/scenarionet

ScenarioNet: Scalable Traffic Scenario Management System for Autonomous Driving

Primary LanguagePythonApache License 2.0Apache-2.0

ScenarioNet

Documentation Status build GitHub license

Open-Source Platform for Large-Scale Traffic Scenario Simulation and Modeling

[ Webpage | Code | Video | Paper | Documentation ]

Colab example for running simulation with ScenarioNet: Open In Colab

Colab example for reading established ScenarioNet dataset: Open In Colab

ScenarioNet allows users to load scenarios from real-world datasets like Waymo, nuPlan, nuScenes, l5 and synthetic dataset such as procedural generated ones and safety-critical ones generated by adversarial attack. The built database provides tools for building training and test sets for ML applications.

Powered by MetaDrive Simulator, the scenarios can be reconstructed for various applications like AD stack test, reinforcement learning, imitation learning, scenario generation and so on.

system

Installation

The detailed installation guidance is available at documentation. A simplest way to do this is as follows.

# create environment
conda create -n scenarionet python=3.9
conda activate scenarionet

# Install MetaDrive Simulator
cd ~/  # Go to the folder you want to host these two repos.
git clone https://github.com/metadriverse/metadrive.git
cd metadrive
pip install -e.

# Install ScenarioNet
cd ~/  # Go to the folder you want to host these two repos.
git clone https://github.com/metadriverse/scenarionet.git
cd scenarionet
pip install -e .

API reference

All operations and API reference is available at our documentation. If you already have ScenarioNet installed, you can check all operations by python -m scenarionet.list.

ScenarioNet dataset and Scenario Description

Please refer to the Scenario Description section in MetaDrive documentation for a walk-through.

Citation

If you used this project in your research, please cite:

@article{li2023scenarionet,
  title={ScenarioNet: Open-Source Platform for Large-Scale Traffic Scenario Simulation and Modeling},
  author={Li, Quanyi and Peng, Zhenghao and Feng, Lan and Liu, Zhizheng and Duan, Chenda and Mo, Wenjie and Zhou, Bolei},
  journal={Advances in Neural Information Processing Systems},
  year={2023}
}