Machine Learning for Playable Room Generation: A Modular Approach with VAE and PCG

  • Juan Paes UEA
  • Roberto Junio UEA
  • Luis Cuevas Rodriguez UEA

Resumo


Introduction: The growing complexity of roguelike games demands smarter content generation, but uncontrolled PCG can result in unbalanced and frustrating gameplay. Objective: This paper proposes a modular system that integrates PCG with Machine Learning (ML) to generate optimized, playable rooms for roguelike games. Methodology: The approach combines Cellular Automata for initial generation and a Variational Autoencoder (VAE) to refine layouts based on user-defined criteria. Results: The system achieved (85%) player approval for VAE-generated rooms, reduced generation time by (58%), and improved diversity, consistency, and gameplay accessibility.
Palavras-chave: Procedural Content Generation, Machine Learning, Roguelike, Variational Autoencoder, Map Generation

Referências

Craddock, D. L. (2021). Dungeon Hacks: How NetHack, Angband, and Other Rougelikes Changed the Course of Video Games. CRC Press.

Gisslén, L., Eakins, A., Gordillo, C., Bergdahl, J., e Tollmar, K. (2021). Adversarial reinforcement learning for procedural content generation. In 2021 IEEE Conference on Games (CoG), pages 1–8. IEEE.

Mateus Omena (2025). Indústria de games no Brasil ’vira o jogo’ no cenário global — e com apoio das grandes marcas.

Minini, P. e Assuncao, J. (2020). Combining constructive procedural dungeon generation methods with wavefunctioncollapse in top-down 2D games. Proceedings of SBGames, pages 27–29.

Rogers, S. (2014). Level Up! The guide to great video game design. John Wiley & Sons.

Se Yeon KIM e Kim, Seokkyoo (2024). A Research on Machine Learning Agent in Rogue-like game. Journal of The Korean Society for Computer Game, 37(1):33–39.

Shaker, N., Togelius, J., e Nelson, M. J. (2016). Procedural Content Generation in Games. Computational Synthesis and Creative Systems. Springer International Publishing, Cham.

Silva, B. C. D., Rodrigues Maia, J. G., e Viana De Carvalho, W. (2024). Procedural content generation in pervasive games: state of affairs, mistakes, and successes. International Journal of Pervasive Computing and Communications, 20(3):345–364.

Werneck, M. e Clua, E. W. (2020). Generating procedural dungeons using machine learning methods. In 2020 19th Brazilian Symposium on Computer Games and Digital Entertainment (SBGames), pages 90–96. IEEE.
Publicado
30/09/2025
PAES, Juan; JUNIO, Roberto; RODRIGUEZ, Luis Cuevas. Machine Learning for Playable Room Generation: A Modular Approach with VAE and PCG. In: SIMPÓSIO BRASILEIRO DE JOGOS E ENTRETENIMENTO DIGITAL (SBGAMES), 24. , 2025, Salvador/BA. Anais [...]. Porto Alegre: Sociedade Brasileira de Computação, 2025 . p. 609-618. DOI: https://doi.org/10.5753/sbgames.2025.10172.