Inspiration
We came to the conclusion we all had a passion for playing video games and curiosity for artificial intelligence (AI). So we decided to combine the two topics and teach AI to master Flappy Bird
What it does
First you have the AI run through hundreds of iterations to gather data. With this data we then can allow the AI to run through the game using the data collected.
How we built it
We used Pygame to run Flappy Bird (we did not actually program the game itself) then we used an Agent and Model to where during a 'training' run, it would save data in an outside file. If we do a 'play' run, the game would render and use the file full of data from the training runs to function and play the game.
Challenges we ran into
The biggest challenge we came across was our lack of understanding AI and Neural Networks.
Accomplishments that we're proud of
Our biggest accomplishment was being able to have a better grasp of AI and Deep Learning, allowing the AI to play a long duration of time with minimal failure.
What we learned
We had the opportunity to learn to code in Python as well as have a better understanding of how AI functions.
What's next for Machine Learning Flappy Bird
Being able to let Machine Learning Flappy Bird master the game in less iterations than the previous run through.
Log in or sign up for Devpost to join the conversation.