileo, Russian GLONASS, and Chinese BeiDou constellations. In LiDAR operations, GNSS provides the position component of the georeferencing pipeline — typically combined with an IMU for full six-degree-of-freedom trajectory reconstruction.

I

IMU (Inertial Measurement Unit)

A device containing accelerometers and gyroscopes that measures the orientation, angular velocity, and linear acceleration of a platform. In LiDAR systems, the IMU provides the attitude data needed to georectify point clouds — correcting for the roll, pitch, and yaw of the aircraft, vehicle, or handheld scanner during data acquisition.

InGaAs (Indium Gallium Arsenide)

A III-V semiconductor alloy used to manufacture photodetectors sensitive to 1550nm near-infrared light. InGaAs detectors — including APDs and SPAD arrays — are the standard receiver technology for eye-safe long-range LiDAR. InGaAs fabrication is more complex and expensive than silicon, which is why 1550nm systems cost more than 905nm systems at equivalent volumes.

InP (Indium Phosphide)

A semiconductor substrate material used alongside InGaAs in the fabrication of 1550nm laser sources and detector arrays. InP-based edge-emitting lasers are the primary illumination source for long-range and defence-grade LiDAR systems.

Intensity (LiDAR)

A measure of the strength of the laser return signal recorded for each point in a point cloud. Intensity varies based on the reflectivity of the target surface, the angle of incidence, and the range. Intensity data adds a layer of information beyond geometry — enabling material classification, road marking detection, and vegetation health analysis.

ISR (Intelligence, Surveillance, and Reconnaissance)

Military and intelligence operations focused on gathering information about adversary activity and terrain. LiDAR — particularly Geiger-mode and 1550nm systems — plays a critical role in ISR by providing high-resolution 3D mapping from long stand-off distances, through obscurants, and at night.

L

LAS / LAZ File Format

The standard file formats for storing LiDAR point cloud data. LAS is the uncompressed format defined by the American Society for Photogrammetry and Remote Sensing (ASPRS). LAZ is the losslessly compressed version, typically 7–20× smaller. Both formats store XYZ coordinates, intensity, return number, classification, timestamp, and other attributes per point.

Last-Mile Fibre

The final segment of a fibre-optic network connecting the backbone infrastructure to the end user. While not a LiDAR term, last-mile fibre is relevant because LiDAR survey data is increasingly used to plan and validate fibre deployment routes — and because LiDAR data processing at scale depends on high-bandwidth network connectivity to cloud compute infrastructure.

LiDAR (Light Detection and Ranging)

A remote sensing technology that measures distance by emitting laser pulses and recording the time it takes for each pulse to return after reflecting off an object. By collecting millions of these measurements per second, a LiDAR sensor constructs a three-dimensional point cloud of the surrounding environment. LiDAR is used in autonomous vehicles, aerial mapping, defence imaging, industrial automation, forestry, agriculture, and dozens of other applications.

M

MEMS Mirror

A micro-electromechanical system (MEMS) mirror is a tiny oscillating mirror used to steer a laser beam across the field of view. MEMS-based beam steering offers a balance between the long range of mechanical scanning and the compact form factor of solid-state systems. Several automotive LiDAR companies use MEMS mirrors as their primary scanning mechanism.

Mesh

A 3D surface model constructed from a point cloud by connecting adjacent points into triangular or polygonal faces. Meshes provide a continuous surface representation that can be textured with imagery for photorealistic visualisation. Mesh generation is a standard post-processing step in photogrammetry and LiDAR workflows for architectural, industrial, and cultural heritage applications.

Mobile Mapping

Collecting geospatial data — typically LiDAR point clouds and imagery — from sensors mounted on a moving vehicle. Mobile mapping systems combine LiDAR scanners, cameras, GNSS, and IMU to capture high-density 3D data of road corridors, rail lines, and urban environments at driving speed. The technology is fundamental to HD map creation for autonomous vehicles.

Multi-Return

The ability of a LiDAR system to record multiple return signals from a single transmitted pulse. When a laser pulse passes through semi-transparent objects (like tree canopy), it reflects off multiple surfaces at different ranges. Multi-return capability allows the system to record all of these reflections — capturing both the canopy surface and the ground beneath it in a single scan.

N

NDVI (Normalised Difference Vegetation Index)

A spectral index used to assess vegetation health. While NDVI is traditionally derived from multispectral imagery, LiDAR intensity data at specific wavelengths can complement NDVI analysis — particularly in forestry and precision agriculture where canopy structure (from LiDAR) and chlorophyll content (from NDVI) are both needed.

Noise (Point Cloud)

Unwanted or erroneous data points in a LiDAR point cloud caused by atmospheric particles, multipath reflections, sensor artefacts, or electronic noise in the detector. Noise filtering is a critical step in point cloud processing — sophisticated algorithms distinguish between valid returns and artefacts based on spatial consistency, intensity, and statistical patterns.

