Open X-Embodiment (OXE)

Aggregate of 60 datasets from 34 labs (project page); paper abstract states 22 robots, 21 institutions, 527 skills, 160,266 tasks. Hours/frames/modalities not stated on fetched pages; hosted on Google Cloud Storage, not Hugging Face.

基本事实

机构: Open X-Embodiment Collaboration (Google DeepMind + partner labs)
形态标签: 多本体
本体（按原文）: 22 different robots
轨迹数: 1,000,000
小时数: 待核实
帧数: 待核实
任务: 160266
场景: 待核实
模态: 待核实
格式: RLDS
许可证: Software: Apache License 2.0; all other materials: Creative Commons Attribution 4.0 International (CC-BY) (repository README); constituent datasets are cited/licensed individually via the dataset spreadsheet
商用: 待核实
发布: 2023-10
体积: 待核实

来源 · 11 条来源

name https://robotics-transformer-x.github.io/ 官方页面 · 抓取 2026-09-18 · 尚未存快照 Page: '22 different robots'; 'pooling 60 existing robot datasets from 34 robotic research labs around the world'

org https://robotics-transformer-x.github.io/ 官方页面 · 抓取 2026-09-18 · 尚未存快照 Page: '22 different robots'; 'pooling 60 existing robot datasets from 34 robotic research labs around the world'

url https://robotics-transformer-x.github.io/ 官方页面 · 抓取 2026-09-18 · 尚未存快照 Page: '22 different robots'; 'pooling 60 existing robot datasets from 34 robotic research labs around the world'

embodiments https://robotics-transformer-x.github.io/ 官方页面 · 抓取 2026-09-18 · 尚未存快照 Page: '22 different robots'; 'pooling 60 existing robot datasets from 34 robotic research labs around the world'

form_tags https://robotics-transformer-x.github.io/ 官方页面 · 抓取 2026-09-18 · 尚未存快照 Page: '22 different robots'; 'pooling 60 existing robot datasets from 34 robotic research labs around the world'

episodes https://robotics-transformer-x.github.io/ 官方页面 · 抓取 2026-09-18 · 尚未存快照 '1M+ real robot trajectories' recorded as floor 1,000,000

tasks https://arxiv.org/abs/2310.08864 官方页面 · 抓取 2026-09-18 · 尚未存快照 Abstract: '527 skills (160266 tasks)'; also on project page

format https://github.com/google-deepmind/open_x_embodiment 官方页面 · 抓取 2026-09-18 · 尚未存快照 README: each episode 'is represented using the RLDS episode format'; data at gs://gdm-robotics-open-x-embodiment/

license https://github.com/google-deepmind/open_x_embodiment 官方页面 · 抓取 2026-09-18 · 尚未存快照 README: 'All software is licensed under the Apache License, Version 2.0' / 'All other materials are licensed under the Creative Commons Attribution 4.0 International License (CC-BY)'; per-dataset licenses not enumerated, so commercial_ok left null

commercial_ok https://github.com/google-deepmind/open_x_embodiment 官方页面 · 抓取 2026-09-18 · 尚未存快照 README: 'All software is licensed under the Apache License, Version 2.0' / 'All other materials are licensed under the Creative Commons Attribution 4.0 International License (CC-BY)'; per-dataset licenses not enumerated, so commercial_ok left null

release_date https://arxiv.org/abs/2310.08864 官方页面 · 抓取 2026-09-18 · 尚未存快照 arXiv v1 submitted 2023-10-13 (latest v9 2025-05-14)

索引里的相关本体

按写明的机器人名或形态匹配；匹配只说明形态覆盖，不代表某个模型在其上训练过。

—

方向清晰，判断有据。