EMBODIED AGENTS

IN AUGMENTED & VIRTUAL REALITIES

Course E6998-004, Dept. of Computer Science, Columbia University, Fall 2002
Prof. Kris Thórisson, Ph.D.
 
 
 

 

LECTURE NOTES

LECTURE 1 PART 1

September 5, 2002

 

 
     









0

Concepts We'll Cover in This Course

 
 

Embodiment

 
 

Agents

 
 

Virtual & Augmented Worlds

 
 

Multimodal Knowledge

 
 

Perception

 
 

Decision

 
 

Action & Motor Control

 
 

Architecture

 
 

 

 









1

Concepts

 
 

Agent

 
 

Embodied Agent

 
 

Virtual World

 
 

Augmented World

 
     








2

Agent

 

 

  • Typical

"An autonomous or semi-autonomous program that acts on behalf of someone or something else."
 
 
Focus on: Delegation

 

 

 

  • Our

"Any (partially or fully) autonomous system that approximates how people or animals behave, in that it has (1) a clearly distinguished perceptual apparatus to get information from the world, a (2) mind-like program for understanding and planning in the world, and (3) communicative & manipulative apparatus for conveying information to and manipulating the world.
 
 
Focus on: Perception, thought & communication
 
     






3

Embodied Agent

 

Embodiment = Physical Appearance

 
 

Embodiment connects to the outside world
- via its perceptual and motor apparatus

 
 

Example: Gandalf

 
 

 
 
Related: Humanoid, Virtual Agent, Artificial Agent, Artificial Character, Virtual Human, Autonomous Character
 
     





 





4

The Case for Embodiment

 
 

Machine intelligence (A.I.) -> Communication

 
 

Communication -> Human-like communication

 
 

Humanoid communication -> Mutlimodal, Real-time, Embodied Humanoids

 
 

Multimodal, Real-time, Embodied Humanoids -> Psychological models

 
 

Psychological Models -> Machine intelligence (A.I.)

 
     








5

The Case for Embodiment

 
 

The "Star Wars™ Bar Problem":

 
 

How do you communicate with these characters?

 
 
 
     









6

Virtual World

 
 

Graphically drawn "world"
- fools the senses into believing in some alternative to the real world

 

Related: Virtual Reality, Synthetic Reality,

 

Example: LEGO Castle World

 

 
 
 
  1998©LEGO - Wizard Group  








7

Augmented Reality

 
 

Reality with a "graphics overlay"

 

- can include virtual worlds

 

Examples:

 
 
 
 

 
  Augmented Reality: A New Way of Seeing. STEVEN K. FEINER.
Scientific American, April 2002 issue
 

 

 
 
 
  1998©LEGO Wizard Group  








8

More Concepts

 
 

Multimodal Communication

 
 

Massively Interactive

 
 

Artifiical Intelligence

 
 

 

 







9

Multimodal Communication

 
 

Multimodal: Speech, gesture, co-spatiality & co-temorality

 
 

Eyes: Attentional and deictic functions, both during speaking and listening.

 
 

Speech: Prosody — the timing and sequence of speech-related sounds and intonation

 
 

Speech content: What we say
(not how we say it -- but beware of interactions!)

 
 

Body: Direction of head and trunk — e.g. when user turns away to talk to someone else, and position of hands in body space (hand position relative to trunk and head).

 
 

Hand gestures: Deictic, iconic, emblematic, pantomimic, beat.

 
 

Turn-taking signals: Various feature-based analysis of combinations of co-occurring multimodal events, such as intonation, gaze, hand position and head direction.

 
     






10

"Massively Interactive"

 
 

Real-time: Timely, Interruptable, Sensitive to outside stimuli

 
 

Relative to human real-time

 
 

- Turn-taking

 
 

- Full perception-action loop

 
     







11

Artificial Intelligence

 
 

From Handbook of A.I.: "Artificial Intelligence is a field of science and engineering concerned with the computational understanding of what is commonly called intelligent behavior, and with the creation of artifacts that exhibit such behavior."

 
 

Minsky: Typically, 'Intelligence' is everything that people can do and machines cannot do, at any point in time

 
 

Alpha-Beta, A* - not A.I.!

 
 

Future of A.I.: Large, "ecological"-like systems with multiple knowledge bases, representations, planning systems, perception schemes, motor control schemes, and communication means.

 
     





PART 2 ->


2002©K.R.Thórisson