Publication: Producción de un Videojuego de Acción Táctica con Personajes Controlados mediante Planificación Automática
Loading...
Official URL
Full text at PDC
Publication Date
2022-09
Advisors (or tutors)
Editors
Journal Title
Journal ISSN
Volume Title
Publisher
Citations
Exportar
Abstract
Los videojuegos son un campo de aplicación que goza de un gran éxito económico y aceptación social. A la hora de producirlos, los estudios más pequeños y los desarrolladores independientes cuentan a día de hoy con herramientas potentes y bastantes asequibles como las que proporcionan plataformas como Unity o Unreal Engine. Estas herramientas suelen actualizarse con frecuencia y en general suelen incorporar muchas novedades interesantes, lo que se traduce en una mayor facilidad para crear nuevos y mejores videojuegos. Sin embargo, en algunas áreas como la Inteligencia Artificial, la existencia de herramientas estándar hace que muchos desarrolladores recurran siempre a los mismos conceptos y las mismas técnicas que ya llevan bien instaladas en la industria desde hace décadas.
El objetivo de este proyecto es aplicar un producto de inteligencia artif icial preexistente que actualmente permite desarrollar personajes no jugadores definiendo su comportamiento mediante un sistema de planificación automática sobre un proyecto real de producción de un videojuego. Esta herramienta, llamada GOAP NPC, nunca había sido utilizada anteriormente para desarrollar un juego, y en este trabajo es puesta a prueba en la producción de un corte vertical de videojuego para ordenador con un género y un estilo definidos: sigilo y acción táctica en formato 3D retro, típico de los años 90 y la primera Playstation.
En este proceso el producto ha sido mejorado pensando en aquellos que van a trabajar con él, compartiendo una nueva versión de código libre y publicándola en el mercado más habitual para llegar a desarrolladores de Unreal Engine. Para probar tanto el resultado del videojuego creado, como indagar acerca de la eficacia percibida en esta nueva herramienta, se ha utilizado un experimento de prueba A/B con dos pequeños grupos de jugadores que arroja luz sobre las posibilidades de este cambio de paradigma en la inteligencia artificial de los videojuegos.
Video games are a field of application that enjoys great economic success and social acceptance. When it comes to producing them, smaller studios and independent developers nowadays have powerful and quite affordable tools such as those provided by platforms like Unity or Unreal Engine. These tools tend to be updated frequently and generally incorporate many interesting new features, which translates into greater ease in creating new and better video games. However, in some areas such as Artificial Intelligence, the existence of standard tools means that many developers always resort to the same concepts and techniques that have been well established in the industry for decades. The objective of this project is to apply a pre-existing artificial intelligence product that currently allows the development of non-player characters by defining their behavior through an automatic planning system on a real video game production project. This tool, called GOAP NPC, has never been used before to develop a game, and in this work it is tested in the production of a vertical cut of a computer game with a defined genre and style: stealth and tactical action in retro 3D format, typical of the 90’s and the first Playstation. In this process the product has been improved thinking about those who are going to work with it, sharing a new open source version and publishing it in the most common market to reach Unreal Engine developers. To test both the result of the video game created, and to inquire about the perceived effectiveness of this new tool, an A/B test experiment with two small groups of players has been used to shed light on the possibilities of this paradigm shift in artificial intelligence in video games.
Video games are a field of application that enjoys great economic success and social acceptance. When it comes to producing them, smaller studios and independent developers nowadays have powerful and quite affordable tools such as those provided by platforms like Unity or Unreal Engine. These tools tend to be updated frequently and generally incorporate many interesting new features, which translates into greater ease in creating new and better video games. However, in some areas such as Artificial Intelligence, the existence of standard tools means that many developers always resort to the same concepts and techniques that have been well established in the industry for decades. The objective of this project is to apply a pre-existing artificial intelligence product that currently allows the development of non-player characters by defining their behavior through an automatic planning system on a real video game production project. This tool, called GOAP NPC, has never been used before to develop a game, and in this work it is tested in the production of a vertical cut of a computer game with a defined genre and style: stealth and tactical action in retro 3D format, typical of the 90’s and the first Playstation. In this process the product has been improved thinking about those who are going to work with it, sharing a new open source version and publishing it in the most common market to reach Unreal Engine developers. To test both the result of the video game created, and to inquire about the perceived effectiveness of this new tool, an A/B test experiment with two small groups of players has been used to shed light on the possibilities of this paradigm shift in artificial intelligence in video games.
Description
Trabajo de Fin de Grado en Desarrollo de Videojuegos, Facultad de Informática UCM, Departamento de Ingeniería del Software e Inteligencia Artificial, Curso 2021/2022.