Towards Robots that Move and Interact Like Humans
Dr. Katsu Yamane
Carnegie Mellon University
April 23, 2009
Troy 2012, 4:00 p.m. to 5:00 p.m.
Refreshments at 3:30 p.m.
In this talk, I will introduce some of my work in humanoid control and
human motion analysis. The goal is to understand human motion control
skills and apply them to humanoid robots to realize realistic behaviors
and interactions with the environment and human.
The first half of the talk deals with synthesizing robot and virtual
character motions from human motion data. I will show several
techniques for adapting captured human motion to robots and environments
that are substantially different from the original. The fundamental
idea is to effectively combine data with dynamics, kinematics and
planning algorithms to synthesize a variety of new motions from
relatively few datasets.
The second half is devoted to our quest towards more thorough
understanding of human motion control through analysis and simulation
using a detailed neuro-musculo-skeletal model with a somatosensory
reflex network. We demonstrate that the parameters identified from
motion capture data are consistent with some of the phenomena known in
biomechanics such as patellar tendon reflex.
Katsu Yamane joined Disney Research, Pittsburgh as a Senior Research
Scientist in October 2008. His research interests are in humanoid robot
control, human motion analysis and simulation, and character animation
synthesis. He received his B.S., M.S. and Ph.D. degrees from University
of Tokyo in 1997, 1999 and 2002 respectively. Before moving to Disney,
he was a faculty member at the University of Tokyo and a postdoctoral
fellow at Carnegie Mellon University. He is a recipient of Best
Transactions Paper Award (2000) and Early Academic Career Award (2004)
of IEEE Robotics and Automation Society, and Young Scientist Award (2005)
of Ministry of Education, Japan.
Hosted by: Dr. Jeff Trinkle (x2510)
Last updated: Feb. 27, 2009