MultiSet AI

What MultiSet AI does

MultiSet AI provides a Visual Positioning System (VPS) that offers a shared spatial coordinate system with 5 cm accuracy and 6-DoF for various devices, including phones, smartglasses, XR headsets, drones, and robots. It is designed for indoor and outdoor environments, supporting multi-floor mapping and dynamic changes without requiring costly re-scans. The platform is scan-agnostic, accepting various input formats like LiDAR, E57, Gaussian Splats, and 360 videos, and offers flexible deployment options including cloud, private cloud, on-premise, or on-device hosting. MultiSet AI aims to be the independent spatial infrastructure layer for Physical AI, enabling precise localization and augmented reality applications across diverse industries.

Products

• MultiSet VPSA Visual Positioning System providing centimeter-accurate, low-drift 6-DoF localization for devices in physical environments.
• Unity SDKA drag-and-drop package with C# helpers and sample scenes for localizing Unity applications.
• Native iOS SDKA Swift Package Manager build that leverages LiDAR for sub-cm accuracy on iOS devices.
• Native Android SDKA lightweight AAR with Kotlin examples for Android devices.
• WebXR KitA zero-install browser runtime for launching spatial experiences from a QR code.
• Quest SDKA native Oculus package with passthrough samples for industrial AR on Meta Quest Series.
• MultiSet Mapper (iOS)An iOS application for real-time mapping and capturing 3D data.
• MapSetA feature for stitching maps across floors, buildings, and outdoor environments, providing multi-floor persistence and auto-registered re-maps.

Key capabilities

• 5 cm accuracy, low drift, instant 6-DoF relocalization
• Scan-agnostic input (LiDAR, E57, Gaussian Splats, 360 videos, textured meshes, iPhone scans, drone captures)
• Multi-floor persistence and auto-stitching of maps
• Change-tolerant for dynamic environments
• Seamless indoor ↔ outdoor transitions
• Flexible deployment (public cloud, private cloud, on-prem, on-device/offline)
• 24/7 engineer-led SLAs and SDKs for Unity, iOS, Android, WebXR, Quest
• One pipeline for any scan modality
• Infinite scale (plant, campus, city)
• One SDK for multiple outputs (mobile, headsets, smartglasses, drones, robots, browsers)
• Complete cloud choice (MultiSet Managed Cloud, Enterprise Private Cloud, On Device (Offline))
• Instant 6-DoF Lock
• Adaptive Spatial AI
• Enterprise-Grade Privacy (SSO/RBAC, encrypted private maps)
• ROS2-Ready Localization
• Multi-Map Continuity
• Semantic Safety Zones
• cm-Level Alignment for AR QA & Commissioning
• Scan-Agnostic Ingest for Construction (IFC/Revit, LiDAR/360, photogrammetry, GPR)
• Versioned Anchors
• Store-Scale Mapping for Retail
• Low-Drift Relocalization in dynamic retail environments
• Planogram-Ready (works with XML/CSV planograms, PIM, POS, ESL)
• Markerless AR Mapping, VPS & Object Tracking in One SDK
• Centimetre-Level 6-DoF + Geo HintPosition
• Instant-Launch, Zero-Install Endpoints (WebXR, App Clips, Mini Apps, Meta Quest Browser)
• Developer Toolkit & Open APIs (CLI map-uploader, REST/vps/pose, GraphQL map management)
• Map Versioning - anchors & annotations resilient over map changes
• Web-based spatial map editor
• Digital-twin repository
• Fast cold-start (edge caching)
• Shared anchors & persistence
• Multi-user / Shared AR (session consistency)
• Indoor–Outdoor Handover (GNSS geo-hint)
• Hosting Device Coverage (phones, tablets, HMDs)
• Import Gaussian Splat / PLY
• Import 360° video capture
• Import Matterport scans
• Import NavVis scans
• Import Leica scans
• Import XGRIDs scans
• Import Faro scans
• Import Polycam scans
• Import E57, meshes & MatterPak
• Outdoor GNSS (WGS‑84)
• Beacon / IoT fusion (BLE, UWB, Wi‑Fi RTT)
• Multi-device mapping (team capture & merge)
• No-code AR Authoring / Rapid Prototyping
• Seamless re-mapping (change-tolerant)

