JSON
Structured metadata format used for manifests, annotations, and calibration files.
FORMAT FACETS
Datasets by format
Pick the format your training pipeline expects. RLDS, LeRobot, HDF5, MCAP, ROS bag, and Parquet each carry trade-offs for streaming, schema preservation, and tool compatibility — these facets group datasets by what they actually ship.
DIRECT ANSWER
A dataset's delivery format decides whether a small sample can move from supplier to your training pipeline without a hidden conversion project. Buyers should require samples in the requested format before scaling, since not every format preserves episode boundaries, action streams, or timestamps the same way.
6 FACETS
Browse datasets by format
MP4
Compressed video container used for RGB or egocentric footage.
HDF5
Hierarchical data format commonly used for trajectories, demonstrations, and arrays.
Point cloud
3D scene or object geometry represented as spatial points, often from depth sensors or reconstruction.
Parquet
Columnar data format used by modern robotics dataset hubs for episode metadata, frame tables, and sharded time-series records.
Pickle
Python object serialization format sometimes used by benchmark releases for demonstrations, observations, and robot state dictionaries.
CROSS-CATALOG
Pair with another facet
Combine this facet with a second filter (modality, task, robot, format, license, or commercial-use) on the main dataset catalog to narrow the buyer decision faster.
KEEP DIGGING
Use this record as part of a broader dataset review
A dataset record is only useful when it connects into the rest of the buyer workflow. The next review step is usually not another summary; it is a fit check, rights triage, source comparison, or custom bounty spec that names the missing proof.
For physical AI teams, the hard question is whether the public source can support a specific model objective under real deployment constraints. That requires adjacent dataset records, tools, comparisons, and sourcing paths, plus external references that a reviewer can open and challenge.
Use the links below to keep the review grounded. Start broad when discovery is incomplete, move into profile and comparison pages when the candidate source is known, and switch to custom collection when the blocker is rights, consent, geography, robot embodiment, or target environment coverage.
Where to go next
- Physical AI dataset catalogUse the catalog to compare source-backed dataset profiles by modality, task, rights signal, consent risk, and deployment fit.
- Hugging Face robotics indexScan the broader robotics dataset surface before narrowing into promoted profiles, comparisons, and custom collection specs.
- Dataset changelogTrack source updates, licensing notes, and buyer-readiness changes that should trigger a renewed review.
- Dataset fit checkerScore whether a public source is enough for the model, rights path, modalities, and target environment.
- License risk checkerSeparate source license language from contributor consent, redistribution, private-space risk, and model-use assumptions.
- Data spec generatorTurn a public-source gap into a scoped capture request with sample QA, metadata, and delivery requirements.
- Vendor alternatives hubCompare data providers when the answer is not another public dataset but a better sourcing or capture route.
- Data annotation companiesUse the company index to separate annotation vendors, data engines, marketplaces, and specialist capture teams.
Other places to verify the claims
- Scale AI physical AI data engineMarket context for why physical AI systems need custom, enriched, real-world data beyond generic labeling workflows.
- LeRobot documentationRobotics dataset and tooling context for Hugging Face based collection, sharing, conversion, and training workflows.
- Open X-EmbodimentA cross-embodiment robotics dataset reference for comparing trajectory scale, robot diversity, and VLA training assumptions.
- DROID datasetA large in-the-wild robot manipulation dataset reference for real-world trajectory capture and deployment transfer risk.
TRUELABEL ROUTING
Want a dataset converted to a different format?
Tell us the source dataset and your target schema. Our partners can deliver pre-validated conversions with manifest, checksums, and conversion-loss notes.