Blog · 2026-08-17
iPhone LiDAR Indoor Mapping: The Industries Using It and the Accuracy They Get
iPhone LiDAR indoor mapping is the capture of measured digital maps of interior spaces - floor plans, 3D meshes, parametric room models - using the laser scanner built into iPhone Pro and iPad Pro devices. A room takes minutes to scan and holds centimeter-class accuracy at close range. Construction, real estate, insurance, facilities management, and warehousing all run it in production.
That is the capability. The more useful questions are who actually runs it in production, what accuracy they get, and what stands between a raw scan and a map another system can use. This post answers all three, with sourced numbers.
How does iPhone LiDAR map indoor spaces?
The sensor is a direct time-of-flight (dToF) infrared laser array that measures true hardware depth, fused with the camera stream and inertial sensors through visual-inertial odometry. Indoors this fusion matters more than outdoors: it initializes instantly, handles the blank white walls that defeat camera-only AR, and keeps working in dim back rooms and unlit basements.
Developers get three layers to build on. The Depth API returns per-pixel depth in meters. Scene reconstruction returns a live 3D mesh with surface classification - floor, wall, ceiling, seat. And RoomPlan returns a parametric room model - walls, doors, windows, furniture - with dimensions attached. One nuance worth knowing: the laser itself works in total darkness, but RoomPlan's camera pipeline needs at least 50 lux, per Apple's documentation. The full production envelope of that API is covered in our RoomPlan deep dive.
Which industries use iPhone LiDAR indoor mapping?
Six industries run iPhone LiDAR indoor mapping in production today: construction and AEC, real estate, insurance and restoration, facilities management, interior design and renovation, and warehousing. Accessibility - indoor navigation for blind and low-vision users - is the emerging seventh, still waiting for B2B packaging. Each replaces a slower manual measurement workflow with a scan that takes minutes.
| Industry | What gets mapped | Best-sourced number |
|---|---|---|
| Construction and AEC | As-builts, scan-to-CAD/BIM | 2 h measuring + 1.5 h data entry cut to ~30 min of scanning (Twindo customer case) |
| Real estate | Listing floor plans | 5M+ floor plans from 5-10 minute scans (CubiCasa, company-reported) |
| Insurance and restoration | Interior damage, estimates | On-site estimates cut from 2-3 h to ~30 min (magicplan customer cases) |
| Facilities and digital twins | Whole-building records | 50.7B sq ft under management (Matterport, public filings) |
| Interior design and renovation | Existing conditions, quoting | ~50% time savings (magicplan customer case) |
| Warehousing | Parcel and pallet dimensions | Avoids $15,000-80,000 per fixed dimensioning station |
| Accessibility | Indoor navigation for blind users | Apple ships Door and People Detection; B2B packaging remains an open gap |
The full vertical-by-vertical breakdown, with the economics behind each, lives in the 2026 industry map. The pattern that repeats across every row: the business is the deliverable - the floor plan, the CAD file, the claim package - never the scan itself.
How accurate is iPhone LiDAR for indoor mapping?
A peer-reviewed study in Nature Scientific Reports measured iPhone LiDAR at ±1 cm on objects larger than 10 cm (Luetzenburg, Kroon & Bjørk, 2021). Surfaces at 0.3-3 m converge on 1-2 cm across studies and practitioner benchmarks, and at building scale, iPhone 13 Pro comparison studies report 10-20 cm at 95% confidence.
| Scenario | Measured accuracy | Source |
|---|---|---|
| Objects larger than 10 cm, room scale | ±1 cm | Peer-reviewed: Luetzenburg, Kroon & Bjørk, Nature Scientific Reports 2021 |
| Surfaces at 0.3-3 m indoors | 1-2 cm | Converging result across studies and practitioner benchmarks |
| Whole-building indoor scans | 10-20 cm at 95% confidence | iPhone 13 Pro comparison studies |
A room scan and a whole-floor scan are different measurement problems - indoor error accumulates with scale, and no single number describes both.
The variable that dominates indoors is not the sensor but the app. An independent benchmark scanned identical scenes on identical hardware through two popular apps and measured a 5 cm mean deviation for Polycam against 44 cm for Scaniverse - roughly a 9x spread from software alone. The reason is drift: the phone estimates its own motion frame by frame, and indoors there is no GNSS signal to correct it, so registration, loop closure, and outlier filtering carry the entire accuracy budget. The full scenario-by-scenario table is in our accuracy roundup.
What are the limits of indoor mapping on a phone?
Four constraints are physics and platform, not app quality:
- Glass and mirrors. Indoor spaces are full of the surfaces LiDAR handles worst - glass partitions, mirrored wardrobes, polished floors. The laser scatters or passes through and records false depth behind the real surface. Production apps segment and prune these regions rather than trusting them.
- Thin objects. Point spacing of several millimeters to a centimeter means cables, table legs, and pipe runs blur or vanish from the mesh.
- One room per session. RoomPlan is scoped to roughly 9x9 m and sessions under about five minutes (Apple's documented guidance) before thermal limits bite; whole-floor jobs are a segment-and-stitch problem, and multi-room registration is where indoor mapping products earn or lose their accuracy claims.
- Pro-only hardware. LiDAR ships on iPhone Pro models (since 2020) and iPad Pro only. The common pattern in working deployments is a two-tier design: LiDAR capture for field staff, universal viewing for everyone else.
What turns a scan into a usable indoor map?
A pipeline, and it is most of the engineering. Raw capture needs cleanup - downsampling, outlier removal, surface reconstruction - and semantic correctness: wall thickness, openings, which side of a wall a surface belongs to. The result then has to export into the formats the destination system accepts - DXF, IFC, and RVT for CAD and BIM, E57 for point clouds, USDZ for review. A mesh that looks right on screen and imports wrong into Revit is a failed map. We walk that conversion gap in the scan-to-CAD/BIM post.
Our team ships a production LiDAR scanning app for a surveying technology company - iPhone capture with a per-frame constraint engine that rejects frames on velocity, movement delta, and GNSS fix quality, rendering live point clouds up to 15 million points at 60fps. Outdoors, an RTK correction stream anchors the scan to sub-5 cm georeferenced accuracy. The indoor lesson from that work is blunt: the moment you step inside, the GNSS anchor disappears, and everything rides on drift discipline - quality gating at the frame level, anchors and loop closure, and a capture flow that coaches the operator instead of trusting them. Indoor mapping is the unforgiving case of the rule that the pipeline, not the sensor, is the product.
How should you evaluate an indoor mapping build?
Three questions do most of the work. Which deliverable does the buyer already pay for - a floor plan, a BIM model, a claim package - and in which format must it land? Does centimeter-class accuracy at room scale actually cover the job, or does error at building scale break it? And who operates the capture - trained staff on managed Pro devices, or end users who need coaching built into the flow?
We build this category of system - ARKit and RoomPlan capture, Metal point-cloud rendering, drift and quality gating, and the export pipelines that make indoor data land where it earns money. If you are scoping an indoor mapping capability and want the accuracy budget or architecture pressure-tested, talk to us or start with our AR and spatial engineering services.