Využití strojové učení v počítačových hrách

Abstract

This thesis is focused on the research and implementation of artificial intelligence in computer games with the ambition to utilize machine learning in the greatest possible extent. We have created a simple, procedural generation of non-playable characters, their subsequent learning to walk with the help of reinforcement learning, creating such characters in the simulation environment, which was also extended by a system of experiences and adaptive behavior of characters moving in the environment. The result of this thesis is a presentation of a successful utilization of machine learning in the implementation of behavior of non-playable characters in the simulation environment.