A Low-Cost Framework for Individualized Interactive Telerehabilitation Chetan Jadhav Advisor: Dr. Venkat Krovi Mechanical and Aerospace Engineering Department

A Low-Cost Framework for Individualized Interactive Telerehabilitation

Chetan Jadhav

Advisor: Dr. Venkat KroviMechanical and Aerospace Engineering Department

State University of New York at Buffalo

http://mechatronics.eng.buffalo.edu/home_new.html

Chetan JadhavJuly 7th, 2004

Slide 2

Presentation Overview

• Motivation & Background• Description of Framework• Implementation• Biomechanical Parameter Identification• Exercise Assistance• Conclusion & Future Work


Motivation Framework Implementation Identification Assistance ConclusionChetan JadhavJuly 7th, 2004

Slide 3

Motivation

• Stroke Statistics [1]• U.S. alone, each year over 737,000 people experience

a new or recurrent stroke

• $17 billion direct cost (hospitals, physicians, rehabilitation etc.)

• $13 billion indirect cost (lost productivity etc.)

• Increase of 31.3% in number of cases from 1979 to 2000

• Severely disrupts activities of daily living• Suitable motor-rehabilitation regimen can facilitate

significant functional recovery [2]



Slide 4

Rehabilitation Therapy

• Requirements • Early and accurate diagnosis of the disease coupled with

careful characterization of the level of functional impairment• Functional recovery linked to the duration, frequency,

regularity and intensity of therapy• Attention and monitoring required from a therapist

• Trends• Computer-enhanced Therapy• Home based Therapy



Slide 5

Computer-enhanced Rehabilitation Therapy

• Computerized Exercise Systems• Leverages the ubiquitous computational power

• Measures and records patient’s performance

• Interacts and adjusts the therapy regimen

• Robotic therapy devices • Allows more complex therapies like Constraint Induced

Therapy [3]

• Guides the patient through the intensive, repetitive practice of functional movement



Slide 6

Example of Computer-enhanced Rehabilitation

MIT MANUS [4]• A planar, two-revolute-joint robot• Assists patients in sliding their arms across a tabletop• Significant improvements in motor recovery• Measurement tool to track disease progress



Slide 7

Limitations of Computer-enhanced Rehabilitation

• Suitable for outpatient units• Specialized device and not readily available• Therapist’s or expert’s interventions is required• Prohibitive cost for personal use



Slide 8

Home Based Rehabilitation

• Flexibility in intensity and duration of rehabilitation regimen

• Comparable effects with hospital attendance in terms of functional gains [4]

• Therapist visits the patient’s home periodically• Lack of structured and monitored exercises can

mitigate achievable benefits• Use of specialized exercise machines is limited• Not much cost benefits due to logistic issues

associated



Slide 9

Proposed Telerehabilitation Framework



Slide 10

Research Issues

Requirements/Needs

• Home based• Low-cost• Use of COTS

• Automated decision support• Quantitative data capture • Quantitative assessment of disability and performance

• Individualized• Update exercise based on patient model



Slide 11

Virtual Driving Environment

• Not just a driving simulator or trainer

• Illustrative example to integrate the multiple facets of telerehabilitation

• Helps identify the issues with development, implementation and deployment

• Serves to enhance one higher activities of daily living



Slide 12

Contemporary Telerehabilitation

• Use video conferencing technology• High bandwidth internet connection required• Multiple camera views are necessary to recognize

patient’s movement patterns• Lack of quantitative assessment capabilities• Suitable for outpatient units• Prohibitive cost for home use



Slide 13

Focus of this Thesis

Part A Implementation of VDE

Part B Biomechanical Identification

Part C Manipulation Assist



Slide 14

Hardware

Force Feedback (FFB) Gamming Wheel

Rate-Gyro

Part A



Slide 15

Implementation

Within MATLAB/Simulink environment• Various toolboxes available

• Application Program Interface (API) to extend capabilities

• Fast prototyping environment

Part A



Slide 16

Software

MATLAB FFB wheel interface• Does not posses ability to access such devices

• Implemented using API, C++, DirectX

Part A



Slide 17

Software (cont.)

