‍This repository will be part of the Autonomy.Hub Ecosystem

As part of the Autonomy.Hub Ecosystem, Autonomy.Datasets enables the Automated Driving community to easily test their automated driving building blocks across different datasets:

🔄 Unified ROS 2 Interface: Work with multiple datasets using the benefits of the ROS 2 ecosystem
📊 Comprehensive Benchmarks: Use the provided datasets with Autonomy.Benchmarks to benchmark building blocks across different automated driving tasks
⚡ Efficient Data Pipeline: Preprocessed Rosbag files ensure fast execution during development
🐳 Dockerized Environment: Reproducible setup with all dependencies included
🔌 Modular Architecture: Easy integration with other ROS 2 packages

Supported Datasets

This repository supports various automated driving datasets.

‍Contributions adding more datasets are welcome

Dataset	Release	Countries	Samples
Waymo Open Dataset	August 2019	United States	158.081 Training39.987 Validation
nuScenes	March 2019	United States (Boston), Singapore	28.130 Training6.019 Validation
MAN TruckScenes	July 2024	Germany	747 scenes of 20 seconds each, annotated at 2 Hz with 6 lidars, 6 radars and 4 cameras
NVIDIA Physical AI AV Dataset (Alpamayo)	October 2025	United States, Germany, France, Italy, Sweden, Spain, Portugal, Greece, Austria, Finland, Croatia, Netherlands, Denmark, Slovenia, Estonia, Slovakia, Belgium, Czechia, Lithuania, Poland, Romania, Luxembourg, Latvia, Hungary, Bulgaria	approx. 17.016.400 samples from 85.082 clips, each 20 seconds (10 Hz) with 1 lidar, 7 cameras and up to 10 radars
DrivIng	January 2026	Germany (Ingolstadt)	3 sequences (day, dusk, night) at 10 Hz with 1 lidar and 6 cameras
TUM Traffic	April 2022	Germany (A9 motorway and S110 intersection near Munich)	Roadside infrastructure subsets (releases `R00` to `R02`) with up to 4 cameras and 2 lidars per sensor station
Zenseact Open Dataset	May 2023	14 European countries (Sweden, Germany, Poland, Italy, ...)	100.000 annotated frames, 1.473 sequences of 20 seconds and 29 drives of a few minutes, each at 10 Hz with 3 lidars and 1 camera
FZI-AURA	September 2026	Germany (southern)	2.473 scenes of 20 seconds at 10 Hz with up to 12 lidars, 8 cameras and 3 radars; 4.106.333 3D boxes and 30 billion semantic lidar points on 2 Hz keyframes

🚀 Quick Start • 💻 Development • 📝 Documentation

🚀 Quick Start

The autonomy_datasets package is available in a pre-compiled Docker image. Start a container mounting your local dataset directory. Alternatively, use VS Code to open this repository in a Devcontainer.

‍Follow the instructions in the Supported Datasets section to obtain the dataset.

xhost +local:  # allow graphical output for RViz visualization
DATASET_DIR="$HOME/datasets"  # adapt this to your dataset location
docker run --rm -it --gpus all --env=DISPLAY --volume=/tmp/.X11-unix:/tmp/.X11-unix:rw --volume $DATASET_DIR:/datasets ghcr.io/thinking-cars/autonomy_datasets:latest bash

Run the following command in the container to visualize samples from the FZI-AURA dataset:

hf auth login # login with your HuggingFace account

ros2 launch autonomy_datasets autonomy_datasets.launch.py

This will download all selected scenes sequentially, write samples into Rosbags at $DATASET_DIR/nvidia_physicalai_av_dataset/bags/<version> while visualizing samples in Rviz. Rosbags are stored in a subfolder named after the version of the dataset conversion. Existing Rosbags of the current version are replayed instead of being generated again; a new version generates its Rosbags into its own subfolder.

💻 Development

Set up Development Environment

Clone the repository.
git clone https://github.com/thinking-cars/autonomy_datasets.git

Initialize the .openads-dev-environment submodule containing development environment configuration.
cd autonomy_datasets

git submodule update --init --recursive

Open the repository in Visual Studio Code.
code .

Install the recommended VS Code extensions.

‍Ctrl+Shift+P / Extensions: Show Recommended Extensions / Install Workspace Recommended Extensions (Cloud Download Icon)

Reopen the repository in a Dev Container. Ctrl+Shift+P / Dev Containers: Rebuild and Reopen in Container

Build

‍Ctrl+Shift+B

colcon build

Run Tests

‍Ctrl+Shift+P / Tasks: Run Test Task

colcon build --cmake-args -DCMAKE_EXPORT_COMPILE_COMMANDS=1
colcon test
colcon test-result --verbose

📝 Documentation

Package and node interfaces are documented in the respective package READMEs listed below. Implementation details are found in the Source Code Documentation.

Package	Description
autonomy_datasets	Integrates automated driving datasets into the ROS 2 ecosystem
autonomy_datasets_msgs	Message definitions for dataset meta information that has no representation in perception_msgs
autonomy_datasets_rviz_plugins	RViz plugins to control the playback of the datasets published by autonomy_datasets

⚖️ Licensing

The source code in this repository is licensed under Apache-2.0, see [LICENSE](LICENSE). Container images provided by this repository may contain third-party software shipped with their own license terms.

‍⚠️ IMPORTANT DATASET LICENSE DISCLAIMER

This repository provides tools and interfaces for working with autonomous driving datasets. The actual datasets (nuScenes, Waymo Open Dataset, etc.) are NOT included and must be obtained separately.

Before using any dataset, you MUST:

Register and accept the terms of use for each dataset you wish to use

Download the datasets from their official sources

Comply with all licensing terms and conditions of the respective dataset providers

Dataset-specific requirements:

nuScenes: Register at nuScenes.org and agree to the nuScenes Terms of Use

Waymo Open Dataset: Register at Waymo Open Dataset and agree to their License Agreement

NVIDIA Physical AI Autonomous Vehicles Dataset: Register at HuggingFace and agree to the NVIDIA Autonomous Vehicles Dataset License Agreement

DrivIng: Downloaded automatically from Harvard Dataverse; usage is subject to CC BY-NC-ND 4.0

MAN TruckScenes: Downloaded automatically from the AWS Open Data registry; usage is subject to CC BY-NC-SA 4.0

TUM Traffic: Register at a9-dataset.innovation-mobility.com, agree to the license, and download the archives manually; usage is subject to CC BY-NC-ND 4.0

Zenseact Open Dataset: Apply for access to receive a personal download link, which the adapter uses to download the dataset; usage is subject to CC BY-SA 4.0 and the dataset is not intended for military use

FZI-AURA: usage is subject to CC BY-SA 4.0, which permits commercial use provided the dataset is attributed and adaptations are shared under the same license. The contextual metadata additionally derives from OpenStreetMap (ODbL 1.0) and OpenWeather; see THIRD_PARTY_NOTICES.MD for their attribution requirements

🙏 Acknowledgements

This project is maintained by Thinking Cars. We appreciate contributions and are happy to discuss potential collaborations.