/DexRep_Isaac

Primary LanguagePython

Introduction

The repository implements the grasping task of DexRepNet++ in the Isaac simulator.

Dependencies

  • Create a conda environment

    conda create -n your_env_name python==3.8
conda activate your_env_name

  • Install torch

    pip install torch==1.13.0+cu117 torchvision==0.14.0+cu117 torchaudio==0.13.0 --extra-index-url https://download.pytorch.org/whl/cu117

  • Install IsaacGym

    1. Download isaacgym
    2. Extract the downloaded files to the main directory of the project
    3. Use the following commands to install isaacgym
      cd isaacgym/python
  pip install -e .

  • Install DexRep

    cd dexgrasp
pip install -e .

The above commands show how to install the major packages. You can install other packages by yourself if needed.

Run the scripts

cd dexgrasp
python train.py --task=ShadowHandGraspDexRep --algo=ppo1 --seed=0 --rl_device=cuda:0 --sim_device=cuda:0 --logdir=logs/dexrep

Notes:

  • If you do not want to open the simulator windows, add --headless
  • More parameters can be found in dexgrasp/cfg/shadow_hand_grasp_dexrep.yaml