TCP/IP implementation• Connects JACK to MATLAB application

• JACK model can be controlled from MATLAB to recreate exercise session

• Winsock2 libraries used in S-Function

Part A



Slide 18

Software (cont.)

User interfaces

Two-D Dial Implementation

Fully Immersive VE

Part A



Slide 19

Biomechanical Parameter Identification

Assessing performance of movement tasks• Background

• Muscle activation or Electro Myographs (EMG)• Frequency or time normalization• Pathological motor patterns identified by comparison• Bilateral comparison between limb

• Joint kinetics• Complexity of modeling• Computational Tools like SIMM and Anybody• need to be customized to match the specific individual

• Joint kinematics• Forms a sound basis for kinetic model• Visualization needs at Therapist Interface

Part B

• Quantitative data capture • Quantitative performance• Individualized



Slide 20

Upper-limb Kinematic Model

Part B



Slide 21

Calibration of Planar Four-Bar Mechanism

• Used techniques from ROBOTICS to identify link lengths

• Simulated using known four-bar mechanism

2 2 1 1 3 3 4 4

2 2 1 1 3 3 4 4

cos cos cos cos

sin sin sin sinmeasured nominal

l l l lX X X

l l l l

Part B



Slide 22

Calibration of Planar Four-Bar Mechanism (cont.)

Taylor series expansion of is,

For k measurements

Solving above system by pseudoinverse method

X

3 4 2 2 3 3 4 43 4 2 3 4

3 4 2 2 3 3 4 4

cos cos sin sin sin

sin sin cos cos cos

Tl l lX l l

l l l

1 1

k k

X

X

X

1T TX

Part B



Slide 23

Calibration of Planar Four-Bar Mechanism (cont.)

Considering the use of COTS components, random noise simulated

0 10 20 30 40 50 60 70-2

-1.5

-1

-0.5

0

0.5

1

1.5

2L3L4

0 10 20 30 40 50 60 70 80 90 100-2

-1.5

-1

-0.5

0

0.5

1

1.5

2L3L4

0 20 40 60 80 100 120-2

-1.5

-1

-0.5

0

0.5

1

1.5

2L3L4

5% 10% 25%

Lengths % Length Error (5 % Noise)

% Length Error (10 % Noise)

% Length Error (25 % Noise)

l3 4 -2 4

l4 0.3 -1 6

Iterations 43 99 103

Part B



Slide 24

Calibration of Spatial Four-Bar Mechanism

• Shoulder approximated to spherical

• Elbow approximated as universal

• Wrist and grip approximated as single universal

• Steering joint is revolute

• DOF = 2

Part B



Slide 25

Calibration of Spatial Four-Bar Mechanism (cont.)

Two-Link trial

• Fast convergence (3 iterations)

Parameter Initial Calibrated Actual % Error

L1 1.2991 0.5942 0.6 0.96

L2 0.5 0.9972 1.0 0.28

Part B



Slide 26

Manipulation Assist

• Background• Classification based on use of equipment

• Unassisted Exercises• Machine Assisted Exercises• Robotic-Assisted Exercises

• Classification based on Nature of Assist• Passive Resist• Active Resist

• Classification based on Type of Manipulator Assist• Motion Assistance• Force Assistance• Combined Motion Force Assistance

Part C

• Quantitative data capture • Quantitative performance• Individualized



Slide 27

Exercise Assistance

Model of the interaction of the patient with the driving wheeland the virtual environment

Part C



Slide 28

Exercise Assistance (cont.)

• Dynamics of the system can be written as,

• In state-space form

• Where states are

1 2 1 1 2 1 1 2 1 act h

J B C

J J B B C C

2

32 1 2

3

1 2 3 1 2 3

0

0

10

, , , ,

C Bu

J J J

f g u

1 2 31 1

Part C



Slide 29

Exercise Assistance (cont.)

