Generalization in Reinforcement Learning: Successful Examples Using Sparse Coarse Coding