Transcript Slide 1

Slide 1 of 41
Musculoskeletal Modeling of
Smilodon Fatalis for Virtual
Functional Performance Testing
Kiran Konakanchi
Advisor: Dr. Venkat Krovi
Mechanical & Aerospace Engineering
State University of New York at Buffalo
Kiran Konakanchi
August 19, 2005
Automation, Robotics and Mechatronics Lab, SUNY at Buffalo
Slide 2 of 41
Agenda
Goal : Study the functional and behavioral performance of extinct &
extant animals
 Our Idea & Introduction
 Goal & Issues
 Literature Review & Available Tools
 Musculoskeletal Modeling in AnyBody
 Case Studies
 Conclusion & Future Work
Kiran Konakanchi
August 19, 2005
Automation, Robotics and Mechatronics Lab, SUNY at Buffalo
Slide 3 of 41
Idea
In our thousands of years of evolution, there are many unanswered
questions
Why do present members of feline family not have saber teeth?
Did the Smilodon use its saber teeth during hunting?
Why ?
Tiger
Smilodon
Musculoskeletal
Biomechanical
Model
Kiran Konakanchi
August 19, 2005
+
+
Engineering
Analysis
Possible
Solution
Introduction Issues Background Modeling Case studies Conclusion
Automation, Robotics and Mechatronics Lab, SUNY at Buffalo
Introduction
Slide 4 of 41
Virtual Prototyping (VP)
“Virtual prototype is a computer simulation of a physical product that can
be presented, analyzed, and tested from concerned product life-cycle
aspects such as design /engineering, manufacturing, service, and
recycling as if on a real physical model. The construction and testing of a
virtual prototype is called Virtual Prototyping.”
Allows the designer to realistically , accurately and quantitatively test
multiple models within virtual environment
Kiran Konakanchi
August 19, 2005
Introduction Issues Background Modeling Case studies Conclusion
Automation, Robotics and Mechatronics Lab, SUNY at Buffalo
Slide 5 of 41
Introduction
Biomechanical modeling
Modeling procedure based on the principles of biomechanics
Four types of models
Conceptual
Model
Analytical
Model
CAD Model
Musculoskeletal
Model
Ye
Xe
L2
2
Input
Output
L1
Black
Box
1
L1 cos(1 )  L2 cos( 2 )  X e
L1 sin(1 )  L2 sin( 2 )  Ye
Complexity of Modeling
Kiran Konakanchi
August 19, 2005
Introduction Issues Background Modeling Case studies Conclusion
Automation, Robotics and Mechatronics Lab, SUNY at Buffalo
Project Goal
Slide 6 of 41
Examine various aspects of systematic musculoskeletal model building
with the help of detailed examples
Explore various issues pertaining to the modeling and analysis of such
systems and provide possible solutions
Case Studies
 Muscle force calculation
 Bite force analysis
 Determination of optimal muscle location points.
Kiran Konakanchi
August 19, 2005
Introduction Issues Background Modeling Case studies Conclusion
Automation, Robotics and Mechatronics Lab, SUNY at Buffalo
Slide 7 of 41
Issues
Redundancy
The number of actuators (muscles Nm) is greater than the number of
degrees (Nd) of freedom of the system
x
F1
m
F2
mg
mx  F1  F 2
Reduction Method: Grouping Muscles until Nm = Nd
Addition Method : Adding Constraints
Dynamically Determinate One Sided Constrained Method (DDOSC):
Problem is divided into series of dynamically determinate problems
Optimization Techniques
Kiran Konakanchi
August 19, 2005
Introduction Issues Background Modeling Case studies Conclusion
Automation, Robotics and Mechatronics Lab, SUNY at Buffalo
Slide 8 of 41
Issues
Geometric Complexity
 Modeling Phase
 Analysis Phase
Coarse Model
Individual Muscle Model
Fine Model
Group Muscle Model
Neural In
MUSCLE
Velocity
or Force
Single Muscle
Kiran Konakanchi
August 19, 2005
Multiple Muscles
Introduction Issues Background Modeling Case studies Conclusion
Automation, Robotics and Mechatronics Lab, SUNY at Buffalo
Background
Slide 9 of 41
Why Bite force?
