Plenary talks

RO-MAN2009 will organize three plenary talks, scheduled at 9:00-10:00 on September 29, 30 and October 1.

September 29, 2009, 9:10-10:00

• Learning visually guided autonomous interaction
• Prof. Edgar Koerner (President, Honda Research Institute Europe EmbH)

September 30, 2009, 9:00-10:00

• Motivating Human Interaction with Playful Modular Technology
• Prof. Henrik Lund
  (Prof., Technical University of Copenhagen)

October 1, 2009, 9:00-10:00

• Human-Robot Interaction and Subjective Time
• Prof. Yoshihiro Miyake (Prof., Tokyo Institute of Technology)

Abstract

Brains are basically control systems for organizing behavior in natural environment, a problem domain where AI has been least successful up to now. Therefore, understanding essential principles of how the brain achieves autonomous adaptation to a changing world may enable us to provide our technical artifacts with the needed flexibility and robustness we admire in living beings. We argue that in the brain there are no dedicated systems for each cognitive capability, but dynamical systems that can rapidly configure themselves under control of static information previously acquired. The architecture of nervous systems represents the genetically encoded a-priori knowledge about the hierarchical structure of its interaction with the environment. However, natural environments are too complex for static rules, therefore, brains have evolved to learn to predict and react to environmental changes. This control architecture for the organization of incremental learning and for utilization of the acquired information to configure the processing architecture on the fly may be the key to achieve autonomous interaction comparable to living beings. While behavior success is visible at the systems level, the many lower level of systems organization are equally important to understand how the interactive and emergent properties of the subsystems contribute to the successful adaptation to environmental challenges. Therefore, we at HRI-EU simultaneously study multiple levels of brain control architecture and integrate insights from several levels of systems complexity into a coherent systems architecture. Step by step we implemented nested control loops for reflexes, attention modulated behavior, incremental on-line learning of objects from sensory experience, and generating predictions for interaction. This enables ASIMO to learn to recognize objects through the interaction with humans, to learn associations between acoustic and visual objects, as well as to associate sound with behavioral concepts for interaction, learning from imitation, and to demonstrate first steps of prediction driven behavior.

Biography

Edgar Koerner studied control engineering, and received his Dr.-Ing. in the field of biomedical cyber-netics in 1977 and the Dr. sci. techn. (habilitation) in biocybernetics in 1984, both from Ilmenau Institute of Technology. From 1984 to 1987, he joined the Bioholonics Project of JRDC (Tokyo) dealing with brain-like vision systems. Back at the Ilmenau Institute of Technology, he continued research in biological vision and neurofuzzy control systems as an associate professor. In 1988, Dr. Koerner was appointed full professor for biocybernetics and head of the Department of Neurocomputing and Cognitive Systems. In 1992 he moved to Japan to join Honda R&D’s Wako Research Center near Tokyo, focusing on the brain-like intelligence research. In 1997 he started research in computational neuroscience, evolutionary technology, and cognitive robotics at Honda R&D Europe, where he served as an executive vice president and head of the Future Technology Research Division. Since 2003, Dr. Koerner serves as the president of the Honda Research Institute Europe GmbH. Since October 2007, he additionally serves as a co-director of the Research Institute for Cognition and Robotics at the University Bielefeld. His research interest covers brain-like intelligence, with a special focus on self-organization of knowledge representation and autonomous robots.

Abstract

Playware is intelligent hardware and software that creates play and playful experiences for users of all ages. Playware research seeks to understand play dynamics and play forces in order to implement them in play tools. Playware is of course not the only type of products which can create play and motivate users to perform actions, but digital technology contains new and expanded possibilities, e.g. when developed with embodied artificial intelligence. Playware-tools are tools with a behaviour that initiates play force (e.g. a motion, in the case of sensorimotor play) via interaction. This is the basis for the play dynamic to emerge through which the users are brought into a state of playing. Embodied artificial intelligence can be used to design behaviours of the play tools, e.g. by providing means for creating adaptive play tools. The understanding of play dynamics can help guiding this design of behaviours to be used specifically to create playful and motivating tools for a variety of play interactions, well-knowing that there are both similarities and differences in the play dynamics of different users, environments and activities. Using a modular approach inspired by behavior-based robotics gives opportunity to create modular playware that allows any user to create activities in a flexible manner, regardless of the cognitive and physical abilities of the user. Indeed, the modular approach allows a generalization over users, environments and activities as well as a commercial possibility of mass-production for customization. For Human-Robot Interaction, when considering a modular approach, we are often interested in the interactivity and the opportunities for the human interaction, so instead of developing self-reconfigurable modular robotics, we may describe the above systems in terms of user-configurable modular robotics. In this talk, I will show numerous, specific examples of how such an approach of modular playware (in the form of modular interactive tiles and cubes) facilitates generalization over users, environments and activities in the fields of playgrounds, cardiac rehabilitation, stroke rehabilitation, elderly home care, autism therapy, dementia treatment, soccer training, dancing, music concerts, etc., and how the playful approach provides motivation for users to interact with the modular technological solutions in these fields. Videos will feature use in rehab, play, sport, music, and dance.

