Razvoj inteligentnega agenta za igro z uporabo emulatorja PyBoy

Herksel Japelj, Samo (author), Klemenc, Bojan (mentor)

Diplomsko delo obravnava razvoj inteligentnega agenta, ki se bo sposoben učiti in igrati izbrano igro z uporabo strojnega učenja. Problem se nanaša na ustvarjanje sistema, ki se lahko samostojno uči in izboljšuje svoje zmogljivosti pri igranju igre. Za rešitev tega problema je uporabljen pristop spodbujevalnega učenja in nevronskih mrež, kjer agent skozi poskuse in napake pridobiva izkušnje in izboljšuje svojo strategijo igranja. Najpomembnejši rezultat tega dela je uspešen razvoj agenta, ki lahko konkurenčno igra izbrano igro in se prilagaja novim situacijam. Stremimo k temu, da bo končen izdelek ne le funkcionalen, temveč tudi koristen drugim, ki želijo pristopiti k izdelavi svojega agenta.

• Language: Slovenian
• Keywords: umetna inteligenca, strojno učenje, igre, q-učenje, PyBoy, spodbujevalno učenje
• Work type: Bachelor thesis/paper
• Organization: FRI - Faculty of Computer and Information Science
• Year: 2024
• PID: 20.500.12556/RUL-161479
• Publication date in RUL: 11.09.2024

Secondary language

• Language: English
• Title: Development of an Intelligent Game Agent Using the PyBoy Emulator

Abstract

The thesis focuses on developing an intelligent agent capable of learning and playing a specific game using machine learning techniques. The problem addressed involves creating a system that can autonomously learn and enhance its performance in gameplay. The solution approach utilizes reinforcement learning and neural networks, where the agent acquires experience and improves its gameplay strategy through trial and error. The most significant outcome of this work is the successful development of an agent that can competitively play the chosen game and adapt to new situations. The hope is that the final product will not only be functional but also useful to others who wish to approach the creation of their own agent.

• Keywords: artificial intelligence, machine learning, games, Q-learning, PyBoy, reinforcement learning