Update README.md
Browse files
README.md
CHANGED
@@ -1,3 +1,38 @@
|
|
1 |
-
---
|
2 |
-
license: apache-2.0
|
3 |
-
---
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
---
|
2 |
+
license: apache-2.0
|
3 |
+
---
|
4 |
+
|
5 |
+
# RoboDual
|
6 |
+
|
7 |
+
This repo contains checkpoints for the specialist policy of our paper: \
|
8 |
+
**Towards Synergistic, Generalized and Efficient Dual-System for Robotic Manipulation**
|
9 |
+
|
10 |
+
|
11 |
+
## Evaluation with our checkpint
|
12 |
+
> First set your ```CALVIN_ROOT``` environment variable wtih:
|
13 |
+
```bash
|
14 |
+
export CALVIN_ROOT=/path/to/your/calvin_root_path
|
15 |
+
```
|
16 |
+
- Start evaluation on CALVIN (multi-GPU is also supported):
|
17 |
+
```bash
|
18 |
+
torchrun --standalone --nnodes 1 --nproc-per-node 1 vla-scripts/evaluate_calvin.py \
|
19 |
+
--generalist_path "/path/to/calvin_generalist" \
|
20 |
+
--specialist_path "/path/to/calvin_specialist" \
|
21 |
+
--with_depth \ # use depth input
|
22 |
+
--with_gripper \ # use gripper-view inputs (both RGB and depth)
|
23 |
+
--with_cfg \ # enable classifier-free guidance
|
24 |
+
--log_dir calvin
|
25 |
+
```
|
26 |
+
> Please refer to ```vla-scripts/evaluate_calvin.py``` in our [github repo](https://github.com/OpenDriveLab/RoboDual) for all evaluation options.
|
27 |
+
|
28 |
+
|
29 |
+
## Citation
|
30 |
+
If you find our code or models useful in your work, please cite [our paper](https://arxiv.org/abs/2410.08001):
|
31 |
+
|
32 |
+
```bibtex
|
33 |
+
@article{bu2024robodual,
|
34 |
+
title={Towards Synergistic, Generalized, and Efficient Dual-System for Robotic Manipulation},
|
35 |
+
author={Qingwen Bu and Hongyang Li and Li Chen and Jisong Cai and Jia Zeng and Heming Cui and Maoqing Yao and Yu Qiao},
|
36 |
+
journal={arXiv preprint arXiv:2410.08001},
|
37 |
+
year={2024}
|
38 |
+
}
|