• Non-linear system

• Feedback linearization technique is used

Part C



Slide 30

Input-Output Linearization for Motion Assistance

• Output equation can be written as,

• The relative degree of the system is three and it is input-state linearizable and the decoupling matrix is,

• Nonlinear state feedback can be computed as,

where,

• Solving for and

1

2 1 2 3

3

1 0 0 , ,y h

u

1J

3 and fL h I

1 2 3BC B BC

J J J

J

Part C



Slide 31

Input-Output Linearization for Motion Assistance (cont.)

• States are transformed via a diffeomorphism

• Linearized system is written as,

2

1 1

2 2

3 3

1 0 0

0 1 0

1

f fz T h L h L h

z

z

z C B

J J J

1 1

2 2

3 3

0 1 0 0

0 0 1 0

0 0 0 1

z z

z z

z z

Part C



Slide 32

• Control system is designed using pole-placement techniques

• Stepping back through the diffeormorphism

• Desired linearized sates (zd) can be calculated using diffeomorphism

3 3 3 2 2 1 12 1 2

d d d dz K z z K z z K z z

3

2 1 2 3 3 3 2 2 1 12 1 2 d d d d

u

C BC B J z K z z K z z K z z

J J J


Part C



Slide 33

• Simulation Results


Part C



Slide 34

• Output equation can be written as,

• The relative degree of the system is three and it is input-state linearizable

• We set y=0, then θ3 = 0 and associated zero dynamics is,

which is asymptotically stable

• The control law may be computed as,

Input-Output Linearization for Force Assistance

1

2 1 2 3

3

0 0 1 , ,y h

1 1 1B C

J J

1f

g

u L h khL h

Part C



Slide 35

• k is chosen such that the polynomial K(s) = s+k has all its roots in left half plane. If we let k = -1, then,

• The control law can be written as,

Input-Output Linearization for Force Assistance (cont.)

2

31 20 0 1 0

0

f

C BL h

J J J

3 310 ( 1)

1u

3 3 3d dpu K

Part C



Slide 36

• Simulation Results

Input-Output Linearization for Force Assistance (cont.)

Part C



Slide 37

Conclusion

• Successful implementation

• Efficient Kinematic Calibration method for biomechanical parameter identification

• Motion and Force assistance using non-linear control



Slide 38

Future Work

• Global calibration method which can take arbitrary parameters as initial guesses

• Recursive least square for on-going calibration

• Real time implementation in Windows environment



Slide 39

Questions ?



Slide 40

Thank You



Slide 41

References

[1] Anon. (2003). National Stroke Association.

Available: http://www.stroke.org

[2] P. Langhorne, R. Wagenaar, and C. Partridge, "Physiotherapy after stroke: More is better?," Physiother Res Int, vol. 1, no. 2, pp. 75-88, 1996.

[3] S. L. Wolf, D. E. Lecraw, L. A. Barton, and B. B. Jann, "Forced use of hemiplegic upper extremities to reverse the effect of learned nonuse among chronic stroke and head-injured patients," Exp Neurol, vol. 104, no. 2, pp. 125-32, 1989.

[4] M. L. Aisen, H. I. Krebs, N. Hogan, F. McDowell, and B. T. Volpe, "The effect of robot-assisted therapy and rehabilitative training on motor recovery following stroke," Arch Neurol, vol. 54, no. 4, pp. 443-6, 1997.

[5] J. B. Young and A. Forster, "The bradford community stroke trial: Results at six months," Bmj, vol. 304, no. 6834, pp. 1085-9, 1992.



Slide 42

Virtual Patient Model

Back



Slide 43

Patient Interface

Back



Slide 44

Steering Interface with MATLAB

Back



Slide 45

Matlab Jack Interface

Back



Slide 46

VDE

Back


Documents

A Low-Cost Framework for Individualized Interactive Telerehabilitation Chetan Jadhav Advisor: Dr. Venkat Krovi Mechanical and Aerospace Engineering Department