Inspiration
We thought the Case Closed challenge seemed was very interesting and seemed thematic and exciting, especially since we wanted to do something that was hard.
What it does
The agent plays the game aggressively, seeking to use its boosts early and then forcing a head on collision in order to take advantage of how the game rules work regarding head on collisions and trail length differences.
How we built it
We mostly used regular python and the given code. AI helped a lot with learning and understanding the starter code and debugging.
Challenges we ran into
We also wanted to use pytorch and DQL and spent a lot of time trying to train out agent but it didn't quite work out and one of our teammates had to leave early as well.
Accomplishments that we're proud of
We learned a lot of new skills and tried something different even if it didn't turn out how we wanted it. We're also really happy with how the UI we created turned out (even if it was extra).
What we learned
Python, Pytorch, machine learning theory, Flask
What's next for Cool Agent
Try to debug the pytorch code and successfully train the agent through DQL.
Log in or sign up for Devpost to join the conversation.