A way to model choices, results, and rewards over time.
MDP is like picking moves in a video game level. Grab the shiny coin now, or dodge the lava so future-you survives.
It tracks each situation and each move. It is the usual setup for RL and robot control.
RL
RL often turns the environment and goal into an MDP.
Bellman Equation
Bellman Eq describes how value moves through an MDP.
Policy Gradient
Policy Gradient optimizes a policy inside an MDP.
Q-Learning
Q-Learning uses an MDP, then learns the value of each action.