O

Object Detection

The AI task of identifying and localising objects within a LiDAR point cloud — vehicles, pedestrians, cyclists, buildings, signs, lane markings, etc. Object detection is the core perception function in autonomous driving. Modern detectors use deep learning models trained on millions of labelled point cloud frames, typically running on GPU-accelerated edge hardware inside the vehicle.

OPA (Optical Phased Array)

A solid-state beam steering technology that directs a laser beam by controlling the phase of light emitted from an array of closely spaced waveguides or emitters. OPAs have no moving parts — the beam is steered entirely electronically, enabling extremely fast scan rates and programmable scan patterns. OPA-based LiDAR is still emerging but is considered a potential long-term replacement for mechanical and MEMS scanning architectures.

Orthophoto

An aerial photograph that has been geometrically corrected to remove lens distortion and terrain displacement, producing a uniformly scaled image where measurements can be taken directly. Orthophotos are frequently combined with LiDAR-derived elevation models to create accurate, visually rich 2D maps of large areas.

P

Perception (Autonomous Systems)

The ability of a machine to understand its environment using sensor data. In autonomous vehicles, perception encompasses object detection, tracking, classification, lane detection, free-space estimation, and prediction of other road users’ behaviour. LiDAR is typically the primary geometric input to the perception stack, fused with camera and radar data.

Photogrammetry

The science of making measurements from photographs. In 3D mapping, photogrammetry uses overlapping images captured from multiple angles to reconstruct the geometry and texture of a scene. Structure from Motion (SfM) algorithms compute camera positions and generate dense point clouds or meshes. Photogrammetry is often used alongside LiDAR — the LiDAR provides geometric accuracy while photogrammetry provides colour and texture.

Photonic Integrated Circuit (PIC)

A chip-scale device that integrates multiple optical functions — laser generation, beam splitting, modulation, detection — onto a single semiconductor substrate. PICs are enabling the next generation of miniaturised, low-cost LiDAR by replacing discrete optical components with monolithic integration. Silicon photonics PICs are of particular interest for FMCW LiDAR architectures.

Point Cloud

A dataset consisting of millions or billions of individual points, each with 3D coordinates (XYZ), and typically additional attributes like intensity, return number, classification, and timestamp. Point clouds are the primary output of LiDAR scanning and the foundation for all downstream applications — from terrain modelling to autonomous navigation to BIM.

Point Density

The number of LiDAR points per square metre in a dataset. Higher point density provides finer spatial detail and enables detection of smaller features. Typical densities range from 1–4 points/m² for wide-area aerial surveys to hundreds or thousands of points/m² for terrestrial and mobile mapping. Point density is determined by scan rate, flying height, speed, and overlap.

PPK (Post-Processed Kinematic)

A GNSS positioning technique where raw satellite observations are recorded during data collection and processed afterwards against a base station to achieve centimetre-level accuracy. PPK is the standard positioning method for drone-based LiDAR and photogrammetry where real-time RTK corrections are impractical or unreliable.

R

Range

The maximum distance at which a LiDAR sensor can reliably detect and measure a target. Range depends on laser power, detector sensitivity, target reflectivity, and atmospheric conditions. Specifications typically state range for a target with 10% reflectivity (e.g. dark asphalt) and 80% reflectivity (e.g. white surface). Range is one of the most important differentiators between LiDAR architectures.

Range Resolution

The minimum distance separation at which a LiDAR system can distinguish two objects along the same line of sight. Range resolution is determined by the pulse width (in ToF systems) or the chirp bandwidth (in FMCW systems). Finer range resolution enables detection of thin structures, closely spaced objects, and multi-layer vegetation.

Registration

The process of aligning multiple point clouds captured from different positions into a single, unified coordinate system. Registration uses overlapping features, control points, or IMU/GNSS data to compute the transformation between scans. Accurate registration is essential for building complete 3D models from multiple scan positions — whether from a tripod-mounted scanner, a mobile mapping vehicle, or a drone.

RTK (Real-Time Kinematic)

A GNSS positioning technique that provides centimetre-level accuracy in real time by using correction data from a nearby base station or network. RTK is used in terrestrial LiDAR surveying, mobile mapping, and drone operations where the operator needs immediate confirmation of positional accuracy during field work.

S

Scan Rate

The number of individual range measurements a LiDAR sensor makes per second. Modern automotive LiDAR sensors typically achieve 300,000 to 3 million points per second. Survey-grade aerial systems can exceed 2 million pulses per second. Higher scan rates enable denser point clouds at a given travel speed.

Scan-to-BIM

The workflow of converting a LiDAR point cloud of an existing building or facility into a Building Information Model (BIM). Scan-to-BIM involves registration, cleaning, classification, and manual or semi-automated modelling of architectural, structural, and MEP elements. It is the standard method for creating accurate as-built BIM models for renovation, retrofit, and facility management projects.

