10.1 Glossary

How closely a point cloud's position aligns with real-world geographic coordinates. Requires external control: RTK, PPK, or surveyed ground control points. The L2 Pro and K1 achieve 3.0 cm (1.2 in) RMSE absolute accuracy when RTK or GCPs are applied correctly. A scan with no georeferencing has no defined absolute accuracy because it has no connection to a coordinate system.

A workflow that combines drone-captured imagery with ground-level scan data into a single 3D model. Requires LCC Scan app version 1.2.0_p1 or higher and firmware V3.2.3 or higher. The drone and ground scanner share connection points collected at takeoff and landing locations to establish a shared reference frame.

The electronics inside each Lixel battery that monitor cell voltage, temperature, and state of charge. The BMS draws a small amount of power even when the device is off. If a battery is stored below 10% charge, the cells can drop below the BMS safety floor voltage and become unchargeable. Store batteries at 40 to 60% charge for any storage period exceeding one week.

A fixed GNSS base station that broadcasts real-time correction data over the internet. CORS networks allow RTK positioning without setting up your own base station. State-operated networks (such as the Utah Geospatial Reference Network) provide free access. Commercial networks (Trimble VRS, Swift Skylark) require a subscription. CORS data is delivered to the scanner via NTRIP.

The mathematical framework that defines how coordinates relate to locations on Earth. Each CRS is identified by an EPSG code (for example, NAD83 UTM Zone 12N is EPSG:26912). The CRS must be set in LixelStudio before exporting georeferenced data. A mismatch between the scanner's CRS and the receiving software's CRS offsets the dataset by meters or more.

The gradual divergence between the scanner's estimated position and its true position, caused by accumulated small errors in IMU and odometry calculations. Drift is the fundamental challenge of all SLAM-based scanning. It compounds over distance and cannot be eliminated, only managed through loop closure, RTK, and ground control points. Feature-poor environments (long corridors, tunnels, open fields) accumulate drift faster than feature-rich ones.

An L2 Pro operating mode in which the scanner is mounted to a DJI M300 RTK, M350 RTK, or M400 MTX drone instead of carried by hand. In Drone Mode, the drone provides power and RTK positioning. The L2 Pro's own RTK module must be removed before mounting. Data is processed in LixelStudio as a point cloud, not a 3DGS model. Recommended flight parameters: 3 m/s (10 ft/s) speed, 30 m (98 ft) altitude.

An open file format for storing 3D point cloud data, standardized by ASTM International. E57 files embed coordinate system metadata and are accepted by most professional AEC software including Autodesk ReCap, Leica Cyclone, and Bentley ContextCapture. LixelStudio supports E57 export. Critical limitation: E57 and LAS filenames longer than 20 characters cause a silent export failure in LixelStudio — the software completes without an error message but the output file is corrupt or missing.

A numeric identifier for a specific coordinate reference system, maintained by the EPSG Geodetic Parameter Dataset. Used to unambiguously specify the CRS when exporting or importing georeferenced point cloud data. Common US examples: NAD83 UTM Zone 12N = EPSG:26912 (Utah, Colorado, New Mexico), NAD83 UTM Zone 13N = EPSG:26913 (eastern Colorado, Wyoming), WGS84 = EPSG:4326 (global GPS default).

An RTK positioning state in which the receiver has satellite lock but has not resolved the integer ambiguity in the carrier phase measurement. Float accuracy is typically 20 to 50 cm (8 to 20 in), insufficient for engineering-grade georeferencing. Only Fixed solution status is acceptable for RTK-based absolute positioning. Float commonly occurs when fewer than 10 satellites are visible, HDOP exceeds 2.0, or NTRIP corrections are not arriving.

An RTK positioning state in which the receiver has resolved the integer ambiguity in carrier phase measurements, achieving centimeter-level accuracy. Fixed is the only RTK status suitable for engineering-grade absolute positioning. LixelGO displays the current RTK status. Minimum conditions for a stable Fixed solution: 10 or more satellites, HDOP below 1.5, NTRIP corrections arriving with age below 1 second.

The angular range a sensor can detect at any given moment. The L2 Pro LiDAR covers 360° horizontal and 270° vertical. The K1 LiDAR covers 360° horizontal and 59° vertical (from -7° to +52°). A narrower vertical FOV means features above or below that range are not captured. The K1's recommended 15-20° downward tilt during scanning improves ground coverage but reduces upward FOV available for ceiling-mounted infrastructure.

A physical marker placed at a surveyed location with known real-world coordinates. GCPs constrain SLAM drift across a dataset and provide absolute accuracy when RTK is unavailable or insufficient. Each GCP is scanned during the field session and its coordinates are entered in LixelStudio during processing. GCPs must be non-collinear (not all in a straight line) and distributed spatially across the project area. Minimum recommended count: 3 to 4 for small sites under 10,000 sq ft (929 sq m), 6 to 8 for larger sites.

