Loading presentation...

Present Remotely

Send the link below via email or IM


Present to your audience

Start remote presentation

  • Invited audience members will follow you as you navigate and present
  • People invited to a presentation do not need a Prezi account
  • This link expires 10 minutes after you close the presentation
  • A maximum of 30 users can follow your presentation
  • Learn more about this feature in our knowledge base article

Do you really want to delete this prezi?

Neither you, nor the coeditors you shared it with will be able to recover it again.



No description

Michael Vo

on 28 April 2010

Comments (0)

Please log in to add your comment.

Report abuse

Transcript of Cogniron

Cogniron Project Presented by Michael Vo Cogniron Overview "Cognitive robot companion"
EU Integrated Project
Real life Rosie
Funding active 2002-2006
What do people want? Assistant
Friend Video of Objective "What role should a future
robot companion have?" Do people even want
a robot companion? Do you like the idea of
a robot companion? Summary of other results human-like communication important
human-like appearance/behavior not as much
predictability highly important
Multimodal dialogues different modes (modalities) of communication
speech always important but ...
Gaze, gesture, expression, as well as haptic info
lots of info because of shared environment Two main strands First: nonverbal
modelling human behavior
social interaction
Second: conversational
mostly speech
evaluated by performance BIRON: Bielefeld Robot Companion Main focus: tie two strands
At time, only speech
Can track gazes, focus on objects, store multimodal
Now has face, but didn't for study
Video of home tour Experimental setup Home tour: person bought new
robot, now showing it around

naive user
14 subjects age 20 to 37
Average >7 minute of interaction
Asked to fill questionnaire, but results based on third party observations "Meta-Commentators" Biron can rate interaction and give feedback
"I know it is sometimes difficult with me, but don't feel discouraged!"
Effective for most, but some would get mad
apology deadlock

Meta-commentators Problem:reciprocity, obtrusiveness

Solution: non-verbal
no strong obligations to reply
make face more "sorrowful"as interaction quality gets worse Quiet Speakers Biron doesn't always know if user is there or willing to listen
PTA (Person tracking and attention system)
distance, posture, gaze, loudness, reception
male voices Video on tracking Quiet Speakers Female subject with quiet voice
Biron didn't recognize her, so she got confused and moved away
Lots of internal state changes, but no external sign Quiet Speakers Two solutions
report all state changes
constant routine output

Both impractical with speech, but not with non-verbal Grounding grounding: agents try to reach common ground, or a mutual understanding about situation

Presentation: initiate something
Acceptance: show understanding

Presentor keeps going until other agent accepts Multimodal Grounding Pre-conditions of interaction
visual access
motivation to talk
shared environment

Mutual understanding also means understanding interactional information (raising hand asks for turn)
Mutual Understanding I raise my hand while B is talking (P)

If B gives me my turn, common ground established (A)

Otherwise, I have to be more forceful til he figures it out (more P)

Interactional Unit Two Layers
Motivation Layer (intentional or not)
Behavior Layer (verbal, non-verbal) Types of non-verbal Reinforcement
Adding-information Implementation Exchange: Pair of Presentation and Acceptance

Exchange only grounded when both appear

Stack of ungrounded exchanges

Pop when grounded Implementation on BIRON Already implemented at study, but no ability to respond nonverbally

Meta-commentators: shows a sad face, which has implicit acceptance

Weakness: not great with non-verbal behaviors with no clear "meaning" Social Behavior and Embodied Interaction Need to interact socially and appropriately to humans
Personal Space
Uncanny Valley First observed by Masahiro Mori in 1970
More human -> better emotional response
... until a certain point, at which point there is a huge drop
Then it comes back up again
MacDorman 2005 MacDorman's Study Measured reactions to these images by asking questions about seemingly unrelated things

At end, finally came out and asked COGNIRON Chief Cook Jido Biron People bot
Full transcript