Publication: Desarrollo de un Sistema Multiagente basado en creencias, deseos e intenciones para modelar personajes autónomos en videojuegos utilizando Jason y Unity
Loading...
Files
Official URL
Full text at PDC
Publication Date
2018
Authors
Advisors (or tutors)
Editors
Journal Title
Journal ISSN
Volume Title
Publisher
Citations
Exportar
Abstract
Los videojuegos siempre han querido incorporar personajes autónomos como parte de sus mundos de ficción. El problema de la IA en personajes de videojuego no es tanto el ser avanzada sino que requiere credibilidad, y ser suficientemente expresiva como para que el jugador “entienda” lo que hace el personaje. Un modelo interesante de IA es el que considera lo que el personaje cree saber sobre el mundo, los deseos que tiene y las intenciones que es capaz de desarrollar para satisfacerlos en forma de planes.
Usar técnicas simbólicas de IA y herramientas pensadas para la depuración de sistemas multiagente con este tipo de arquitecturas puede ayudar a desarrollar y convencer a la industria de que es posible tener personajes inteligentes y capaces de “explicar” sus comportamientos.
Con el fin de explorar nuevas tecnologías para crear personajes en videojuegos, se propone utilizar el modelo de Creencias-Deseos-Intenciones y una combinación de plataformas (Unity como entorno de desarrollo de videojuegos, y Jason como plataformas para desarrollar y depurar sistemas multiagente BDI) para conseguir personajes inteligentes.
Se propone para ello un conector UniJason, basado en otro TFG (Sánchez-López, Romero, y Martín-Solís, 2016), con un protocolo mejor definido de mensajes que se intercambian de manera bidireccional, y un diseño genérico aplicable a cualquier tipo de videojuego.
Para probar este sistema se ha desarrollado un videojuego, Miners of the Broken Planet (en homenaje a la superproducción española de Mercurysteam, Raiders of the Broken Planet). En éste, el jugador puede manejar a distintos personajes, cada uno con un rol, tareas y personalidad diferentes.
Las pruebas con usuarios reales realizadas confirman que el conector funciona, y que el juego también es comprensible para las personas que lo juegan.
Video games have always wanted to incorporate autonomous characters as part of their fiction worlds. The problem with AI in video game characters is not being advanced, but to be credible, and to be sufficiently expressive for the player to ’understand’ what the character does. An interesting model of AI is the one that considers what the character thinks he knows about the world, the desires he has in that world and the intentions he is able to develop to satisfy them in the form of plans. Using symbolic AI techniques and tools designed for debugging multi-agent systems with this type of architecture can help develop and convince the industry that it is possible to have intelligent characters capable of ’explaining’ their behaviour. In order to explore new technologies for creating characters in videogames, it is proposed to use the Beliefs-Wishes-Intentions model and a combination of platforms (Unity as a videogame development environment, and Jason as platforms for developing and debugging multi-agent BDI systems) to achieve intelligent characters. A UniJason connector based on another TFG is proposed for this purpose. A one with a better defined protocol of messages that are exchanged bidirectionally, and a generic design applicable to any type of videogame. To test this system, a video game has been developed, Miners of the Broken Planet (an homage to Mercurysteam’s Spanish superproduction, Raiders of the Broken Planet). In this game, the player can manage different characters, each with a different role, tasks and personality. Testing with real users confirms that the connector works, and that the game is understandable to the people who play it.
Video games have always wanted to incorporate autonomous characters as part of their fiction worlds. The problem with AI in video game characters is not being advanced, but to be credible, and to be sufficiently expressive for the player to ’understand’ what the character does. An interesting model of AI is the one that considers what the character thinks he knows about the world, the desires he has in that world and the intentions he is able to develop to satisfy them in the form of plans. Using symbolic AI techniques and tools designed for debugging multi-agent systems with this type of architecture can help develop and convince the industry that it is possible to have intelligent characters capable of ’explaining’ their behaviour. In order to explore new technologies for creating characters in videogames, it is proposed to use the Beliefs-Wishes-Intentions model and a combination of platforms (Unity as a videogame development environment, and Jason as platforms for developing and debugging multi-agent BDI systems) to achieve intelligent characters. A UniJason connector based on another TFG is proposed for this purpose. A one with a better defined protocol of messages that are exchanged bidirectionally, and a generic design applicable to any type of videogame. To test this system, a video game has been developed, Miners of the Broken Planet (an homage to Mercurysteam’s Spanish superproduction, Raiders of the Broken Planet). In this game, the player can manage different characters, each with a different role, tasks and personality. Testing with real users confirms that the connector works, and that the game is understandable to the people who play it.
Description
Universidad Complutense, Facultad de Informática, curso 2017/2018