Seminar

2012.05.21 Qin Zhang (TUAT, GV Lab, Tokyo, Japan)
Evoked EMG-based torque prediction for muscle fatigue tracking and closed-loop torque control in functional electrical stimulation
13:30, Blg 9 room 505

Abstract: Functional electrical stimulation (FES) is a potential technique to provide active improvement to spinal cord injured (SCI) patients in terms of mobility, stability and side-effect prevention. FES-elicited muscle force is required to be appropriate and persistent to perform intended movement or maintain a posture balance. However, muscle state changes such as muscle fatigue degrade the performance of FES. In addition, most of complete SCI patients don’t have sensory feedback to detect the fatigue and in-vivo joint torque sensor is not available yet. Conventional FES control systems are either in open-loop or not robust to muscle state changes. Therefore, the development of joint torque prediction and feedback control method is important to enhance the joint torque control of FES in terms of accuracy, robustness, and safety to the patients.

Biography: Qin Zhang (張琴)is now a research fellow at TUAT awarded by JSPS. She received the M.S. degree in graduate school of engineering at Huazhong University of Science and Technology in 2003, and the PhD degree in automatic and microelectronic systems at Montpellier University II in 2011. She was a lecturer at Wuhan Institute of Technology 2003-2008. Her research interests include biosignal processing and application, rehabilitation technology.


2012.02.17 Kanako Miura (AIST, Humanoid Research group, Tsukuba, Japan)
HRP-4C: a humanoid with appearance similar to humans, and generating its motions
11:00, Blg 9 room 505

Abstract:
It is so hard to figure out the practical application of current biped humanoid robots. It is probable that one practical application for them is the entertainment industry such as exhibitions and fashion shows, provided the robots can move very realistically like humans. "HRP-4C" was developed for such use, which has the appearance and shape of a human being, can walk and move like one.
In the course, several studies to realize HRP-4C to move like a human (or a woman) will be introduced, including walking, and turning using its toes. Referring human motions obtained by motion capture system, the motion of HRP-4C is generated. However there are difficulties to manage its balance and “human-likeness” simultaneously. So the approaches to solve such problems will also be presented.

Biography:
Kanako Miura(三浦郁奈子) is a researcher at the National Institute of Advanced Industrial Science and Technology (AIST). She received her PhD in information science from Tohoku University in 2004 and in “Electronique, Electrotechnique, Automatique” from Universite Strasbourg 1 (France) in 2004. She was a Pos-doc researcher in the fields of visual servoing in Tohoku University 2004-2005, and she has also worked on a bilateral control system and an augmented reality system in NTT Docomo 2005-2007, then she became a researcher of Humanoid Research Group in AIST. Her research interests include human movement analysis and motion planning for humanoid robots.


2011.11.14 Ritta Baddoura (Universite Paul Valery - Montpellier 3 ? CRISES: Interdisciplinary Research Center in Human and Social sciences, France)
Being familiar with the robot: emotional and mental expressions of familiarity in a human interacting with a robot
15:00, Blg 9 room 505

Abstract:
In this research we are interested in the most basic emotional and mental states experienced by a human sharing a common space ? be it public or private- with a robot. In this perspective, the concept of familiarity can provide us with an interesting and essential approach to the various applications and uses of human/ robot interaction. Even though most of us have already experienced (and are familiar with) being familiar or unfamiliar with a person, object, animal or place; familiarity is still a very poor concept theory wise, and lacks a clear and solid definition. This research project aims to learn more about familiarity and get an insight on the way it’s experienced and recognized by a person: is it identified as an emotional state or/and as a mental one? It is also about understanding which aspects of the robot’s non verbal communication: gait, movement, physical appearance, sound … are perceived as familiar or unfamiliar and how does this perception influence human feelings of security, empathy and -on a more observable level- their readiness to interact and cooperate with the robot. The research will be lead with a quantitative and a qualitative approach.

