Our code implemented on Ubuntu 20.04 and Conda virtual environment.
Please follow below instruction.
git clone --recursive https://github.com/gist-ailab/uop-net.git
cd ./uop-net
# create conda env
conda create -n uop_net python=3.8
conda activate uop_net
# install python requirements
pip install --upgrade pip setuptools wheel
pip install trimesh plotly pyyaml open3d point_cloud_utils natsort tqdm scipy
pip install warmup-scheduler
# install pytorch with compatible CUDA version
conda install pytorch==1.12.1 torchvision==0.13.1 torchaudio==0.12.1 cudatoolkit=11.3 -c pytorch
uop data generation code built in CoppeliaSim and PyRep. Please install CoppeliaSim and pyrep first PyRep github
# download coppelia sim(check ubuntu version)
wget https://www.coppeliarobotics.com/files/V4_1_0/CoppeliaSim_Edu_V4_1_0_Ubuntu20_04.tar.xz --no-check-certificate
tar -xvf CoppeliaSim_Edu_V4_1_0_Ubuntu20_04.tar.xz
# at first, you must run with GUI monitor setup
cd CoppeliaSim_Edu_V4_1_0_Ubuntu20_04/
./coppeliaSim.sh
# edit ~/.bashrc file
export COPPELIASIM_ROOT=EDIT/ME/PATH/TO/COPPELIASIM/INSTALL/DIR
export LD_LIBRARY_PATH=$LD_LIBRARY_PATH:$COPPELIASIM_ROOT
export QT_QPA_PLATFORM_PLUGIN_PATH=$COPPELIASIM_ROOT
# install pyrep
cd setups/thirdparty/PyRep
pip3 install -r requirements.txt
pip3 install .
cd ../../..
You shold follow below thing to run and visualize our simulation scenes
-
video compressin library issue CoppeliaSim issue
- build vvcl and copy them to coppeliasim root
# start from uop-net folder cd setups/thirdparty/videoRecorder/vvcl mkdir build cd build cmake .. make # apply to CoppeliaSim cp ./libvvcl.so $COPPELIASIM_ROOT
-
custom simulation parameter warning(gui)
- For visualize simulation (headless=False), you should check the below gui box three times for each environments (./uop_sim/coppeliasim/*.ttt)
uop sim data use shapenet, 3dnet, ycb with watertight method Manifold.
# start from uop-net folder
cd setups/thirdparty/Manifold
mkdir build
cd build
cmake .. -DCMAKE_BUILD_TYPE=Release
make