Controlling Energy-Propagation-Based Dungeon Generation using Genetic Algorithms

• Rafael D. M. dos Santos UNIFESP
• Rodrigo C. Contreras UNIFESP / USP
• Gabriel B. Guimarães USP
• Monique S. Viana UNESP
• Rodrigo C. Guido UNESP

DOI: https://doi.org/10.5753/sbgames_estendido.2026.25589

Resumo

Introduction: Procedural Content Generation plays an important role in game development, especially in scenarios that require control over random processes. However, balancing stochastic behavior with rigid Level Design constraints remains a challenging problem. Objective: This work investigates the use of classic metaheuristics, specifically a Genetic Algorithm (GA), to control dungeon generation in a high-entropy environment. Steps: The approach consists of comparing the performance of the Genetic Algorithm with a Random Search strategy within the same generative environment. Expected Results: Experimental results indicate that, despite the high variance of the environment, the GA is able to find high-quality solutions more frequently and with lower computational cost, highlighting advantages and limitations of evolutionary approaches in noisy optimization scenarios.

Palavras-chave: Procedural Content Generation, Genetic Algorithms, Level Design, Metaheuristics, Dungeon Generation

Referências

Amato, A. (2017). Procedural content generation in the game industry. Game and Media Technology, 1:1–9.

Dias, M. V. P., Franco, A. d. O. d. R., da Silva, J. W. F., e Maia, J. G. R. (2025). Procedural map generation by energy propagation. In Proceedings of the XXIV Brazilian Symposium on Computer Games and Digital Entertainment (SBGames), Salvador, BA, Brazil.

Goldberg, D. E. (1989). Genetic Algorithms in Search, Optimization, and Machine Learning. Addison-Wesley, Reading, MA.

Zhang, Y., Zhang, G., e Huang, X. (2022). A survey of procedural content generation for games. In 2022 International Conference on Culture-Oriented Science and Technology (CoST), pages 186–190.