Cross-dataset Training

LEAD supports cross-dataset training as a side-feature, allowing you to train on datasets beyond CARLA.

Table of Contents

Setup Waymo Data
Setup NavSim Data
Training on Waymo
Training on NavSim

Setup Waymo Data

Follow this step if you want to use Waymo E2E and Waymo Perception datasets.

Install Google Cloud CLI

cd ..
curl -O https://dl.google.com/dl/cloudsdk/channels/rapid/downloads/google-cloud-cli-linux-x86_64.tar.gz
tar -xf google-cloud-cli-linux-x86_64.tar.gz
./google-cloud-sdk/install.sh
gcloud init

Install Additional Python Packages

pip install waymo-open-dataset-tf-2-12-0==1.6.7
pip install numpy==1.26.0 # Reinstall the correct numpy

Download Waymo Data

TODO: Add Waymo download instructions

Setup NavSim Data

Follow this step if you want to use the navtrain split.

Download Data

First download data and organize them:

cd 3rd_party/navsim_workspace/dataset
bash 3rd_party/navsim_workspace/navsimv1.1/download/download_navtrain_parallel.sh
bash 3rd_party/navsim_workspace/navsimv1.1/download/download_test_parallel.sh
bash 3rd_party/navsim_workspace/navsimv2.2/download/download_navhard_two_stage.sh
bash 3rd_party/navsim_workspace/navsimv1.1/download/download_maps.sh

Build Cache

conda activate navsimv1.1
sbatch scripts/tools/data/navsim/001_navtest_cache.sh
conda activate navsimv2.2
sbatch scripts/tools/data/navsim/002_navhard_cache.sh

Training on Waymo

TODO: Add Waymo training instructions

Training on NavSim

TODO: Add NavSim training instructions

Cross-Dataset Evaluation

TODO: Add cross-dataset evaluation instructions