Repository logo
 

Procedural Content Generation Using Meta Heuristics Approaches: A Comparison Study

Files

Thesis(4.97 MB)

Date

2024

Authors

Supervisor

Nand, Parma
Sinha, Roopak

Item type

Thesis

Degree name

Doctor of Philosophy

Journal Title

Journal ISSN

Volume Title

Publisher

Auckland University of Technology

Abstract

Procedural content generation (PCG) has emerged as a powerful approach for automating game content creation, offering significant benefits in terms of cost reduction and time efficiency, compared to traditional game design and development processes. The use of PCG algorithms enables the automatic generation of various game elements including terrain, maps, stories, dialogues, quests, and characters. Although genetic algorithms (GAs) have been widely employed in PCG, there is a growing interest in exploring alternative metaheuristic algorithms that can further enhance content generation capabilities. In recent years, particle swarm optimization (PSO) and artificial bee colonies (ABC) have gained attention as effective metaheuristic algorithms capable of delivering high-quality solutions and efficient optimization in diverse problem domains. However, their application in PCG remains relatively limited, with most studies focusing on GAs. This study challenges the conventional 'one-size-fits-all' approach to PCG by assessing the effectiveness of PSO and ABC, specifically for race track and map generation tasks. The objective is to showcase their task-specific advantages over GAs, with the potential to enhance efficiency and content quality within these defined domains. To achieve this, a comparative analysis was conducted among three metaheuristic algorithms, GA, ABC, and PSO. The objective is to assess the effectiveness and performance characteristics of these algorithms in generating game levels. Comprehensive experiments were conducted by applying GA, ABC, and PSO to generate diverse levels, such as race tracks and map layouts. Metrics, such as convergence speed and content quality, were employed to evaluate the generated game content. Convergence speed measures how quickly the algorithms reach optimal or near-optimal solutions, whereas content quality assesses the aesthetic appeal, playability, and overall suitability of the generated content for gameplay. The findings of this study indicate that both ABC and PSO demonstrate advantages over traditional GA implementations in terms of race track generation. This highlights the potential benefits of integrating alternative metaheuristic algorithms into PCG workflow. By leveraging a diverse range of algorithms, PCG can not only improve content creation efficiency and effectiveness but also enhance the overall gaming experience. Furthermore, this study underscores the importance of a "Task-Specific Considerations" approach when selecting algorithms for PCG tasks. This approach emphasizes a meticulous analysis of the inherent complexity of each task. Factors such as the desired level of content quality and the required solution speed are crucial considerations in the algorithm selection. The "Task-Specific Considerations'' approach encourages exploration beyond traditional GAs and acknowledges the distinct advantages offered by alternative metaheuristic algorithms like ABC and PSO. Each algorithm possesses unique strengths and weaknesses in key areas such as exploration, exploitation, consistency, and solution quality. Therefore, the judicious selection of the most suitable algorithm based on the specific requirements of a PCG project is essential for its successful implementation.

Description

Keywords

Source

DOI

Publisher's version

Rights statement

Permanent link

http://hdl.handle.net/10292/18846

Collections

Doctoral Theses