This repository contains learning-based locomotion control research from OpenRobotLab, currently including Hybrid Internal Model & H-Infinity Locomotion Control.
- [2024-04] Code of HIMLoco is released.
- [2024-04] We release the paper of H-Infinity Locomotion Control. Please check the 👉 webpage 👈 and view our demos! 🎇
- [2024-01] HIMLoco is accepted by ICLR 2024.
- [2023-12] We release the paper of HIMLoco. Please check the 👉 webpage 👈 and view our demos! 🎇
- Release the training code of HIMLoco, please see
rsl_rl/rsl_rl/algorithms/him_ppo.py
. - Release deployment guidance of HIMLoco.
- Release the training code of H-Infinity Locomotion Control.
- Release deployment guidance of H-Infinity Locomotion Control.
We test our codes under the following environment:
- Ubuntu 20.04
- NVIDIA Driver: 525.147.05
- CUDA 12.0
- Python 3.7.16
- PyTorch 1.10.0+cu113
- Isaac Gym: Preview 4
- Create an environment and install PyTorch:
conda create -n himloco python=3.7.16
conda activate himloco
pip3 install torch==1.10.0+cu113 torchvision==0.11.1+cu113 torchaudio==0.10.0+cu113 -f https://download.pytorch.org/whl/cu113/torch_stable.html
- Install Isaac Gym:
- Download and install Isaac Gym Preview 4 from https://developer.nvidia.com/isaac-gym
cd isaacgym/python && pip install -e .
- Clone this repository.
git clone https://github.com/OpenRobotLab/HIMLoco.git
cd HIMLoco
- Install HIMLoco.
cd rsl_rl && pip install -e .
cd ../legged_gym && pip install -e .
Note: Please use legged_gym and rsl_rl provided in this repo, we have modefications on these repos.
- Train a policy:
cd legged_gym/legged_gym/scripts
python train.py
- Play and export the latest policy:
cd legged_gym/legged_gym/scripts
python play.py
If you find our work helpful, please cite:
@inproceedings{long2023him,
title={Hybrid Internal Model: Learning Agile Legged Locomotion with Simulated Robot Response},
author={Long, Junfeng and Wang, ZiRui and Li, Quanyi and Cao, Liu and Gao, Jiawei and Pang, Jiangmiao},
booktitle={The Twelfth International Conference on Learning Representations},
year={2024}
}
@misc{long2024hinf,
title={Learning H-Infinity Locomotion Control},
author={Junfeng Long and Wenye Yu and Quanyi Li and Zirui Wang and Dahua Lin and Jiangmiao Pang},
year={2024},
eprint={2404.14405},
archivePrefix={arXiv},
}
This work is under the Creative Commons Attribution-NonCommercial-ShareAlike 4.0 International License.
- legged_gym: Our codebase is built upon legged_gym.