Sparrow-V3.0: A Reinforcement Learning Friendly Simulator for Multiple Mobile Robots

Dependency

numpy>=1.24.4
scipy>=1.10.1
pygame>=2.5.2
torch>=2.2.0 # GPU version is recommended
torchaudio>=2.2.0
torchvision>=0.17.0

Play with keyboard

python play_with_keyboard.py

Play with trained model

python main.py

Train your own model

Please refer to ColorDynamic

Documentations

• Please refer to this repo.

Main differences from Sparrow-V2

• Multiple robots are supported

The Sparrow families

• Sparrow-V1: Single Robot, Static environments
• Sparrow-V2: Single Robot, Dynamic/Static environments
• Sparrow-V3: Multiple/Single Robot, Dynamic/Static environments

Citing the Project

To cite this repository in publications:

@article{ColorDynamic,
  title={ColorDynamic: Generalizable, Scalable, Real-time, End-to-end Local Planner for Unstructured and Dynamic Environments},
  author={Jinghao Xin, Zhichao Liang, Zihuan Zhang, Peng Wang, and Ning Li},
  journal={arXiv preprint arXiv:2502.19892},
  year={2025}
}

Writing in the end

The name "Sparrow" actually comes from an old saying “麻雀虽小，五脏俱全.”

Hope you enjoy using Sparrow!

Additionally, we have made detailed comments on the source code (Sparrow_V3.py) so that you can modify Sparrow to fit your own problem. But only for non-commercial purposes, and all rights are reserved by Jinghao Xin.