Personalised Gaming Using Learning Technology

Personalised Gaming Using Learning Technology

Boellaard, B.
Bolier, W.
Mairet, V.
Peeters, T.

Bulling, N. (mentor)

Electrical Engineering, Mathematics and Computer Science

Computer Science

Bachelor Eind Project

2015-07-31

This Bachelor Thesis is part of the Bachelor Computer Science and Engineering program at Delft University of Technology. During this project the team had to create an extension to an existing game framework. This extension consisted of improved artificial intelligence for the antagonist and a maze generator that could be re-used as a library. The original framework consisted of a game that was implemented in Unity. The game featured a bug that had to traverse several mazes that were interconnected by portals. These portals would allow it to travel between mazes. The mazes also featured an antagonist that will pursue the player and attempt to kill it. The intention of the game is to be challenging and difficult, and allow the framework to easily modify the difficulty settings. The game has an atmosphere that resembles a horror game and should be scary to a certain extent. The implementation of the artificial intelligence was done with Rain AI, a freely distributed library that is used in combination with Unity. This library allows for the creation and modification of the behaviour of the antagonist through use of a behaviour tree. This behaviour tree allows the antagonist to react differently depending on its state. The implementation of the maze generator was created separately from the main framework. This allows for it to be easily taken out as a whole, thus effectively acting as a library. It depends on various parameters to generate maze data, which the framework can provide. The framework can also specify a specific difficulty, making the maze generator use default parameters for that difficulty. The maze data is converted to actual Unity 3D objects in the framework. In order to provide the parameters for the antagonist AI and the maze generator, a player-observer was made. This observer tracks the decisions made by the player and utilises this to generate statistics regarding the player. These are then converted to parameters that are used by the antagonist and the maze generator. This conversion depends on the difficulty settings set for the player. In addition there have been improvements to the graphical design and gameplay. The gameplay now features more items that can be used by the player as well as improved graphics and the usage of sound.

Gaming
Machine Learning
PortalBug
Personalised
Divireo

http://resolver.tudelft.nl/uuid:2159c3a7-5db6-45a5-b8f0-c5e3a1af4c06

Student theses

bachelor thesis

(c) 2015 Boellaard, B.; Bolier, W.; Mairet, V.; Peeters, T.