Bite force: The amount of force that can be exerted by the jaw adductor
musculature and realized at the tooth row as a function of jaw geometry.
Meers et al.(2003) established prey/predatory relationships among
Triceratops horridus, Tyrannosaurus Rex and other dinosaurs.
Verwaijan et al.(2002) found that head and body size have considerable
impact on bite force magnitude.
Kiran Konakanchi
August 19, 2005
Introduction Issues Background Modeling Case studies Conclusion
Automation, Robotics and Mechatronics Lab, SUNY at Buffalo
Available Tools
Slide 10 of 41
SIMM
 Popularly used musculoskeletal modeling software
 Require Bone, Joint, Muscle data in ‘C’ language format
 Requires substantial programming knowledge
SimMechanics
 Good for Mechanical systems
 Can define rigid bodies (bones), joints and drivers
 Difficult to define mathematical muscle models
LifeMod
 Can interact with environment
 Our data format does not suit the software requirements
Kiran Konakanchi
August 19, 2005
Introduction Issues Background Modeling Case studies Conclusion
Automation, Robotics and Mechatronics Lab, SUNY at Buffalo
Available Tools
Visual Nastran
Slide 11 of 41
Vertebrate Analyzer
 Good for mechanical
 Visualize and experiment with accurate
systems simulation
biomechanically constrained models
 Perform motion and stress
 Capability to attach muscles, ligaments,
analysis
tendons etc.
 Issue of muscle recruitment
 Proposed future work could be an ideal
pattern
package for functional performance
 Can not solve the problem
testing
of redundancy
Development of a Computational Toolkit for Biomechanical
Analysis and Simulation : The Vertebrate Analyzer
Kiran Konakanchi
August 19, 2005
Introduction Issues Background Modeling Case studies Conclusion
Automation, Robotics and Mechatronics Lab, SUNY at Buffalo
Musculoskeletal Modeling in AnyBody
Slide 12 of 41
What is AnyBody software? What will it do?
AnyBody is a musculoskeletal modeling software used for developing
detailed multi body biomechanical systems
Applications
 Therapy/Medical rehabilitation
 Ergonomic Design in the fields of
automotives, sports etc.
 Functional performance studies
 Training tools for surgeons when
combined with virtual environment
Kiran Konakanchi
August 19, 2005
Introduction Issues Background Modeling Case studies Conclusion
Automation, Robotics and Mechatronics Lab, SUNY at Buffalo
Software Interface
Kiran Konakanchi
August 19, 2005
Slide 13 of 41
Introduction Issues Background Modeling Case studies Conclusion
Automation, Robotics and Mechatronics Lab, SUNY at Buffalo
Modeling Procedure
Slide 14 of 41
5 stage process
Define Global fixed
Fixedframe
frameand
andcorresponding
correspondingfixed
Fixedpoints
points
Define Segments or rigid elements (like
( likebones
bonesetc.)
etc.)
Define
DefineJoints
various
(Joint
Joints
Constraints)
(Joint Constraints)
& Drivers
Define
Drivers
to move
the Joints
Define
Muscles
(Actuators)
Run the
Define Muscles
andAnalysis
run the Analysis
Kiran Konakanchi
August 19, 2005
Introduction Issues Background Modeling Case studies Conclusion
Automation, Robotics and Mechatronics Lab, SUNY at Buffalo
Slide 15 of 41
Analysis
Different types of Studies / Analysis
Set up Initial Conditions
Kinematic analysis
 Main emphasis is on system motion
 No forces in the system are calculated.
 Obtain position, velocity and acceleration information
Muscle calibration analysis
 Adjusts the lengths of tendons
 Optimal length is in the middle of simulation
 Not required for simple muscle models.
Kiran Konakanchi
August 19, 2005
Introduction Issues Background Modeling Case studies Conclusion
Automation, Robotics and Mechatronics Lab, SUNY at Buffalo
Slide 16 of 41
Analysis
Inverse Dynamic Analysis (IDA)
IDA can be thought as the heart of the software system.
