![[PNG] icub-macsi-head](view.php/icub-macsi-head2.png "[PNG] icub-macsi-head")
OPTIMALITY PRINCIPLES AND ADAPTATION IN HUMANOID ROBOTIC CONTROL
IROS 2012 WORKSHOP
ORGANIZERS
- Serena Ivaldi *, ISIR - UPMC, France
- Bastien Berret °, RBCS - IIT, Italy
- Francesco Nori °, RBCS - IIT, Italy
- Olivier Sigaud *, ISIR - UPMC, France
* Institut des Systèmes Intelligents et de Robotique, Université Pierre et Marie Curie. Pyramide – Tour 55 - Boîte courrier 173. 4 Place Jussieu, 75252 PARIS cedex 05 – France
° Department of Robotics Brain and Cognitive Sciences, Istituto Italiano di Tecnologia. Via Morego 30, 16163, GENOVA - Italy
WHEN & WHERE
This full-day workshop is held during IROS 2012 in Vilamoura, Algarve (Portugal) on October 7th 2012, room Fenix 3.
CONTENT
The availability of new mechatronic platforms such as humanoid robots has driven robotics control research to new paradigms where optimal control of movements and adaptation to perturbations or unforeseen situations are becoming the key challenges. In these new lines of research, inspiration from human motor control and learning capability is playing a central role. But the corresponding frameworks come with very large computational costs so a lot of effort is required in designing practical methods that can address meaningful control and learning problems up to the whole body control of humanoids.
OBJECTIVES
The goal of this workshop is to get an accurate picture of the current state of the art and discuss the hot questions of the application of optimal control and adaptation models of human motor control to robotics:
• The inverse optimal control being ill-posed, how can we determine the cost function underlying human movements?
• Do we truly behave optimally in some sense?
• Shall we address whole body control with standard optimal control approaches, or are there some higher level organization principles to be considered?
• What are the next challenges to be addressed?
TOPICS
• Optimal control methods and their derivatives (OFC, SOC, SOFC...)
• Reinforcement learning methods for robotics
• Practical implementations of the above methods in robots
• Humans to humanoids transfer
• Models of human movements based on optimality : successes and limitations
INTENDED AUDIENCE
The workshop will bring together roboticists, optimal control and machine learning experts interested in the use of these methods in robots, as well as computational neurosciences researchers interested in modeling these processes in animals and humans, so as to generate a fruitful discussion between both communities.
CONFIRMED SPEAKERS
- Jan Babic (Josef Stefan Institute, Ljubjana, Slovenia)
- Mike Mistry (University of Birmingham, UK)
- Etienne Burdet (Imperial College, London, UK)
- Martin Butz (University of Tübingen, Germany)
- Jean-Paul Laumond (LAAS-CNRS, Toulouse, France)
- Katja Mombaur (Ruprecht-Karls-Universität Heidelberg, Germany)
- Alexander Terekhov (University Pierre et Marie Curie, Paris, France)
- Jan Peters (Technische Universitaet Darmstadt, Germany)
- Stephen Scott (Queen's University, Kingston, Canada)
- David Franklin (University of Cambridge, UK)
- David Braun (University of Edinburgh, UK)
- Evangelos Theodorou (University of Southern California, USA)
PROGRAM
| Time | Speaker | Title |
| 08.30 | Olivier Sigaud | Introduction |
| 08.45 | Stephen Scott | Putting Sensory back into Voluntary Motor Control |
| 09.15 | David Franklin | Adaptive feedforward and feedback strategies in sensorimotor control |
| 09.45 | Etienne Burdet | A theory of interactive motor behaviors |
| 10.15 | poster teasers |
|
| 10.30 | coffee break | |
| 11.00 | Alexander Terekhov | Can we learn what the brain optimizes? |
| 11.30 | Evangelous Theodorou | Robustness of optimal control predictions to robotic and neuromuscular control |
| 12.00 | Jan Babic | Robot skill synthesis through human sensorimotor learning |
| 12.30 | lunch break |
|
| 14.00 | Katja Mombaur | Identifying optimality principles in human locomotion for humanoid robots |
| 14.30 | Jean-Paul Laumond | The geometric dimensions of action |
| 15.00 | Mike Mistry | Exploiting Redundancy to Optimize the Task Space |
| 15.30 | coffee break |
|
| 16.00 | David Braun |
Optimality in Compliant Actuation: Theory and Practice |
| 16.30 | Jan Peters | Machine Learning of Motor Skills for Robotics |
| 17.00 | Martin Butz | How Do We Interact With Objects: Optimal Control or Weighted Integration of Multiple Biases? |
| 17.30 | Serena Ivaldi, Bastien Berret, Francesco Nori, Olivier Sigaud | Conclusions |
Download the final program and the contributed papers here.
SOCIAL LUNCH
We are organizing a social lunch together in a restaurant close to the hotel on Vilamoura Marina. If you are interested in participating to the lunch, and continue discussions informally, please send an email to the organizers and sign up to this doodle list !!
Details about the restaurant will follow soon.
CALL FOR CONTRIBUTIONS
The workshop will also feature a session for posters and short presentations.
Prospective participants are required to submit an extended abstracts (4 pages maximum), optionally accompanied by a video. Submissions must be in pdf, following the IEEE conference style (two-column). All contributed workshop presentations are subject to a peer review process.
Authors wishing to give a talk should mention it when sending their abstract: upon available time, selected contributions may have the opportunity to be presented in the oral session.
Other selected contributions will be presented as posters during dedicated sessions.
Important dates:
July 10, 2012: Extended abstracts submission deadline
July 30, 2012: Notification of acceptance
August 31, 2012: Final paper due
October 7, 2012: workshop
Submission:
Please send your contribution via e-mail directly to the Workshop organizers ( h2h_iros12@isir.upmc.fr ) with the tag IROS2012.
ACKNOWLEDGMENTS & REFERENCES
This workshop is organized under the support of the ANR Project MACSI.
Dernière modification le 09/10/2012