Biography:
Ritta Baddoura is a psychologist and a writer. She received her Vocational Master degree in clinical & pathological psychology from St-Joseph University, Lebanon in 2006 and her Research Master degree in psychoanalysis & aesthetics from Universite Montpellier 3, France in 2009. She has worked in the fields of child and adolescent psychology, addiction & harm reduction, gender studies, art therapeutic mediation and post traumatic stress disorder. She is currently enrolled in a PHD program in psychology (3rd year) at Montpellier University 3. Her thesis tackles Human Robot Interaction and studies: the human desire involved in building androids, the robot as an ancient and futuristic figure, and the emotional dimension of HRI. Baddoura’s work is rooted in a creative and multidisciplinary approach to research and finds inspiration in her literature and art practice.


2011.09.16 Tomomichi Sugihara (Osaka University, Motor Intelligence Laboratory, Japan)
Dynamics morphing toward robust and autonomous biped control
Time to be defined between 13:00 and 16:00, Blg 9 room 505

Abstract:
"Dynamics morphing" is a paradigm to design an autonomous, namely, non-time-slaved controller for various motor skills which require complicated body manipulation under continuously or discontinuously varying dynamical constraints. Bipedalism is a typical example of such motor skills to locate and locomote the body in the world with many uncertainties in environments and events. In this talk, a unique controller which unifies biped standing, stepping and steady walking even under disturbances is introduced. A dynamically morphing property between regulatory and oscillatory behaviors enables general bipedalism without any help of pre-planned motion trajectory.

Biography:
Tomomichi Sugihara is an associate professor at Department of Adaptive Machine Systems, Graduate school of engineering, Osaka University since 2010. He received PhD from University of Tokyo in 2004. He was a research associate in UT from 2004 to 2005 and became an assistant professor. He moved to Kyushu University as a program associate professor in 2007. His research interests include kinematics, dynamics computation, motion planning, control, hardware design and software development of anthropomorphic robots. He is a principal investigator of Motor Intelligence Laboratory.


2011.07.19 Mitsuhiro Hayashibe (INRIA, DEMAR project Team, Montpellier, France)
Human sensory motor system and rehabilitation robotics
9:45-11:15 L0016

Abstract:
In this course, we overview the basics of human sensory motor system. How robotics can be applied for functional rehabilitation is discussed along with some examples of ongoing research.
This seminar is part of the master course on "multibody dynamics".

Biography:
Mitsuhiro Hayashibe received the B.S. degree in mechano-aerospace engineering from Tokyo Institute of Technology in 1999. M.S. and Ph.D. degrees from University of Tokyo, graduate school of engineering in 2001 and 2005 respectively. He was an assistant professor at Jikei University School of Medicine, Department of Medicine, Research Center for Medical Sciences from 2001 to 2006, and a postdoctoral fellow at INRIA Sophia Antipolis and LIRMM from 2007. Since 2008, he has been a research scientist with INRIA and LIRMM, Computational Medicine and Neurosciences, DEMAR project. His research interests include modeling and identification of neuromuscular dynamics and biomechanics. He received Best Paper Award from Journal of Japanese Society for Computer-aided Surgery and CAS Young Investigator Award, Gold Prize from Hitachi Medical Systems. He is a member of IEEE Engineering in Medicine and Biology Society, and french side leader of Japan-France Integrated Action Program AYAME supported by JSPS and INRIA (2010-2012)


2011.04.25 Yuka Ariki (AIST - Digital Human Research Center, Japan)
Acquisition of common expressions between human and robots to solve problems in imitation learning
15:00-16:00 Blg 9 room 505

Abstract:
For humanoid robots with many degrees of freedom, a considerable amount of time is required to prepare multiple motions in advance since the number of combinations of joint angle trajectories is quite large. Imitation learning is considered as a suitable approach to initialize parameters in the vast search space. However, direct use of the instructor’s motion trajectories often fails because of the difference of physical properties between the instructor and the robot. For example, a humanoid robot can falls over or hits own body by own hand if the robot directly copy the corresponding joint trajectories of an instructor’s behavior. In this talk, I introduce our approaches to deal with two major difficult problems involved in the imitation learning paradigm. 1) kinematic properties of a demonstrator and an imitator are different, 2) Dynamical characteristics of a demonstrator and an imitator are different. For the first problem, we first find a shared low-dimensional latent space between demonstrator's and imitator's postures.Then, as a result, the imitators will be able to acquire the corresponding demonstrator's movements from the shared low-dimensional latent space. For the second problem, we estimate ground reaction force from captured demonstrator’s movements so that an imitator can generate physically consistent imitated behaviors.

