Multi-Agent Assignment via State Augmented Reinforcement Learning
Abstract
We address the conflicting requirements of a multi-agent assignment problem through constrained reinforcement learning, emphasizing the inadequacy of standard regularization techniques for this purpose. Instead, we recur to a state augmentation approach in which the oscillation of dual variables is exploited by agents to alternate between tasks. In addition, we coordinate the actions of the multiple agents acting on their local states through these multipliers, which are gossiped through a communication network, eliminating the need to access other agent states. By these means, we propose a distributed multi-agent assignment protocol with theoretical feasibility guarantees that we corroborate in a monitoring numerical experiment.
Cite
Text
Agorio et al. "Multi-Agent Assignment via State Augmented Reinforcement Learning." Proceedings of the 6th Annual Learning for Dynamics & Control Conference, 2024.Markdown
[Agorio et al. "Multi-Agent Assignment via State Augmented Reinforcement Learning." Proceedings of the 6th Annual Learning for Dynamics & Control Conference, 2024.](https://mlanthology.org/l4dc/2024/agorio2024l4dc-multiagent/)BibTeX
@inproceedings{agorio2024l4dc-multiagent,
title = {{Multi-Agent Assignment via State Augmented Reinforcement Learning}},
author = {Agorio, Leopoldo and Van Alen, Sean and Calvo-Fullana, Miguel and Paternain, Santiago and Bazerque, Juan Andrés},
booktitle = {Proceedings of the 6th Annual Learning for Dynamics & Control Conference},
year = {2024},
pages = {1202-1213},
volume = {242},
url = {https://mlanthology.org/l4dc/2024/agorio2024l4dc-multiagent/}
}