Institutional Repository [SANDBOX]
Technical University of Crete
EN  |  EL

Search

Browse

My Space

Generating personalized/balanced racing games via rolling horizon evolution

Ziskas Christos

Full record


URI: http://purl.tuc.gr/dl/dias/325C3ACB-AF77-4839-94D8-9389EC3F8B70
Year 2022
Type of Item Diploma Work
License
Details
Bibliographic Citation Christos Ziskas, "Generating personalized/balanced racing games via rolling horizon evolution", Diploma Work, School of Electrical and Computer Engineering, Technical University of Crete, Chania, Greece, 2022 https://doi.org/10.26233/heallink.tuc.91693
Appears in Collections

Summary

In recent years, game development has been heavily dedicated to the advancement of Procedural Content Generation (PCG). A much-discussed topic of study for game developers is the autonomous generation of levels for video games. Evolutionary Algorithm (EA)s have seen extensive uses due to the stability in general computational problems and the demand for artificial intelligence techniques. In this dissertation, we test the ability of an innovative algorithm to offer personalized experiences online in the racing video game genre. We use a recent stochastic planning algorithm named Rolling Horizon Evolution Algorithm (RHEA), which generated content (parts of a race track) based on the difficulty of the level and the player’s in-game performance. The algorithm is tested against Artificial Intelligence (AI) and human players; AI racing players define the bounds of the flow channel within which the human players are assessed. The algorithm then attempts to bring the level of difficulty to match player performance through its fitness function. Results suggest that the algorithm is operational and that the experience of various human players is optimized based on their skill level.

Available Files

Services

Statistics