The question of the anthropomorphism is central

The Japan population is declining. And thus aging. This nation is concerned to find the century with such proportion of elderly people that it would more find enough manpower to help them, support them in their end of life. This speech is the official justification for an extensive program of research on the anthropomorphic robots. Robots like us, with two arms, two legs Seen from a distance, it is the impression that the comments on this draft: because it's to help tired human, or even disabled, these robots should look like the us. On-site, most robotics experts believe that this is a shortcut, or even of an oxymoron. The first thing that one should expect a robot to help us, it is that, as each of us an excellent understanding of the human body. Know first "capture" the man, rather than the singer.

"The question of the anthropomorphism is central... because we are anthropomorphistes" summarizes Yoshihiko Nakamura. This researcher from the University of Tokyo, Director of a laboratory which bears his name, is one of the best specialists in the world of the simulation and the interpretation of the human body. "We are equipped to interpret the body of our fellow, guess its efforts, its sensations", he said.

Yoshihiko Nakamura and his team have developed a comprehensive model of the human body and algorithms capable of "reading" the dynamic - gestures, postures, approach-d-' a quidam, on "motion capture" information (motion capture) obtained via a camera. The rapid progress in this area, spurred by the urgent request of Hollywood, led to little time for the revival of the movement by the detection of marks on the body reconstruction 3D without marks, by artificial vision. Voila that allows to consider that a robot could collect, "feel" what a man nothing that by looking at.

Interpret movements

Mathematical models developed by the University of Tokyo researchers know not only interpret movements, categorize it, to name them (the service of the tennis player, Boxer... hook), but also determine what muscles are put to contribution, what tension or torsion are seen in every muscle and joint. Conversely, other software are reproduce these gestures by a robot, to mimic the human gesture. This robot can learn as we all, by example, by reproducing the gesture of the master.

Yoshihiko Nakamura works more generally on the "dialogue of the body", body interaction. He has thus made a kind of virtual combat between a human and a robot, the latter interpreting each stroke of the adversary to him the same parade. The researcher is now mobilized by a futuristic application. It prepares the arrival on the Moon of the first Japanese: a robot controlled from Earth by humans simply to show him what he should do.

The example of R2D2

"When I see someone wearing a heavy object, explains gentian Venture, I already know about how much it weighs, what is its density, which is its centre of gravity. I guess the effort, the gesture to. "This young researcher of the ECA, French joined in 2004 Nakamura lab. Since last year, she has her own laboratory at the Tokyo University of Agriculture and Technology. His research on human-robot interaction are intended to provide the means to the latter to interpret the behavior of the first to help it effectively. It seeks to collect clues to capture emotions in the "language" of the human body. Thus a robot could attend a patient or senior according to its States of soul - fear, anger, boredom... Should one such robot have human appearance to be useful "What is R2D2, in"Star wars", a need of arms and legs to get essential", replied the researcher...

The more humanoid robots that can see can be found at Tsukuba, JRL (Joint Robotics Laboratory), a joint venture between CNRS (including LAAS in Toulouse and the Montpellier LIRMM) and AIST (National Institute of Advanced Industrial Science and Technology). Its Director, the French Abderrahmane Kheddar, assisted by Eiichi Yoshida. Should a line of androids named HRP-1 (directed by Honda) to the JRL to HRP - 4 c. But it is still HRP - 2 (manufactured by Kawada) which serves as a platform for most of the research.

Human-Robot collaboration here is the major topic. "Something as simple as wearing a table with a human is a very delicate task for a robot", admits Eiichi Yoshida. And a subject of research for some time, can we suggest viewing the video showing HRP-2 at work. He does what he can, and it is said that it is infinitely. His arms and legs are illusion, but its perception, coordination are embryonic, compared to what nature with us.

More difficult still, body-to-body. Help for example a senior walk, without the rush, assumes a subtle perception of the body of the robot, and adapted, calibrated reactions... Brief of the algorithms that will require further years of research.

As a robot, must be said, it is above all of the software. A lot of software. And sensors, lot of sensors. Finally engines, scrap metal. This is not wheelchair with mechanical, by passing to humans that they are not, the Androids will travel useful. And well before that artificial creatures in our image from entering our lives, gear wheels and pincers us will no doubt very useful, provided that they know capture our humanity.