Below is a guide of the contents of this repository:
Base game and training model can be found in the air_hockey_game folder - playable with instructions in the folder's README.md file.
pong_v2 is a proof of concept. It is a mesh of pong and air hockey that we trained a dqn model on.
The models folder contains various models that were trained on the air hockey game, with different model architectures, opponents, reward systems, and other hyperparameters.
The speed_multiplier folder contains alternate versions of the training process with changes to increase the training speed.
The browser folder was a failed attempt to create a web-based version of the game. It is currently not functional.
Our presentation is in the presentation folder and report in the report folder.
The play folder contains a version of the AI that can be played against in the air hockey game.