Top AI Repos — open-source AI, indexed and scored
Top AI Repos tracks AI repositories on GitHub and answers two different questions about each one: is it moving right now, and would you bet a product on it.
Top AI Repos tracks AI repositories on GitHub and answers two different questions about each one: is it moving right now, and would you bet a product on it.
UMI-3D SLAM and Data Processing Pipeline: https://umi-3d.github.io/
| Date | Stars |
|---|---|
| 2026-07-30 | 259 |
| 2026-08-09 | 264 |
| 2026-08-17 | 265 |
| 2026-08-18 | 266 |
| 2026-08-20 | 267 |
| 2026-08-22 | 268 |
| 2026-08-25 | 269 |
| 2026-08-26 | 270 |
| 2026-09-01 | 271 |
| 2026-09-03 | 272 |
| 2026-09-09 | 274 |
| 2026-09-10 | 267 |
| 2026-09-11 | 270 |
| 2026-09-12 | 271 |
| 2026-09-13 | 273 |
| 2026-09-14 | 275 |
| 2026-09-15 | 274 |
| 2026-09-16 | 276 |
| 2026-09-17 | 277 |
| 2026-09-25 | 278 |
| 2026-10-02 | 279 |
| 2026-10-14 | 276 |
| 2026-10-16 | 277 |
| 2026-10-17 | 278 |
| 2026-10-18 | 280 |
| 2026-10-22 | 281 |
| 2026-11-04 | 281 |
Today
— stars today
This week
— stars this week
This month
+2 stars this month
Momentum
0.0
growth rate 0.00%/day
# UMI-3D SLAM and Data Processing
<div align="center">
<h3>
🌐 <a href="https://umi-3d.github.io/">UMI-3D Project Homepage</a>
</h3>
<table>
<tr>
<td align="center" width="33%">
<a href="https://github.com/Physical-Intelligence-Laboratory/UMI-3D-Hardware">
<b>🔧 UMI-3D Hardware</b>
</a>
</td>
<td align="center" width="33%">
<a href="https://github.com/hku-mars/UMI-3D">
<b>🛰️ UMI-3D SLAM Pipeline</b>
</a>
</td>
<td align="center" width="33%">
<a href="https://github.com/Physical-Intelligence-Laboratory/UMI-3D-Policy">
<b>🤖 UMI-3D Policy</b>
</a>
</td>
</tr>
<tr>
<td align="center">
<a href="https://github.com/Physical-Intelligence-Laboratory/UMI-3D-Hardware">
<img src="docs/assets/nav_hardware.jpg" width="100%" alt="UMI-3D Hardware"/>
</a>
</td>
<td align="center">
<a href="https://github.com/hku-mars/UMI-3D">
<img src="docs/assets/nav_processing.jpg" width="100%" alt="UMI-3D SLAM Pipeline"/>
</a>
</td>
<td align="center">
<a href="https://github.com/Physical-Intelligence-Laboratory/UMI-3D-Policy">
<img src="docs/assets/nav_policy.jpg" width="100%" alt="UMI-3D Policy"/>
</a>
</td>
</tr>
<tr>
<td align="center">
Hardware design, BOM, CAD, 3D-print parts
</td>
<td align="center">
SLAM, synchronization, calibration, and data processing
</td>
<td align="center">
Policy training, deployment, inference<br><br>
<a href="https://github.com/Physical-Intelligence-Laboratory/UMI-3D-Dataset">
<b>📦 Dataset & Models</b>
</a>
</td>
</tr>
</table>
</div>
## Overview
UMI-3D provides a complete end-to-end pipeline, transforming **raw rosbag recordings** into **training-ready datasets** for embodied manipulation learning:
```
Collected rosbag Files
↓
Calibration
↓
SLAM
↓
Aligned Demos
↓
Dataset Pipeline
↓
Zarr Dataset (for policy training, e.g. Diffusion Policy)
```
## 0. Complete Data Collection
To build the UMI-3D data collection system, please follow the hardware assembly and sensor setup instructions in:
👉 https://github.com/Physical-Intelligence-Laboratory/UMI-3D-Hardware
UMI-3D collects two types of data:
1. **Demonstration data**
Human-guided manipulation trajectories captured during task execution.
2. **Gripper calibration data**
Slowly open and close the gripper for approximately 5 cycles to estimate gripper motion range.
<div align="center">
<img src="docs/assets/cup_demo_gif.gif" width="64%" />
<img src="docs/assets/gripper.gif" width="34%" />
</div>
All data are recorded as **rosbag files**, including:
- LiDAR point clouds
- IMU measurements
- Camera images
These recordings serve as the raw input for the full UMI-3D data processing pipeline.
## 1. Sensors Calibration
<div align="center">
<img src="sensors_calibration/fisheye_intrinsics/pics/fisheye_calib_guide.jpg" width="76%"/>
<img src="sensors_calibration/fisheye_intrinsics/pics/checkboard_6x9.jpeg" width="15%"/>
</div>
### 1.1 Fisheye Camera Intrinsic Calibration
**Step 1 — Collect calibration images**
- Use a checkerboard (**6 × 9 inner corners**, square size = **0.1 m**, configurable in script)
- Capture ≥ 100 images with different positions (center / edges / corners), orientations, and distances
- Save all images to `fisheye_intrinsics/images/`
**Step 2 — Run calibration**
```bash
cd fisheye_intrinsics
python3 calibrate_fisheye_intrinsics.py \
--image_glob "images/*.png" \
--checkerboard_cols 6 \
--checkerboard_rows 9 \
--square_size 0.10 \
--output_dir calib_output
```
Calibration results will be saved to: `fisheye_intrinsics/calib_output/`
### 1.2 LiDAR–Camera Extrinsic Calibration
<div align="center">
<img src="sensors_calibration/livox2cam_calibration/src/pics/livox2cam_guide.jpg" width="90%"/>
</div>
This step estimates the rigid transformaExcerpt of 13,155 characters
Read on GitHubWould you bet a product on this? Bounded 0–100 and slow moving.
matched fp:ce2fe712835d2f78, topic:embodied-ai, topic:manipulation, topic:slam