The main requirement for IDA is that, it should be able to cope with
 Statically indeterminate problems
 Limits on forces in the problem
Equilibrium
equations
+
Optimality criteria
involving muscle
forces
Unique
solution
Body attempts to use its muscles in such a fashion that minimum fatigue
condition is obtained.
Kiran Konakanchi
August 19, 2005
Introduction Issues Background Modeling Case studies Conclusion
Automation, Robotics and Mechatronics Lab, SUNY at Buffalo
Slide 17 of 41
Analysis
The muscle recruitment is performed according to the following optimal
criteria
Minimize (Maximum muscle activity )
+ e1*(sum of activities)
+ e2*(sum of squared activities)
Subject to
Equilibrium equations are fulfilled
Muscles are not allowed to push.
Where,
e1: RecruitmentLpPenalty
e2: RecruitmentQPenalty
Weber’s hypothesis: Muscle recruitment is performed in a way such that
muscular effort is minimized during routine activities.
Kiran Konakanchi
August 19, 2005
Introduction Issues Background Modeling Case studies Conclusion
Automation, Robotics and Mechatronics Lab, SUNY at Buffalo
Project Implementation
Slide 18 of 41
CT / MRI Data
MIMICS
CAD Model
(ASCII STL File)
Musculo-skeletal model with approximate
Muscles using “MATLAB” GUI
Inverse Dynamic Analysis to
calculate Muscle forces
Calculate Bite Force
Change the Muscle locations
according to constraints
Kiran Konakanchi
August 19, 2005
Yes
No
Optimality attained?
End
Introduction Issues Background Modeling Case studies Conclusion
Automation, Robotics and Mechatronics Lab, SUNY at Buffalo
CAD Model from Point data
Slide 19 of 41
MIMICS (Materialise’s Interactive Medical Image Control System) is
used to develop a CAD model from CT/MRI data
Kiran Konakanchi
August 19, 2005
Introduction Issues Background Modeling Case studies Conclusion
Automation, Robotics and Mechatronics Lab, SUNY at Buffalo
CAD Model from Point data
Windowing
: Adjusting the grey scale values
Thresholding
: Selection between soft and hard bone
Region growing : Reduce noise and separate structures
Editing
: Remove artifacts
Slide 20 of 41
Mandibl
e
Skull
Kiran Konakanchi
August 19, 2005
Introduction Issues Background Modeling Case studies Conclusion
Automation, Robotics and Mechatronics Lab, SUNY at Buffalo
MATLAB Interface
Slide 21 of 41
 To provide user friendly easy to use interface
 Eliminates the necessity to learn the programming language
 Use other MATLAB features and functions for easy analysis of
results
Kiran Konakanchi
August 19, 2005
Introduction Issues Background Modeling Case studies Conclusion
Automation, Robotics and Mechatronics Lab, SUNY at Buffalo
Global Reference Frame
Slide 22 of 41
Fixed
Points
Inertial
frame
Kiran Konakanchi
August 19, 2005
Introduction Issues Background Modeling Case studies Conclusion
Automation, Robotics and Mechatronics Lab, SUNY at Buffalo
Segment(s) addition
Tiger
Kiran Konakanchi
August 19, 2005
Slide 23 of 41
Sabertooth tiger
Introduction Issues Background Modeling Case studies Conclusion
Automation, Robotics and Mechatronics Lab, SUNY at Buffalo
Slide 24 of 41
Joints
Kiran Konakanchi
August 19, 2005
Introduction Issues Background Modeling Case studies Conclusion
Automation, Robotics and Mechatronics Lab, SUNY at Buffalo
