Egocentric household robotics video
CommissionableFirst-person video of real household work — cooking, cleaning, laundry, tidying — captured in lived-in homes rather than staged environments, against a defined task list with dispatched and recovered wearable rigs.
Specifications
| Dataset ID | KV-EGO-001 |
|---|---|
| Availability | Commissionable |
| Intended uses | Humanoid and embodied-AI policies · Manipulation learning · World models |
| Modality | Head- or chest-mounted egocentric video |
| Source category | Distributed contributor networks in real homes |
| Geography | Pending verification — Defined per program across the operating footprint |
| Languages | Pending verification — Paired narration language defined per program |
| Volume / capacity | Pending verification — Defined per program: fixed monthly hours against a defined task list |
| Timeline | Pending verification — Stated at specification; confirmed before activation |
| Formats & schema | Pending verification — Resolution, frame rate and schema agreed in the capture specification |
| Processing | Raw, or annotated (task segmentation, object and action labels) to buyer specification |
Why this data is novel
Humanoid programs commission training corpora at very large scale, and every hour requires a real person in a real home wearing a camera. Existing egocentric datasets skew toward Western homes; this program class captures kitchens, appliances and objects rare in those corpora.
Source methodology and collection context
Contributors within established networks opt into a defined task list. Wearable rigs are dispatched and recovered through in-market physical presence. Capture follows a per-task specification with defined session lengths.
Composition (organic-data statement)
Human-generated, non-synthetic egocentric capture. Paired audio is collected only where every party opts in and jurisdiction review clears capture. AI-assisted annotation, where used, is disclosed in the dataset card.
Public availability and prior licensing
Pending verification — Commissioned output is new capture; prior-licensing status stated per program
Duplication, overlap and contamination
Pending verification — Overlap with public egocentric corpora assessed and reported per program
Rights and permitted uses
Consent at task acceptance for the declared purpose, with participant compensation in the payment flow. Permitted model uses and exclusivity documented per program.
Privacy, PII and de-identification
Homes are private spaces: bystander rights, in-home PII and jurisdiction-specific requirements are reviewed before any collection begins. De-identification options are defined in the capture specification.
Quality, acceptance criteria and known limitations
Two-stage verification against written acceptance criteria: automated integrity checks plus independent adjudication. Task-level acceptance reporting.
Delivery
Secure transfer with versioned manifests, checksums, consent attestation and rejection analysis.
Commercial structure
Commissioned program (buyer-specific). Exclusivity where required, priced as a rights grade.
Evaluation pack
Buyer diligence is productized. A qualified request receives an evaluation pack containing:
- Representative stratified sample
- Dataset card
- Data dictionary and schema
- Provenance summary
- Rights and consent summary
- Privacy and PII assessment
- Quality and novelty report
- Known limitations
- Security and delivery sheet
- Version and checksum manifest
Samples are never cherry-picked: each pack states whether its sample is representative, illustrative, anonymized, or structurally simulated.