Welcome to the official repository of BestMan, a mobile manipulator simulator (with a wheel-base and arm) built on PyBullet.
- Pull the repository and update the submodule
git clone https://github.com/AutonoBot-Lab/BestMan_Pybullet.git
cd BestMan_Pybullet
git submodule init
git submodule update
First install Anaconda
or minconda
on linux system and then perform the following steps:
- Run the following script to add the project to the PYTHON search path
cd Install
chmod 777 pythonpath.sh
bash pythonpath.sh
source ~/.bashrc
- Install ffmpeg to enable video record
sudo apt update && sudo apt install ffmpeg
- Configure related libraries and links to support OpenGL rendering (If it already exists, skip this step.)
sudo apt update && sudo apt install -y libgl1-mesa-glx libglib2.0-0
sudo mkdir /usr/lib/dri
sudo ln -s /lib/x86_64-linux-gnu/dri/swrast_dri.so /usr/lib/dri/swrast_dri.so
- Install gcc/g++ 9 (If it already exists, skip this step.)
sudo apt install -y build-essential gcc-9 g++-9
sudo update-alternatives --install /usr/bin/gcc gcc /usr/bin/gcc-9 9
sudo update-alternatives --install /usr/bin/g++ g++ /usr/bin/g++-9 9
sudo update-alternatives --config gcc # choice gcc-9
sudo update-alternatives --config g++ # choice g++-9
# Make sure gcc and g++ versions are consistent (conda enviroment don't install gcc to prevent problems caused by inconsistent versions)
gcc -v
g++ -v
- Configure mamba to speed up the conda environment construction (Optional, skip if installation is slow or fails)
conda install mamba -n base -c conda-forge
- Create basic conda environment
conda(mamba) env create -f basic_env.yaml
conda(mamba) activate BestMan
- Install torch
conda(mamba) env update -f cuda116.yaml
- Install lang-segment-anything
pip install -U git+https://github.com/luca-medeiros/lang-segment-anything.git
- Install AnyGrasp
Note: you need export MAX_JOBS=2
in terminal; before pip install if you are running on an laptop due to this issue.
# Install MinkowskiEngine
conda install pytorch=1.13.1 -c pytorch --force-reinstall
pip install -U git+https://github.com/NVIDIA/MinkowskiEngine -v --no-deps --global-option="--blas_include_dirs=${CONDA_PREFIX}/include" --global-option="--blas=openblas"
# Install graspnetAPI
pip install graspnetAPI
# Install pointnet2
cd third_party/pointnet2
python setup.py install
# Force reinstall to ensure version
pip install --force-reinstall opencv-python==4.1.2.30 numpy==1.23.5
-
AnyGrasp License
You need to get anygrasp license and checkpoint to use it.
- Pull docker image from tencentyun
docker pull ccr.ccs.tencentyun.com/4090/bestman:v1
- Create docker container
docker run -it --gpus all --name BestMan ccr.ccs.tencentyun.com/4090/bestman:v1
-
Install VcXsrv Windows X Server, Start and keep running in the background.
-
Execute
echo $DISPLAY
inside the container, Make sure the result ishost.docker.internal:0
so that it can be visualized on the host machine, if not:
export DISPLAY=host.docker.internal:0
- TBD
First, Enter directory Examples:
cd Examples
Below are some examples and their rendering in Blender
☘️ Navigation
python navigation_basic.py
navigation_basic.mp4
☘️ Manipulation
- Open Fridge
python open_fridge.py
open_fridge.mp4
- Open microwave
python open_microwave.py
open_microwave.mp4
- Grasp bowl on table use vacuum_gripper
python grasp_bowl_on_table_vacuum_gripper.py
grasp_bowl_on_table_sucker.mp4
- Grasp lego on table use gripper
python grasp_lego_on_table_gripper.py
grasp_lego_on_table_gripper.mp4
- Move bowl from drawer to table
python move_bowl_from_drawer_to_table.py
move_bowl_from_drawer_to_table.mp4
open microwave
demo with blender render:
open_microwave.mp4
We have improved the pybullet-blender-recorder to import pybullet scene into blender for better rendering
If you want to enable pybullet-blender-recorder, please:
-
Set
blender: Ture
in Config/default.yaml -
After running the demo, a pkl file will be generated and saved in Examples/record dir
-
Install the
pyBulletSimImporter.py
plugin under Visualization/blender-render dir in blender (test on blender3.6.5) , and enalbe this plugin
- Import the pkl files into blender
If you find this work useful, please consider citing:
@inproceedings{ding2023task,
title={Task and motion planning with large language models for object rearrangement},
author={Ding, Yan and Zhang, Xiaohan and Paxton, Chris and Zhang, Shiqi},
booktitle={2023 IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS)},
pages={2086--2092},
year={2023},
organization={IEEE}
}
@article{ding2023integrating,
title={Integrating action knowledge and LLMs for task planning and situation handling in open worlds},
author={Ding, Yan and Zhang, Xiaohan and Amiri, Saeid and Cao, Nieqing and Yang, Hao and Kaminski, Andy and Esselink, Chad and Zhang, Shiqi},
journal={Autonomous Robots},
volume={47},
number={8},
pages={981--997},
year={2023},
publisher={Springer}
}