a new way of myoelectric control a new way of myoelectric control myopro project daphne van baal...

Post on 31-Mar-2015

213 Views

Category:

Documents

0 Downloads

Preview:

Click to see full reader

TRANSCRIPT

A new way of myoelectric A new way of myoelectric controlcontrol

MyoPro project

Daphne van BaalHans Rietman, Laura Kallenberg

25 september 2009

Roessingh Research and Development, Enschede, the Netherlands2

IntroductionIntroduction

Low usage myoelectric prosthesis

Advantages DisadvantagesNon-invasiveNo straps or harnessesUse of remaining muscles

Limited selectivityLack of intuitive controlLack of sensory feedback

Roessingh Research and Development, Enschede, the Netherlands3

GoalGoal

Improve the control ofa myoelectric forearm prosthesis

by increasing the number of degrees of freedom

using multi-channel surface electrodes

Roessingh Research and Development, Enschede, the Netherlands4

Experimental protocolExperimental protocol

• Ten healthy male subjects• One forearm amputee• Forty small electrodes in grid• Eight isometric contractions

Roessingh Research and Development, Enschede, the Netherlands5

Isometric contractionsIsometric contractions

Flexion of wrist Extension of wrist Flexion of fingers Extension of fingers

Abduction of wrist Adduction of wrist Supination of forearm

Pronation of forearm

Roessingh Research and Development, Enschede, the Netherlands6

Data analysisData analysis

• Data selection of active contractions• Time periods of 200 milliseconds• Root mean squared (RMS) value• Subject specific classification• Simple robust classifier (kNN)

Roessingh Research and Development, Enschede, the Netherlands7

RMS-level imagesRMS-level images

Roessingh Research and Development, Enschede, the Netherlands8

Robust classifierRobust classifier

Training set Test set

Roessingh Research and Development, Enschede, the Netherlands9

ResultsResults

OutputInput FF FE P S AB AD WE WF

Finger flexion99.9

40.03 0.01 0.02

Finger extension

0.0199.8

90.01 0.01 0.08

Pronation99.9

60.04

Supination 0.01 0.0199.8

50.03 0.02 0.08 0.01

Wrist abduction

0.02 0.0199.9

40.02 0.01

Wrist adduction

0.0199.9

70.01

Wrist extension

0.02 0.03 0.02 0.10 0.0199.8

30.01

Wrist flexion 0.02 0.05 0.0199.9

3

Roessingh Research and Development, Enschede, the Netherlands10

Continuous resultsContinuous results

Roessingh Research and Development, Enschede, the Netherlands11

RMS-level images RMS-level images amputeeamputee

Roessingh Research and Development, Enschede, the Netherlands12

Results amputeeResults amputee

OutputInput

FF FE P S AB AD WE WF ABP ADS

Finger flexion 100.0

Finger extension

99.28

0.48 0.25

Pronation 99.62

0.02 0.13 0.23

Supination 99.64

0.13 0.06 0.17

Wrist abduction

99.73

0.06 0.21

Wrist adduction

0.08 0.25 0.0499.5

60.02 0.06

Wrist extension

0.02 0.17 0.1199.6

40.06

Wrist flexion 0.10 0.0299.8

9

Wrist ab & pro 0.19 0.34 0.0699.4

1

Wrist ad & sup 0.6199.3

9

Roessingh Research and Development, Enschede, the Netherlands13

ConclusionConclusion

• Offline classification – normally limbed: 99.9%– amputee patient: 99.6%

• Online classification– normally limbed: 93.4%

• Up to ten movements can be distinguished (five degrees of freedom)

Roessingh Research and Development, Enschede, the Netherlands14

FutureFuture

• Validation: more measurements with amputees

• Additional patient specific contractions• Random-ordered contractions• Simultaneous control of different

contractions• Force level analysis for proportional control

Roessingh Research and Development, Enschede, the Netherlands15

Questions?Questions?

For more information: www.myopro.nlFor more information: www.myopro.nl

top related