Procedural generation of biomes using a MAP-Elites algorithm applied to the context of a serious game
Resumo
Introduction: The Atlantic Forest is one of the most biodiverse and threatened biomes in the world, requiring new approaches to promote awareness and understanding of its ecological dynamics. Procedural Content Generation (PCG), combined with evolutionary algorithms, offers a promising way to simulate and represent such complex environments in interactive systems. Objective: This work aims to develop a MAP-Elites evolutionary algorithm capable of procedurally generating terrains that resemble Atlantic Forest biomes, and to evaluate its effectiveness within an educational game context. Methodology or Steps: The proposed algorithm generates biomes based on six physical attributes, using a Dirichlet distribution and mutation strategies to guide evolution toward predefined input values. The system was integrated into an educational game (GAIA), enabling both computational evaluation and user-based assessment. Results: The algorithm satisfactorily converges to desired attributes. Pilot study (N=18) shows positive perceptions of generated environments. Given small exploratory sample, this is a computational proof-of-concept for terrain generation, not a validated educational intervention. Positive findings suggest potential, but educational efficacy needs future controlled studies.
Referências
Brasil. Ministério do Meio Ambiente e Mudança do Clima (n.d.). Biodiversidade. Página institucional do governo brasileiro. Acesso em: 24 abr. 2026.
Cardoso, J. T. (2016). A mata atlântica e sua conservação. Revista Encontros Teológicos, 31(3).
Fischer, R., Dittmann, P., Weller, R., e Zachmann, G. (2020). Autobiomes: procedural generation of multi-biome landscapes. The Visual Computer, 36(10):2263–2272.
Kung, F. Y., Kwok, N., e Brown, D. J. (2018). Are attention check questions a threat to scale validity? Applied Psychology, 67(2):264–283.
Ng, K. W., Tian, G.-L., e Tang, M.-L. (2011). Dirichlet and Related Distributions: Theory, Methods and Applications. Wiley Series in Probability and Statistics. John Wiley & Sons, Hoboken, NJ, USA.
Nordmoen, J., Nygaard, T. F., Samuelsen, E., e Glette, K. (2021). On restricting real-valued genotypes in evolutionary algorithms. In Castillo, P. A. e Jiménez Laredo, J. L., editors, Applications of Evolutionary Computation, pages 3–16, Cham. Springer International Publishing.
Pei, W. et al. (2020). Attention please: Your attention check questions in survey studies can be automatically answered. In Proceedings of The Web Conference 2020, page 1182–1193. Association for Computing Machinery.
Pereira, L. T., de Souza Prado, P. V., Lopes, R. M., e Toledo, C. F. M. (2021). Procedural generation of dungeons’ maps and locked-door missions through an evolutionary algorithm validated with players. Expert Systems with Applications, 180:115009.
Pereira, L. T., Viana, B. M. F., e Toledo, C. F. M. (2024). A system for orchestrating multiple procedurally generated content for different player profiles. IEEE Transactions on Games, 16(1):64–74. DOI = 10.1109/TG.2022.3213781.
Sepúlveda, G. K., Besoain, F., von Brand, S., e Barriga, N. A. (2023). Biome generation tool: A mixed-initiative software for the procedural generation of biomes. Applied Sciences, 13(14):8070.
Silber, H., Roßmann, J., e Gummer, T. (2022). The issue of noncompliance in attention check questions: False positives in instructed response items. Field Methods, 34(4):346–360.
