๐ค Physical AI (Robotics & Embodied AI) ๋ถ์ผ์ ์คํ์์ค ๋ชจ๋ธ, ๋ฐ์ดํฐ์ , ์๋ฎฌ๋ ์ดํฐ๋ฅผ ์ฒด๊ณ์ ์ผ๋ก ์ ๋ฆฌํ ํ๋ ์ด์ ๋ฆฌ์คํธ. A curated list of open-source models, datasets, and simulators for Physical AI (Robotics & Embodied AI).
๐ ์ธํฐ๋ํฐ๋ธ ๋์๋ณด๋์์ ํํฐ๋ง ๋ฐ ์๊ฐํ ๋ณด๊ธฐ | View Interactive Dashboard
์คํ ์ ๊ธฐ์ค ๋ด๋ฆผ์ฐจ์ ์ ๋ ฌ | Sorted by GitHub stars (auto-updated weekly)
| Name | Organization | Year | Category | Hardware | Learning | โญ Stars | Links |
|---|---|---|---|---|---|---|---|
| LeRobot | Hugging Face | 2024 | manipulation | manipulator | IL, diffusion, VLA | 24,952 | ๐ ๐ค |
| ฯ0 (pi-zero) | Physical Intelligence | 2024 | manipulation, whole-body | manipulator, humanoid | VLA, diffusion | 12,331 | ๐ ๐ค |
| GR00T N1 | NVIDIA | 2025 | manipulation, whole-body | humanoid | VLA, IL | 7,335 | ๐ ๐ค |
| OpenVLA | Stanford / UC Berkeley | 2024 | manipulation | manipulator | VLA | 6,421 | ๐ ๐ค |
| Mobile ALOHA | Stanford | 2024 | manipulation, navigation, whole-body | mobile, manipulator | IL | 4,439 | ๐ |
| Diffusion Policy | Columbia University | 2023 | manipulation | manipulator | IL, diffusion | 4,270 | ๐ |
| ACT (Action Chunking with Transformers) | Stanford | 2023 | manipulation, dexterous | manipulator | IL | 2,001 | ๐ |
| Octo | UC Berkeley / Stanford / CMU / others | 2023 | manipulation | manipulator, mobile | IL, VLA | 1,669 | ๐ ๐ค |
| HumanPlus | Stanford | 2024 | manipulation, whole-body | humanoid | IL | 849 | ๐ |
| RoboVLMs | Multiple | 2025 | manipulation | manipulator | VLA, IL | 473 | ๐ ๐ค |
| RoboFlamingo | ByteDance | 2023 | manipulation | manipulator | VLA | 434 | ๐ |
| GR-1 | BAAI / Beijing Academy of AI | 2024 | manipulation | manipulator | VLA, IL | 312 | ๐ |
| CrossFormer | UC Berkeley / others | 2024 | manipulation | manipulator, mobile | IL, VLA | 283 | ๐ ๐ค |
| CogACT | Tsinghua University / Microsoft Research Asia | 2024 | manipulation | manipulator | VLA, diffusion | 0 | ๐ ๐ค |
์คํ ์ ๊ธฐ์ค ๋ด๋ฆผ์ฐจ์ ์ ๋ ฌ | Sorted by GitHub stars (auto-updated weekly)
| Name | Organization | Year | Category | Source | Modality | Trajectories | โญ Stars | Links |
|---|---|---|---|---|---|---|---|---|
| ManiSkill2 | UC San Diego / Shanghai AI Lab | 2023 | manipulation, dexterous | simulation | rgb, rgbd, proprioception | 36,000 | 2,996 | ๐ ๐ค |
| LIBERO | UMass Amherst / Bosch | 2023 | manipulation | simulation | rgb, proprioception | 130,000 | 1,941 | ๐ ๐ค |
| Open X-Embodiment (OXE) | Google DeepMind / RT-X Team | 2023 | manipulation | real, teleoperation | rgb, rgbd, proprioception | 1,000,000 | 1,883 | ๐ ๐ค |
| Meta-World | Stanford / Berkeley | 2019 | manipulation | simulation | proprioception | โ | 1,836 | ๐ |