Sensor Fusion

The process of combining data from multiple sensor types — LiDAR, camera, radar, ultrasonic, IMU — into a single coherent perception model. No individual sensor is sufficient for safety-critical autonomous systems. Fusion algorithms weight each sensor’s strengths (LiDAR for geometry, cameras for colour and classification, radar for velocity and weather penetration) to produce a perception output more robust than any sensor alone.

SiPM (Silicon Photomultiplier)

A solid-state photon-counting detector made from an array of SPAD elements on a silicon substrate. SiPMs offer single-photon sensitivity at the 905nm wavelength and can be manufactured using standard CMOS processes — making them a strong candidate for low-cost, high-volume automotive LiDAR receivers.

SLAM (Simultaneous Localisation and Mapping)

An algorithm that allows a sensor platform to build a map of an unknown environment while simultaneously tracking its own position within that map. SLAM is fundamental to indoor mapping, handheld LiDAR scanners, and autonomous robots operating in GPS-denied environments. LiDAR-based SLAM uses point cloud matching to compute relative motion between frames.

Solid-State LiDAR

A LiDAR system with no mechanical moving parts. Solid-state architectures include flash, OPA, and some MEMS-based designs. The primary advantages are reliability (no spinning components to wear out), compact form factor, and suitability for automotive-grade vibration and temperature requirements. The trade-off is typically reduced range and FOV compared to mechanical systems.

SPAD (Single-Photon Avalanche Diode)

A photodetector designed to detect individual photons with picosecond timing precision. When a single photon is absorbed, it triggers an avalanche of charge carriers that produces a measurable electrical pulse. SPAD arrays — made in silicon (for 905nm) or InGaAs (for 1550nm) — are the detector technology behind both Geiger-mode LiDAR and next-generation direct time-of-flight (dToF) sensors.

SfM (Structure from Motion)

A photogrammetry technique that reconstructs 3D geometry by analysing the apparent motion of features across a sequence of 2D images captured from different viewpoints. SfM algorithms compute camera positions and a sparse point cloud, which is then densified and meshed. SfM is the computational engine behind most modern photogrammetry software.

T

TDC (Time-to-Digital Converter)

A circuit that converts the time interval between a laser pulse emission and its return detection into a digital value. TDC precision — typically measured in picoseconds — directly determines the range resolution of a time-of-flight LiDAR system. High-performance TDCs are implemented on ASICs co-packaged with detector arrays.

Terrestrial Laser Scanning (TLS)

Capturing 3D point clouds using a stationary, tripod-mounted laser scanner. TLS produces extremely dense and accurate data — typically millimetre-level — and is used for building survey, industrial plant documentation, forensic scene capture, and heritage preservation. Multiple scan positions are registered together to create a complete 3D model of the site.

Time-of-Flight (ToF)

The most common LiDAR measurement principle. A laser pulse is emitted, reflects off a target, and returns to the detector. The round-trip time is measured and converted to distance using the speed of light. ToF systems can be pulsed (discrete pulses) or continuous (phase-based). Pulsed ToF is the dominant architecture in automotive and aerial LiDAR.

U

UAV / UAS (Unmanned Aerial Vehicle / System)

A drone used as a platform for LiDAR sensors, cameras, and other payloads. UAV-based LiDAR has transformed surveying, inspection, and mapping by providing rapid, safe, and cost-effective data acquisition for areas that are difficult or dangerous to access on foot. Regulations vary by jurisdiction — the system (UAS) includes the aircraft, ground control station, and communication links.

V

VCSEL (Vertical-Cavity Surface-Emitting Laser)

A semiconductor laser that emits light vertically from its surface, enabling dense 2D array fabrication on a single wafer. VCSELs are the dominant illumination source for 905nm LiDAR systems because they can be manufactured at extreme scale using standard GaAs processes. VCSEL arrays provide the structured illumination for flash LiDAR and the scanning illumination for many solid-state and MEMS-based architectures.

Voxel

A volumetric pixel — a unit cube in 3D space. In point cloud processing, voxelisation converts irregular point data into a regular 3D grid, simplifying storage, search, and computation. Voxel representations are widely used in autonomous vehicle perception and 3D deep learning, where convolutional neural networks process voxelised point clouds for object detection and segmentation.

W

Waveform LiDAR

A LiDAR system that records the full shape of the returned signal over time rather than just discrete peaks. Full-waveform data captures information about the vertical structure of the target — the density of vegetation layers, the slope of terrain within the beam footprint, or the presence of semi-transparent surfaces. Waveform LiDAR provides richer information than discrete-return systems but generates significantly larger datasets.

Z

Zenith Angle

The angle measured from the vertical (directly overhead) to a line pointing toward a specific direction. In LiDAR surveying, zenith angles are used to describe the orientation of scan lines and the geometry of aerial acquisition. A zenith angle of 0° points straight up; 90° points to the horizon.

Z-Coordinate

The vertical component of a point’s position in a 3D coordinate system. In LiDAR point clouds, the Z-coordinate typically represents elevation above a reference surface (e.g. sea level, geoid, or local datum). Z-accuracy is critical for terrain modelling, flood analysis, and construction grade verification.