Biography:
Yuka Ariki (有木 由香) received her M.E degree and Ph.D. degree in information science from Nara Institute of Science and Technology, Japan in 2007 and 2010 respectively under Prof. Mitsuo Kawato. From 2009 to 2010, She worked under Katsu Yamane, Jessica Hodgins as an intern student at the Disney Research Pittsburgh. She is currently a Post-doctoral Fellow at the National Institute of Advanced Industrial Science and Technology (AIST) Digital Human Research Center.


2011.03.28 Emel Demircan (Stanford University, USA)
Reconstruction and Analysis of Dynamic Motions in Human Musculoskeletal Systems
Due to the situation in Japan the talk has been cancelled and will be rescheduled in Automn

Abstract:
The synthesis of human motion is a complex procedure that involves accurate modeling and simulation of musculoskeletal kinematics, dynamics and actuation, and characterization of reliable performance criteria. Motivated by the previous robotics research applied to these problems, the present research aims to leverage the task-level control strategies to reconstruct and synthesize human dynamic motions. For this purpose, our existing robotic tools were applied to reconstruct and analyze human dynamic skills, introducing the approaches of (i) human motion reconstruction by direct control marker trajectories, (ii) task-driven muscular effort minimization criterion and (iii) characterization of human dynamic skills.
Dynamic motion reconstruction is achieved through the operational space-control of a simulated human musculoskeletal model to track the captured marker trajectories in real-time. The reconstruction of human dynamic motion is performed by successive projections into the null spaces of all tasks that are above it in the hierarchy formed in marker space. The task-level control method is an efficient way to map motion patterns to accurate musculoskeletal models and provides real-time motion dynamics. Additionally, task-driven human performance criteria, including the muscular effort expenditure and the feasible set of operational space accelerations are developed and implemented to characterize various athletic skills based on the musculoskeletal physiology.
The methodology followed incorporates extensive motion capture experiments on elite athletes and martial art players, musculoskeletal modeling, real-time simulations and task-oriented control techniques. This robotics-based approach of exploring natural human skills provides generalized motion analysis techniques and physiologically accurate performance predictions applicable in the domains of sport medicine, rehabilitation and physical therapy.

Biography:
Emel Demircan (デミルカン　エメル) is a Mechanical Engineering PhD student in Professor Oussama Khatib's robotics group within Artificial Intelligence Laboratory. Her research focuses on the application of dynamics and control theory for the simulation and analysis of biomechanical and robotic systems. She is broadly interested in human motion analysis to investigate high performance skills for athletic training and performance improvement and rehabilitation of patients with impaired sensory-motor function. Emel received her MS('07) in Mechanical Engineering from Stanford University and BS('06) in Mechanical Engineering and Industrial Engineering from Bogazici University.


2010.10.14 Prof. Dana Kulic (Waterloo University, Adaptative Systems Laboratory, Canada)
Human Movement Observation and Analysis for enabling Humanoid Robot Learning
17:00-18:00 Blg 9 room 505

Abstract:
As robots move to human environments, the ability to learn and imitate from observing human behaviour will become important. The talk will focus on our recent work on designing humanoid robots capable of continuous, on-line learning through observation of human movement. Learning behaviour and motion primitives from observation is a key skill for humanoid robots, enabling humanoids to take advantage of their similar body structure to humans. First, approaches for designing the appropriate motion representation and abstraction will be discussed. Next, an approach for on-line, incremental learning of whole body motion primitives and primitive sequencing from observation of human motion will be described. The second half of the talk will overview recent work on learning controllers for robot movement and imitation, and accurate human movement observation from portable sensors. The talk will conclude with an overview of preliminary experimental results and a discussion of future research directions.

