Create A Reinforcement Learning Agent That Can Solve Any Maze