This is the first version of World4Drive (End-to-End Autonomous Driving via Intention-aware Physical Latent World Model).
a. Create a conda virtual environment and activate it.
conda create -n w4d python=3.8 -y
conda activate w4db. Install PyTorch and torchvision following the official instructions.
pip install -r requirements.txt
pip install torch==1.9.1+cu111 torchvision==0.10.1+cu111 torchaudio==0.9.1 -f https://download.pytorch.org/whl/torch_stable.htmlc. Install gcc>=5 in conda env (optional).
conda install -c omgarcia gcc-6 # gcc-6.2c. Install mmcv-full.
pip install mmcv-full==1.4.0
d. Install mmdet and mmseg.
pip install mmdet==2.14.0
pip install mmsegmentation==0.14.1e. Install timm.
pip install timmf. Install mmdet3d.
conda activate w4d
git clone https://github.com/open-mmlab/mmdetection3d.git
cd /path/to/mmdetection3d
git checkout -f v0.17.1
python setup.py developg. Install nuscenes-devkit.
pip install nuscenes-devkit==1.1.9
pip install yapf==0.40.1h. Install other dependencies.
pip install -r requirements.txtFollow VAD to organize nuscenes and download the pickle
$WORK_DIR/data/nuscenes/For details, please refer to docs/prepare_dataset.md
./tools/nusc_my_train.sh w4d/default 8 ./tools/dist_test $CONFIG $CKPT $NUM_GPU