A carregar...
Publicação
Deep learning aplicado aos videojogos
| Nome: | Descrição: | Tamanho: | Formato: | |
|---|---|---|---|---|
| 5.01 MB | Adobe PDF |
Autores
Orientador(es)
Resumo(s)
Existem videojogos para todos os gostos e com vários tipos de jogabilidade, no entanto, a interação do jogador com o jogo varia pouco, ou nada, de utilizador para utilizador. Os percursos são, na grande maioria dos casos, lineares, as experiências uniformizadas ou, na melhor das hipóteses, existem pequenas variações de jogabilidade que pouco divergem umas das outras.
Atualmente, as novas tecnologias, particularmente a inteligência artificial e uma das suas subcategorias, o Deep Learning, trazem novas possibilidades para desenvolver experiências não lineares aos jogadores. Neste contexto, o sistema pode distinguir que tipo de desafios, recompensas, cenários e comportamentos dos NPC provocam reações emocionais e comportamentais no jogador e, desta forma, aumentar o seu engagement com o jogo. Este trabalho tem o objetivo de desenvolver um modelo que ajude o Designer de Interação a criar novas possibilidades de interação dos jogadores com NPC para jogos do tipo RPG. Este modelo permitirá alimentar os sistemas IA com Deep Learning, uma tecnologia emergente nos jogos digitais. O modelo proposto foi desenvolvido utilizando o Kansei, uma técnica que permite a criação de um motor preditivo que, neste projeto, relaciona o perfil do utilizador (reações emocionais e comportamentais) em função do comportamento dos NPC (expressões faciais e ações motoras).
O modelo elaborado mostra ser eficaz para ser integrado num jogo que tenha incorporado um sistema de IA com Deep Learning, permitindo adaptar automaticamente os comportamentos dos NPC, em função do contexto de jogo e das reações comportamentais e emocionais dos jogadores.
There are video games for all tastes with various types of gameplay, but on the other hand, the interaction the player has with the game does not change, and it is always the same from user to user. The paths are mostly linear, the experience all the same or, in the best case scenario, they have few gameplay variations that differ a little from one another. Currently, new technologies, particularly Artificial Intelligence and one of its subfields, Deep Learning, bring new possibilities to develop new nonlinear experiences to the players. Within this context, the system can distinguish what type of challenges, rewards and different types of NPC behavior can create emotional and behavioral player’s reactions, and this way increase the engagement during the game. This project has the objective of creating a model that can help the Interaction Designer develop new ways of player interactivity, with the NPC in RPG games. This model will provide IA systems with Deep Learning, an emergent technology in digital games. The proposed model was developed using the Kansei method, a technique that allows the creation of a predictive engine which, in this project, connects the user profile (emotional and behavioral reactions) depending on the NPC reaction (facial and bodily expressions). The developed model shows efficiency and can be integrated in a video game which has Artificial Intelligence with integrated Deep Learning, allowing to automatically adapt the NPC behavior within the game context and player behavior.
There are video games for all tastes with various types of gameplay, but on the other hand, the interaction the player has with the game does not change, and it is always the same from user to user. The paths are mostly linear, the experience all the same or, in the best case scenario, they have few gameplay variations that differ a little from one another. Currently, new technologies, particularly Artificial Intelligence and one of its subfields, Deep Learning, bring new possibilities to develop new nonlinear experiences to the players. Within this context, the system can distinguish what type of challenges, rewards and different types of NPC behavior can create emotional and behavioral player’s reactions, and this way increase the engagement during the game. This project has the objective of creating a model that can help the Interaction Designer develop new ways of player interactivity, with the NPC in RPG games. This model will provide IA systems with Deep Learning, an emergent technology in digital games. The proposed model was developed using the Kansei method, a technique that allows the creation of a predictive engine which, in this project, connects the user profile (emotional and behavioral reactions) depending on the NPC reaction (facial and bodily expressions). The developed model shows efficiency and can be integrated in a video game which has Artificial Intelligence with integrated Deep Learning, allowing to automatically adapt the NPC behavior within the game context and player behavior.
Descrição
Palavras-chave
videojogos deep learning NPC RPG kansei
Contexto Educativo
Citação
Editora
Faculdade Arquitetura, Universidade Lisboa