Use cases

• Front Line Operations - Instructions & Assets
• Human–Robot Ground Truth for AMRs, Drones & Cobots
• AR QA & Commissioning for Construction & Infrastructure
• Retail Planogram Compliance & Spatial Commerce
• AR work instructions
• Inspection checklists
• Indoor navigation
• Localization for robots, drones, and cobots
• Aligning BIM and as-builts with reality for layout checks
• Commissioning and utility locating in construction
• Automated planogram compliance in retail
• Faster BOPIS (Buy Online, Pick Up In Store)
• Precise AR promotions
• Warehouse navigation
• Asset inspection
• Robot SLAM hand-off

AI approach

MultiSet AI provides a Visual Positioning System (VPS) that acts as a spatial infrastructure layer for Physical AI. It enables precise 6-DoF localization for various devices (phones, smartglasses, XR headsets, drones, robots) against a shared, scan-agnostic map of the physical world. The system uses AI to process various 3D mapping data formats (LiDAR, E57, Gaussian Splats, 360 videos, textured meshes, etc.) and provides robust localization even in challenging environments. It also includes features like object tracking, multi-floor persistence, and handling dynamic environments. The company emphasizes its proprietary VPS technology and SDKs.

Tech named: Visual Positioning System, 6-DoF localization, 3D Mapping, Augmented Reality, Visual Positioning System, Location Persistence, Cloud Anchors, On Device Hosting, Self Hosting, ROS 2, Localization, Mapping, Reconstruction, 3D Gaussian Splat, 360 capture, LiDAR, E57, Gaussian Splats, textured meshes, iPhone Pro LiDAR, Matterport, NavVis, Leica, Faro, XGrids, GLB/GLTF, GNSS, RTK, IoT, UWB, BLE, Wi-Fi RTT, QR Code, BIM, IFC/Revit, photogrammetry, GPR, SKU detections, CV/SKU recognition

Industries served

• Software Development
• Industrial
• Construction
• Robotics
• Retail
• Manufacturing
• Warehousing & Logistics
• Utilities & Energy
• Data Centers
• Pharma & Biotech
• Airports & Seaports
• AEC (Architecture, Engineering, Construction)
• Grocery
• Convenience
• Pharmacy
• Big-Box
• DIY/Hardware
• Specialty Retail

What it says sets it apart

• Independently ranked the most robust VPS by AREA Research
• Scan-agnostic, no proprietary capture pipeline or vendor lock-in
• Map ops layer handles remapping, multi-floor MapSet stitching, and environment changes without taking the system offline
• Unified workflow to map any facility (multi-floor, indoor, outdoor, low light, visually identical objects)
• Anchors and annotations re-register themselves as the world changes
• Provides a neutral, privacy-controlled spatial map for humans and robots
• Supports cm-level alignment for point-of-work overlays
• Flexible hosting options: MultiSet Cloud, Private Cloud, On-Prem, On-Device (offline)
• One SDK for Unity, Web, iOS, Android, Quest, ROS2, deploying to mobile, headsets, smartglasses, drones, robots, and browsers from the same build and ground truth
• Instant 6-DoF Lock with stable, precise poses in seconds
• Adaptive Spatial AI for context-aware steps and role-based views
• Enterprise-grade privacy with SSO/RBAC and encrypted private maps
• ROS2-ready localization with client SDK and TF alignment for AMRs and drones
• Seamless yard → dock → aisle transitions with low drift
• Semantic safety zones for no-go/slow-down lanes and emergency routes
• Precise overlays suitable for tolerance checks in construction
• Versioned anchors to keep pace with weekly site changes and phases
• Store-scale mapping with bay/shelf semantics for retail
• Low-drift relocalization in dynamic, high-traffic retail aisles
• Planogram-ready, working with existing planograms, PIM, POS, and ESL
• Zero-install endpoints for WebXR & iOS App Clips to launch AR from a QR code
• Maintains sub-cm accuracy as users tilt, rotate or switch sensors
• Accepts HintPosition (WGS84 latitude/longitude) for faster, more accurate poses and reduced device CPU/memory usage
• Comprehensive developer toolkit with sample scenes, CLI map-uploader, REST/vps/pose, and GraphQL map management
• Active Discord community for support

Funding rounds we track

From the AI funding tracker — rounds as reported by the linked publications.