Project Abstract

Play-Testing is a complex and time consuming part of video game development. Smaller game development studios often struggle to find the time and resources needed for thorough play-testing. This project aims to investigate the practicality of automated play-testing, for smaller independent developers, and adapt existing techniques to better suit their needs. By focusing on the practical application and the specific needs of indie studios this research takes a new approach to automatic play-testing that previous papers have not explored. The practice oriented approach of this research has uncovered several insights into the capabilities and short comings of automated play-testing agents. These insights have led to potential new applications of automated agents for balance testing, and revealed new avenues for future research. In this research an Evolutionary Algorithm was used to optimise play within a card game still under development. Play data was collected on the algorithm as it evolved and developed an optimised strategy. This data was then visualised and scrutinised to detect potential imbalances within the game. This technique was successful in highlighting several imbalances within the game underdevelopment. However, applying this methodology highlighted several problems within the algorithms design, that negatively effected its generalisability, training time and utility within a game development setting.

Keywords: Evolutionary Algorithm, Game Development, Practice Oriented

 Conference Details

Session: Presentation Stream 34 at Presentation Slot 5

Location: CoFo 002 at Wednesday 8th 13:30 – 17:00

Markers: Mark Jones, Galileo Sator (GTA)

Course: MSci Computer Science, Masters 4th Year

Future Plans: I’m continuing studies