Skip to main content
Physical AI Video 🟢 Actively collecting · continuously growing

Off-the-Shelf Egocentric Data for Physical AI

🟢 Available now and continuously growing — first-person (egocentric) video of real commercial and residential tasks for embodied and Physical-AI training. Video only; no audio.

  • Rights-cleared
  • Consent documented
  • 🇮🇳 India data residency
View license tiers
Multi-sensor humanoid capture and 3D annotation illustrating the Off-the-Shelf Egocentric Data for Physical AI physical-AI dataset
Languages
Video only (no audio / no language track)
Scale
🟢 Actively collecting · continuously growing
Quality & methodology metrics

Measured, audited, reproducible

Modality: first-person (head-mounted) egocentric video only. No audio is captured, uploaded, or delivered with this data.

Capture standard: hands and task area kept in frame, camera level and forward-facing (chin ≤30° down); recorded on approved flagship devices (iPhone 11+, Pixel 6+, Galaxy S21+).

Integrity: zero-tolerance 2-strike QA regime — automated duplicate-user, sensitive-content and filler detection; payment on valid accepted hours only.

Availability: off-the-shelf now and expanding weekly; new accepted hours are added continuously across categories and geographies.

Provenance & consent

Recorded by employed frontline workers capturing their own real workplace tasks, under employer / site-manager approval plus colleague consent. Sessions run 1 min–4 hrs. Consented for commercial reuse; provenance documented per accepted hour.

Recorded by householders performing their own real home tasks, with household-member awareness and guardian consent where minors may appear. Sessions run ≥30 min (recommended 1–2 hrs, 4 hr max). Consented for commercial reuse; provenance documented per accepted hour.

Methodology

Consented contributors capture first-person video of tasks they genuinely perform — nothing staged — on head-mounted flagship phones, across two tracks: Commercial (employed workers, their own workplace) and Residential (householders, their own home). Every accepted hour passes a zero-tolerance 2-strike QA regime.

Note: this is egocentric video data — the collection app does not upload audio, so no speech or audio track is present.

Data preview

What's inside each record

A representative schema for Off-the-Shelf Egocentric Data for Physical AI. The full data card ships the complete field dictionary, value ranges and annotation rubric.

Field Type Example
id string (uuid) "txt_a3f1…"
prompt string "When will this feature ship?"
response string "…"
label enum "preferred" | "rejected"
language string (ISO 639) "vid"
annotator_id string (hashed) "anr_7f3…"
qa_status enum "passed"
consent_ref string "cns_2024_…"

Representative schema — exact fields and value ranges are documented in the dataset card shipped with every licence.

Licensing

License tiers

Choose the tier that matches your use case. Every tier ships with the full data card, provenance log, and quality report.

Non-Exclusive License

Contact us

License accepted egocentric video hours across categories. Video only — no audio.

Request access

Custom Collection

Contact us

Commission egocentric hours in specific categories, environments or geographies to spec.

Talk to sales

Related datasets

See all Physical AI datasets

Explore more Physical AI data

Physical AI datasets Data services Sovereign delivery

Ready to license Off-the-Shelf Egocentric Data for Physical AI?

A senior data PM will scope access, residency, and licensing terms and respond within one business day.

Request access Browse the catalog