MetaDrive

MetaDrive Simulator

To facilitate the research of generalizable reinforcement learning, we develop an open-source, highly efficient and flexible driving simulator MetaDrive, which holds the following key features:

• Modular
• Lightweight
• Customizable
• Realistic

We construct a variety of RL tasks and baselines in both single-agent and multi-agent settings, including benchmarking generalizability across unseen scenes, safe exploration, and modeling multi-agent behaviors.

Empowered by ScenarioNet, all features of MetaDrive can be applied to the virtual environments reconstructed from the open-source dataset, such as Waymo Open Dataset, nuPlan, and L5.

Reference

Please refer to the technical report or the code repo for more information.

@article{li2021metadrive,
  title={Metadrive: Composing diverse driving scenarios for generalizable reinforcement learning},
  author={Li, Quanyi and Peng, Zhenghao and Feng, Lan and Zhang, Qihang and Xue, Zhenghai and Zhou, Bolei},
  journal={IEEE Transactions on Pattern Analysis and Machine Intelligence},
  year={2022}
}