Passive Parameters of the Human Ankle in Downhill Walking

1
Passive Parameters of the Human Ankle in
Downhill Walking

• Jonathan K. Holm,
• Sang Wook Lee, and John Jang

The University of Illinois at Urbana-Champaign
Midwest Graduate Student Biomechanics Symposium
April 1, 2006
2
Biomimetic Ankles for Passive Robots
Unactuated passive-dynamic robots can walk down
sloped surfaces powered only by gravity. To
date, ankle designs for passive-dynamic robots
have been ad hoc or inflexible.
McGeer, Yamakita, Spong, Collins, Wisse, Ruina.
Kuo
Dynamics of human ankle during level walking
reveals hysteresis loops showing no net loss of
energy through the ankle for level walking at
self-selected speed.
These characteristics imply that the human ankle
joint could be effectively replaced with a
rotational spring and damper for slow to normal
walking speeds.
Hansen et al.
3
Goals of the Study
Determine whether the ankle joint can be
effectively modeled as a spring and damper during
slope walking. Determine the spring and damper
coefficients and how these parameters change as
slope angle varies.
?
4
Experimental Methods
Ten healthy male subjects. (ages 19-27, mean
mass 74.56.0kg, mean height 1734.3cm) Self-sele
cted walking speed on 3m ramp. Two successful
trials on each of seven angles (0, 2, 3, 4,
5, 6, 8). Six VICON cameras, embedded AMTI
forceplate.
5
Spring-Damper Ankle Model
Model for ankle torque
6
Fitting Model to Ankle Torque
5 slope, trial 2
5 slope, trial 1
HC
TO
TO
HC
7
Model RMS Error
With the exception of outliers, RMS error
remained low (10Nm) across subjects and
angles. Conclusions Spring and damper
effectively model the behavior of the human ankle
joint during the stance phase of downhill
walking. Best fit occurs during ankle
dorsiflexion model breaks down during impact and
push-off (plantarflexion).
8
Model Coefficients
constant torque offset
Two-Way Analysis of Variance
Difference between subjects and slope angle have
significant effect on this coefficient.
9
Model Coefficients
Two-Way Analysis of Variance
Difference between subjects and slope angle have
significant effect on this coefficient. Spring
coefficient decreases as ramp angle increases.
10
Model Coefficients
damper coefficient
kd ltlt kp
damping is negligible
11
Conclusions
Spring-damper model is effective during stance
phase dorsiflexion, but breaks down during stance
phase plantarflexion (impact and
push-off) Spring behavior only (negligible
damping) at self-selected speed. Spring
coefficient decreases as ramp angle increases.
12
Acknowledgements and References
Thanks to the graduate students who also
contributed to this study Jonas Contakos, Eric
Dudley, Timothy Filipiak, Kelly McHugh, Arun
Ramachandran, David Lim. Special thanks to the
faculty of UIUCs interdisciplinary locomotion
group Karl S. Rosengren, Elizabeth T.
Hsiao-Wecksler, John D. Polk. Funding provided
by the Departments of General and Mechanical
Industrial Engineering at UIUC.
Bauby CE and Kuo AD. J Biomechanics, 2000
331433-1440. Collins S, Ruina A, Tedrake R, and
Wisse M. Science, 2005, 3071082-1085. Hansen
AH, Childress DS, Miff SC, Gard SA, and Mesplay
KP. J Biomechanics, 2004, 371467-1474.
McGeer T. Int J Robotics Research, 1990,
9(2)62-82. Spong MW and Bhatia G. Int Conf
Intelligent Robots Systems, 2003. Yamakita M
and Asano F. Advanced Robotics, 2001,
15(2)139-168.
13
(No Transcript)
14
Model RMS Error
Two-Way Analysis of Variance
ANOVA Difference between subjects and slope
angle have significant effect. Subject
difference has greater significance.
15
Peak Ankle Power During Propulsion
Two-Way Analysis of Variance
ANOVA Difference between subjects has
significant effect on the peak power. Slope
angle has no significant effect on the peak power.
16
Ankle Propulsion Energy
Two-Way Analysis of Variance
ANOVA Difference between subjects has
significant effect on the propulsion
energy. Slope angle has no significant effect on
the propulsion energy.