Adaptive Locomotion
in Biological Systems Ð How
it can be Modeled and Understood to
build Legged Robots
Department of Biology
and Neuroscience and Cognitive Science and
Institute of Systems
Research
University of
Maryland
Locomotion in
all its various forms is an oscillatory phenomenon. The limbs swing
periodically, or the body moves periodically if limbless. Any form of
locomotion is a function of 1) the mechanics of the body, plus the environment,
2) the neural control of the muscles, and 3) the sensory feedback from the body
and the environment. All are necessary for fully functional and adaptive
locomotion.
The
spinal circuits for locomotion are distributed along the spinal cord. Each
segment has its own oscillatory neural circuit. Successful locomotion requires
that the segmental oscillators be coordinated properly. For example, the
coupling must produce a constant relative phase among the segments virtually
regardless of the frequency of the oscillation. Over the past 20 years, my
colleagues and I have been modeling the coupling among the segmental
oscillators. The original model sets forth a simple periodic sinusoidal
coupling function of the phase difference among the oscillators. We now have
evidence that this model is insufficient for the spinal control of swimming in
lampreys, primitive vertebrates that we study experimentally and that inspired
the original modeling.
This
talk will describe the mathematical models weÕve developed, including the new
data. We will also examine how the use of some known biological principles of
control for locomotion, including the neural oscillatory circuits, can be used
to build and control legged robots.
3:30 Ð 4:30 p.m.