EP #00000 · real capture

Open X-Embodiment

Cross-embodiment robot learning

Indexed from an external sourceReal-worldUpdated 2024-05-29Open Source

Published by

Open X-Embodiment Collaboration

RobotData indexes this dataset's metadata. Attribution and licensing stay with the original publisher.

The pooled cross-embodiment corpus: 60 datasets from 34 labs, over 1M real robot trajectories on 22 embodiments covering 527 skills and 160,266 tasks, standardized into RLDS. It is the pre-training mixture behind RT-X, Octo and OpenVLA.

Reported structure

Values below come from the dataset's own documentation. Missing fields are shown as not provided rather than estimated.

Observation

3 cameras

State

7 DoF

Action

10 Hz

Result

Not provided

Data modalities

RGB VideoDepthJoint PositionsRobot StateActionsForce/TorqueAudioLanguage Instructions

Dataset Explorer

Real episodes from this dataset, with every camera stream playing in sync.

1.0M episodes indexed

Episode #00000

real capture

instruction: “sweep the green cloth to the left side of the table”

clip 10.0s71 frames @ 5 Hz3 camera streams
CAMERA 1 · SCENE
WRIST CAMERA
DEPTH

Playback

source lerobot/berkeley_autolab_ur5 · CC BY 4.0

original episode source

Linked models, robots & papers

Resolving related models, datasets and papers…

Related datasets