 |
Email: Homepage: http://wwwhome.ewi.utwente.nl/~jonkman
|
|
| Beschrijving: |
The goal of this project is to develop an embodied presenter agent that can make use of nonverbal behavior (head movement, gaze, facial expressions, gestures etc.) that is appropriate for storytelling. Currently the Presenter Agent in the Virtual Storyteller reads texts of a generated story. The Presenter Agent displays some emotional expressions, based on tags that have been manually placed in the story. The end result does not provide an optimal "entertaining" experience. In the current situation the embodied agent is looking towards you when telling his story.
For this project we want to see if an embodied agent reading the story from a book enhances the experience. The embodied agent will use gazing and head movement as its main methods of nonverbal behavior. By analyzing video material of a "real" storyteller the different types of gazing and head movement will be extracted, accompanied by their point in time. Subsequently Elckerlyc (a verbal and nonverbal behavior realizer, by means of the Behavior Markup Language) will be used as an existing framework to model the nonverbal behavior and to recreate the setting in which stories are told by a "real" storyteller.
For example towards the end of the Dutch version of Sesame Street a story is told by a storyteller who is surrounded by a group of children. A survey will be used to see what the advantages and disadvantages are of the introduced nonverbal behavior, and it will be questioned if the newly created setting (with the introduced nonverbal behavior) provides an "entertaining" story.
|
|
| Beschrijving: |
The Virtual Storyteller is a system that can simulate a story world in which agents play out the roles of their character in a story. This simulation is discrete: each round, every character agent may select and perform an action. This creates interaction problems since no attention is paid to the fact that agents may want to respond directly to each other’s actions.
The purpose of this assignment is to make a model for turn management that allows the character agents to have better ways of interrupting and responding to each other’s actions. Several turn-based games, such as collectible card games (e.g., Magic: the Gathering) and tabletop RPG’s have sophisticated formal rules for turn management that have demonstrated their success. Also, the multi agent systems (MAS) literature might have a thing or two to say about turn management. Another source of inspiration can be the turn taking mechanisms that take place in human conversations. If time allows, the model can be implemented and tested within the Virtual Storyteller framework, or within a small demo application.
|
|
|