Biography

Henrik Hautop Lund is head of the Center for Playware at Technical University of Denmark (DTU Elektro). His interdisciplinary research center comprises staff in the fields of engineering, humanities, arts, and music. Professor Henrik Hautop Lund is known world-wide for his work in bringing robotics to use in novel ways. His approach is to combine modular robotics and modern artificial intelligence to create novel solutions to problems that occupy the citizens of the World, e.g. obesity, rehabilitation, and 3rd World development. He has recently founded the Center for Playware to focus even further on how playful aspects of robotics may provide motivation for any citizen to perform different kinds of interaction with the robots of our future daily life. He chaired the Robots at Play festivals in the open city areas where researchers, artists, entertainers, and citizens meet through playful hands-on experience with robotics in the daily life of the citizens. In all cases, Lund has shown how the combination of a modern artificial intelligence, modular robotics and entertainment may provide novel opportunities in play, rehabilitation, sport, music, teaching, third World development, etc., by trying to allow non-expert users easy access to the technology in a playful and motivating way. Professor Henrik Hautop Lund has published more than 125 scientific articles in the field of robotics, he has been a member of the Danish Research Council, and he has been invited to present his robotic work in numerous occasions, for instance for the Emperor of Japan at Akasaka Palace in Tokyo. He founded and headed the LEGO Lab in 1997-2000. He invented the RoboCup Junior robot football game for children, and his Adaptronics group won the RoboCup Humanoids Free Style World Championship 2002. Further, he developed the RoboMusic in collaboration with World Music Award winner, remix musician Funkstar De Luxe. Professor Lund’s work has received world-wide interest from news media such as CNN, WIRED, etc. and he was nominated for the award for the best entertainment robots and systems research at IROS. He has led numerous research projects and projects involving collaboration between academic institutions and private companies. He founded RoboCluster, an industrial network and interest organisation for the robotics industry comprising more than 100 industrial interested parties in Denmark. Henrik Hautop Lund is director of the spin-off company Entertainment Robotics.

Abstract

The expression as a group can be improvised in realtime through human's various cooperative communications, without being limited to the fields of dance and sports. This experience is so familiar that we do not usually give thoughts to its mechanism. However, a hidden interesting problem of subjective time is found when questioning it, "Why?". It is that subjective time, unlike objective time, is not necessarily shared interpersonally in advance. For example, you cannot assume that five subjective seconds for one person corresponds to five subjective seconds for another person. However, human communications seem to have overcome this problem quite easily. Then, the question arises, by what kind of mechanism are subjective areas connected, and does cooperation as a group become possible? We call such subjective time “Ma”, and search for its shared interpersonal mechanism. If we could jump to a conclusion, it is not that it connects detached subjective worlds but that it generates a subjective area together. In other words, it is a co-creation of “Ma”. Necessity of co-creation in communication exists here. In order to examine this mechanism, we have focused on a very simple experiment: a synchronized tapping task. This is a task that involves pressing a button with one's finger in-sync with the rhythmic sound stimulation, such as the metronome periodically presented. See Fig.1. The horizontal axis indicates time. You can see that the time indicated by the dotted line corresponds to the time of stimulation of rhythmic sound. The vertical axis, indicating frequency of time when the button is pressed, is shown as a histogram. At this point, despite the sound stimuli and timing of button-pressing being synchronous as the subject’s subjective time, the timing of pressing the button and the sound stimuli apparently differ objectively. Moreover, motion of the finger occurs before the sound. This means a gap exists in subjective simultaneity and objective simultaneity. Moreover, it indicates that subjective time "Now" is generated to the area of future in an objective sense. We regard it as the generation of "Ma" as subjective time. We are examining the mechanism of generation of such subjective time and the mechanism of how it is shared interpersonally. We are also working on applying it to human-robot interaction. I assume everyone has experienced naturally walking at a pace synchronized with the pace of the person you are walking with. Walking paces synchronize naturally without being conscious of it. While such phenomenon of interpersonally synchronized “Ma” is observed in various phases of life, we focused on walking rehabilitation. It is a walking training where a patient tries to walk in the same pace as the pace of a therapist. We have tried establishing such cooperative walking with simultaneous “Ma” between human and robot. This system is called "Walk-Mate" [1], and it co-creates subjective time “Ma” while a human and a virtual robot mutually adapt their timing of walking. In other words, it is a walking robot that adapts "Ma". We especially aim at the co-creation of walking function recovery in cooperation with a robot as a therapist. Some characteristics of co-creation such as generation of sense of togetherness and stabilization of walking are observed in cooperative walking with synchronized “Ma”. At this symposium, I introduce the potential of humanrobot interaction technology as a co-creation system, as well as recent application to a walking assist robot [2] seen in Fig.2.

References
[1] Miyake, Y., "Interpersonal synchronization of body motion and the Walk-Mate walking support robot," IEEE Transactions on Robotics, 25-3, 638-644 (2009)
[2] http://www.honda.co.jp/ASIMO/assist/rhythm/index.html

Biography

Yoshihiro MIYAKE has strong interest to the relationship between co-creative communication and human’s subjectivity. He received the Ph.D. in 1989 from The University of Tokyo. Since 1996, he has been an associate professor in the dept. of Computational Intelligence and Systems Science at Tokyo Institute of Technology. Since 1999, he is also a guest professor in the Human Science Center at The University of Munich, Germany.