Slide 25 of 41
Drivers
Kiran Konakanchi
August 19, 2005
Introduction Issues Background Modeling Case studies Conclusion
Automation, Robotics and Mechatronics Lab, SUNY at Buffalo
Muscle Models
Kiran Konakanchi
August 19, 2005
Slide 26 of 41
Introduction Issues Background Modeling Case studies Conclusion
Automation, Robotics and Mechatronics Lab, SUNY at Buffalo
Muscle Route types
Tiger
Kiran Konakanchi
August 19, 2005
Slide 27 of 41
Sabertooth tiger
Introduction Issues Background Modeling Case studies Conclusion
Automation, Robotics and Mechatronics Lab, SUNY at Buffalo
Process flow of case studies
Slide 28 of 41
Issue 1
Issue 2
Musculoskeletal
Model
Muscle
Location
Design
Geometry
Drivers
Resolved Inverse
Dynamic
Analysis
Muscle
parameters
Bite Force
Calculation
Forces
Motion
Muscle Force
Calculation
Optimization
Criteria
Kiran Konakanchi
August 19, 2005
Introduction Issues Background Modeling Case studies Conclusion
Automation, Robotics and Mechatronics Lab, SUNY at Buffalo
Revisiting our project flow chart
Slide 29 of 41
CT / MRI Data
MIMICS
CAD Model
(ASCII STL File)
Muscle forces
obtained at
this point
Musculo-skeletal model with approximate
Muscles using “MATLAB” GUI
Inverse Dynamic Analysis to
calculate Muscle forces
Calculate Bite Force
Change the Muscle locations
according to constraints
Kiran Konakanchi
August 19, 2005
Yes
No
Optimality attained?
End
Introduction Issues Background Modeling Case studies Conclusion
Automation, Robotics and Mechatronics Lab, SUNY at Buffalo
Muscle force calculation
Slide 30 of 41
Min/Max criteria is used for muscle force calculation
Using bound formulation we can easily solve the min/max problems
By choosing artificial variable β and artificial function B(β) such that
B(β) = β
The min/max criteria can be reformulated as
Minimize 
F ,
Subject to:
Cf=d
FM ,i
Ni
  , i  1,....n
FM ,i  0, i  1,....n
Kiran Konakanchi
August 19, 2005
Introduction Issues Background Modeling Case studies Conclusion
Automation, Robotics and Mechatronics Lab, SUNY at Buffalo
Slide 31 of 41
Case studies
Depending on 3 factors
Case
Specimen Feline
(Tiger member
or Sabertooth
Routing
tiger)algorithm
Muscle Type
Case1
Tiger
VPMuscle
Simple
 Muscle route algorithm (Via Point muscle or Shortest Path muscle)
Case2
Tiger
VPMuscle
Complex
Case3

Type of muscleTiger
(Simple or Hill SPMuscle
muscle)
Case4
Tiger
SPMuscle
Case5
Case6
Case7
Case8
Simple
Complex
Saber Tiger
VPMuscle
Simple
Saber Tiger
VPMuslce
Complex
Saber Tiger
SPMuslce
Simple
Saber Tiger
SPMuscle
Complex
Complex Muscle
Simple Muscle
SPMuscle
VAMuscle
Tiger
Saber Tiger
Kiran Konakanchi
August 19, 2005
Introduction Issues Background Modeling Case studies Conclusion
Automation, Robotics and Mechatronics Lab, SUNY at Buffalo
60
Masseter11 Force(Newtons)
0.01
0.008
20
0.004
0.002
0
Kiran Konakanchi
August 19, 2005
0
0.5
1
Simulation Time
1.5
Masseter11(left) force Vs Time
60
40
20
0
0.006
Pterygoid11 Force(Newtons)
Muscle Activities
0.012
0
0.5
1
Simulation Time
1.5
Pterygoid11(right) force Vs Time
40
30
20
10
0
0
0
0.2
0.4
0.8
0.5
1 0.6
1.5
Simulation Time Simulation Time
Masseter21 Force(Newtons)
60
40
Slide 32 of 41
Temporalis21(right) force Vs Time
Muscle Activities Vs Time80
0.016
0.014
Temporalis21 Force(Newtons)
