1. Press the flag to start. 2. Click on tiles to change them from one of three types of tiles: a. Start State (S) b. Wall (grey) c. Goal (flag) 3. Press SPACE to make the cat try to solve your maze and explore. Once the cat completes it, press SPACE again to see if he completes it faster with what he's learned. You can edit your maze at any time.
I did not bug test to see what happens if you edit the cat into a wall or what happens if you try to start (SPACE) without first placing a starting state (S). Yes, the cat can get a little stuck, but if you give him some time he'll get there. He has some randomness built in so he won't always be optimal. Q-Learning is a Reinforcement Learning algorithm that you can look more into with the following citation: Watkins, C.J.C.H., Dayan, P. Q-learning. Mach Learn 8, 279–292 (1992). https://doi.org/10.1007/BF00992698
