Code for paper: "iPad: Iterative Proposal-centric End-to-End Autonomous Driving"

Install

conda create -n ipad python=3.8
conda activate ipad
pip install torch==2.1.0 torchvision==0.16.0 torchaudio==2.1.0 --index-url https://download.pytorch.org/whl/cu121
pip install -e ./Bench2DriveZoo
pip install -e ./nuplan-devkit
pip install -e .

Set environment variable

set the environment variable based on where you place the PAD directory.

export NUPLAN_MAP_VERSION="nuplan-maps-v1.0"
export NUPLAN_MAPS_ROOT="$HOME/pad_workspace/dataset/maps"
export NAVSIM_EXP_ROOT="$HOME/pad_workspace/exp"
export NAVSIM_DEVKIT_ROOT="$HOME/pad_workspace/navsim"
export OPENSCENE_DATA_ROOT="$HOME/pad_workspace/dataset"
export Bench2Drive_ROOT="$HOME/pad_workspace/Bench2Drive"

Navsim training and evaluation

1. download the navtrain dataset and map as Navsim

bash download/download_maps.sh
bash download/download_navtrain.sh
bash download/download_navtest.sh

Put the downloaded maps in "dataset/maps", and dataset in "dataset/navsim_logs" and "dataset/sensor_blobs"

2. cache training data and metric

python navsim/planning/script/run_training_metric_caching.py
python navsim/planning/script/run_dataset_caching.py

Change "cache_data=True" in pad_agent during running "run_dataset_caching.py".

3. train navsim model

python navsim/planing/script/run_training.py

4. test navsim model

Change the checkpoint path in agent_config

python navsim/planing/script/run_create_submission_pickle.py

Then, submit the created "submission.pkl" to the official leaderboard on HuggingFace. Pretrained model

Bench2drive

1. download the base dataset as Bench2Drive

huggingface-cli download --repo-type dataset --resume-download rethinklab/Bench2Drive --local-dir Bench2Drive-Base

2. download and setup CARLA 0.9.15

    mkdir carla
    cd carla
    wget https://carla-releases.s3.us-east-005.backblazeb2.com/Linux/CARLA_0.9.15.tar.gz
    tar -xvf CARLA_0.9.15.tar.gz
    cd Import && wget https://carla-releases.s3.us-east-005.backblazeb2.com/Linux/AdditionalMaps_0.9.15.tar.gz
    cd .. && bash ImportAssets.sh
    export CARLA_ROOT=YOUR_CARLA_PATH
    echo "$CARLA_ROOT/PythonAPI/carla/dist/carla-0.9.15-py3.7-linux-x86_64.egg" >> YOUR_CONDA_PATH/envs/YOUR_CONDA_ENV_NAME/lib/python3.8/site-packages/carla.pth # python 3.8 also works well, please set YOUR_CONDA_PATH and YOUR_CONDA_ENV_NAME

3. Prepare Dataset

4. cache training data and metric

python Bench2Drive/leaderboard/pad_team_code/b2d_datacache.py
python Bench2Drive/leaderboard/pad_team_code/gen_mapinfo.py

5. train Bench2drive model

python navsim/planning/script/run_b2d_training.py

6. closeloop evaluation

cd Bench2Drive
python leaderboard/leaderboard/pad_eval.py
python tools/merge_route_json.py
python tools/ability_benchmark.py
python tools/efficiency_smoothness_benchmark.py