Exercise 2.1.1
Markov decision process: states, actions, rewards for traffic light control
Reinforcement Learning › Exercises › Lecture 2 · The RL Problem and Finite MDPs › 2.1 Markov decision process
Describe how the concept of Finite Markov Decision Processes (FMDPs) can be applied to design an intelligent system for adaptive traffic light control. Include in your discussion how states, actions, and rewards could be defined, and how FMDPs facilitate the learning of optimal policies to improve traffic flow.