Captures rotation only: pitch, roll, and yaw. Tells you which way something is pointing, but not where it is. Used in early VR headsets and basic compass-based systems.

Adds three dimensions of position (left/right, up/down, forward/back) on top of rotation. The full picture of where something is and how it is oriented, corresponding to latitude, longitude, and altitude in geospatial terms. Required for most modern AR and robotics applications.

A real-world location that has been mapped, indexed, and made ready for geospatial applications. The foundation of location-based AR experiences, precise navigation, and any application requiring reliable spatial context. The denser and more accurate the underlying data, the richer the experience.

Fixing a digital object to a specific location in the real world.

Overlays digital content onto the real world, visible through a phone camera or headset. The physical environment remains the primary experience; digital elements like labels or characters are added on top. AR headsets are considered to have transparent lenses, unlike MR or VR headsets.

A visual representation of geographic information designed to communicate spatial relationships, features, and data about a specific area. Uses symbols, colors, and scale to depict terrain, boundaries, and other relevant details for navigation, analysis, or storytelling.

Two ends of a spectrum describing positional accuracy. Coarse location, derived from GPS, cell towers, or Wi-Fi, places a device within a few meters at best. Precise location, the kind VPS delivers, can resolve position down to centimeters and includes accurate orientation, not just coordinates.

A virtual replica of a real-world object, environment, or system that stays synchronized with its physical counterpart over time. Unlike a static 3D model, a digital twin reflects changes in the real world. Used across manufacturing, infrastructure, urban planning, and facility management to analyze current reality and model potential changes.

A 3D scene representation using volume rather than surfaces. Composed of millions of fuzzy blobs (Gaussians), each carrying position, size, orientation, and color. Splats capture lighting nuances like translucency and view-dependent color shifts well, and render in real time on modern phones, computers, and VR headsets. Particularly efficient to generate from raw scan data.

Technology that captures, analyzes, and acts on location data tied to the real world. Encompasses GPS navigation, satellite imagery, and augmented reality. Modern platforms are designed to make the real world legible to machines, powering autonomous robotics and high-precision wayfinding.

Data with three dimensions plus a spatial component — information that can be played relative to a user and what the user is looking at in augmented reality. More than just adding 3D.

Systems for capturing, storing, analyzing, and presenting spatial and geographic data.

The umbrella term for all satellite-based positioning systems worldwide, including GPS (US), GLONASS (Russia), Galileo (Europe), and BeiDou (China). Modern devices typically use signals from multiple constellations simultaneously to improve accuracy and reliability.

A satellite network that allows devices to determine their location anywhere on Earth by measuring signals from multiple satellites simultaneously. Accuracy is typically within a few meters; struggles indoors, underground, and in dense urban environments.

The horizontal direction an object is facing, measured as an angle relative to north: the compass direction, independent of tilt. Can be reported relative to either true or magnetic north; consistency matters when integrating multiple systems.

An electronic device that measures and reports specific force, angular rate, and sometimes magnetic field, using accelerometers, gyroscopes, and magnetometers. In XR, IMUs track user motion and orientation in 3D space, vital for low-latency tracking of head and device movements. Often fused with visual data via Visual Inertial Odometry (VIO).

The two coordinates that together pinpoint any location on Earth's surface. Latitude measures north-south position between the equator and the poles; longitude measures east-west position relative to the prime meridian in Greenwich, London. Nearly every geospatial system anchors its data to lat/long at some level.

A large continuous geographic region processed as a unified spatial asset.

A form of AI that understands the physical world from training on massive amounts of imagery and reconstructions. Like a large language model can recognize and generate language, an LGM can interpret live scenes and anticipate what is not yet seen — for example, estimating the appearance of a newly constructed building from a refreshed satellite image.

Large Geospatial Model: Localization

Large Geospatial Model: Reconstruction

Large Geospatial Model: Semantics

The process of determining where something is in the world: a device, robot, or person. Absolute localization pins a position to global coordinates; relative localization determines position in relation to nearby objects, which can afford higher precision. Not to be confused with l10n, the practice of adapting products for specific markets.

