Cubey's Exam (Reinforcement Learning)
The Project
This project is an 'engine' that lets you draw out a level and get an AI agent to play it and try to beat it. The AI is mildly dumb, so I recommend running it on Turbo Mode when the AI starts playing. I've previously seen people on YouTube teach AI to play games, so I thought I'd make this engine for an AI agent to play a simple tiled platformer. It wasn't particularly hard, once I understood what Q learning is and how it works. (Note:- I won't hide the fact that I have used Google Gemini for help, but I have only used it to understand how Q learning works, and the Bellman equation. All the code, art, music, sound effects etc. are 100% original.)
