A collection of simulation benchmarks for robot policy research

Currently we included robocasa, robomimic, rlbench, metaworld, libero. They all supports asynchronous vectorized policy evaluation. Find more instructions on each environment below. Also check conda_yaml for installation commands.

For MetaWorld

1. Add the following lines to ~/.bashrc

# mujoco
export LD_LIBRARY_PATH=$LD_LIBRARY_PATH:${HOME}/.mujoco/mujoco210/bin
export LD_LIBRARY_PATH=$LD_LIBRARY_PATH:/usr/lib/nvidia
export LD_LIBRARY_PATH=$LD_LIBRARY_PATH:/usr/local/cuda/lib64
export MUJOCO_GL=egl

2. Make it count

source ~/.bashrc

3. Download and install mujoco binary to location ${HOME}/.mujoco

$ cd ~/.mujoco

$ wget https://mujoco.org/download/mujoco210-linux-x86_64.tar.gz -O mujoco210.tar.gz --no-check-certificate

$ tar -xvzf mujoco210.tar.gz

4. Generate dataset with

python scripts/gen_data.py metaworld --help

For RLBench

1. Add the following lines to ~/.bashrc

# coppelia sim
export COPPELIASIM_ROOT=${HOME}/.coppeliasim
export LD_LIBRARY_PATH=$LD_LIBRARY_PATH:$COPPELIASIM_ROOT
export QT_QPA_PLATFORM_PLUGIN_PATH=$COPPELIASIM_ROOT

2. Make it count

source ~/.bashrc

3. Download and install CoppeliaSim binary to location ${HOME}/.coppeliasim

$ wget https://downloads.coppeliarobotics.com/V4_1_0/CoppeliaSim_Edu_V4_1_0_Ubuntu20_04.tar.xz

$ mkdir -p $COPPELIASIM_ROOT && tar -xf CoppeliaSim_Edu_V4_1_0_Ubuntu20_04.tar.xz -C $COPPELIASIM_ROOT --strip-components 1

$ rm -rf CoppeliaSim_Edu_V4_1_0_Ubuntu20_04.tar.xz

4. Generate dataset with

python scripts/gen_data.py rlbench --help

For LIBERO

1. find a desired path and clone the repo

git clone https://github.com/Lifelong-Robot-Learning/LIBERO.git

2. install

cd LIBERO
pip install -e .

3. Download datasets

python benchmark_scripts/download_libero_datasets.py --datasets libero_[spatial/object/goal/100]

4. Check scripts/convert_libero_dataset.py

For RoboCasa

1. Download assets with

python -m robocasa.scripts.download_kitchen_assets

2. Configure macros

python -m robocasa.scripts.setup_macros

3. Edit DATASET_BASE_PATH in robocasa/macros_private.py to desired data folder for data downloading
4. Download datasets

python -m robocasa.scripts.download_datasets --tasks [...] --ds_types [mg_im/human_im/human_raw]

5. Check scripts/convert_robocasa_dataset.py

For RoboMimic

1. Download datasets

python -m robomimic.scripts.download_datasets --download_dir data/robomimic/hdf5_datasets --tasks [...] --dataset_types ph --hdf5_types raw

2. Check scripts/convert_robomimic_dataset.py

Notes

• RoboCasa requires nightly robosuite, install this version

git+https://github.com/Chaoqi-LIU/robosuite.git@trajtok#egg=robosuite

• LIBERO requires robosuite==1.4.0, install with

pip install robosuite==1.4.0