Temporalis11(left) force Vs Time
80
Pterygoid11 Force(Newtons)
Temporalis11 Force(Newtons)
Case 1: Tiger – VPMuscle - Simple
40
20
0
60
Temporalis11
Temporalis21
Masseter11
0.5Masseter211
Simulation Time
Pterygoid11
Massaeter21(left) force Vs Time
Pterygoid21
1.5
40
20
0
0
0.5
1
Simulation Time
1.5
Pterygoid11(left) force Vs Time
1
0.5
0
01
1.2
0.5
1.4 1
Simulation Time
1.5
Introduction Issues Background Modeling Case studies Conclusion
Automation, Robotics and Mechatronics Lab, SUNY at Buffalo
0.08 360
340
0.07 320
300
0.03
0.02
0.01
0
Kiran Konakanchi
August 19, 2005
0
0.5
1
Simulation Time
1.5
1.5
Masseter21 Force(Newtons)
0.04
0
1.5
Pterygoid11 Force(Newtons)
Masseter11 Force(Newtons)
0.05
Pterygoid11 Force(Newtons)
Muscle Activities
0.06 280
Masseter11(left) force Vs Time
4
3
2
1
0
0
0.5
1
Simulation Time
Pterygoid11(left) force Vs Time
8
6
4
2
0
Slide 33 of 41
Muscle Activities Vs Time
Temporalis21(right) force Vs Time
Temporalis11(left) force Vs Time
0
0.5
1
0.2Simulation Time
0.4
Temporalis21 Force(Newtons)
Temporalis11 Force(Newtons)
Case 8: Sabertooth tiger – SPMuscle - Complex
150
100
50
0
0
0.5
1
Simulation Time
Temporalis11
Temporalis21
Masseter11
Masseter21
Pterygoid11
1.5
Pterygoid21
Massaeter21(right) force Vs Time
1.5
1
0.5
0
0
0.5
1
Simulation Time
1.5
Pterygoid11(right) force Vs Time
300
250
200
150
0
0.6
0.8
Simulation Time
0.5
1
1
Simulation
Time1.2
1.5
1.4
Introduction Issues Background Modeling Case studies Conclusion
Automation, Robotics and Mechatronics Lab, SUNY at Buffalo
Bite force calculation
Slide 34 of 41
Bite force is calculated from the following equilibrium equation.
 Fe   $ˆ   FM 
Where,
$ˆ
is the line vector matrix that depends on the muscles line of
action (calculated from plucker coordinates by knowing the
positions of origin and insertion of muscle)
FM
is the column vector representing the muscle force (obtained above)
Fe
is the external force or Bite force
Kiran Konakanchi
August 19, 2005
Introduction Issues Background Modeling Case studies Conclusion
Automation, Robotics and Mechatronics Lab, SUNY at Buffalo
Optimality Criteria
Slide 35 of 41
Minimize: –Bite force ( X )
Subject to:
Cf=d
FM ,i
  , i  1,....n
Ni
FM ,i  0, i  1,....n
X L  X  XU
Where,
X Vector representing design variables i.e. muscle origin and
insertion coordinates
X L and X U represent the lower and upper limits respectively
Kiran Konakanchi
August 19, 2005
Introduction Issues Background Modeling Case studies Conclusion
Automation, Robotics and Mechatronics Lab, SUNY at Buffalo
Slide 36 of 41
Parametric sweep studies
Tiger-SPMuscle-Simple
Variation of Masseter
pterygoid
insertion
insertion
Temporalis
Pterygoidorigin
origin
insertion
origin
Bite
Force
Bite
Force
Vs Vs
Iterations
Bite
Force
Vs Iterations
Iterations
440
283
288
264.6
300
380
force
Bite
Bite force
of
plot
Surface
Surface
plot
offorce
Surface
plot of
Bite
-25
-27.5
-28.2
-26.8
-25
Bite force
Bite force
Bite force
Bite force
Net Bite Force (Newtons)
Net Bite
Bite Force
Force (Newtons)
(Newtons)
Net
-30
-28.3
-28
-26.85
-28
-30
-28.4
-28.5
-26.9
-35
-28.5
-29
-28.5
-35
-28.6
-26.95
-29.5
-40
-28.7
-29
-40
0.05
-27
-30
-28.8
264.4
420
286
282
295
360
264.2
400
284
264
380
281
290
340
282
263.8
360
280
