AN EXPLORATION OF VISION-BASED HUMAN-MACHINE INTERFACES

Pietro Perona - California Institute of Technology

If humans could communicate with machines as they communicate with
other humans the world would be a better place. Current human-machine
interfaces (pointing devices, keyboards, knobs and buttons) are
mediocre: they require training, they must be touched, they are bulky
and slow.  Vision and speech engineers now have the opportunity to
provide computer and software designers with better input devices.

I will report on recent research at Caltech on exploring the
opportunities offered by single-camera computer vision for building
better HMIs. We have developed techniques and working systems for
measuring the position and orientation of the head/face of the user,
the 3D pose and motion of the body, as well as handwriting.  We have
also developed a simple interface that allows the user to capture the
shape of 3D objects. Our experiments are encouraging, in that accurate
and useful vision-based HMIs appear to be achievable in the near
future.

Joint work with E. DiBernardo, L. Goncalves, M. Munich, M. Weber,
M. Welling, W. Einhauser.

---
Pietro Perona is Professor of Electrical Engineering at the California
Institute of Technology where he is also a member of the Computation
and Neural Systems program. He received his PhD at UC Berkeley
in 1990. Dr. Perona's research interests are computer vision and human
vision.
---