Razvoj agenta za društvenu igru Catan

master's thesis

• Title (english): Development of an agent for the Catan boardgame
• Author: Ivan Skorić
• Mentor: Marko Đurasević (mentor)
• Committee member: Marko Đurasević (predsjednik povjerenstva)
• Committee member: Domagoj Jakobović (član povjerenstva)
• Committee member: Željka Mihajlović (član povjerenstva)
• Granter: University of Zagreb; Faculty of Electrical Engineering and Computing; Zagreb
• Defense date and country: 2022-07-04, Croatia
• Scientific / art field, discipline and subdiscipline: TECHNICAL SCIENCES; Computing

Abstract

Ovaj rad opisuje razvoj šest računalnih igrača, odnosno agenata za stratešku društvenu igru Catan. Pravila i koncepti igre detaljno su opisani, kako bi se lakše obrazložile ideje iza pojedinih igrača. Za potrebe razvoja i igranja na računalu, isprobana su tri radna okvira za Catan, i analizirani su u sklopu ovog rada. Agenti su razvijeni oko različitih metoda za odabir poteza, poput pretraživanja prostora stanja i neuronskih mreža. Kod nekih se koriste i tehnike strojnog učenja za optimizaciju parametara. Njihove sposobnosti uspoređene su na velikim uzorcima odigranih igri, a njihovo ponašanje je evaluirano igranjem protiv njih kroz grafičko sučelje radnog okvira Catanatron. Catanatron implementira nekoliko postojećih igrača, te su novi agenti uspoređeni i s njima. Na kraju rada, iznesene su i razne ideje za budući razvoj te napredak projekta.

Abstract (english)

This paper describes the development of six computer players, i.e., agents for the Catan strategy board game. The rules and concepts of the game are described in detail, to help explain the ideas behind individual players. For the purposes of development and testing, three Catan frameworks have been tested, and are analyzed as part of this paper. Agents have been developed around various methods for selecting their actions, such as state space search and neural networks. With some of them, machine learning techniques have been used to optimize their parameters. Their capabilities have been compared on large samples of games played, and their behavior has been evaluated by playing against them through Catanatron framework’s graphical user interface (GUI). Catanatron implements several existing players, and the new agents have been compared to those as well. At the end of the paper, various ideas for future development and improvements of the project are presented.

• Keywords: Catan; društvena igra; umjetna inteligencija; agent; strategija; strojno učenje; heuristika; pretraživanje prostora stanja; neuronska mreža; genetski algoritam; neuroevolucija
• Keywords (english): Catan; boardgame; artificial intelligence; agent; strategy; machine learning; heuristics; state space search; neural network; genetic algorithm; neuroevolution
• Language: croatian
• URN:NBN: urn:nbn:hr:168:309640
• Study programme: Title: Computing; Study programme type: university; Study level: graduate; Academic / professional title: magistar/magistra inženjer/inženjerka računarstva (magistar/magistra inženjer/inženjerka računarstva)
• Type of resource: Text
• File origin: Born digital
• Access conditions: Closed access
• Created on: 2023-01-14 09:56:08