Learning to Coordinate with Anyone

This repository contains official implementation for Learning to Coordinate with Anyone.

Environment Installation

Build the environment by running:

pip install -r requirements.txt

Install the Level Based Foraging (LBF) environment by running:

pip install -e src/envs/lb-foraging

Install the Predator-Prey (PP) environment by running:

pip install -e src/envs/mpe/multi_agent_particle

Install the StarCraft Multi-Agent Challenge (SMAC) environment by running:

pip install -e src/envs/smac

Run an experiment

python3 src/main.py --config=[Algorithm name] --env-config=[Scenario name]

The config files act as defaults for an algorithm or scenario. They are all located in src/config. --config refers to the config files in src/config/algs including Macop-VDN and Macop-QMIX. --env-config refers to the config files in src/config/envs, including the LB-Foraging environment (https://github.com/semitable/lb-foraging), the Predator Prey and the Cooperative Navigation environments (https://github.com/openai/multiagent-particle-envs), and the StarCraft Multi-Agent Challenge environment (https://github.com/oxwhirl/smac).

All results will be stored in the results folder.

For example, run Macop-VDN on LBF1 scenario:

python3 src/main.py --config=vdn --env-config=lbf1

Publication

If you find this repository useful, please cite our paper:

@inproceedings{macop,
  title     = {Learning to Coordinate with Anyone},
  author    = {Lei Yuan and Lihe Li and Ziqian Zhang and Feng Chen and Tianyi Zhang and Cong Guan and Yang Yu and Zhi-Hua Zhou},
  booktitle = {Proceedings of the Fifth International Conference on Distributed Artificial Intelligence},
  year      = {2023}
}