README.md

CEURL

This is the Official implementation for "PEAC: Unsupervised Pre-training for Cross-Embodiment Reinforcement Learning" (NeurIPS 2024)

State-based DMC & Image-based DMC

Installation

The code is based on URLB

You can create an anaconda environment and install all required dependencies by running

conda create -n ceurl python=3.8
conda activate ceurl
pip install -r requirements.txt
pip install torch==1.8.0+cu111 torchvision==0.9.0+cu111 -f https://download.pytorch.org/whl/torch_stable.html

Instructions

The simplest way to try PEAC in three embodiment distributions of state-based DMC by running

cd DMC_state
chmod +x train_finetune.sh

./train_finetune.sh peac walker_mass 0
./train_finetune.sh peac quadruped_mass 0
./train_finetune.sh peac quadruped_damping 0

The simplest way to try PEAC in three embodiment distributions of image-based DMC by running

cd DMC_image
chmod +x train_finetune.sh

./train_finetune.sh peac_lbs walker_mass 0
./train_finetune.sh peac_lbs quadruped_mass 0
./train_finetune.sh peac_lbs quadruped_damping 0

./train_finetune.sh peac_diayn walker_mass 0
./train_finetune.sh peac_diayn quadruped_mass 0
./train_finetune.sh peac_diayn quadruped_damping 0

Citation

If you find this work helpful, please cite our paper.

@article{ying2024peac,
  title={PEAC: Unsupervised Pre-training for Cross-Embodiment Reinforcement Learning},
  author={Ying, Chengyang and Hao, Zhongkai and Zhou, Xinning and Xu, Xuezhou and Su, Hang and Zhang, Xingxing and Zhu, Jun},
  journal={arXiv preprint arXiv:2405.14073},
  year={2024}
}