Playing Carcassonne with Monte Carlo Tree Search

TL;DR

This paper investigates the application of vanilla Monte Carlo Tree Search and MCTS-RAVE to the game of Carcassonne, demonstrating their effectiveness over a domain-specific heuristic-based algorithm in developing long-term strategies.

Contribution

It is the first to compare vanilla MCTS and MCTS-RAVE against a heuristic-based algorithm in Carcassonne, highlighting their strategic advantages.

Findings

MCTS outperforms Star2.5 in Carcassonne gameplay.

Vanilla MCTS shows more robustness than MCTS-RAVE.

Long-term strategy development is enhanced by MCTS methods.

Abstract

Monte Carlo Tree Search (MCTS) is a relatively new sampling method with multiple variants in the literature. They can be applied to a wide variety of challenging domains including board games, video games, and energy-based problems to mention a few. In this work, we explore the use of the vanilla MCTS and the MCTS with Rapid Action Value Estimation (MCTS-RAVE) in the game of Carcassonne, a stochastic game with a deceptive scoring system where limited research has been conducted. We compare the strengths of the MCTS-based methods with the Star2.5 algorithm, previously reported to yield competitive results in the game of Carcassonne when a domain-specific heuristic is used to evaluate the game states. We analyse the particularities of the strategies adopted by the algorithms when they share a common reward system. The MCTS-based methods consistently outperformed the Star2.5 algorithm given their ability to find and follow long-term strategies, with the vanilla MCTS exhibiting a more robust game-play than the MCTS-RAVE.