Skip to main content
LivestreamHear more about how Niantic Spatial is working with NVIDIA and Flexion to close the Sim-to-Real gap!Watch now
Scaniverse appCapture and reconstruct real places.価格Plans for teams and enterprises.
お問い合わせ
導入事例

Where people and machines work in the real world.

Field operator at a refinery石油・ガスIntelligent field operations. Aerial view of a coastal site, measured from drone imagery防衛Mission rehearsal and GPS-denied operations. Robots in a real-to-sim workshopEmbodied AIReal-to-sim training for robots.
Company

The team, partners and resources behind Niantic Spatial.

Niantic Spatial について
Niantic Spatial についてイベント採用情報
Partners
デベロッパー・パートナー・プログラムAcademic and community partnersパートナー
Guides and support
Scaniverse FAQ - よくあるご質問Niantic Spatial SDK ドキュメントデベロッパーフォーラム
News研究
A family of foundation models, trained on real-world data to establish ground truth for physical AI.
ReconstructionMetric-scale splats and aligned meshes of real places, from a single room to city-scale.LocalizationPrecise vision-based positioning, even where GPS doesn’t work.UnderstandingObjects, change over time and answers at every 3D point of a real place.
Where people and machines work in the real world.
石油・ガスIntelligent field operations.防衛Mission rehearsal and GPS-denied operations.Embodied AIReal-to-sim training for robots.
The team, partners and resources behind Niantic Spatial.
Niantic Spatial について
Niantic Spatial についてイベント採用情報
Partners
デベロッパー・パートナー・プログラムAcademic and community partnersパートナー
Guides and support
Scaniverse FAQ - よくあるご質問Niantic Spatial SDK ドキュメントデベロッパーフォーラム
Scaniverse appCapture and reconstruct real places.価格Plans for teams and enterprises.
お問い合わせ

Query the world, track changes, and predict what's next.

We’re building a semantic layer that turns 3D reconstructions into spatial knowledge so people and machines can understand what’s there, how objects relate, and what actions are possible.

お問い合わせDiscover our research

One spatial data layer that connects what a place looks like with what it contains, how it works, and how it changes.

ReconstructionSplats and meshes of a place
Captures over timeFor change detection
QuestionsIn plain language
Your dataAsset lists and records
Understanding
VocabularyOpen-set, no retraining
MeaningAt every 3D point
InterfaceNatural language and API
Per-point semanticsGeometry, material, context
Located answersTied to a place
AssetsFound remotely
ChangeWhat moved, appeared or degraded

From 3D geometry to real-world understanding

  1. Spatially grounded semantics
    Each point carries a descriptor for geometry, material and context. Understand objects, materials, and relationships in their real-world context, with insights tied to precise 3D locations and traceable to the underlying geometry and imagery.
  2. Open-vocabulary understanding
    Move beyond predefined categories to query specific objects and concepts in natural language, and reason about their spatial relationships, not just recognize their presence.
  3. Change over time
    Compare captures to investigate what appeared, disappeared, or moved, grounding each change in its location and surrounding context so teams can understand more than a visual difference.

What spatial understanding makes possible

Get answers from your environment in plain language“Where does vegetation encroach on power lines?” “Which valves are above two meters?” Answers come back tied to a location.
Locate and inspect assets from anywhereLocate equipment, inspect it in 3D, and give remote experts and field crews the same view without unnecessary site visits.

Industries we support

石油・ガスRemote inspection, asset finding and change detection for plants, pipelines and utilities.Explore oil and gas 防衛A shared operational picture of terrain and sites for mission-critical decisions.Explore defense Embodied AIScene understanding so robots act on what things are, not just where they are.Explore embodied AI

Related

ReconstructionUnderstanding builds on a reconstruction. Start there, or bring one you have.LocalizationAnswers are tied to real positions, so people and robots can act where they stand.

Frequently Asked Questions

What is semantic 3D understanding?

It adds meaning to a reconstructed environment, not just geometry. Each 3D point carries information about geometry, material and context, creating maps that people, machines and AI agents can query.

What are open-set semantics?

They let a system recognize concepts beyond a fixed list of categories, adapting to new concepts without retraining. That suits open-ended questions about real environments.

Can I query a 3D model in natural language?

Yes. Ask open-ended questions, such as where vegetation encroaches on infrastructure or which terrain is traversable, and each answer is tied to a location in the model.

How is this different from object detection?

Object detection assigns predefined labels to regions of 2D images. 3D semantic understanding represents meaning at individual points in a measured environment, relates objects to each other and supports open-ended queries.