Skip to content

chengyutang/GreenLight

Repository files navigation

Source code for paper "GreenLight: Green Traffic Signal Control using Attention-based Reinforcement Learning on Fog Computing Network".

Prerequisites

  • SUMO >= 1.19.
  • Python >= 3.11 (older versions are not tested).

Install

  1. Clone or download this repository.
  2. Install dependencies.
    pip install -r requirements.txt

Usage

The model can be trained using train.py. The user will need to specify the path to a .sumocfg file using the --sumocfg argument. The program will use the corresponding .net.xml file to build the environment that the RL agent will interact with.

Use python train.py --help to see all accepted arguments that can be used to control the environment, RL model, and training.

After a model is trained, the test.py script can be used to evaluate the model. The testing simulation scenario should use the same network used for training. Use python test.py --help to see all accepted arguments.

For more technical details, please refer to the paper.

Citing GreenLight

@INPROCEEDINGS{tang2024greenlight,
  author={Tang, Chengyu and Baskiyar, Sanjeev},
  booktitle={2024 IEEE 15th International Green and Sustainable Computing Conference (IGSC)}, 
  title={GreenLight: Green Traffic Signal Control using Attention-based Reinforcement Learning on Fog Computing Network}, 
  year={2024},
  pages={129-134},
  doi={10.1109/IGSC64514.2024.00032}
}

About

Traffic signal control framework using deep reinforcement learning and self-attention.

Resources

License

Stars

Watchers

Forks

Contributors

Languages