Control-Tutored Reinforcement Learning: an application to the Herding Problem

This work addresses the problem of safe and efficient reinforcement learning in continuous environments for robotics or control systems, but it appears incremental as it builds on existing Q-learning and control methods.

The authors tackled the challenge of developing model-based and safe reinforcement learning for continuous state spaces by introducing a control-tutored Q-learning approach, which they validated on a multi-agent herding control problem.