Markov Decision Process (MDP) almost defines the standard format of the reinforcement learning. With the symbolic representation, almost any problem can be transfered to the MDP. Learning the reinforcement, MDP is the best start.