We wanted to explore using Reinforcement Learning to create a Galalga Enemy that uses novel attack patterns powered by Neural Networks.

What it does

We trained enemies and user controlled ships with reinforcement learning against each other. The enemies are then powered by what that model has learned.

How we built it

We built it mostly in Python, with online elements (database monitoring scores, website that displays top scores)

Challenges we ran into

Reinforcement learning has been studied a lot with old games, however our approach to power the game with AI against the user has not been well researched our documented. Powering the enemies with AI required a ground up approach to building the game. We started our approach with a Python game called Alien Invasion by @goswami-rahul

Accomplishments that we're proud of

A platform that works and that can bring people together. We worked as a team through countless servings of caffeine and built an application we are really proud of!

What we learned

We learned many new technologies and methods, Cloud, Neural Networks, Reinforcement Learning, We have a working training model for phase 1 of our plan.

What's next for GalAIga

releasing the game as a phone app.

Share this project: