Automatic tuning of myoelectric prostheses

Journal of Rehabilitation Research andDevelopment Vol . 35 No . 3, July 1998Pages 294-304

Automatic tuning of myoelectric prostheses

Claudio Bonivento, Prof, PhD ; Angelo Davalli, MS; Cesare Fantuzzi, PhD; Rinaldo Sacchetti, MS;Sabina Terenzi, MSDEIS-LAR, University of Bologna, 40136 Bologna, Italy ; Centro per la Sperimentazione ed Applicazione diProtesi e Presidi Ortopedici, INAIL, Vigorso (Bologna), Italy ; University of Ferrara, Department of Engineering,44100 Ferrara, Italy

)partment ofarans Affairs

Abstract—This paper is concerned with the development of asoftware package for the automatic tuning of myoelectricprostheses . The package core consists of Fuzzy Logic ExpertSystems (FLES) that embody skilled operator heuristics in thetuning of prosthesis control parameters.

The prosthesis system is an artificial arm-hand systemdeveloped at the National Institute of Accidents at Work(INAIL) laboratories . The prosthesis is powered by an electricmotor that is controlled by a microprocessor usingmyoelectric signals acquired from skin-surface electrodesplaced on a muscle in the residual limb of the subject. Thesoftware package, Microprocessor Controlled Arm (MCA)Auto Tuning, is a tool for aiding both INAIL expert operatorsand unskilled persons in the controller parameter tuningprocedure.

Prosthesis control parameter setup and subsequent recur-rent adjustments are fundamental for the correct working ofthe prosthesis, especially when we consider that myoelectricparameters may vary greatly with environmental modifica-tions . The parameter adjustment requires the end-user to go tothe manufacturer's laboratory for the control parameters setupbecause, generally, helshe does not have the necessaryknowledge and instruments to do this at home . However, thisprocedure is not very practical and involves a waste of timefor the technicians and uneasiness for the clients.

The idea behind the MCA Auto Tuning package consistsin translating technician expertise into an FLES knowledgedatabase . The software interacts through a user-friendly

This material is based on work supported by a collaboration contractbetween DEIS (Dipartimento di Elettronica Informatica e Sistemistica),University of Bologna, and Centro per la Sperimentazione ed Ap-plicazione di Protesi e Presidi Ortopedici, INAIL, Vigorso, Italy.

Address all correspondence and requests for reprints to : Prof. ClaudioBonivento, DEIS-LAR, University of Bologna, Viale Risorgimento,2 - 40136 Bologna, Italy . email : cbonivento@deis .unibo.it .

graphic interface with an unskilled user, who is guidedthrough a step-by-step procedure in the prosthesis parametertuning that emulates the traditional expert-aided procedure.

The adoption of this program on a large scale may yieldconsiderable economic benefits and improve the servicequality supplied to the users of prostheses . In fact, the timerequired to set the prosthesis parameters are remarkablyreduced, as is the technician's working time . This isinterpreted as minor costs for prostheses manufacturers andsuppliers.

Key words : fuzzy logic, human-machine interface, prosthet-ics.

INTRODUCTION

The preliminary studies about the possibility ofusing residual muscles of a residual limb after amputa-tion to move a prosthesis dated from the beginning ofthis century, but it was not until the late 1940s that theimportance of myoelectricity in this field was under-stood (1-4) . The first myoelectric hand was realized byReiter in Germany, while significant improvements inthese studies were made by Battye, Nightingale, andWhillis in England and by Popov in Moscow (5,6).

In the early sixties, Hannes Schmidl promoted theexperimental application of myoelectric control at TheNational Institute for Accidents at Work (INAIL),Prosthesis Centre of Vigorso (Bologna, Italy) . On thebasis of this work, the first Italian myoelectric prosthe-sis was made in 1965 (7) . At present, the INAILProsthesis Centre is one of most important clinics in

294

295

BONIVENTO et al . Tuning of Myoelectric Prostheses

Europe that is producing myoelectric hands and prosthe-

required by the wearer for home assistance will beses (see Figure 1) .

the communication hardware.The INAIL Research and Development department

is now working to realize a hand prototype with

Some work has already been carried out on thefeedback sensors and to provide the user with tele-

methods of increasing prosthesis performances usingassistance and telediagnosis services (8,9) . Some other

computer systems (10) . Our work is a further attempt inresearch concerning "intelligent" electrodes, as well as

this field toward high performance and ease in prosthe-studies about lower limbs, is also being carried out . This

sis use. The key idea was to incorporate technicians'article is concerned with the most recent prototype

expertise of manual tuning of control parameters into arealized by the Centre : the MCA Auto Tuning system .

