Genetic Encoding of Synergies: Coevolving Traits, Behaviors, and Strategies for Game AI

• Lucas Schurer UFSM
• Luis A. L. Silva UFSM
• Cesar T. Pozzer UFSM
• Joaquim V. C. Assunção UFSM

DOI: https://doi.org/10.5753/sbgames.2025.10159

Resumo

Introduction: Several games use data to train machine learning models, which requires extensive gameplay before creating the model. Moreover, these models are designed to be strong in the game, rather than adapting to the player’s level. Objective: To address this problem, this work focuses on coevolution through the embedded use of Genetic Algorithms (GAs) to adapt the behaviors of game characters to those of human players. Methodology: We describe how the various elements (characters’ traits, weapons, and skills) of a developed 3D game were projected and implemented, showing a concrete instance of how to develop similar games. Then, this work details the mapping of these elements to the GAs so that groups of characters represent populations, and generations represent the rounds of a match. To evaluate the coevolution of agents with the proposed adaptation method, the combined use (or synergies) between weapons, traits, behaviors, and strategies are analyzed in diverse scenarios. Results: The experiments show that characters generated with GAs adapt effectively to different types of opponents and exhibit coherent strategic behavior, confirming the potential of the coevolution method to produce dynamic and responsive game agents.

Palavras-chave: Coevolution, Genetic algorithms, Game Adaptation, Games

Referências

Cook, M., Eladhari, M., Nealen, A., Treanor, M., Boxerman, E., Jaffe, A., Sottosanti, P., e Swink, S. (2016). Pcg-based game design patterns.

Dockhorn, A. e Kruse, R. (2017). Combining cooperative and adversarial coevolution in the context of pac-man. pages 60–67, New York, NY.

Elfeky, E., Elsayed, S., Marsh, L., Essam, D., Cochrane, M., Sims, B., e Sarker, R. (2021). A systematic review of coevolution in real-time strategy games. IEEE Access, 9.

Faria, F. G., Pereira, L. T., e Toledo, C. F. M. (2019). Adaptive generation of characters for tabletop role playing games. pages 39–46.

Galway, L., , Charles, D. K., e Black, M. M. (2008). Machine learning in digital games: A survey. Artificial Intelligence Review, 29:123–161.

Garcia, B. E. R., Crocomo, M. K., e Andrade, K. O. (2018). Dynamic difficulty adjustment in a whac-a-mole like game. pages 88–888.

Justesen, N., Bontrager, P., Togelius, J., e Risi, S. (2019). Deep learning for video game playing.

Krall, J. e Menzies, T. (2012). Aspects of replayability and software engineering: Towards a methodology of developing games. Journal of Software Engineering and Applications, 5(7):459–466.

Liu, S., Louis, S., e Ballinger, C. (2016). Evolving effective micro behaviors in real-time strategy games. IEEE Transactions on Computational Intelligence and AI in Games, 8.

Pereira, L. T., Viana, B. M. F., e Toledo, C. F. M. (2021). Procedural enemy generation through parallel evolutionary algorithm. pages 126–135.

Rawal, A., Rajagopalan, P., e Miikkulainen, R. (2010). Constructing competitive and cooperative agent behavior using coevolution. pages 107–114.

Segundo, C. V. N., Calixto, K. E. A., e De Gusmão, R. P. (2016). Dynamic difficulty adjustment through parameter manipulation for space shooter game.

van den Nouweland, A. (2007). Rock-paper-scissors a new and elegant proof. Economics Bulletin, 3(43):1–6.

Vie, A., Kleinnijenhuis, A. M., e Farmer, D. J. (2021). Qualities, challenges and future of genetic algorithms: a literature review.

Weber, M. e Notargiacomo, P. (2020). Dynamic difficulty adjustment in digital games using genetic algorithms. pages 62–70.

Zaidan, T. e Góes, L. F. W. (2016). Evolvestone an evolutionary generator of balanced digital collectible card games. In Brazilian Symposium on Computer Games and Digital Entertainment.