Inspiration
Inspired to build this project because it is one of my favorite games, and I wanted to see what kinds of strategies it would come up with.
What it does
It is an AI agent that plays against you and gets better every time.
How we built it
Build through reinforcement learning, slowly improving each game through a reward policy.
Challenges we ran into
One of the main challenges was getting data from the current state of the game to make an informed decision, especially for a game so niche.
Accomplishments that we're proud of
Since there haven't been many projects like this for this game, I had to build a dataset of cards and images, and then train it myself in order to use machine vision in real-time to feed information to the RL AI Agent.
What we learned
Learned a lot about data annotation and how important it is to have a clean dataset.
What's next for Kings Gambit AI Agent
To keep training and hopefully become better than humans.
Log in or sign up for Devpost to join the conversation.