Learning Eco-Driving Strategies at Signalized Intersections

This repo contains the codebase for the paper titled "Learning Eco-Driving Strategies at Signalized Intersections" published in European Control Conference (ECC) 2022.

Paper | Website

Installation

1. The code has been tested on Ubuntu 20.04, and MacOS 13.0 Ventura.
2. Follow instructions here to install Miniconda, likely wget https://repo.anaconda.com/miniconda/Miniconda3-latest-Linux-x86_64.sh followed by bash Miniconda3-latest-Linux-x86_64.sh.
3. Follow instructions in SUMO website to install SUMO simulator. Try running sumo in the terminal to confirm the successful installation.
4. Install PyTorch from pytorch.org.
5. Make a clone of the repo.
6. Set the environment variable 'F' to the code directory.
7. Install dependencies pip install -r requirements.txt.
8. If you want to use fast libsumo instead of traci, set LIBSUMO os variable to true.

Instructions

<agent_type> is the type of agent that can be used to control CAVs. Available Options: RL, IDM

<res_dir> is the result directory, which is where the model checkpoints, training logs, and training csv results will be saved. Add --test as an argument to run the simulation in inference mode. E.g., python pexps/<script>.py --agent <agent_type> --res <res_dir> --test.

Eco-driving in 1x1 intersection

python pexps/main.py --agent <agent_type> --res <res_dir>

Citation

If you are using this codebase for any purpose, please cite the following paper.

@INPROCEEDINGS{ecodrive2022jayawardana,
  author={Jayawardana, Vindula and Wu, Cathy},
  booktitle={2022 European Control Conference (ECC)}, 
  title={Learning Eco-Driving Strategies at Signalized Intersections}, 
  year={2022},
  pages={383-390},
  doi={10.23919/ECC55457.2022.9838000}}