Mastering Terra Mystica: Applying Self-Play to Multi-agent Cooperative Board Games

TL;DR

This paper applies a modified AlphaZero algorithm, called AlphaTM, to the complex strategy game Terra Mystica, aiming to develop an AI that can rival human players by leveraging novel state representations and analyzing strategies.

Contribution

The paper introduces AlphaTM, a novel adaptation of AlphaZero for Terra Mystica, with new state representations and initial results demonstrating its potential against baselines and human scores.

Findings

AlphaTM shows promising performance against baselines.

The approach highlights strengths and limitations of applying self-play to Terra Mystica.

Initial analysis suggests potential for human-level play.

Abstract

In this paper, we explore and compare multiple algorithms for solving the complex strategy game of Terra Mystica, hereafter abbreviated as TM. Previous work in the area of super-human game-play using AI has proven effective, with recent break-through for generic algorithms in games such as Go, Chess, and Shogi \cite{AlphaZero}. We directly apply these breakthroughs to a novel state-representation of TM with the goal of creating an AI that will rival human players. Specifically, we present the initial results of applying AlphaZero to this state-representation and analyze the strategies developed. A brief analysis is presented. We call this modified algorithm with our novel state-representation AlphaTM. In the end, we discuss the success and shortcomings of this method by comparing against multiple baselines and typical human scores. All code used for this paper is available at on \href{https://github.com/kandluis/terrazero}{GitHub}.