AI Playtesting

Using AI Agents to solve Card Game balance

Our Method

Step 1 : We train the AI agent using Deep Reinforcement Learning algorithms.

Step 2 : The AI Agent plays against itself over and over again. This generates a large quantity of statistical data.

Step 3 : We use the generated statistics to identify imbalance and update the rules of the game.

Dev Blog

December 17, 2020

The last week involved a lot of work on bug fixes and getting ready for the festival, the final presentation and the final playthrough. Our instructor Mike worked with us to find hidden bugs which was very helpful in ensuring that our application ran smoothly. We also began work on...

Week 15 : A Good End

December 7, 2020

This week started off with the softs opening on Monday which was received well. We handed out our playable prototype to the faculty but because of the lengthy nature of the entire process of using our app, no one really playtested the application. However, we were able to show a...

Week 14 : Final Bug Fixes

December 1, 2020

This week involved more polishing of our app in order to prepare for softs. We identified several bugs through the course of last which needed to be ironed out. We also added some minor features aimed at improving user experience. Last but not the least, we also worked towards creating...

Week 13 : Preparation for Softs

November 23, 2020

This week saw refinement of features in our app. We playtested with a CMU faculty for three consecutive days to get a good understanding of how our app feels and if the data it offers is valuable. We got some good suggestions and have kept working on those suggestions to...