Transcript Presentation - Colorado Space Grant Consortium

Biomechanical Modeling and
Analysis of Human Motion
Cole, Joshua
Knapp, Austen
University of Colorado at Colorado Springs, Department of Mechanical
Engineering
Advisor: Prof. Steve Tragesser
The Objective
Create a low cost, non-invasive, portable
system for analysis of the internal forces
and torques required for human motion
Video Capture
Kodak PlaySport
Resolution: 720 X 1280
Frame Rate: 60 fps
Image Processing with
MATLAB
•
•
•
•
•
Manually input initial marker
positions
Filter all colors but green/blue
for each frame
Search based on previous
centroid location
Two colors distinguish
between markers
Output Cartesian Coordinates
Dynamic Modeling
• Modeled body as
series of rigid links
• Convert video
capture output to
angular position
• Two methods of
analysis
Point Mass Analysis
• Compute position, velocity, and acceleration of center of mass
(CoM) of the three – link model
• Calculate horizontal and vertical components of ground
reaction force
– Calculated using Newton’s 2nd Law for CoM
Position Data
Outputted Video
Capture System
Finite Difference
Method
F=ma to attain
GRFx & GRFy
Three – Link Analysis
Application of Newton’s 2nd Law and Euler’s
Law to attain 9 equation’s of motion (EOM’s)
– Three EOM’s per link
• 2 EOM’s through 𝐹 = 𝑚𝑎
• 1 EOM through 𝑀𝐶 = 𝐻𝐶
Three – Link Analysis
Equations of Motion:
Three – Link Analysis
Angular Position,
Velocity, and
Acceleration
EOM’s
Internal Forces
and Torques
Results and Validation
Point Mass Analysis
• GRFy vs. Experimental
Data from Force Plate
• Force Plate =
Validation of dynamic
analysis
• GRFx << GRFy which is the
expected outcome
• Data from 0.0 to ~0.2
seconds is invalid due to
filtering of video capture
data
• Squat performed
after 0.2 sec which
allows invalid data to
be neglected
Results and Validation
Three-Link Analysis:
• GRFy vs. Experimental Data
from Force Plate
• GRFy is less accurate
with this model than
point mass model
• GRFx increases to a large
magnitude which is not
correct outcome
• Computations with threelink model more sensitive
to measurement errors
than point mass analysis
Results and Validation
Three – Link Analysis:
• Figure, top to bottom:
• Calculated horizontal
component of internal
knee force
• Calculated vertical
component of internal
knee force
• Calculated horizontal
component of internal
hip force
• Calculated vertical
component of internal
hip force
• Amplitudes appear to be
within a realistic range
Results and Validation
Three – Link Analysis:
• Ankle, knee, and hip
torques
• Result of subject’s
muscle actuation
• Results qualitatively match
expected torque behavior
• Values not consistent
with results in
literature
Figure 6: Three-Link Calculated Internal Torques
Conclusions and Recommendations
• Partially demonstrated feasibility of system
– GRF more accurate than internal forces and
torques
• Sources of error
– Markers
– Clothing Color
– Approximations
– Frame rate/camera resolution
Future Research and Limitations
• Long-term goal: Analyze broad range of human motion
and impact of such motion
– Possible through calculation of internal forces and torques
– Examples of human motion which could be analyzed:
• Prosthetic patient gait analysis
• Sports performance analysis
• Low-gravity analysis
• Three-dimensional video capture system
• Multi-link model with an arbitrary number of links
• Progress EOM’s to allow analysis where feet leave
contact with ground
Thank you
We appreciate you taking your time to listen
– Questions/comments/concerns?