Curious Meta-Controller: Adaptive Alternation between Model-Based and Model-Free Control in Deep Reinforcement Learning | IEEE Conference Publication | IEEE Xplore