exploring the utility of the concept of “rheostat activators” of the forearm and hand muscles...

21
Exploring the Utility of the Concept of “Rheostat Activators” of the Forearm and Hand Muscles for Modeling Hand Movements Institution: University of Toronto Authors: Winnie Tsang, Karan Singh, Nancy McKee, Anne Agur

Upload: myles-eaton

Post on 28-Dec-2015

216 views

Category:

Documents


0 download

TRANSCRIPT

Exploring the Utility of the Concept of “Rheostat Activators” of the Forearm and Hand Muscles

for Modeling Hand Movements

Institution: University of TorontoAuthors: Winnie Tsang, Karan Singh,

Nancy McKee, Anne Agur

The Human Hand…a remarkable biomechanical deviceProblem How do forearm and hand muscles produce hand motion? Muscle redundancy Problem: # of muscles > # of joint DOF.

Our Ultimate GoalTo determine a set of muscle excitations that produce a desired movement.

Tools MRI, Ultrasound, EMG measurement. Biomechanical model of muscle (Hill's muscle model). Musculoskeletal model of the hand. Numerical Optimization Theory.

Hand Models…for the study of hand motionWhat our Hand model can do Forward simulation: given muscle excitations motion Inverse simulation: given hand motion muscle excitations Explore muscles’ function by changing their usability

Other works in human motion Scott Delp's SIMM (Software for Interactive Musculoskeletal

Modeling). Anderson and Pandy's work in Human Walking. Yamaguchi, Zajac, Hardt, Teran, An and many others.

Comparison of two dominant techniques to analyze motionStatic Optimization body motion and external force measurements muscle forces Advantages:

computational efficient. Disadvantages:

Highly dependent on the accuracy of measurements. Muscle physiology is difficult to include. A model of the goal of the motor task cannot be included.

Dynamic Optimization muscle excitations body motion Advantages:

Uses system of equation to describe the force-motion relationship. Muscle physiology is easy to include. A model of the goal of the motor task can be included.

Disadvantages: computationally expensive.

Forward Simulation (Dynamic Optimization)

∫ ∫

Compute body motion given muscle excitations. Involves implicit integration (a stable method) of integrating

system of equations.

Hill's Model input: excitation of muscle m output: force of muscle m

Forward Simulation (Dynamic Optimization)

∫ ∫

Compute body motion given muscle excitations. Involves implicit integration (a stable method) of integrating

system of equations.

Torque = Lever Arm X Forceinput: force of muscle m output: torque of muscle m

Forward Simulation (Dynamic Optimization)

∫ ∫

Compute body motion given muscle excitations. Involves implicit integration (a stable method) of integrating

system of equations.

input: Torque of muscle moutput: Angular Acceleration of joint j

Inverse Simulation (Parameterized Dynamic Optimization) Compute muscle excitations from body motion. parameterize input: rheostat actuations (muscle excitations). convert to a parameter optimization problem.

Error Term Efficiency Term

Objective Function

Constraints

Inverse Simulation (Parameterized Dynamic Optimization)

Model AssumptionsMusculotendon Dynamics only force-length property of muscle no force-velocity property of muscle Muscles’ peak isometric force and corresponding fiber length from

[Brand 81]

Musculoskeletal Geometry 42 musculotendon units. Muscle and tendon origins and insertions are points not areas. Ignored pennation angle. Lever arm approximated from [Brand 99] moment arms.

Skeletal Dynamics masses, geometric measurements of bone and muscle from

[Biryukova and Yourovskaya]

ResultsForward Simulation…

Implementation on Maya platform. Simulation results on a P4, 2.79 GHz, 1.00G RAM machine. Forward Simulation Clip 1: FDS index set with excitation of 0.5 Forward Simulation Clip 2: FDS index set with excitation of 0.5 and

usability of 0.2

ResultsInverse Simulation… Implementation on Maya platform. Simulation results on a P4, 2.79 GHz, 1.00G RAM machine. Inverse Simulation Clip 1: Use Forward Simulation result as input Inverse Simulation Clip 2: Use Forward Simulation result as input

and FDS index usability of 0

So What Have We Done?

Our Contribution An anatomical Model of Hand A neuromusclotendon model:

muscle excitations motion (forward simulation) A parameter optimization approach:

motion muscle excitations (inverse simulation) Parameterized muscles to model strength or utility Error based performance measurement

More Work to be Done…Limitations Approximations:

averaged data muscle's line of action. lever arm.

Ignore external forces (i.e.: gravity). Ignore some muscle physiology (force-velocity relation). No actual muscle activation measurements for verification. Optimization technique is computationally intensive.

Future Challenges More Sophisticated Hand Model Validation and Calibration of Hand Model

References ALBRECHT I., HABER J., SEIDEL H.: Construction and animation of anatomically based

human hand models. In Proceedings of SIGGRAPH Symposium for Computer Animation 2003 (2003), vol. 22, ACM Press / ACM SIGGRAPH, pp. 98–109.

AGUR A. M. R., LEE M.: Grant’s Altas of Anatomy, 10 ed. Lippincott Williams & Wilkins, Baltimore, Maryland, USA, 1999.

ANDERSON F. C., PANDY M. G.: Dynamic optimization of human walking. Journal of Biomechanical Engineering 123, 3 (2001), 381–390.

ANDERSON F. C., PANDY M. G.: Static and dynamic optimzation solutions for gait are practically equivalent. Journal of Biomechanics 34 (2001), 153–161.

BRAND P., BEACH R., THOMPSON D.: Relative tension and potential excursion of muscles in the forearm and hand. Journal of Hand Surgery 6, 3 (1981), 209–219.

BRAND P., HOLLISTER A.: Clinical Mechanics of the Hand., 3 ed. Mosby - Year Book, Inc., St. Louis, MO, 1999.

BIRYUKOVA E., YOUROVSKAYA V.: A model of hand dynamics. In Advances in the Biomechanics of Hand and Wrist (1994), Plenum Press, New York, pp. 107–122.

NG-THOW-HING V.: Anatomically-Based Models for Physical and Geometric Reconstruction of Humans and Other Animals. PhD thesis, University of Toronto, 2001.

ZAJAC F. E.: Muscle and tendon: properties, models, scaling, and application to biomechanics and motor control. Critical Reviews in Biomedical Engineering 17 (1989), 359–411.

THELNE D. G., ANDERSON F. C., DELP S. L.: Generating dynamic simulations of movement using computed muscle control. vol. 36, pp. 321–28.

See Me for More…

Thanks For Listening!Any Questions?