Fuzzy Logic Expert System, which is then integrated inThe project aims to develop a software package for

a software tool.aiding both skilled technicians and unskilled users in the

Fuzzy Logic methodology (11—14) has been usedsetup process of a prosthesis control system .

in this task because of its capability to blend humanAt present, individuals are required to go to the

qualitative knowledge into formal algorithms . Thesemanufacturer's laboratory for the prosthesis control

"fuzzy" algorithms may then be efficiently imple-parameter setup. This procedure is performed manually

mented in computer programs.by factory technicians together with the prospective

Basically, a Fuzzy Logic Expert System is made upusers . The operator interacts with the client asking

of a database of rules and an inference engine (seehim/her to perform muscle contractions while the

Figure 2).technician records the corresponding sensor signals .

The database of rules consists of a set of linguisticUsing the results of the above tests and past experience,

statements ; such as, "if a premise is fulfilled, thenthe technician sets the control system parameters to

execute a corresponding action ."optimize the performances of the prosthesis . The setup

This structure makes it easy to translate humanprocedure should be performed several times during the

knowledge, often expressed via common languagelife-time of the prosthesis, mainly because of environ-

phrases, into the fuzzy rule database . These rules aremental (e .g ., temperature, humidity) changes, which still

then applied by means of an inference engine to thecause discomfort in users .

actual information gained from the wearer . In this step,The key idea of this project was to incorporate the

the actual premise is compared to all the rule premisesskill of expert technicians in an automated system based

stored in the database . The rules corresponding to theon easy-to-use software . The three most promising usesof the system are :

350-

Myoelectric prostheses number1 . The system can be used by the prosthesis manu-

shoulderfacturer technicians to automate the initial setup,

3001

permitting significant timesaving in the whole

Elbow-arm

process of prosthesis delivery .

250

Forearm

2. The software will be distributed to about 100

200

INAIL-affiliated centers stationed all over Italy, inwhich suitable hardware will be provided for

150users. In this way, persons with amputation canreceive local assistance for prosthesis setup.

loo-

3. The most ambitious aim of the project is toprovide an automated setup procedure directly in

50the home of the wearer . The system requires a

it.

personal computer (PC) and simple communica- R,'a

NC:,

0000 00

00 00

00

00

CN

CN

ONrn

•1'

tion hardware to establish the PC-prosthesis link, CN

CN

ON

ON0N

°

° °

as we will explain later . Since home computingsystems are becoming as common as domestic

Figure 1.appliances, it can be foreseen that the only thing

Production of prostheses by INAIL Centre from 1983 to 1994 .

296

Journal of Rehabilitation Research and Development Vol . 35 No . 3 1998

Figure 2.General scheme of a Fuzzy Logic-Based Expert System .

premises that agree, even partially, are chosen and thecorresponding actions are merged together to obtain thefinal output.

The MCA Automatic Tuning software works asfollows:

1. The client connects the prosthesis hardware to apersonal computer via the serial port and runs theprogram: MCA Auto Tuning.

2. The program acquires both sensor signals as wellas client input . For example, the client is asked tosustain muscle contractions with "high," "me-dium," and "low" intensity, while the programacquires the corresponding sensor signals.

3. The program combines the above qualitative andnumerical information with the expert humanknowledge stored in the fuzzy logic database tocompute the prosthesis parameter values.

pocket

4. At the end of the parameter tuning procedure, theprogram enables the new parameter values to bedownloaded into the prosthesis control-system

Electrical

memory.motor

Fore ^ - 1 link

Figure 3.Myoelectric prosthesis general layout.

Figure 4.The INAIL myoelectric prosthesis .

METHODS

INAIL Prosthesis CharacteristicsThe INAIL myoelectric prosthesis considered (see

Figures 3 and 4) is a new generation multifunctionprosthesis with three degrees of freedom, correspondingto the opening/closing movement of the hand, wristpitch, and elbow flexion/extension (15) . As the controlof these degrees of freedom differs only in minordetails, we concentrate on the description of handcontrol ; that is the most crucial in terms of performance.In the more general case (i .e ., when all the functions areenabled), the user may select which degree of freedom(hand/wrist/elbow) is under control:

• Using a double contraction of the EMG drivingmuscles

• Using a third EMG sensor (see Figure 3).

The prosthesis is controlled by surfaceelectromyographic (EMG) signals from remnantmuscles . EMG signals are acquired by skin-surfaceelectrodes and processed by a small custom electronicboard inserted in the prosthesis . The board is governedby the industry-standard microprocessor INTEL87C196KC with input signal analog to digital (A/D)converter and pulse width modulation (PWM) for motor

297

BONIVENTO et at . Tuning of Myoelectric Prostheses

control facilities on chip, that enables low boardspace occupancy . The software parameters are stored inan electrically erasable programmable read-onlymemory (EEPROM), allowing either on-line or off-linetuning.

The system provides two kinds of movement . Thefirst is linear proportional control, in which the fingervelocity is kept proportional to muscle twitch . Thismode allows a fine control of the hand motion, enablingthe wearer to have a "natural" grasp of objects . Thesecond kind of movement corresponds to "on-off"control of the hand, in which the motor moves at fixedvelocity in the presence of a non-zero EMG signal . Thismovement is easily controllable by users throughremnant muscle contraction, but fine grasps cannot beachieved.

Several parameters enable a customization ofprosthesis control in connection to user characteristics(such as muscular tone or amputation level) on whichthe EMG signal magnitude depends, and accordingly,the control action on the motor that actuates the hand.

As each user has his or her own characteristics, an"ad hoc" tuning session is necessary before finaldelivery of the prosthesis . Moreover, it is often neces-sary to repeat the setup procedure during the lifetime ofthe prosthesis, due to changes in environmental condi-tions and wear and tear of mechanical components.

The following subsections introduce the parametersthat mainly characterize the macroscopic prosthesisbehavior and which will be considered in the auto-tuning procedure.

Noise (n0)This value is a measurement of electromagnetic

noise that is coupled to the electrodes . The macroscopiceffect of this noise is an unintentional prosthesismovement ; therefore, the maximum noise value isrecorded, and this value is subtracted from the AIDconverter output in order to eliminate such undesirablebehavior . In practice, the expert operator manually setsnO equal to zero, then asks the wearer to keep theresidual limb muscle that drives the prosthesis motion-less . The expected EMG signal should be null and,therefore, a non-null measured value has to be added tonO .

The MCA Auto Tuning program performs thesame task: first the program user interface instructs theclient to remain motionless, then the software acquiresthe EMG signals values and sets nO equal to a proper

value based on the EMG signal maximum, which isnothing other than the electromagnetic noise.

It is very important to set nO equal to a propervalue, otherwise the prosthesis could vibrate even if theclient remains motionless, and the motion control couldbe difficult.

Inactivity Threshold (1)

The 1 parameter consists of an EMG signalthreshold below which the acquired signal is notprocessed . Generally, owing to physiological reasons,there is a cross-influence between the muscles on whichthe EMG electrodes are attached and the neighboringmuscles that may generate spurious EMG signals, andtherefore, unwanted movements.

This spurious signal, also known as physiologynoise, should be distinguished from noise nO, becausethe first is related to an unwanted cross-relation betweenneighboring muscles, while the second is related mainlyto electromagnetic noise or unwanted contraction of themuscle that directly drives the prosthesis.

Usually, this spurious signal is somewhat lowerthan that generated by intentional muscle contraction;therefore, the solution simply consists of using signalthreshold under which the signal is cut off.

In order to tune this parameter, the client is askedto move the muscles that are not attached to EMGsensors, then the MCA software computes the I valuedepending on the magnitude of the correspondingspurious signals . The details of this procedure, which isbased on fuzzy logic, are explained in the next section.

Minimum Threshold (m)

The m parameter represents the minimum value ofenergy required to move the electric motor . Due tomotor or system friction asymmetry, two m thresholds(opening and closing) have been used.

In other words, these parameters are an "offset"on the PWM motor control signal that compensate forthe dry friction of the whole system (covering glove andmechanical resistance) . Without these parameters, aconsiderable effort would be required by the user toeffectively move the hand . These parameters are evalu-ated by means of a step-by-step procedure for openingand closing actions . In this procedure, the m value (e .g .,the opening one) is progressively increased until theprosthesis moves . During this operation, the prosthesisworks in "minimum modality," in which the userdrives only the movement direction and the system

298


calculates the m parameter . The MCA visual interfacehelps the user to set the m proper value through slidingbars, after informing him/her about the movementshe/she has to achieve.

Maximum Threshold (M)

The M value assigns the upper power limit abovewhich the motor gives the maximum power value . Thethreshold enables the user to reach the maximumprosthesis velocity even if the signal is weak . In order tocalculate this value, the software acquires the user'sEMG signals generated during maximum effort condi-tions.

Extensor Gain (E) and Flexor Gain(F)

These parameters assign the gains to be applied toextensor and flexor signals, respectively, and are used toscale out the EMG signals so that similar opening andclosing actions of the prosthesis are achieved withapproximately equal muscle-contraction magnitude.

In order to set these parameters, the client is askedto extend and, subsequently, to bend the prosthesis drivemuscle with the same intensity . The MCA software thenprocesses the signals acquired, using a fuzzy algorithmto increase the gain corresponding to the lower signal,to finally obtain equal behavior in the extension andflexion actions.

The Auto-Tuning SystemThe software package has been developed using

Visual Basic 4 .0 programming language for Windows95 operating system. Visual Basic is an object-orientedprogramming language and an event-driven language.Other features in the graphic management enable theprogrammer to easily build user-friendly applications.

The system is hosted by a Personal Computer (PC).The minimal system requirements are an 80386 proces-sor (Pentium is suggested), 4 Mbytes of memory (16Mbytes suggested), 20 Mbytes free on the PC hard disk,with one RS232-serial port free.

The PC is connected to the prosthesis controlsystem through the serial line . The connection has beentested for a 2-meter cable connector, which enables thesubject to sit comfortably in front of the PC screen andto perform the required movements with the prosthesis.

The system is easy to use and enables the client to

deal with the prosthesis setup procedure without helpfrom an expert operator . The package is also useful tothe expert technician, who can tune the prosthesis

parameters in less time than it would take if it weredone manually, and with evident economic advantages.

Congruity checking of the parameters is performedboth after the automatic setup procedure (off-line) andduring usual prosthesis working (on-line) . If one ormore parameters fall outside a validity range, theoperator is informed that an error has occurred. Thesetup procedure also provides the "load default param-eters" function, which allows the operator to loadsystem memory with a default set of parameters thatassures minimal prosthesis performance.

The MCA Auto Tuning software system is madeup of two parts: one deals with general variables andprocessing routines; the second with the graphic-interface management . The general logical flow isrepresented in Figure 5, while Figure 6 shows theinteraction between program windows.

In particular, the global blocks refer to FuzzyLogic system implementation and PC-microcontrollerprotocol management, respectively . The most relevantroutines included in the first block are as follows : 1)Define Membership Functions : it allows Gaussianfuzzy sets (mean values and standard deviations) to bedefined, uniformly distributed from minimum to maxi-mum value representing linguistic variables (FLC inputand output) ; 2) Infer : it implements the input/outputFLC relation between input and output.

The most important routine included in the secondblock is Acquire, which carries out acquisition frommicrocontroller to PC . The most innovative part of thesoftware is the one relating to Fuzzy Logic structure,which implements the Fuzzy Expert System. This expertsystem is called by two modules dedicated to calculat-ing inactivity threshold I, and gains E and F, respec-tively.

Inactivity Threshold Fuzzy Logic TunerThe Inactivity Threshold Fuzzy Logic Tuner is a

single-input, single-output mapping : its input is a valueresulting from spurious signal acquisition and filteringprocess while the output is the increment to be assignedto the parameter value.

The entity of the spurious signal and the incrementare coded in the linguistic terms "very weak,""weak," "medium," "strong," and "very strong ."These linguistic terms are defined using a fuzzy set withGaussian-shaped membership functions (see Figure 7).

Similarly, the output variable (increment) is repre-sented by five Gaussian fuzzy sets, which are labeled as"very small," "small," "medium," "big," and "very

299


Start WindowSoftware

producersInformation

SetupWindow

SystCmset up

:.V.: ..ae{

State ofprosthesis

Figure 5.MCA Auto Tuning logical flow.

big," respectively (see Figure 8) . The Fuzzy ExpertSystem is based on a set of rules that represents expertoperator knowledge (Table 1).

Extensor Gain and Flexor Gain Fuzzy Logic Tuner

The Fuzzy Logic system implemented to tune Eand F gains has two inputs, the EMG signals e and frespectively, and one output : the increment to beassigned to the gain corresponding to lower signals sothat the final signals have similar magnitudes. Every

Store intoEEPROM

Figure 6.The MCA Auto Tuning graphical interface windows.

input variable was described by three fuzzy sets (Figure9) .

Output variable (increment) is represented by fiveGaussian fuzzy sets, which are labeled as "very small,""small," "medium," "big," and "very big," respec-tively (Figure 10) . The set of rules that resulted fromexpert operator knowledge is presented in Table 2,while the fuzzy control mapping surface obtained ispresented in Figure 11.

Parameter CouplingSome of the above-described control parameters

are coupled; therefore, the setup procedure has beencarefully designed to minimize such interactions.

The first parameter considered in the MCA soft-ware procedure is the noise nO because it is the "free"property (every other parameter depends on it, while itis independent of all of them) . After evaluating thenoise nO, it is possible to acquire the noise-free EMGsignals and then evaluate the Minimum threshold m andInactivity threshold I parameters . After m and I arecomputed, it is possible to calculate the E and F gainsto equalize extensor and flexor signals . After the signalequalizing, it is then possible to acquire signalscorresponding to sustained forceful contractions andthen calculate the optimal Maximum threshold.

Software TestingThe testing was carried out in three different ways:

by using a proper electrical circuit to simulate constant

300


very1weak weak medium strong1

I' '

( ./~ ®

f

%

a0.8

0.6

.ri 0.4

0.2

1 .2

2 .5

3 .7

5 Volts0

. . .

;`"0

very strongr

3

Figure 8.Membership function degree to represent inactivity increment(output variable) .

1 .2

2.5

3 .7

5 Volts

EMG signal

Figure 9.Membership function degree to represent EMG signals (inputvariables).

Figure 10.Membership function degree to represent increment (output vari-ables).

Spurious s i g n a l

m degree to represent spurious signal (input

big

146

292 mVIncrement

Figure 7.Membership funcvariable) .

ur

i1f i

jit

40

60

80

100 {GalIncrement

0

EMG signals, then placing skin-surface electrodes on anINAIL technician and, finally, by carrying out experi-mental trials with prosthesis users.

First, a dedicated electrical circuit was used inorder to simulate constant EMG signals . The resultswere satisfactory except in the case of E and F gains.The incorrect behavior is evident when EMG signals aresimilar and very high because the expected output iszero while the real output is different from zero, asshown in Figure 11. Therefore, the rule database, aswell as the fuzzy sets obtained by expert heuristics, wasrefined during experimental setup . The new system has

five fuzzy terms ("very weak," "weak," "medium,""strong," and "very strong") describing the inputvariables; consequently, the number of rules is increasedfrom 9 to 25 . With respect to Table 2, the rule databasechanges, as shown in Table 3.

This finer granularity gave good results during thesoftware test . The new mapping surface of the fuzzytuner is shown in Figure 12.

The most important experimental test was thatrealized by using the same electrodes that are used inthe prosthesis itself. The results from a test performedonan INAIL technician are shown in Table 4, in which

301


Figure 11.Control law surface of E and F gains.

the comparison between manual tuning and automatictuning is shown. Differences between automatic andmanual tuning are comparable with those achieved withmanual tuning by different INAIL operators ; therefore,these results are to be interpreted as positive perfor-mance of the system.

Further experiments on subjects are shown in theAppendix . The comprehensive results show good per-formance of the MCA software in the auto tuning of theprosthesis control parameters .

Table 1.

Fuzzy rule set for the inactivity Fuzzy Login Tuner.

If

Then

1. spurious signal is very weak increment is very small2. spurious signal is weak

increment is small3. spurious signal is medium

increment is medium4. spurious signal is strong

increment is big5. spurious signal is very strong increment is very big

CONCLUSIONS

The software package presented (MCA Auto Tun-ing System) is a tool that may be useful for both expertoperators and persons with amputation . The experttechnicians may exploit the user-friendly environmentprovided by the MCA software in their work, whileclients are no longer required to come to the prosthesismanufacturers' laboratories for control system setup.

Moreover, the MCA software project will becarried on toward directions other than those sketchedhere. First, the software will enable the acquisition of asoftware database in which every prosthesis workingcondition, notably the user amputation level, will berelated to "optimal" prosthesis control parameters . Thisdatabase may be used for factory parameter initializa-tion before performing prosthesis setup, allowing a timereduction in the tuning procedure.

Another direction of work will consist in thedistribution of MCA software to all persons usingtele-assistance services on data-transmission networks,in order to decrease mobility needs and assistance costs.

Although the most relevant control parameters aretreated in the auto-tuning procedure, there are secondaryparameters that are not considered in the MCA soft-ware. A further improvement of the project will be theextension of the automatic tuning procedure to thoseminor parameters.

As a final comment, one can note that the area ofmyoelectric prostheses, and in particular myoelectrichands, appears rather active both for new controlapproaches (16,17) and advanced robot interface appli-cations (18,19). This may be very interesting inconsideration of technology transfer and knowledgeintegration between heterogeneous research areas.

ImNamirmiu.

+•+.+ / / /s!!/ / /H/ /U

1wow

•":=•I

R

%i,,

IN,t\\~% ,),

3.2

.6

vv.

f signal

0 0

Figure 12.Modified control law surface of E and F gains FLC .

5 Volts

(Gain)100.

80..

d 60.

40.

20 ..

05 Volts

3 .26

e signal

5 Volts

302


Table 2.

Fuzzy rule set for the Extensor and Flexor gains fuzzy logic tuner.

If

And

Then

1. e signal is weak2. e signal is weak3. e signal is weak4. e signal is medium5. e signal is medium6. e signal is medium7. e signal is strong8. e signal is strong9. e signal is strong

f signal is weakf signal is mediumf signal is strongf signal is weakf signal is mediumf signal is strongf signal is weakf signal is mediumf signal is strong

increment is very smallincrement is smallincrement is very bigincrement is smallincrement is very smallincrement is bigincrement is very bigincrement is bigincrement is very small

Table 3.

Fuzzy rule set for the improved fuzzy logic tuner of the Flexor and Extensor gains.

If

And

Then

1. e signal is very weak2. e signal is very weak3. e signal is very weak4. e signal is very weak5. e signal is very weak6. e signal is weak7. e signal is weak8. e signal is weak9. e signal is weak

10. e signal is weak11. e signal is medium12. e signal is medium13. e signal is medium14. e signal is medium15. e signal is medium16. e signal is strong17. e signal is strong18. e signal is strong19. e signal is strong20. e signal is strong21. e signal is very strong22. e signal is very strong23. e signal is very strong24. e signal is very strong25. e signal is very strong

f signal is very weakf signal is weakf signal is mediumf signal is strongf signal is very strongf signal is very weakf signal is weakf signal is mediumf signal is strongf signal is very strongf signal is very weakf signal is weakf signal is mediumf signal is strongf signal is very strongf signal is very weakf signal is weakf signal is mediumf signal is strongf signal is very strongf signal is very weakf signal is weakf signal is mediumfsignal is strongfsignal is very strong

increment is very smallincrement is smallincrement is mediumincrement is bigincrement is very bigincrement is mediumincrement is smallincrement is smallincrement is mediumincrement is bigincrement is mediumincrement is smallincrement is very smallincrement is smallincrement is very smallincrement is bigincrement is mediumincrement is smallincrement is very smallincrement is smallincrement is very bigincrement is bigincrement is mediumincrement is smallincrement is very small

Table 4.Parameter values resulting from automatic and manualtuning by using electrodes.

