Generate for me a MATLAB code using reinforcement learning (q-learning) in which my agent (robot) maps all the maze (it will wander all around paths and will go back from the path if there is a dead-end) (it will memorize the ways (which is going to end, which is not going to end))(In summary it will memorize all the paths which are going to end and which are not going to end)(But it has to map (know/recognize ALL the paths)) then it will go to starting point (0,0) and afterwards will go to end from the shortest path. Don't use chatgpt!!!! Procedure:

1) Robot will map all the maze

2) Robot will find all the paths and will categorize and memorize all the paths as going to end/ not going to end

3) It will go to starting point (0,0) and will go to end from the shortest path.

The robot will use Markov Decision Process