2006 Fiscal Year Final Research Report Summary
Intelligent Game Character Platform Producing Active Drama Experiences
| Project/Area Number |
17300053
|
|---|
| Research Category |
Grant-in-Aid for Scientific Research (B)
|
| Allocation Type | Single-year Grants |
|---|
| Section | 一般 |
|---|
| Research Field |
Perception information processing/Intelligent robotics
|
|---|
| Research Institution | University of Tsukuba |
|---|
Principal Investigator |
HOSHINO Junichi University of Tsukuba, Graduate School of Systems and Information Engineering, Assistant Professor, 大学院システム情報工学研究科, 講師 (40313556)
|
|---|
| Co-Investigator(Kenkyū-buntansha) |
HOSHINO Kiyoshi University of Tsukuba, Graduate School of Systems and Information Engineering, Associate Professor, 大学院システム情報工学研究科, 助教授 (80251528)
KUZUOKA Hideaki University of Tsukuba, Graduate School of Systems and Information Engineering, Professor, 大学院システム情報工学研究科, 教授 (10241796)
|
|---|
| Project Period (FY) |
2005 – 2006
|
| Keywords | game technology / episode action control / imitation learning / second language learning system / reaction synthesis / conversation representation / psychological representation |
|---|
| Research Abstract |
1) Action Control of Game Characters
Constructing a narrative ecosystem is one of the important technologies for many computer games. This paper presents a reactive episode control system that allows for dynamic creation of a long-term story in response to the active interaction between multiple characters and the user's actions. This system makes possible a story world where the user's actions not only create reactions among the characters that are currently present, but also influence the long-term story events. Episode control system stores a number of episode trees (components of the story) and dynamically connects them in a hierarchy, in parallel or in series to generate a variety of possible stories. This model organizes the flow of events into a hierarchy based on AND/OR conditions, creating many possibilities that could happen in a single episode. A single story is composed of a number of episodes organized in a hierarchy, in parallel or in series.
2) Imitation Learning of Action
… More
Game Character
In action games, the computer's behavior lacks diversity and human players are able to learn how the computer behaves by playing the same game over and over again. As a result, human players eventually grow tired of the game. Therefore, this paper proposes a method of imitating the behavior of human players by creating profiles of players from their play data. By imitating what many different players do, a greater variety of actions can be created.
3) Application for Second Language Learning System
In the process of acquiring a language skill by task-based learning, it's important for the learner to try to convey information to one another and reach mutual comprehension through restating, clarifying, and confirming information in the process of communications. Therefore, we construct a model of negotiation of meaning for NPC (Non-player Character). When the learner has a conversation with NPC, NPC may help learner get started or work through a stumbling block by the negotiation of meaning function. The system architecture is designed to support several important internal requirements. The Task Model provides a task based on the leaner and initializes the scenario and the position of NPC in the virtual environment. The Conversation Model includes the Negotiation of Meaning Model that both applications can use. When NPC answers the question from the learner, a speech recognizer that both applications can use, and a natural language parser that can annotate phrases with structural information and refer to relevant grammatical explanations. Less
|
Research Products
(17 results)
