mirzakhalov / coup-ai

Coup: The Dystopian Universe + Reinforcement Learning

This project was an attempt to create an AI bot to play the game of Coup in 5 player setting. Are you wondering what is Coup? It is basically a multi-player game where a player has to manipulate, bluff and damage others with the goal of being the last one alive. It is a fast-paced and highly unpredictable game that we just love to play in our free time. You can learn about the rules more here

Setting up Game Environment

We do this in terms of these processes:

• Rule identification
• Standardization
• Game flow setup
• Testing (random agent)

First step before starting to code up the environment is to strictly define the rules and actions once more. In the context of Coup, we identified all Actions any particular player can perform at any given time. For example, it would include Play card X, Play card Y, Challenge a play etc. Once identified, we through the process of Standardization. Standardization, by our own definition, would involve coming up with a standardized way of representing all actions in a uniform way. This is a crucial step and actually enables a very intuitive transition when we move to Reinforcement Learning phase of our development.