We propose a decentralized reinforcement learning solution for multi-agent shepherding of non-cohesive targets using policy-gradient methods. Our architecture integrates target-selection with target-driving through Proximal Policy Optimization, overcoming discrete-action constraints of previous Deep Q-Network approaches and enabling smoother agent trajectories. This model-free framework effectively solves the shepherding problem without prior dynamics knowledge. Experiments demonstrate our method's effectiveness and scalability with increased target numbers and limited sensing capabilities.
View on arXiv@article{covone2025_2504.02479,
title={ Hierarchical Policy-Gradient Reinforcement Learning for Multi-Agent Shepherding Control of Non-Cohesive Targets },
author={ Stefano Covone and Italo Napolitano and Francesco De Lellis and Mario di Bernardo },
journal={arXiv preprint arXiv:2504.02479},
year={ 2025 }
}