Abstract: Although today's reinforcement learning-based control agents operate effectively in ideal situations, their performance can significantly decrease under changing conditions. This paper ...