sexlies

Uploaded from authorPOINTLite
Views:
 
Category: Entertainment
     
 

Presentation Description

No description available.

Comments

Presentation Transcript

Sex, Lies and Video Games: An Interactive Storytelling Prototype: 

Sex, Lies and Video Games: An Interactive Storytelling Prototype Marc Cavazza, Fred Charles, Steve Mead University of Teesside Middlesbrough, UK

Other References: 

Other References Marc Cavazza, Fred Charles and Steven Mead “Characters in Search of An Author” (2001) Marc Cavazza, Fred Charles and Steven Mead “AI-Based Animation for Interactive Storytelling” (2001) Marc Cavazza, Fred Charles and Steven Mead “Agents’ Interaction in Virtual Storytelling” (2001) Michael Mateas “An Oz-Centric Review of Interactive Drama and Believable Agents” (1997) http://www-2.cs.cmu.edu/afs/cs.cmu.edu/project/oz/web/papers/CMU-CS-97-156.html Jonathan Gratch “Émile: Marshalling Passions in Training and Education” (2000) John Gratch, Jeff Rickel, Stacy Marsella, William Swartout and Randall Hill “Steve Goes to Bosnia: Towards a New Generation of Virtual Humans for Interactive Experiences” (2001)

Motivation: 

Motivation Extend audience interaction Military training Educational Purposes

Architecture: 

Architecture Multi-Agent System Unreal™ game engine DLL interfaces with the game engine System fully implemented as template C++ classes

Types of Agents: 

Types of Agents Agents: actors or “characters” Can be of two kinds Primary: Usually goal driven Secondary: Purely reactive

Environment: 

Environment Continuous Non-deterministic Episodic Inaccessible Dynamic

Goals: 

Goals Vary from time to time No ultimate drive Programmed into agent by “author” Absence of goals: Purely reactive agent

Sensory Input: 

Sensory Input Auditory: Can hear “nearby” sounds Visual: Conical field

Actions (Plans): 

Actions (Plans) Can be primitive or complex Complex actions built upon primitives Agents use Planning Plans: Ordered sequence of Steps Steps: Preconditions, Actions and Effects Planning: Top-Down or Bottom-Up Action Selection: Based upon agent plan.

Action Selection: 

Action Selection Actor can only react to sensed changes in environment Unless actor has a goal Actors with goals: Use real-time planning All actors compete for “resources” Resources: Time, physical objects Unavailability of a resource necessitates re-planning capabilities

Action Selection from HTN plans: 

Action Selection from HTN plans Solution derived by searching through plan Top-down left-to-right search with backtracking Implemented using real-time variant of AO* Hendler, Tsunato et al. “Plan-Refinement Strategies and Search-Space Size”, Proceedings of the European Conference on Planning, 1997, pp. 414-426.

Agent Goals and Planning: 

Primary characters usually have a definite goal Create a plan towards achieving it Plan represented as a Hierarchical Task Network (HTN). HTN is an AND/OR Graph Tasks from HTN are usually executed from top-down and left to right Backtracking if actions fail Agent Goals and Planning

Hierarchical Task Network (HTN): 

Hierarchical Task Network (HTN)

Dramatic Purpose: 

Dramatic Purpose Dynamic Interaction of characters’ plans (or no plans) leads to humorous situations Illustrated by enactment of sitcom ”Friends™” User can follow story from any perspective (of characters or her own) User can also navigate the virtual set unseen by characters

Actors and “Characters” in “Friends™”: 

Actors and “Characters” in “Friends™” Jennifer Anniston (“Rachel”) Courtney Cox (“Monica”) Lisa Kudrow (“Phoebe”) Matthew Perry (“Chandler”) Matt LeBlanc (“Joey”) David Schwimmer (“Ross”) Prototype restricts itself to: Ross and Rachel (primary actors) Phoebe (secondary actor)

Episode Details: 

Episode Details Ross’ Goal: To ask Rachel out to dinner Rachel’s Goal: None

Ross’ Plan: 

Ross’ Plan To ask Rachel out: Ross must find out Rachel’s preferences Consult her PDA Ask Phoebe Gain Rachel’s affection Buy her gifts Isolate Rachel from the others…..

Ross’ Preferences among Actions: 

Ross’ Preferences among Actions Influenced by personality profile Maybe influenced by “moods” or emotions Personality profile can be built-in Can be changed Moods (emotions): Not implemented in prototype but subject of future

Hierarchical Task Network (HTN): 

Hierarchical Task Network (HTN)

User Intervention: 

User Intervention Act upon physical objects on screen that bear narrative influence Influence actors’ actions by directly “speaking to them” Consequence for actors: Re-planning Re-planning uses bottom-up search of HTN

Re-planning scenarios for actors: 

Re-planning scenarios for actors Emergent situations that cannot be ignored Actors use “situated reasoning” Situated reasoning tries to avoid undesirable future outcomes with respect to actor’s goals Actions of actors in emergent situations impacts future scenario Unavailability of resources User intervention

Friends™: An interactive episode: 

Friends™: An interactive episode Ross enters Rachel’s bedroom

Friends™: An interactive episode: 

Friends™: An interactive episode Unseen by Phoebe who’s preparing coffee

Friends™: An interactive episode: 

Friends™: An interactive episode User intervenes and removes “narrative object” (PDA) from Rachel’s Room

Friends™: An interactive episode: 

Friends™: An interactive episode Ross gets to Rachel’s room and discovers PDA missing

Friends™: An interactive episode: 

Friends™: An interactive episode Ross makes a new decision to ask Phoebe about Rachel’s preferences (Re-planning)

Friends™: An interactive episode: 

Friends™: An interactive episode Ross interrupts Phoebe to ask her about Rachel

Friends™: An interactive episode: 

Friends™: An interactive episode Ross interrupts Phoebe to ask her about Rachel

Friends™: An interactive episode: 

Friends™: An interactive episode Ross asks Rachel out

Emotions in Agents: 

Emotions in Agents Emotions: Related to Agent Plans (Gratch 2000) Outcome of relation of events to agent’s plans and goals (Ex: Fear, Frustration) Outcome of interaction between events and agents’ plans and goals (Ex: Anger, Jealousy)

Back to Emotional Friends™: 

Back to Emotional Friends™ Rachel sees Ross and Phoebe conversing animatedly Rachel “feels” jealous Actors can’t really emote (!!) Alternative: “Mood” T-shirts Emotions affect Action Selection Rachel in a jealous mood would refuse Ross outright

User advising Ross: 

User advising Ross

Conclusions and Future Directions: 

Conclusions and Future Directions Interactive Storytelling is at an early developmental stage Better co-ordination of actors required Emotive aspects of actors need to be worked upon Character-based plot generation cannot really “surprise” the user Plot-based narration and emergent plot generation can lead to more entertaining packages