263.6
285
320
340
263.4
278
320
279
280
300
263.2
276
300
263
278
275
280
0.3
00.05
0.04
0.8
274
280
262.8
-29.5
-30.5
-27.05
-28.9
-45
0
-0.07
-0.09
0.18
0.15
0
0.1
-0.02
0.16
-0.1
-0.075
0.05
-0.04
-0.11
0.14
Temporalis 'Y'
-0.05
Masseter
Temporalis
Pterygoid
'Y' -0.4
0
-0.1
00
0.2
-0.04
-0.1 -0.08
-0.08
-0.05
-0.12
0.12-0.2-0.1
-0.3-0.06
-0.15
-0.1
-0.06
0
Temporalis 'X'
Surface plot
Kiran Konakanchi
August 19, 2005
-0.04
0
-0.05
0.1 0.2
0.4
-0.06
-0.02
0.1
'X''X'
Masseter
Masseter 'X'
Pterygoid
Temporalis
Pterygoid
-0.02
0.02
00.6
0.2
0.3
260
277
272
262.6
270
00
1010
10
2020
20
30 30
40 40
30
40
Iterations
Iterations
Iterations
50 50
50
60
60
60
70
70
70
Line plot
Introduction Issues Background Modeling Case studies Conclusion
Automation, Robotics and Mechatronics Lab, SUNY at Buffalo
Slide 37 of 41
Bite Force graphs
Case 1:Tiger – SPMuscle - Simple
Bite Force Vs Iterations
281.5
281
Net Bite Force (Newtons)
280.5
280
279.5
279
278.5
278
277.5
Kiran Konakanchi
August 19, 2005
0
5
10
15
Iterations
20
25
30
Introduction Issues Background Modeling Case studies Conclusion
Automation, Robotics and Mechatronics Lab, SUNY at Buffalo
Bite Force graphs
Slide 38 of 41
Case 1:Sabertooth tiger – SPMuscle - Simple
Bite Force Vs Iterations
-895.5
Net Bite Force (Newtons)
-896
-896.5
-897
-897.5
-898
Kiran Konakanchi
August 19, 2005
0
5
10
15
Iterations
20
25
30
Introduction Issues Background Modeling Case studies Conclusion
Automation, Robotics and Mechatronics Lab, SUNY at Buffalo
Slide 39 of 41
Conclusion
Developed a user friendly computational framework for testing hypothesis
and various if-then scenarios.
Identified the critical issues pertaining to musculoskeletal modeling like
redundancy, geometric complexity, muscle recruitment pattern etc.
Validated some of the available software packages with regards to
available data.
Conducted a range of virtual experiments on members of feline family
(tiger & sabertooth tiger) with our proposed methodology that can help in
the study of functional performance.
Finally, we presented the biologist with a novel validated toolbox.
Kiran Konakanchi
August 19, 2005
Introduction Issues Background Modeling Case studies Conclusion
Automation, Robotics and Mechatronics Lab, SUNY at Buffalo
Future work
Slide 40 of 41
Optimization routine
Brute force method has been employed for parametric sweep.
More sophisticated algorithm like will be used in future.
Limitations of the software:
MATLAB interface limitations will be resolved in future versions.
Issues pertaining to modeling:
We approximated the muscle origin and insertion as points. The
future work will include the solution strategy to this issue like proving a
curtain of muscles instead of single muscle etc.
Task space redundancy:
The bite force can be calculated at the tip pf one tooth. Task
space redundancy need to be resolved to simultaneously calculate the bite
force at the tips of two teeth. Screw theory (delSignore [36]) in order to
solve the problem of task space redundancy.
Kiran Konakanchi
August 19, 2005
Automation, Robotics and Mechatronics Lab, SUNY at Buffalo
Slide 41 of 41
Thank You!
Questions?
Kiran Konakanchi
August 19, 2005
Automation, Robotics and Mechatronics Lab, SUNY at Buffalo