The processing, aggregation, and transformation of raw scanned data into structured, spatially accurate 3D assets: maps, meshes, or splats. Enables key AR capabilities like VPS-based localization. The pipeline performs cloud or device-based reconstruction, with QA and activation ensuring only validated assets reach real-world applications.

A 3D representation of a surface made up of connected vertices, edges, and faces, typically triangles. The most common format for 3D graphics and game engines, widely used to represent the shape of real-world environments. Captures geometry rather than appearance, making meshes well suited for physics, collision detection, and spatial reasoning.

Goes further than AR by anchoring digital objects so they interact with the real world: like a virtual ball rolling across a real table, or a digital interface attaching to a real wall. MR headsets use front-facing cameras to capture the world from the user's perspective and combine it with digital content.

The fundamental unit of spatial map data. Unlike POIs (which are geographic with lat/long but no spatial information), nodes are spatial, with lat/long, an optional POI, and a localizability score indicating how likely a user can localize there. A node has slots for mesh, VPS map, and splat assets.

The up-down tilt of an object along its side-to-side axis: like the nose of an airplane rising or dipping. Matters when a device is angled, since a phone held flat versus held at eye level reads the world very differently.

A collection of data points in 3D space, each representing a location on the surface of a real-world object or environment. Typically the raw output of LiDAR sensors or photogrammetry pipelines, and the starting material for reconstruction and mesh generation. Widely used in surveying, construction, and precision spatial measurement.

The complete description of an object's position and orientation in space: where it is, which way it is facing, and how it is tilted. Essential for placing digital content accurately in the real world, or for a robot to understand how to interact with its environment.

The process of building a 3D model of a real-world scene from images, video, or sensor data. Algorithms infer geometry and appearance by analyzing differences and similarities between frames. LiDAR data significantly improves accuracy, particularly for scale and depth.

Rotation along the front-to-back axis: like an airplane banking left or right. Often the trickiest axis to resolve accurately, since small amounts of roll can meaningfully shift where digital content appears in an AR scene.

Capturing raw data from a physical environment using devices like mobile phones or drones, where users move through a space to record visual and spatial information. The goal is to efficiently and reliably collect input data for later processing.

Capturing not just what a space looks like geometrically, but understanding what is in it. A mature semantic model can recognize a chair it has never seen or a doorway it has never mapped, generalizing from training data. Crucial for robotic manipulation, indoor navigation, and self-driving cars.

Algorithm that maps an environment and tracks the device's location within it simultaneously. Essential for AR and MR applications.

Information that is anchored to and presented at a specific point of an object, location, or area in the world.

The process of scanning and creating a digital map of the physical environment for interaction with digital objects, localization, and navigation.

Technology that monitors the position and orientation of a headset, controllers, or user's body in real time.

Determines device location by comparing live camera input to imagery with known location, often supplemented by GPS. More accurate than GPS alone, especially indoors, underground, in urban canyons, and in GPS-denied environments. Requires either an internet connection or onboard reference data with local compute.

Applies VPS technology within business contexts, allowing companies to create, modify, and interact with 3D spatial data tailored to their needs, including detailed maps of environments like warehouses, stadiums, and other private spaces.

Replaces the real world entirely with a simulated environment experienced through a headset. Fully immersive and fully synthetic: the user moves in the physical world but sees only the digital one. VR headsets traditionally do not use cameras for the experience itself.

A machine-readable representation of an environment: a dynamic, queryable understanding of what exists and where. In geospatial computing, a continuously updated, reasoned-over representation of the real world as it actually is. Foundational to spatial AI and autonomous robotics.

Provides developers with the 3D position and orientation of a device in geographic coordinates. Combines GPS, compass heading, and AR tracking inputs from ARKit/ARCore for greater accuracy and frame-to-frame stability than GPS and compass alone. Requires the device camera, with results improving as the user moves through the scene.

The umbrella term for any technology that extends, augments, or replaces physical reality with digital content. Shorthand for the entire spectrum from AR to VR and everything in between.

Rotation along the vertical axis: like a spinning top, or a person turning left and right on the spot. Closely related to heading, but yaw describes the rotation itself while heading describes the resulting direction. GPS can only infer yaw from movement; VPS resolves it visually from a single camera frame.