Biography:
Dana Kulic (クリチ　ダナ) received the combined B. A. Sc. and M. Eng. degree in electro-mechanical engineering, and the Ph. D. degree in mechanical engineering from the University of British Columbia, Canada, in 1998 and 2005, respectively. From 2002 to 2006, Dr. Kulic worked with Dr. Elizabeth Croft as a Ph. D. student and a post-doctoral researcher at the CARIS Lab at the University of British Columbia, developing human-robot interaction strategies to quantify and maximize safety during the interaction. From 2006 to 2009, Dr. Kuliｃ was a JSPS Post-doctoral Fellow and a Project Assistant Professor at the Nakamura-Yamane Laboratory at the University of Tokyo, Japan, working on algorithms for incremental learning of human motion patterns for humanoid robots. Dr. Kulic is currently an Assistant Professor at the Electrical and Computer Engineering Department at the University of Waterloo, Canada. Her research interests include robot learning, humanoid robots, human-robot interaction and mechatronics.


2010.07.22 Dr. Claire Dune (Joint Robotics Laboratory (AIST/CNRS), Tsukuba)
Visual servoing for robot walk control
14:45-16:15 Lecture Blg room L0015

Abstract:
In this talk, I will present a visual servoing scheme to control the walk of a humanoid robot using the on board camera. Though most of the existing methods use an perception-plannification-execution scheme, we propose to close the control loop to be robust to error in the model (such as the sliding of the feet on the ground, the compliance of the ankles and neck, calibration error,etc..). The basic idea is to control the robot motion so that the current image features match some desired image features. First I will present the specificities of our control of the HRP2 walking motion. And we will see that the use of the information extracted from the on board camera is challenging due to the peculiar motion induces by this walk. Then I will present our visual control that is based on a 3D visual servoing control scheme. After a short introduction of visual servoing, I will briefly introduce an overview of generalised inverse kinematics to handle several tasks at the same time : for example looking for an object while walking towards it. Finally I will conclude on the difficulties that still need to be solved.

Biography:
Claire Dune (デゥヌ　クレル)In september 2005, I graduated from the Ecole Nationale Superieure de Physique de Strasbourg with major in electrical engineering and computer vision. I also received a master degree in optics, image processing, and robotics from the Universite Louis Pasteur, Strasbourg. Then I received a grant from the CEA and the Brittany council to do PhD on a vision based grasping tool for the disabled with Eric Marchand. This project was a collaboration between the INRIA- Rennes Bretagne Atlantique and the CEA-LIST. My PhD defense took place in April 2009 at the Universite de Rennes 1. Since july 2009 I have been research associate at the JRL lab in Tsukuba with a 1 year fellowship of the Japan Society for Promotion of science.


2009.07.21 Dr. Carson Reynolds (University of Tokyo, Ishikawa laboratory)
Meta-perception: bodily interfaces
11:00-12:00 Blg 9 room 505

Abstract:
Sensors, actuators, wearable computers and wearable robots can do more than simply observe our bodies; these devices can alter and manipulate our perceptions. The Haptic Radar and other research projects are presented as examples of devices that act on phenomena related to the process of perception. Moreover, robotic and sensor systems which move reflexively are discussed.

Biography:
Carson Reynolds （レノーヅ　カーソン）is a project assistant professor in the Department of Creative Informatics of the University of Tokyo. He is co-founder of the Meta-Perception research group which investigates methods for capturing and manipulating information that is normally inaccessible to humans and machines. His work has been discussed (online) in Wired, Make, Engadget, New Scientist, Smart Mobs, and Slashdot (in print) in Boston Globe, Washington Times, the German weekly Focus and (broadcast) on National Public Radio in the US, The Discovery Channel’s Daily Planet and Nippon TV World’s Best Lectures. He holds a Doctor of Philosophy and Master of Science from the Massachusetts Institute of Technology; his research work there was performed at the Media Laboratory in the Affective Computing Group.