FTP-1 Dataset
FTP-1-Dataset is the heterogeneous pretraining corpus behind FTP-1 (A Generalist Foundation Tactile Policy Across Tactile Sensors for Contact-Rich Manipulation).
Source notes
FTP-1-Dataset is the heterogeneous pretraining corpus behind FTP-1 (A Generalist Foundation Tactile Policy Across Tactile Sensors for Contact-Rich Manipulation). It aggregates roughly 3,000 hours of contact-rich manipulation data from 26 sources spanning human demonstrations, dexterous-hand, and gripper robots, recorded with 21 distinct tactile sensors (image-, array-, and state-based). Each episode pairs tactile observations with RGB images, proprioceptive state, and action trajectories, stored in zarr format. Hosted on ModelScope and mirrored on Hugging Face.
- Scale
- 3,000 hours
- Formats
- zarr
- License
- MIT
- Published
- 2026-06-11
Decision summary
teleoperation
Metadata requires review against the official source before publication.
Read the dataset manifest and feature schema.
Release history
Source-backed release timing for this canonical dataset record.
- Release evidence
Dataset series published
Release timing is recorded from official dataset metadata.
Selection readiness
6 evidence dimensions for deciding whether this dataset is ready to inspect, compare, or adopt. This is not a model benchmark.
Catalog evidence · not task performance
73/100
Provisional
4/6 dimensions scored · 48% confidence
Read the evidence behind all 6 dimensions
Access and governance
Access and license are declared by the source.
Schema and signal coverage
No machine-readable schema has been verified yet.
Policy training readiness
Observation and action/state signals are declared; alignment quality still depends on sample verification.
World-model readiness
Temporal observations plus geometry or semantic context are declared; sample alignment remains to be audited.
Failure and recovery readiness
No verified failure/recovery annotation evidence is available yet.
Download and processing readiness
Scale is declared; transfer and processing estimates are not measured.
Review unresolved evidence and next checks
Signal gaps
- Gaze / attention. Not enough evidence is available to classify this signal.
- Feedback / correction / failure. Not enough evidence is available to classify this signal.
- Sim-real pairing. Not enough evidence is available to classify this signal.
Next checks
- Read the dataset manifest and feature schema.
- Run a bounded sample audit before assigning Strong readiness.
Loop signals
4/7 present or partial
No decision-grade evidence captured yet.
unknownNo decision-grade evidence captured yet.
unknownView 5 more signal categories
video · FTP-1-Dataset is the heterogeneous pretraining corpus behind FTP-1 (A Generalist Foundation Tactile Policy Across Tactile Sensors for Contact-Rich Manipulation). It aggregates roughly 3,000 hours of contact-rich manipulation data from 26 sources spanning human demonstrations, dexterous-hand, and gripper robots, recorded with 21 distinct tactile sensors (image-, array-, and state-based). Each episode pairs tactile observations with RGB images, proprioceptive state, and action trajectories, stored in zarr format. Hosted on ModelScope and mirrored on Hugging Face.
presentFTP-1-Dataset is the heterogeneous pretraining corpus behind FTP-1 (A Generalist Foundation Tactile Policy Across Tactile Sensors for Contact-Rich Manipulation). It aggregates roughly 3,000 hours of contact-rich manipulation data from 26 sources spanning human demonstrations, dexterous-hand, and gripper robots, recorded with 21 distinct tactile sensors (image-, array-, and state-based). Each episode pairs tactile observations with RGB images, proprioceptive state, and action trajectories, stored in zarr format. Hosted on ModelScope and mirrored on Hugging Face.
partiallanguage · has-success-labels
presentNo decision-grade evidence captured yet.
unknownOpen · MIT · zarr
partialEvidence details and provenance
Official signal claims
Schema and annotations
No machine-readable schema facts are captured.
Sample verification
Pending. Metadata does not prove sample coverage, alignment, or file integrity.
curated official source
Integration notes
- Metadata requires review against the official source before publication.