MOTIONATLASThe Global Dataset for Physical AI
Home

Documentation

Data card

The Global Dataset for Physical AI. This page states purpose, method, and limits for technical evaluators. It does not invent scale.

Purpose
Motion Atlas builds structured, synchronized multi-camera human-motion datasets and motion intelligence for robotics and Physical AI.
Mission
To map human motion at meaningful scale and transform it into motion intelligence that advances robotics and Physical AI for the benefit of people and society.
Intended use
Licensed research and model training. Sample access under NDA. Not a public bulk download of original media.
Platform workspaces
Nexus is the public gateway (AVAILABLE). Dataset Explorer is the inspection UI (AVAILABLE). Motion Atlas Studio is a separate application for Episode editing (PILOT). Nexus launches the existing Studio deployment through one centralized destination; this tree does not embed Studio or connect to its database.
Contract
Platform Contract 1.0.0. Metadata schema 1.0.0. Architecture baseline NEXUS-CONVERGENCE-1.1.
Capture
CameraStreams on Master Episode Time. Launch configurations include one, three, and six cameras. Count is per Episode. Measured sync offset is published per release — not claimed on this demonstration.
Media
Web playback uses PROXY assets. ORIGINAL media is not exposed here. Thumbnails are THUMBNAIL-role assets.
Pose
Demonstration uses 21-point image-plane hand keypoints. Metric 3D only when a release documents frame, units, origin, and method.
ECoT
ECoT is structured, observable task-execution context: observation, goal, action plan, decision rule, and success check. It is not private reasoning or hidden thought.
Taxonomy
Capture domain → workflow label → task label → L2 skill → L1 atomic → contact → state → outcome. Free-text labels are not canonical Task IDs.
QA / provenance
First-class records. This catalog uses a demonstration QA profile. Privacy status UNKNOWN unless a Guardian review exists. Origin on this catalog is SYNTHETIC.
Care domain
Care and clinical support activities. Not a regulated medical-procedure dataset.
Limitations
Pilot coverage. Contributor and country counts unpublished until verified. Camera photometric specs unpublished until measured.
Licensing
Sample by request. Production by contract. Package varies with configuration.

Schema, taxonomy, and ECoT versions publish per release. Request sample data.

Quality gate

Demonstration

Motion Atlas quality gate.

Scale alone is not sufficient. Quality should be measurable. Provenance should be traceable. Originals should remain protected. Capture quality and annotation integrity are checked before a clip is training-eligible. This report is a demonstration of the gate, not a production certificate.

  • Temporal synchronizationPASS
  • Camera calibrationPASS
  • Frame integrityPASS
  • ExposurePASS
  • Motion blurPASS
  • Hand visibilityPASS
  • Critical object visibilityPASS
  • Pose confidencePASS
  • Annotation completenessPASS
  • Privacy reviewPASS
  • Sequence continuityPASS

Dataset status

Training eligible

Demonstration slice. Production eligibility is per-episode.

Delivery

What you receive.

Exact contents depend on capture configuration and contract. A sample is a package, not a highlight reel.

MA_SAMPLE_001/
├── video/          CameraStreams as captured
├── annotations/    episode, actions, objects, contacts,
│                   states, pose, failures, ecot
├── calibration/    when the release includes it
├── qa/             qa_report.json
└── README.md

Data card

Purpose, method, limits.

  • Purpose: structured human activity for embodied AI
  • Intended use: research and licensed training
  • Capture: per-episode camera count (1 / 3 / 6 and future N)
  • Pose: 21-point image-plane hands unless a release states otherwise
  • ECoT: observable embodied context, not private thought
  • Consent and privacy review before eligibility
  • Known limits documented per version
  • Licensing: sample under NDA; production by contract
View data card