The whole learning experience was based on this crazy idea of teaching my computer to play tictacktoe, so I started with the basics. Then I decided to make the gridworld bigger and with more obstacle and teach my computer to solve a maze. The idea I had in my mind required a neural net so I stopped and instead went with Q learning approach.