sunzhenguo commited on
Commit
9f434d7
·
verified ·
1 Parent(s): 7952d07

Update README.md

Browse files
Files changed (1) hide show
  1. README.md +60 -1
README.md CHANGED
@@ -9,4 +9,63 @@ tags:
9
  - teleoperation
10
  - reinforcement-learning
11
  arxiv: 2602.08594
12
- ---
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
9
  - teleoperation
10
  - reinforcement-learning
11
  arxiv: 2602.08594
12
+ ---
13
+
14
+ # MOSAIC Model
15
+
16
+ [**Project Page**](https://baai-humanoid.github.io/MOSAIC/) | [**Paper**](https://arxiv.org/abs/2602.08594) | [**Code**](https://github.com/BAAI-Humanoid/MOSAIC) | [**Dataset**](https://huggingface.co/datasets/BAAI-Humanoid/MOSAIC_Dataset) | [**Model**](https://huggingface.co/BAAI-Humanoid/MOSAIC_Model)
17
+
18
+ This repository releases **deployment-ready ONNX models** for **MOSAIC**, introduced in:
19
+
20
+ **[MOSAIC: Bridging the Sim-to-Real Gap in Generalist Humanoid Motion Tracking and Teleoperation with Rapid Residual Adaptation](https://arxiv.org/abs/2602.08594)**
21
+
22
+ In MOSAIC, a **general motion tracker** is trained in simulation, and **interface-specific adaptation** is handled via a lightweight **residual adaptor** that injects action-space corrections while preserving the general policy’s capabilities.
23
+
24
+ It includes models:
25
+ 1. **gmt.onnx**: the general motion tracking policy
26
+ 2. **noitom_teleop.onnx**: the adaptor policy for teleoperation using Noitom inertial mocap suit
27
+ 3. **pico_teleop.onnx**: the adaptor policy for teleoperation using PICO VR device
28
+
29
+ ---
30
+ ## How to download
31
+
32
+ ### Option A: Download a single ONNX file
33
+ ```python
34
+ from huggingface_hub import hf_hub_download
35
+
36
+ onnx_path = hf_hub_download(
37
+ repo_id="BAAI-Humanoid/MOSAIC_Model",
38
+ filename="pico_teleop.onnx", # or "noitom_teleop.onnx"
39
+ )
40
+ print("Downloaded to:", onnx_path)
41
+ ```
42
+ ### Option B: Download all files in this model repo
43
+ ```python
44
+ from huggingface_hub import snapshot_download
45
+
46
+ local_dir = snapshot_download(
47
+ repo_id="BAAI-Humanoid/MOSAIC_Model",
48
+ )
49
+ print("Downloaded to:", local_dir)
50
+ ```
51
+ ---
52
+
53
+ ## Usage
54
+
55
+ For constructing the correct deployable observations and mapping model outputs to Unitree G1 control targets, please use the official [MOSAIC codebase](https://github.com/BAAI-Humanoid/MOSAIC) and [RobotBridge Deployment framework](https://github.com/BAAI-Humanoid/RobotBridge).
56
+
57
+ ---
58
+ ## Citation
59
+ If you use these models for your research, please cite our paper:
60
+ ```bibtex
61
+ @article{sun2026mosaic,
62
+ title = {MOSAIC: Bridging the Sim-to-Real Gap in Generalist Humanoid Motion Tracking and Teleoperation with Rapid Residual Adaptation},
63
+ author = {Zhenguo Sun and Bo-Sheng Huang and Yibo Peng and Xukun Li and Jingyu Ma and Yu Sun and Zhe Li and Haojun Jiang and Biao Gao and Zhenshan Bing and Xinlong Wang and Alois Knoll},
64
+ journal = {arXiv preprint arXiv:2602.08594},
65
+ year = {2026}
66
+ }
67
+ ```
68
+ ---
69
+
70
+ ## License
71
+ This dataset is released under Apache-2.0.