Home

PyTSC is a Python library for training reinforcement learning (RL) models for traffic signal control (TSC). It lets you simulate traffic scenarios and train agents using different RL/MARL algorithms. You can either SUMO or CityFlow simulator. The library is modular so you can modify MDP formulations as per as your need. PyTSC is open-source. You can find it on GitHub. Contributions from the community are welcome and appreciated!

Contents:

• Installation
  • Required Simulators
  • Required MARL Libraries
• Architecture
• Usage
  • Add scenario files
  • Create a config file
  • Using PyTSC in your code
  • Using PyTSC with PyMARL/EPyMARL
• API Reference
  • pytsc
  • pytsc package
  • pytsc.backends package
  • pytsc.backends.cityflow package
  • pytsc.backends.sumo package
  • pytsc.common package
  • pytsc.controllers package
  • pytsc.generators package
  • pytsc.scenarios package
  • pytsc.tests package
  • pytsc.wrappers package

Note

Citation

If you use this repository for your research, please cite it using the following BibTeX entry:

@article{bokade2025pytsc,
  title={Pytsc: A unified platform for multi-agent reinforcement learning in traffic signal control},
  author={Bokade, Rohit and Jin, Xiaoning},
  journal={Sensors},
  volume={25},
  number={5},
  pages={1302},
  year={2025},
  publisher={MDPI}
}