Broad Area Colloquium For AI-Geometry-Graphics-Robotics-Vision
The Computation of Human Movement
November 28, 2005, 4:15PM
Hewlett (TCSeq) 200
Human movement appears to be as easy as child's play. Yet it has
proved to be astonishingly difficult to reproduce its elegance and
robustness in either computer animation or robotics. Good
computational models of human movement are also critically important
in neuroscience, bioengineering, and medicine.
I will describe ongoing work in my lab towards developing constructive
models of human movement. I will first provide a brief introduction
to the human musculoskeletal system, its neural control, and its
interaction with the physical environment.
I will then describe how natural human movement with contact can be
unobtrusively measured using a technique we call ``interaction
capture.'' We simultaneously measure contact forces and limb
movements at a high rate and estimate task-specific impedances used by
human subjects. This information can be immediately used for computer
animation, and it generalizes the widely used motion capture
retargeting techniques to the important class of movements with
Finally, I will describe a new approach to constructing detailed
physical models of the neuro-musculo-skeletal system, using fiber-like
3D elastic elements we call ``muscle strands.'' The strand model
offers the efficiency of line-based muscle models and more realistic
handling of neural control and mechanical constraints.
(Joint work with Kry, Sueda, and Wei).
About the Speaker
Dinesh K. Pai is a Professor in the Department of Computer Science at
Rutgers University. Previously, he was a Professor at the University
of British Columbia and a fellow of the BC Advanced Systems Institute.
He received his Ph.D. from Cornell University, Ithaca, NY. His
research interests span the areas of computer graphics, robotics,
multisensory integration, and the neural control of movement. See
http://www.cs.rutgers.edu/~dpai for more details.
Back to the Colloquium Page