Project Description
The relationship between humanoid robotics and human-figure animation
is synergistic, as each discipline provides tools and techniques of
use to the other. Some of these have been mentioned
previously. This work has demonstrated that fundamental motion exemplars for a
robot can be generated using a gain-scheduled variant of the scattered
data interpolation algorithms used for motion
synthesis. Viewing robots as primitive
learners, our results show that a robot can learn to interact
purposefully with its environment through a developmental acquisition
of sensory-motor coordination. Part of this effort has been to extend
the Sensory Ego-Sphere (SES) for sensor fusion.
Additionally,
we are employing manifold learning strategies
to uncover the
robot's underlying sensorimotor state structures and thus to develop better
learning and control
strategies. We have conducted
several sets of experiments on Robonaut, NASA's humanoid robot, to
validate our approach.
.