Humanoids '09 workshop

From Wiki for iCub and Friends
Revision as of 12:56, 3 December 2009 by Robotics (talk | contribs)
(diff) ← Older revision | Latest revision (diff) | Newer revision → (diff)
Jump to navigation Jump to search


the iCub
shaking hands

Developmental Psychology Contributions to Cooperative Human Robot Interaction

A Workshop of the The 9th IEEE-RAS International Conference on Humanoid Robots http://humanoids2009.org/

Objectives and Topics

The Human ability to cooperate defines one of the ultimate goals for robot cognition. Human cognitive development lays out a well documented series of milestones for achieving that goal. The objective of the workshop is to provide concrete illustrations of how data from human developmental psychology has/can influenced robot cognitive systems to foster cooperation with humans. A corresponding objective is to begin to bring these two research communities together in a cooperative dialog. Presentations should ideally include concrete cases of developmental psychology or human cognitive science being applied to (or posing well characterized challenges to) robotics.

Joint Workshop with Workshop 4

Our workshop (Workshop 1) and Workshop 4 share common interests, and we thus decided to have a common component in the morning with talks from both workshops. In the afternoon the two groups will break-up into their specific sessionns. We trust that this integrative effort will have a strong synergistic effect!

Joint Workshop Schedule

Introduction to Joint Workshop

9:00 – 9:15 Introduction and Motivation for joint iTalk-CHRIS Cooperation – Peter Ford Dominey (INSERM)

CHRIS Contributions

Steps Towards Unified Concepts and Empirical Tests for the Study of Cooperation in Human Children and Humanoid Robots

9:15-9:45 Introduction and Human Developmental Studies – Felix Warneken* (Harvard, MPG)

9:45–10:15 Robotic Implementation of Shared Planning Using Vision and Language– Stephane Lallee*, Carol Madden, Michel Hoen, Peter Dominey (INSERM)

Human Perspective taking as a model for solving ambiguities in HRI

10:15-10:45 Katharina Hamann*,Jasmin Steinwender (MPG), Raquel Ros Espinoza*, Akin Sisbot (LAAS)

Break

iTalk Contributions

Distributional Information in Grounded Language Learning

11:00-11:30 Kerstin Fischer*, University of Southern Denmark

Social Interaction for Acquiring Word Meaning and Concepts of Negation in Robots

11:30-12:00 Joe Saunders, Frank Förster, Chrystopher L. Nehaniv

The Integration of Action and Language in Cognitive Robots

12:00 – 12:30 Angelo Cangelosi (University of Plymouth) Lunch

Afternoon – CHRIS Workshop Program

Invited Talks, and Work:

The afternoon session will include two invited talks (45 minutes), and then a focused discussion concerning (1) what are the developmental cognitive phenomena that we are currently not addressing in cognitive robotics, and (2) what is the roadmap for addressing these issues? Please feel free to send me alternate topics of discussion.

Afternoon Schedule:

Tuning and signaling: lessons for human and robot learning from the Speechome Project.

14:30-15:15 Brandon Roy (MIT)

Developmental issues in lifelong assistive robotics

15:15-16:00 Yiannis Demiris (Imperial College London)

Discussion

16:00-17:00 All participants

17:00 - End of Workshop


Abstracts for Invited talks

Tuning and signaling: lessons for human and robot learning from the Speechome Project. Brandon Roy (MIT) Naturalistic longitudinal recordings of child development promise to offer new opportunities and perspectives for researchers in cognitive science and robotics alike. In this talk, I present our work exploring early language development by “stepping into the shoes” of a child as part of the Human Speechome Project. Toward this end, the home of a young child was outfitted with a custom audio-video recording system to unobtrusively capture roughly 80% of the child’s waking life, from birth to age three. With a corpus of more than 230,000 hours of audio-video recordings, we have developed new tools and methods of analysis. Our current work focuses on the relation between caregiver speech and early word learning, showing the child to be sensitive to both frequency and prosodic cues in the input. Similarly, the caregivers show a surprising sensitivity to the child’s linguistic abilities and tune their behavior accordingly. Preliminary observations of gestural, situational and other signals further suggest that the caregiver-child dyad is a well-tuned system, enabling communication and creating learning opportunities. We believe the natural structure of the child’s learning environment should prove fertile ground for developmental robotics. However, in online learning a key challenge for robotics will be engaging with the human in a similar interactive, tuned-in manner. I will present an overview of the Speechome Project, our techniques for large-scale annotation and analysis of audio and video, and some initial results.

Developmental issues in lifelong assistive robotics Yiannis Demiris (Imperial College London)

Children and adults with sensorimotor disabilities can significantly increase their autonomy through the use of assistive robots. As the field progresses from short-term, task-specific solutions to long-term, adaptive ones, new challenges are emerging. In this talk a lifelong methodological approach is presented, that attempts to balance the immediate context-specific needs of the user, with the long-term effects that the robot¹s assistance can potentially have on the user’s developmental trajectory.


Intended Audience

This workshop should be of interest to at least two groups of people:

  • Those working in robot cognitive systems, as well as robot planning and HRI.
  • Cognitive scientists and developmental psychologists to see how work from their field can interact with and be tested by robot implementation.


Organizers