This project uses PyGame to simulate 1-sided Pong. The goal is to use Q-Learning(With Tensorflow) to teach an RL agent to play the game.
I was able to succesfully train the agent to play perfectly when I rewarded it at all times for being closer to the ball. I haven't yet been able to train the agent to play only rewarding it with succesful hits. I anticipate that this is an issue of my patience in letting the model train, and not actually anything with the implementation.