Analyzing deforming mechanical systems using classical numerical methods
is computationally demanding. The sheer complexity of realistic systems
and their interactions makes low-latency simulation difficult. This
frustrates important human-computer applications such as surgical
simulation and multi-modal interaction using sound or haptic force
feedback.
In this talk, I will describe a variety of sub-linear time algorithms
that use a two-stage approach to compile discrete simulation models into
optimized representations. The first stage "preprocesses away"
system complexity using precomputation, model reduction, compression,
and novel system parameterizations. The second stage performs
interactive computations using sub-linear time algorithms that reason
efficiently about the complete system by exploiting the precomputed
information. I will propose sub-linear time algorithms for a range of
problems including output-sensitive collision detection; subspace
dynamics for large-deformation models; haptic force-feedback rendering;
efficient mesh animation using rotation-sequence clustering; and
precomputed acoustic transfer for output-sensitive sound generation from
geometrically complex vibration sources.
ABOUT THE SPEAKER: Doug L. James has been an Assistant Professor of
Computer Science and Robotics at Carnegie Mellon University since Fall
2002. He received his Ph.D. from the Institute of Applied Mathematics at
the University of British Columbia, Vancouver, Canada, advised by Dinesh
K. Pai. Doug received an NSF Early Career Development Award for his work
on "Precomputing Data-driven Deformable Systems for Multimodal
Interactive Simulation," he was one of Popular Science magazine's
"Brilliant 10" young scientists for 2005, and is an Alfred P. Sloan
research fellow.