Push Battle

Win for Blue

Inspiration

I liked machine learning, and we all wanted to experiment with different algorithms and models, so each of our teammates used a different method of solving the problem.

What it does

This plays a game of Push Battle, in which the model tries to win against other players by making optimal moves to achieve three pieces in a row. Every time a piece gets put next to another piece, it will push the piece back unless there is another piece behind it. This makes the game full of strategy and thinking ahead.

How we built it

We used heuristic-based approaches to combine if-else statements to generate solutions and optimal moves to get to 3 pieces in a row while fighting against other players.

Challenges we ran into

We ran into a lot of challenges getting our different models to work. For one, it was really difficult to get anything to work after training the models. We had a lot of errors with what was happening since the model kept trying to make invalid moves that wouldn't work in the game, leading to a forfeit.

Accomplishments that we're proud of

We are most proud that we managed to solve the puzzle. It took a lot of learning where we stepped out of our comfort zone and learned many new ideas and concepts we didn't even know existed. Because we got this experience, we got to learn a lot to become better problem solvers and coders.

What we learned

We learned how to use different models such as NEAT, Monte Carlo, different heuristic-based algorithms, and even neural networks to try to solve the problem. Even though we didn't end up using any of those methods to ultimately solve the challenge, getting through it was extremely fun since we got some new experience with things we don't usually get to work with. Hopefully, the next time we use these programs, it'll be better designed and put together.

What's next for Push Battle

Hopefully, Push Battle will improve in the future by actually using AI models and algorithms, making it more accurate than our current methods. It'll be more trained, and eventually, it'll be able to see wins more accurately, and jump on them whenever it's on the board.