AAU Student Projects - visit Aalborg University's student projects portal
A master's thesis from Aalborg University
Book cover


Optimization of behavioral parameters of Artificial Intelligence agents created by UtilityAI tool

Term

4. term

Publication year

2019

Submitted on

Pages

35

Abstract

The goal of this project is to improve the development workflow of an Artificial Intelligence agent created using UtilityAI tool published by Apex Game Tools. Such agent contains multiple numeric parameters that need to be carefully optimized by the designer of the agent in order for the agent to behave in the intended way. Optimization of these parameters can be aided by definition of a reward function and employing optimizers that are then able to converge the parameters to values maximizing defined reward. Bayesian optimization was chosen as the optimization algorithm for this project and was evaluated on the Survival Shooter game and two known constructed functions.