The equipotential surface of Earth's gravity field that approximates mean sea level. The geoid is irregular because Earth's mass is unevenly distributed. The difference between an ellipsoidal height (GPS-derived) and an orthometric height (elevation above sea level) is called the geoid undulation or height anomaly. In the US, the GEOID18 model is the current standard. LixelStudio requires this value when converting GPS heights to real-world elevations.

The process of establishing a mathematical relationship between a point cloud's local coordinate frame and a real-world coordinate system. Georeferencing requires external control: RTK, PPK, or surveyed GCPs. A georeferenced point cloud can be overlaid with survey data, registered to other datasets, and used for area and volume calculations tied to real-world coordinates. Georeferencing must be configured before the scan begins. It cannot be applied retroactively without control that was captured during the scan.

A rendering technique that represents a scene as millions of semi-transparent, elliptical 3D shapes called Gaussian splats. Unlike point clouds, 3DGS produces photorealistic visual output optimized for real-time viewing and web delivery. LCC Studio processes scan data into 3DGS models. The output is a visual asset, not a geometric measurement tool. 3DGS files cannot be converted into point clouds after processing. Choosing 3DGS versus point cloud is an irreversible pipeline decision that must be made before the scan begins.

A registration algorithm that aligns two point clouds by iteratively minimizing the distance between corresponding points. Used in LixelStudio for fine-tuning the registration between segments after an initial alignment. ICP can fail when the initial alignment is too far off (local minima) or when the overlap zone lacks sufficient unique geometry. Always verify residuals after ICP registration. Results above 2 cm (0.8 in) RMSE indicate a failed or poor alignment that requires manual intervention.

A sensor package containing accelerometers and gyroscopes that measures the scanner's acceleration and rotational rate. The IMU provides motion data between LiDAR scan frames, allowing the SLAM system to estimate position even when the scanner is moving. Rapid rotations, shaking, or swinging motions can saturate the IMU, causing SLAM tracking to degrade or fail. The IMU also determines the scanner's level, introducing a small angular error (approximately 0.07°) that compounds into elevation discrepancy over large areas.

The startup sequence during which the scanner establishes its starting position and stabilizes the IMU before a scan begins. For L2 Pro and K1: hold the scanner stationary on a flat, level surface for the countdown period (typically 15 seconds) plus an additional dwell period until the point cloud preview appears in LixelGO. For PortalCam: use a tripod or flat surface for the first 25 seconds. Moving the scanner during initialization causes trajectory errors that propagate through the entire scan session.

Ingress Protection rating, a standardized classification of a device's resistance to solid particles and liquids. The L2 Pro and K1 are rated IP54: the 5 indicates protection against dust particles large enough to affect operation, and the 4 indicates protection against water splashing from any direction. IP54 permits use in light rain and dusty construction environments but not immersion or heavy water exposure. The PortalCam carries the same IP54 rating.

LAS is the industry-standard binary file format for lidar point cloud data, defined by ASPRS. Each point stores X, Y, Z coordinates, intensity, return number, and optional color. LAZ is a lossless compressed version of LAS, typically 80 to 90% smaller. LixelStudio exports both. Use LAS or LAZ for survey deliverables, archive storage, and import into CloudCompare, Global Mapper, or ArcGIS Pro. LAS filenames longer than 20 characters cause a silent export failure in LixelStudio.

A remote sensing technology that measures distances by emitting laser pulses and timing the return. In the XGRIDS devices, rotating LiDAR sensors emit Class 1, 905 nm laser pulses to capture hundreds of thousands of range measurements per second. Combined with the scanner's position data from the SLAM system, each range measurement becomes a 3D point in space. The result is a point cloud representing the measured environment.

The event that occurs when the SLAM system detects that the scanner has returned to a previously visited area. At the point of recognition, the algorithm measures the accumulated drift error and distributes a correction back across the trajectory. Loop closure is the primary mechanism for keeping a scan accurate over long distances. Two conditions must both be met: you must physically return to a previously scanned area, and your viewing angle must be within 40 degrees of your original angle at that location. Walking back through a room facing the opposite direction may not trigger a valid closure.

The processing workflow that aligns multiple scan segments into a single registered dataset. Used when a project requires more than one scan session due to battery limits, RAM limits, site access constraints, or area size. Hard limits that apply to all fusion jobs: maximum 10 segments, maximum 200 minutes total scan duration, maximum 20 minutes per segment, same device model across all segments, minimum 15 m (49 ft) of shared overlap between adjacent segments. Adjacent segments must connect through RTK alignment or matching control point names.

XGRIDS' implementation of simultaneous localization and mapping that fuses data from three sensor types: LiDAR, visual cameras, and the IMU. Each sensor type contributes different strengths. LiDAR provides geometric structure. The cameras provide visual texture and feature tracking. The IMU provides continuous motion data between frames. Fusing all three makes the system more robust than any single sensor alone, particularly in environments where one sensor type is unreliable (for example, low-light environments where visual tracking degrades but LiDAR remains reliable).

A LixelStudio processing mode optimized for tunnels, mine shafts, and long corridors where the standard SLAM algorithm performs poorly due to the repetitive, low-feature geometry. Must be selected before processing begins. Using Narrow Scene Mode on standard environments causes processing failure. Using standard mode on tunnels produces severe drift. The mode must be chosen based on where the scan was collected, not as a retry option when standard processing fails.

