Agent mediated interactive storytelling

Abstract

Interactive storytelling in the virtual environment attracts many interests from researchers and educators in recent years. People are engaged in the storytelling as active participants by interacting with virtual characters as well as story director (storyteller), rather than acting as passive audiences in the conventional storytelling. However, it is a great challenge to

create an immersive interactive storytelling system in the virtual environment from scratch with an engaging storyline and believable characters, especially for the novice developers without many programming skills. The current interactive storytelling methodologies (e.g. plot-based interactive storytelling and character-based interactive storytelling) cannot ensure the story plot coherence and character believability simultaneously. The director and

characters are not autonomous and intelligent enough to handle dynamic user interactions

and context changes.

In the thesis, we propose a novel Agent-Oriented Methodology for Interactive Storytelling (AOMIS) to facilitate novices to develop an immersive interactive storytelling system from a story. It generates both the story plot and the character behaviors as agent goals seamlessly in real-time. Human-like agent is a promising solution for the intelligent director and characters in the interactive virtual storytelling with the high-level goal automation besides

the low-level behavior automation. A multi-agent system (MAS) has been constructed for

illustrating the hybrid model, which involves a scriptwriter agent to draft the nonlinear

storylines, a director agent to select the storyline, virtual character agents to perform the story scenes, and other supporting agents.