The study presents a framework for hybrid teams comprising of humans and autonomous systems, or agents. The authors propose an AI manager, equipped with Reinforcement Learning, to monitor the team and select the most suitable control agent at any given time. The AI manager operates under certain constraints and aims to minimize its interventions to avoid complexity. The effectiveness of the AI manager is demonstrated in a simulated driving scenario, where it significantly improved team performance.
Publication date: 9 Feb 2024
Project Page: https://doi.org/XXXXXXX.XXXXXXX
Paper: https://arxiv.org/pdf/2402.05605