The standard protocol for streaming GNSS correction data from a CORS network to a rover over the internet. LixelGO uses NTRIP to connect to an RTK correction source. Configuration requires a caster IP address, port (standard: 2101), mountpoint name, username, and password. Corporate WiFi networks commonly block port 2101. If RTK stays in Float status after entering valid credentials, switch to a phone hotspot before troubleshooting the credentials themselves.

A flexible point cloud and mesh file format used for 3DGS output from LCC Studio. PLY files from LCC Studio contain the Gaussian splat attributes (position, color, opacity, scale, rotation) needed to render a 3DGS scene in compatible viewers. PLY files from 3DGS workflows are substantially larger than LAS files from equivalent scan areas. Not suitable as a survey deliverable. Used for visualization, web embedding via Cesium 3D Tiles conversion, and integration into game engines.

A collection of individual 3D points, each with X, Y, and Z coordinates and optional attributes such as intensity, color, and return number. Point clouds from the L2 Pro and K1 are generated by combining LiDAR range measurements with the scanner's SLAM-estimated position. The resulting dataset represents the physical geometry of the scanned environment. Point clouds are the standard deliverable format for survey, BIM coordination, and engineering documentation workflows.

A GNSS positioning method in which the rover logs raw satellite observations during the field session and a base station simultaneously logs raw observations at a known location. After the session, the two observation files are processed together to compute centimeter-level positions. PPK produces accuracy comparable to RTK but does not require a live internet connection during the scan. Suitable for remote sites, underground areas, or locations with poor cellular service. Requires a GNSS base station with raw data logging capability and post-processing software.

An advanced LCC Studio parameter that controls the weight applied to LiDAR-derived spatial anchors during the 3DGS training process. High PPR forces Gaussian splats to adhere closely to the LiDAR geometry. This improves geometric accuracy but can create visual artifacts in areas where the LiDAR sensor and cameras have a parallax-induced misalignment, because the splats are forced into positions that conflict with what the cameras actually observed. Treat as a debug option rather than a first-pass setting.

How accurately the point cloud represents the internal geometry of the scanned space: dimensions, distances, and shapes. The L2 Pro achieves 1.0 cm (0.4 in) RMSE relative accuracy. The K1 achieves 1.2 cm (0.5 in) RMSE. These figures apply to point-to-point distances within approximately 100 m (328 ft) of path segments with proper technique and loop closure. Relative accuracy is independent of georeferencing and does not require RTK or GCPs to achieve.

The statistical measure used to express accuracy in survey and scanning contexts. RMSE is the square root of the average of squared differences between measured and true values. In XGRIDS documentation, accuracy specifications use the 3-sigma rule (3σ), meaning the published figure represents the distance within which 99.7% of measurements fall. A 3 cm RMSE means that 99.7% of absolute positioning errors are within 3 cm (1.2 in) of true position under the stated conditions.

A GNSS positioning technique that uses real-time correction data from a base station or CORS network to achieve centimeter-level absolute positioning. RTK corrections are delivered via NTRIP over a cellular or WiFi connection. Only Fixed solution status is suitable for engineering-grade georeferencing. Float status is not. RTK data is fused with SLAM positioning in LixelGO during the scan. RTK must be configured and in Fixed status before the scan begins. It cannot be applied after the fact.

A single continuous scan recording from start to stop, stored as one project file on the device. Segments are the building blocks of Map Fusion projects. Each segment must stay under 20 minutes in duration. The recommended target per segment is 12 to 15 minutes to leave buffer for unexpected coverage needs. Segments within a Map Fusion job must all come from the same device model and must have overlapping geometry at their connection zones.

An algorithm class that solves two interdependent problems at the same time: determining the scanner's position within an environment, and building a map of that environment. Position estimation requires knowing the map, and map building requires knowing the position. SLAM resolves this by continuously updating both in real time using incoming sensor data. All XGRIDS devices use Multi-SLAM, combining LiDAR, camera, and IMU inputs. The quality of a SLAM output depends on the environment's geometric and visual feature density and on the operator's scanning technique.

The path the scanner traveled during a scan session, recorded as a time-stamped sequence of positions and orientations. The trajectory is the backbone of a point cloud: each LiDAR return is placed in 3D space by combining its measured range with the scanner's position at that moment from the trajectory. A distorted or drifted trajectory produces a distorted point cloud. SLAM optimization, RTK fusion, and GCP adjustment all work by correcting the trajectory.

Memory on the GPU used for processing and rendering data. LCC Studio requires VRAM for 3DGS training and rendering. The default maximum of 25 million Gaussian splats requires approximately 8 GB VRAM to process. Larger projects or higher splat counts require more VRAM. VRAM shortage causes LCC Studio to fail mid-processing without a clear error. Using the Low-Memory Reconstruction option offloads intermediate data to disk at a processing time cost, enabling larger scenes on systems with limited VRAM.