This repository contains the official implementation of the following paper:

(RA-L 2022) Learning Closed-loop Dough Manipulation Using a Differentiable Reset Module

Carl Qi, Xingyu Lin, David Held

Website / Paper

1. Install python3 -m pip install -e .
2. Install torch (version 1.9.0 tested)
   • We tested pip install torch==1.9.0+cu111 torchvision==0.10.0+cu111 torchaudio==0.9.0 -f https://download.pytorch.org/whl/torch_stable.html on RTX 3090.
3. Install packages for computing the EMD loss:
   • pykeops (1.5) by running pip install pykeops==1.5
   • geomloss by running pip install geomloss
4. Install chester from https://github.com/Xingyu-Lin/chester.
5. Run python imitation/launchers/launch_gen_data.py to run Diff-Reset.

If you find this codebase useful in your research, please consider citing:

@ARTICLE{qi2022dough,
  author={Qi, Carl and Lin, Xingyu and Held, David},
  journal={IEEE Robotics and Automation Letters}, 
  title={Learning Closed-Loop Dough Manipulation Using a Differentiable Reset Module}, 
  year={2022},
  volume={7},
  number={4},
  pages={9857-9864},
  doi={10.1109/LRA.2022.3191239}}