| HumanoidBench | CMU / UC San Diego / MIT | 2024 | locomotion, manipulation, whole-body | simulation | rgb, proprioception | โ | 773 | ๐ |
| RoboAgent Dataset | CMU | 2023 | manipulation | real, teleoperation | rgb, proprioception | 7,500 | 372 | ๐ |
| DROID | UC Berkeley / Stanford / others | 2024 | manipulation | real, teleoperation | rgb, rgbd, proprioception | 76,000 | 371 | ๐ ๐ค |
| Language-Table | 2023 | manipulation | real, simulation, teleoperation | rgb, proprioception | 600,000 | 360 | ๐ ๐ค | |
| BridgeData V2 | UC Berkeley | 2023 | manipulation | real, teleoperation | rgb, proprioception | 60,096 | 282 | ๐ ๐ค |
| RH20T | Shanghai AI Lab | 2023 | manipulation, dexterous | real, teleoperation | rgb, rgbd, tactile, proprioception, audio | 110,000 | 108 | ๐ |
์คํ ์ ๊ธฐ์ค ๋ด๋ฆผ์ฐจ์ ์ ๋ ฌ | Sorted by GitHub stars (auto-updated weekly)
| Name | Organization | Year | Type | GPU | ROS2 | Language | โญ Stars | Paper |
|---|---|---|---|---|---|---|---|---|
| Genesis | Genesis Authors (MIT/CMU/Stanford) | 2024 | physics_engine | โ | โ | python | 29,334 | ๐ |
| MuJoCo | DeepMind / Google | 2022 | physics_engine | โ | โ | c, python, c++ | 13,859 | ๐ |
| Isaac Lab | NVIDIA | 2023 | rl_framework | โ | โ | python | 7,438 | ๐ |
| dm_control | DeepMind | 2018 | benchmark | โ | โ | python | 4,616 | ๐ |
| Brax | 2021 | physics_engine | โ | โ | python | 3,191 | ๐ | |
| Robosuite | Stanford / UT Austin | 2020 | benchmark | โ | โ | python | 2,462 | ๐ |
| RoboCasa | UT Austin | 2024 | full_stack | โ | โ | python | 1,461 | ๐ |
| SimplerEnv | UC Berkeley / Google | 2024 | benchmark | โ | โ | python | 1,087 | ๐ |
| SAPIEN | UC San Diego / others | 2020 | physics_engine | โ | โ | python, c++ | 780 | ๐ |
์ ํญ๋ชฉ์ ์ถ๊ฐํ๋ ค๋ฉด GitHub Issue๋ฅผ ์ด์ด์ฃผ์ธ์. To add a new entry, please open a GitHub Issue:
์ด์๊ฐ ๋ฑ๋ก๋๋ฉด ๋ด์ด ์๋์ผ๋ก PR์ ์์ฑํ๊ณ , ๊ด๋ฆฌ์๊ฐ ๊ฒํ ํ ๋จธ์งํฉ๋๋ค. A bot will automatically create a PR from your issue for admin review.
| Field | Valid Values |
|---|---|
categories |
manipulation ยท locomotion ยท navigation ยท dexterous ยท whole-body ยท aerial |
hardware |
manipulator ยท humanoid ยท quadruped ยท biped ยท mobile ยท drone ยท hand |
learning |
VLA ยท IL ยท RL ยท diffusion ยท world_model ยท sim2real |
framework |
pytorch ยท jax ยท tensorflow |
communication |
ros2 ยท grpc ยท lcm ยท zenoh ยท other |
| Field | Valid Values |
|---|---|
source |
real ยท simulation ยท teleoperation ยท human_demo ยท mocap |
modality |
rgb ยท rgbd ยท depth ยท lidar ยท tactile ยท proprioception ยท audio ยท force_torque |
๐ Stats auto-updated every Sunday via GitHub Actions ยท README auto-generated by scripts/generate_site.py