Proper Reinforcement - an overview | ScienceDirect Topics
But for some problems, the definition of MDP state space is not as straightforward: for instance, to design an RL controller for grid control, the best way to define the state space that can accurately reflect