Datasets
37 of 37
| Links | ||||||||
|---|---|---|---|---|---|---|---|---|
UT Austin · 2023 | 6.5K | – | Simulation | – | CC BY 4.0 (datasets), MIT (code) | |||
DROID Dataset Team (Stanford, UC Berkeley and others) · 2024 | 76K | 350 | Robot | 2 TB | CC BY 4.0 | |||
UC Berkeley (RAIL) · 2023 | 60.1K | – | Robot | – | CC BY 4.0 | |||
Google (Robotics at Google, Everyday Robots) · 2022 | 130K | – | Robot | – | CC BY 4.0 (via Open X-Embodiment) | |||
Google DeepMind and 21 partner institutions · 2023 | 1M | – | Robot | – | CC BY 4.0 (data), Apache-2.0 (code) | |||
RoboTwin Platform · 2025 | 100K | – | Simulation | – | MIT | |||
Beijing Innovation Center of Humanoid Robotics, Peking University, BAAI · 2024 | 107K | 306 | Mixed | – | Apache-2.0 (gated) | |||
Galaxea · 2025 | 100K | 500 | Robot | 3 TB | CC BY-NC-SA 4.0 (gated) | |||
AgiBot, OpenDriveLab (Shanghai AI Lab) · 2025 | 1M | 3K | Robot | 48 TB | CC BY-NC-SA 4.0 + AgiBot World Community License (gated) | |||
UT Austin, NVIDIA · 2026 | 55K | 612 | Simulation | – | CC BY 4.0 (assets and datasets), MIT (code) | |||
InternRobotics (Shanghai AI Lab) · 2025 | 630K | 7.4K | Simulation | 7 TB | CC BY-NC-SA 4.0 (gated) | |||
Hugging Face (LeRobot) and community contributors · 2025 | 22.9K | – | Robot | – | Apache-2.0 (bundle); source uploads carry their own licenses | |||
University of Freiburg · 2021 | – | 24 | Simulation | – | MIT (repo) | |||
Ai2 · 2026 | 38.1K | 184 | Robot | – | Apache-2.0 (manifest and annotations); source repos keep their licenses | |||
Fourier · 2025 | 30K | 140 | Robot | Fourier GR-1Fourier GR-2 | 3 TB | CC BY-NC-SA 4.0 (gated) | ||
Stanford Vision and Learning Lab · 2025 | 10K | 1.2K | Simulation | 2 TB | MIT | |||
NVIDIA · 2025 | – | – | Simulation | – | CC BY 4.0 | |||
Apple · 2025 | 338K | 829 | Egocentric | 2 TB | CC BY-NC-ND | |||
Google Research · 2022 | 442.2K | – | Mixed | – | CC BY 4.0 (via Open X-Embodiment); repo code Apache-2.0 | |||
Shanghai Jiao Tong University · 2023 | 110K | – | Mixed | – | CC BY-SA 4.0 (RH20T-C) / CC BY-NC 4.0 (RH20T-NC) | |||
Beijing Innovation Center of Humanoid Robotics, Peking University · 2025 | 310K | 1K | Mixed | – | Apache-2.0 (ModelScope metadata) | |||
University of Southern California, Toyota Research Institute · 2025 | 10.3K | – | Robot | Unitree G1Unitree H1 | – | Apache-2.0 | ||
NVIDIA, UT Austin · 2024 | 21K | – | Simulation | Bimanual Franka PandaFourier GR-1 | – | Conflicting: CC BY-NC-SA 4.0 (HF tag), NVIDIA non-commercial license (HF file), CC BY 4.0 (GitHub) | ||
Columbia University, Toyota Research Institute, MIT · 2023 | 200 | – | Simulation | Simulated 2D pusher | – | MIT (repo) | ||
Georgia Tech, Stanford, UCSD, ETH Zurich, MIT, Meta and others · 2026 | 80K | 1.4K | Egocentric | – | – | |||
Pengcheng Laboratory, SUSTech, Sun Yat-sen University (ARIO Alliance) · 2024 | 3M | – | Mixed | – | CC BY 4.0 or MIT (own data); converted data keeps its source license | – | ||
AgiBot, OpenDriveLab (Shanghai AI Lab) · 2024 | 92.2K | – | Robot | 9 TB | CC BY-NC-SA 4.0 + AgiBot World Community License (gated) | – | ||
AgiBot · 2026 | – | – | Robot | AgiBot G2 | 18 TB | CC BY-NC-SA 4.0 | – | |
Shanghai AI Laboratory, Northwestern Polytechnical University · 2025 | 100K | 600 | Robot | FastUMI handheld gripper | – | Apache-2.0 | – | |
Unitree · 2025 | – | – | Robot | Unitree G1Unitree Z1 | – | Apache-2.0 | – | |
Stanford Vision and Learning Lab | 20K | – | Simulation | 3 TB | MIT | – | ||
NVIDIA, UT Austin · 2023 | 48K | – | Simulation | – | CC BY 4.0 (datasets) | – | ||
Ant Group · 2026 | – | 2K | Egocentric | 23 TB | Open-AoE Dataset License (research and commercial with attribution); MANO-derived files under the MANO license | – | ||
Ego4D Consortium (Meta and university partners) · 2021 | – | 3.7K | Egocentric | 7 TB | Ego4D License Agreement (gated, no redistribution) | – | ||
Ego4D Consortium (Meta and university partners) · 2023 | – | 1.3K | Egocentric | – | Ego-Exo4D License Agreement (gated, no redistribution) | – | ||
University of Bristol and partners · 2020 | – | 100 | Egocentric | – | CC BY-NC 4.0 | – | ||
TwentyBN (now Qualcomm) | 220.8K | – | Human video | – | Qualcomm Data License Agreement – Research Use (non-commercial) | – |