Application of minimax algorithm in dots and boxes game

Tan, Ah Hwa (2025) Application of minimax algorithm in dots and boxes game. Final Year Project, UTAR.

Preview

PDF Download (2365Kb) | Preview

Abstract

This project addresses the common issue of deficient artificial intelligence (AI) opponents in digital versions of the classic strategy game Dots and Boxes, which often limits gameplay engagement and strategic depth. The core problem lies in developing a strategically competent AI capable of navigating the game's large decision space and computational demands. The methodology involved implementing the Minimax algorithm as the primary decision-making engine for the AI. This was enhanced with several optimization techniques, including Alpha-Beta pruning, transposition tables, killer move heuristics, and quiescence search. For the highest difficulty setting, an iterative deepening Minimax approach was utilized alongside a heuristic evaluation function that considers score difference and strategic board positions like chains and potential opponent scoring opportunities. The research process included designing and developing a functional Dots and Boxes game prototype using C# and the Universal Windows Platform (UWP), featuring a user-friendly interface and customizable settings. Rigorous testing confirmed the application's functionality and the AI's progressive difficulty, with human players winning 70% of games on "Easy," the AI winning 50% on "Medium," and the AI achieving an 80%-win rate on "Hard". AI response times remained acceptable even on larger boards. The project successfully demonstrates the application of an enhanced Minimax algorithm to create a challenging and engaging AI opponent, effectively revitalizing the classic game by offering significant strategic depth through its advanced AI implementation and varied difficulty levels. Area of Study (Minimum 1 and Maximum 2): Artificial Intelligence in Gaming, Game Theory Keywords (Minimum 5 and Maximum 10): Minimax Algorithm, Dots and Boxes, Artificial Intelligence, Game AI, Alpha-Beta Pruning, Heuristic Evaluation, Game Development, Strategic Games, Universal Windows Platform (UWP)

Actions (login required)