Implementation of bots for multi-action adversarial game solving

Abstract

This document presents the report of a Final Degree Work consisting of a project that includes several components related to multi-action adversarial game solving. It involves the development of a multi-action adversarial card game that is capable of delivering relevant results on the performance of different algorithms. It also encompasses the development and testing of bots that play the game using several different algorithms, one of which entails a novel approach to solving these kind of games. Finally, as the main research action of the project, an experiment on the performance of the algorithms within the game was performed. Within each of the chapters of this report attention will be put into going trough all these components of the project. The framework architecture for A Simple MultiAction Card Game (ASMACAG) will be firstly presented. It is the simple but complete card game proposed as a tool to test, develop and debug bots that implement artificial intelligence algorithms in the context of adversarial multi-action games. Then the focus will be on the development process and the specifics of each of the algorithms included in the project, as well as the decisions taken on what parameters to use for them. Finally the experiment carried out within the game will be discussed, to further comprehend what conclusions can be drawn about the implemented bots and the game itself.