rehabilitation robotics using central pattern generators · rehabilitation robot one degree of...

Rehabilitation robotics using Central Pattern Generators

Student: Sarah MOUSSOUNI: ::::

Master ProjectFinal Presentation

June 25th ,2010

Supervisors : Renaud RONSSE

: Mohamed BOURI

Professor : Auke Jan IJSPEERT

ENSIMAG : Florence MARANINCHI

Introduction | Plan | Environment | Model & simulation | Implementation | Conclusion

� Project context : BioRob Lab & LSRO� Goals of the project � Presentation of the Adaptive oscillators � Presentation of the Knee-orthosis

SIMULINK modeling

Plan

2

� SIMULINK modeling � Work on the Knee-orthosis

� Transparent mode� Integration of the adaptive oscillators� Results and discussion

� Conclusion and future work

Rehabilitation robotics using Central Pattern Generators | EPFL | June 2010 | Sarah MOUSSOUNI

Biorobotics & LSRO Lab


3 Rehabilitation robotics using Central Pattern Generators | EPFL | June 2010 | Sarah MOUSSOUNI

Goal :

To provide a new rehabilitation

method to disabled persons

Goal :

Conception of locomotor re-

education and walking

assistance devices

� Issue

Rehabilitation robotics

It is an application of engineering to design and develop technological

solutions for people suffering from movement disorders

� Solution



Goals of the project


Autonomy


Movement



Autonomy




Movement


1) Investigation of a rehabilitation protocol based on the theory of adaptive oscillator.

2) Implementation of the method on the Knee orthosis



9

2) Implementation of the method on the Knee orthosis

3) Test of the method with various movements

4) Validation of the method on healthy people


General Project Schema

Biorobotics Lab


SIMULINK modeling

LSRO Lab


Adaptive oscillators Knee-orthosis


SIMULINK modeling

Implementation and validation

Rehabilitation protocol

Force processing



SIMULINK modeling

LSRO Lab


Knee-orthosis Adaptive oscillators

Biorobotics Lab


SIMULINK modeling



Force processing

� Used in the context of rhythmic movement assistance

� Predict the state evolution in real-time (without delay)

Adaptive oscillators


� Modified Hopf oscillator


� Modified Hopf oscillator

Adaptive oscillators


Signal F(t) = sin (20t)



Knee-orthosis Rehabilitation roboticsAdaptive oscillators

SIMULINK modeling

LSRO Lab


Biorobotics Lab


SIMULINK modeling



Force processing


� Rehabilitation robot� One degree of freedom� Position and force sensors

Environment : Knee Orthosis

15

KneeOrthosis

Torque Γ

Orthosis Position

Measuredforce


Integration of the CPG system


Timeline

MeasurementSys

characterizationControl

Modeling the System

16

Control

Test & validation SimuLink + Matlab

Test & validation FlexWare + Orthosis





SIMULINK modeling

LSRO Lab


Biorobotics Lab


SIMULINK modeling



Force processing



Timeline : MatLab – Simulink implementation

MeasurementSys


Modeling the System

18

Control






Model and simulation

Rehabilitation robotics using Central Pattern Generators | EPFL | June 2010 | Sarah MOUSSOUNI19


Model and simulation


Torque


CPG & torque estimator

Dynamical model + Orthosis Position controller

Adaptive oscillator

Measuredposition

x

y

w

SignalEstimato

Position

Velocity

21

w

Amplitude

Offset

Estimator Acceleration



Simulation & resultsω =2π � ω = 1.5π




SIMULINK modeling

LSRO Lab


Biorobotics Lab


SIMULINK modeling



Force processing


Timeline: Force


MeasurementSys


Modeling the System

26

Control






Implementation: Transparent mode

� 1st Objective:

Make the orthosis transparent for the user

� 5 Steps :

1

Knee orthosis


Force measured

Digital analogue value

(Volt) Calibration

Volt => NewtonInterpolation Filtering

ControlApplied torque

2 34

5


Calibration, Interpolation and filter


Using the following filter :


Implementation: Filtering

Using the following filter :



Measuring the stabilization tension for a set of position





Measured position

31

position

Measured force

+-

Desired force =0

+

PID Controller



Validation of the transparent mode

62.95

80

100

120

Measured force (Newton)


47.17

25.82

38.29

0

20

40

60

Without any implem With transparent mode

neg

pos



SIMULINK modeling

LSRO Lab


Biorobotics Lab


SIMULINK modeling



Force processing


Timeline : Integrating CPG


Processing the force sensors measurement

Modeling the System

34






Validation of the oscillator block



Validation of the system

24.66

30

40

50

60

Human effort

62.95

38.29

80

100

120

Measured force (Newton)


21.35

0

-15.21

0

-20-40

-30

-20

-10

0

10

20

Without any implem

With transparent mode

With force controller and

assistance K= 0.5

Extension

Flexion

47.17

25.82

10.61

38.29

18.29

0

20

40

60

Without any implem

With transparent

mode

With force controller and assistance K=

0.5

Extension

Flexion


DEMO



� Test of the method with various movements

� Validation of the method on healthy people

Future work

38

� Validation of the method on healthy people

� Work on the LAMBDA robot offering 3 DOF

� Design of a preliminary rehabilitation protocol.



Questions ?



Additional slide [1] : Schema of the Knee orthosis system



Additional slide [2] : Controller



Additional slide [3] : Part of the Simulink model



Additional slide [4] : E.g. of the measurement procedure



Additional slide [5] : General schema

Adaptive oscillatorForce

controller

Measured

Measured force

DesiredForce


Torque estimator

Measuredposition

Signal estimation : position, velocity and acceleration

rehabilitation robotics using central pattern generators · rehabilitation robot one degree of...

Documents