Mixed Traffic Control and Coordination from Pixels

• Paper in ArXiV
• ICRA 2024 Presentation Video
• ICRA 2024 Poster

Installation

Instructions for installing and setting up everything for this repo are located here.

Training

Example commands for training each of the five environments from the paper are provided in run.sh. An example command for training on the ring environment is the following:

python configs/train.py ring

More details about training and running the environments are provided in the instructions.

Evaluation

Example commands for evaluating a trained policy are provided in eval.sh. More details for evaluating a trained policy are provided for in the instructions.

Citing

Please cite our work using the following bibtex:

@inproceedings{villarreal2023mixed,
  title={Mixed Traffic Control and Coordination from Pixels},
  author={Villarreal, Michael and Poudel, Bibek and Pan, Jia and Li, Weizi},
  booktitle={2024 IEEE International Conference on Robotics and Automation (ICRA)},
  year={2024}
}

This code base originally comes from the following paper:

C. Wu, A. Kreidieh, K. Parvate, E. Vinitsky, A. Bayen, "Flow: Architecture and Benchmarking for Reinforcement Learning in Traffic Control," CoRR, vol. abs/1710.05465, 2017. [Online]. Available: https://arxiv.org/abs/1710.05465.

Please also consider cite this work.