Kevin Granata
University of Virginia
David Brogan
University of Virginia
Pradip Sheth
University of Virginia
IV World Congress of Biomechanics, 2002.
Abstract
Within the biomechanist community, there is a rhetorical hypothesis that both movement
trajectory and joint torque are modulated or adapted to maintain dynamic stability in bipedal
walking. Not only are these two control variables intricately related, but such additional
objectives as desired speed, stealth, endurance, etc. inevitably contribute to the complex behavior
in animal locomotion. As a consequence, any single objective function used to describe walking
dynamics is necessarily limited. We propose a bipedal walking control algorithm that
simultaneously solves for movement trajectory and joint torque without relying on any a priori
assumptions regarding one or the other. The absence of such assumptions permits the study of
pathologic movement dysfunctions where the desired movements and torques are unknown. Our
technique uses a constraint-based space-time optimization algorithm to compute optimal
movements and torques. Such pathologic constraints as leg-length discrepancy, range-of-motion
limitations, or velocity constraints from spastic hypertonia may be added and this optimization
technique will find non-homogeneous solutions. When this technique is applied to a control task
with a known optimal solution, two-segment downhill walking, it produces identical results to a
torque-free forward-integration approach. Solutions to pathologic behavior conditions were also
demonstrated by limiting swing leg velocities to simulate the neuro-physiologic constraints of
hamstring spasticity.
Paper
|
|
|