Open X-Embodiment: over a million real robot trajectories from 22 embodiments, pooled from 60 datasets, under CC BY 4.0

The document exists. The ledger fetched it at its publisher and quotes it.

Record ID

EAD-2026-0023

Kind

Data supply

Jurisdiction

Global

Last verified

13 September 2026

Added

13 September 2026

• Project page: over one million real robot trajectories, 22 robot embodiments, 60 existing datasets from 34 robotic research labs.
• Paper abstract, arXiv 2310.08864, first submitted 13 October 2023: 22 different robots, 21 institutions, 527 skills and 160,266 tasks.
• Licence, per the repository: code under Apache License 2.0, all other materials under Creative Commons Attribution 4.0 International.

1. 1,000,000 trajectories

   Real robot trajectories, at least, per the project page

   Open X-Embodiment project page, primary source, as of 13 October 2023.

2. 22 embodiments

   Robot embodiments

   Open X-Embodiment paper, arXiv 2310.08864, primary source, as of 13 October 2023.

3. 527 skills

   Skills

   Open X-Embodiment paper, arXiv 2310.08864, primary source, as of 13 October 2023.

This is the open, commercially usable pool, and it is a federation of academic datasets recorded for other purposes, which is why a training ground that records to one specification is worth building. A maker can start from it; nobody on this ledger claims a product was trained on it alone.

1. Open X-Embodiment project pagePrimary · 13 October 2023
2. Open X-Embodiment paper, arXiv 2310.08864Primary · 13 October 2023
3. Google DeepMind, open_x_embodiment repositoryPrimary · 13 October 2023

• VerifiedData supplyEAD-2026-0013

  The largest open real robot dataset is Chinese, over a million trajectories, and licensed for non commercial use only

  China · verified 13 September 2026

  GitHub release notes: Alpha, 92,214 trajectories, 30 December 2024; Beta, 1,003,672 trajectories, about 43.8 terabytes, 1 March 2025.

• VerifiedData supplyEAD-2026-0033

  DROID: 76,000 demonstration trajectories, 350 hours, one robot arm, under CC BY 4.0

  United States · verified 13 September 2026

  Project page: 76,000 demonstration trajectories or 350 hours of interaction data, 564 scenes, 86 tasks; the paper abstract gives 84 tasks.

• VerifiedData supplyEAD-2026-0024

  The largest open supply of hours is human, not robot: Ego4D holds over 3,670 hours of first person video

  Global · verified 13 September 2026

  Ego4D site: over 3,670 hours of daily life activity video, 923 unique participants, 74 locations in 9 countries, released 17 February 2022.

• Open questionOpen questionEAD-2026-0027

  Open question: does a person in a training ground's footage have a say? Three regimes, three different answers, none written for robots

  Global · verified 13 September 2026

  GDPR and the EDPB video guidelines: footage of an identifiable person is personal data; it is biometric only when processed to identify. Workers are owed a lawful basis and information, and consent is a weak basis at work.

• Open questionOpen questionEAD-2026-0026

  Open question: who owns the hours a robot records at a customer's site? Only the EU has a rule, and it is about access, not ownership

  Global · verified 13 September 2026

  Data Act articles 4 and 5: the user may access readily available data and have it shared with a third party; article 6(2)(e) bars the third party from a competing product. The Regulation does not use the word ownership.

• Open questionOpen questionEAD-2026-0025

  Open question: do synthetic hours count? NVIDIA generated the equivalent of 6,500 human hours in eleven, and no standard says what an hour is

  Global · verified 13 September 2026

  NVIDIA newsroom, 18 March 2025: 780,000 synthetic trajectories, the equivalent of 6,500 hours or nine continuous months of human demonstration data, in eleven hours; a 40 percent improvement over real data alone.