conda env create -f environment.yml
conda activate gridworld
Run gridworld_manual.py with arguments for the map size [x=y=map_shape], the number of agents and obstacle density:
python gridworld_manual.py --map_shape 10 --n_agents 2 --obj_density 0.2
To see a command-line printout of the observations add --verbose
Follow comand instuctions to make agents move. For example: 0 1 1 2 s => agent 0 move up ; agent 1 move right ; step
python odmstar_example.py --map_shape 32 --n_agents 20 --obj_density 0.2 --inflation 1.1
python run_experiments.py --name ppo_test
python run_experiments.py --name ic3net_test
python run_experiments.py --name maac_test
python run_experiments.py --name primal_test
Rendering of Gridworld environment:
@misc{gym_minigrid,
author = {Chevalier-Boisvert, Maxime and Willems, Lucas and Pal, Suman},
title = {Minimalistic Gridworld Environment for OpenAI Gym},
year = {2018},
publisher = {GitHub},
journal = {GitHub repository},
howpublished = {\url{https://github.com/maximecb/gym-minigrid}},
}
Code in Agents/MAAC.py and Agents/maac_utils/ modified from: https://github.com/shariqiqbal2810/MAAC
@InProceedings{pmlr-v97-iqbal19a,
title = {Actor-Attention-Critic for Multi-Agent Reinforcement Learning},
author = {Iqbal, Shariq and Sha, Fei},
booktitle = {Proceedings of the 36th International Conference on Machine Learning},
pages = {2961--2970},
year = {2019},
editor = {Chaudhuri, Kamalika and Salakhutdinov, Ruslan},
volume = {97},
series = {Proceedings of Machine Learning Research},
address = {Long Beach, California, USA},
month = {09--15 Jun},
publisher = {PMLR},
pdf = {http://proceedings.mlr.press/v97/iqbal19a/iqbal19a.pdf},
url = {http://proceedings.mlr.press/v97/iqbal19a.html},
}
Code in Agents/maac_utils/vec_env/ from Open AI baselines: https://github.com/openai/baselines/tree/master/baselines/common
Code in Agents/ic3Net.py modified from: https://github.com/IC3Net/IC3Net
@article{singh2018learning,
title={Learning when to Communicate at Scale in Multiagent Cooperative and Competitive Tasks},
author={Singh, Amanpreet and Jain, Tushar and Sukhbaatar, Sainbayar},
journal={arXiv preprint arXiv:1812.09755},
year={2018}
}