DOC PREVIEW
Stanford HPS 154 - Character driven Story Generation in Interactive Storytelling

This preview shows page 1-2 out of 7 pages.

Save
View full document
View full document
Premium Document
Do you want full access? Go Premium and unlock all 7 pages.
Access to all documents
Download any document
Ad free experience
View full document
Premium Document
Do you want full access? Go Premium and unlock all 7 pages.
Access to all documents
Download any document
Ad free experience
Premium Document
Do you want full access? Go Premium and unlock all 7 pages.
Access to all documents
Download any document
Ad free experience

Unformatted text preview:

Character-driven Story Generation in Interactive StorytellingFred Charles, Steven J. Mead, and Marc CavazzaUniversity of Teesside, Middlesbrough, TS1 3BA, UK{f.charles, steven.j.mead, m.o.cavazza}@tees.ac.ukAbstractIn this paper we describe a fully implemented prototype for interactive storytelling usingthe Unreal™ engine. We describe the important mechanisms involved in the variability ofplot instantiations, within a scenario of sitcom genre. We also provide an evaluation of theconcepts of how the dynamic interactions between agents and/or the user influence thegeneration of story, with first results of examples.1. IntroductionIn this paper, we present results from a first version of a fully-implemented storytellingprototype, which illustrate the generation of variants of a generic storyline. These variantsresult from the interaction of autonomous characters with one another, with environmentresources or from user intervention.The development of artificial actors and AI-based animation naturally leads to envisionfuture interactive storytelling systems. A typical interactive storytelling system would bebased on autonomous virtual actors that generate the plot through their real-time interaction.Besides, the user should be allowed to interfere with the ongoing action, thereby altering theplot as it unfolds.Many interactive storytelling models have been proposed: user-centred plot resolution [1],character-based approaches [2] [3], anytime interaction [4] and the need for narrativeformalisms [5]. Previous work has identified relevant dimensions and key problems for theimplementation of interactive storytelling, among which: the status of the user, the level ofexplicit narrative representation and narrative control, the modes of user intervention, therelations between characters and plot, etc.Some of these problems derive from the inherent tension between interaction and narrative[2] [3]. Interactive systems demand user involvement but often at the expense of a realstoryline; on the other hand, a strong narrative dimension is traditionally conceived with auser as spectator rather than being actively involved. Our solution to this problem consists inlimiting the user involvement in the story, though interaction should be allowed at anytime.This is achieved by driving the plot with autonomous characters’ behaviours, and allowingthe user to interfere with the characters’ plans. The user can interact either by physicalintervention on the set or by passing information to the actors (e.g., through speech input).In the next sections, we will introduce the important concepts of character-centredstorytelling as well as a brief description of our implementation. Results of variants in storygeneration are illustrated with different examples.2. System description2.1. Character-based storytellingThe storyline for our prototype is based on a simple sitcom-like scenario, where the maincharacter (“Ross”) wants to invite the female character (“Rachel”) out on a date. Thisscenario tests a narrative element (i.e. “Will he succeed?”) as well as situational elements (theactual episodes of this overall plan that can have dramatic significance, e.g., how he willmanage to talk to her in private if she is busy, etc.). Our system is driven by characters’behaviours. These actually “compile” narrative content into characters’ behaviours, bydefining a superset of all possible behaviours, represented by a plan for each character.Dynamic choice of an actual course of action within this superset is the basis for plotinstantiation [2]. In that sense, this addresses the causality/choice duality described by Raskin[6] in storytelling, though this choice takes place within the limits of the formalism used torepresent possible behaviours, which is a plan-based formalism [2]. This can be illustrated byconsidering the overall plan for the Ross character (Figure 1).Figure 1. Plan-based representation of agent behaviourIn order to invite Rachel, he must for instance acquire information on her preferences, finda way to talk to her, and finally formulate his request (or having someone acting on hisbehalf, etc.). These goals can be broken into many different sub-goals, corresponding tovarious courses of action, each having a specific narrative significance.The initial storyline should actually determine not only the main character's plan, but thoseof other characters as well. The problem of dependencies between characters’ roles hasactually been described within modern narratology, though not to a formal level. Narrativefunctions can be refined into bipolar relations between a couple of actors, emphasising theasymmetry in their roles [7]. We have adopted this framework to define the respectivebehaviours of our two leading characters. We started with the overall narrative propertiesimposed by the story genre (sitcoms). In terms of behaviour definition, this amounts todefining an “active” plan for the Ross character (oriented towards inviting Rachel) and ageneric pattern of behaviour for Rachel (her day-to-day activities).2.2. AI-based characters’ behavioursIndividual agent behaviours are produced by solving the set of sub-plans described in thepreceding section, which are represented by Hierarchical Task Networks (HTN), such as theone of Figure 1. Using formal properties of these plans, it is possible to generate solutionplans by searching directly the AND/OR graph of the HTN with an algorithm such as AO*[8] [9]. In our system, this is done with a “real-time” variant of AO*, which interleavesplanning and execution and supports re-planning that is required when a character’s plan isaltered through interaction with another virtual actor or the user. The terminal actions (e.g.reaching a location, using an object, interacting with other actors) forming the plan areactually played in the graphic environment through their corresponding animations. Thedramatisation of these actions constitutes the story as seen by the user.The “virtual sitcom” prototype described in this paper has been developed using theUnreal™ game engine. The Unreal™ environment provides most of the user interactionfeatures required to support user intervention in the plot, such as navigating about andinteracting with objects within the virtual set and its use has been previously reported inprototyping interactive storytelling [2]. The system has been fully implemented as a set oftemplate C++ classes, which


View Full Document

Stanford HPS 154 - Character driven Story Generation in Interactive Storytelling

Download Character driven Story Generation in Interactive Storytelling
Our administrator received your request to download this document. We will send you the file to your email shortly.
Loading Unlocking...
Login

Join to view Character driven Story Generation in Interactive Storytelling and access 3M+ class-specific study document.

or
We will never post anything without your permission.
Don't have an account?
Sign Up

Join to view Character driven Story Generation in Interactive Storytelling 2 2 and access 3M+ class-specific study document.

or

By creating an account you agree to our Privacy Policy and Terms Of Use

Already a member?