Classical AI planning considers how to change the state of the world from an initial state to a goal state given a set of possible operators. Relatively recently, the area has grown to include probabilistic domains and operators, bridging towards studies of Markov Decision Processes and learning.