Flow is a computational framework for deep RL and control experiments for traffic microsimulation.
See results and videos of the application of Flow to several mixed-autonomy traffic scenarios.
- Ask questions on our mailing list: [email protected].
- Please report bugs by submitting a GitHub issue.
- Submit contributions using pull requests.
If you use Flow for academic research, you are highly encouraged to cite our 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
Flow is created by and actively developed by members of the Mobile Sensing Lab at UC Berkeley: Cathy Wu, Eugene Vinitsky, Aboudy Kreidieh, Kanaad Parvate, Nishant Kheterpal, Saleh Albeaik, Kathy Jang, and Ananth Kuchibhotla. Alumni contributors include Leah Dickstein and Nathan Mandi.