Parameter Manual Tuning Automatic Tuning

no 005 01.4m 100 1.31I 25 20E 100 64F 64 86M 120 147

303

BONIVENTO et at . Tuning of Myoelectric Prostheses

APPENDIXExperimental Results

In this section, we introduce experimental results acquired during a test session on clients at INAIL centre.The results from the first experiment are shown in Table A-l . As the test was definitely the first one performed on a

subject, automatic tuning was executed twice . The difference between the two automatic runs is negligible, and the absolutedifference with respect to manual tuning is considered to be quite good, and close to the difference between the manual tuningby two technicians operating on the same person.

After this first experiment, some more methodical tests were performed on other INAIL centre clients . The results fromthose tests are shown in Tables A-2-A-5. The MCA system has been considered quite good in both performance and reliability.

Table A-1 .

Table A-3.Automatic tuning and manual tuning test on subject with

Subject B.amputation (first case).

ParameterManualTuning

AutomaticTuning 1

AutomaticTuning 2

n0 001 003 011m 100 99 83I 30 31 31E 255 255 255F 128 128 128M 150 230 230

Table A-2.Subject A.

Parameter

Manual Tuning

Automatic Tuning

000

000

166

165013

016064

071064

064100

102

Age : 17 . Sex : Female . Date 8/5/97 . Kind of accident : Congenitalmalformation . Amputation level : third distal humerus .


n0 003 005m 166 165I 025 031E 064 084F 064 064M 130 144

Age : 15 . Sex : Female . Date 8/5/97 . Kind of accident: Domestic accidentwhen aged 2 . Amputation level : Transradial-third proximal . Note : Thesubject sometimes confuses the muscles she should contract to drive theprosthesis . For this reason, the I value for manual tuning is kept lower thanusual, which is the one given by the automated procedure.

Table A-4.Subject C.


n0 005 005m 166 165I 015 016E 085 090F 064 064M 120 133

Age : 28 . Sex : Male . Date 8/27/97 . Kind of accident : Accident when aged25 . Amputation level : Transradial-third distal.

Table A-5.Subject D.


n0 001 001m 166 165I 015 019E 064 064F 085 095M 120 122

Age : N .A. Sex : Male . Date 9/8/97 . Kind of accident : Accident.

n0mIE

F

M

304


REFERENCES

1. Hogan N . A review of the methods of processing EMG foruse as a proportional control signal . Biomed Eng 1976;March :81-6.

2. Kreifeldt J . Signal versus noise characteristics of filteredEMG used as a control source . IEEE Trans Biomed Eng1971 ;BME-18 :(Jan):16-22.

3. Hogan N, Mann R . Myoelectric signal processing : optimalestimation applied to electromyography . IEEE Trans BiomedEng 1980 ;BME-27 :(July) :382-410.

4. DeLuca C . Physiology and mathematics of myoelectricsignals . IEEE Trans Biomed Eng 1979 ;BME-26 :(June) :313-25.

5. Mariani M, Gelati G . Storia della protesi . In: Note sultrattamento riabilitativo medico-ortopedico e tecnico-ortopedico delle amputazioni e delle malformazioni degli artiinferiori e superiori . Bologna : Tipografia Editoriale; 1984.

6. Nader M. The substitution of missing hands with myoelectricprostheses . Clin Orthop Related Res 1990 ;Sept :9-17.

7. Schmidl H . The INAIL-CECA prostheses . Bologna : CentroProtesi INAIL di Budrio ; 1983.

8. Brighetti U . Sistema di teleassistenza per protesi mioelettriche(Master's thesis) . DEIS, University of Bologna, Bologna,Italy, 1995.

9. Terenzi S . Sistema per it tuning automatico di protesimioelettriche basato su tecniche di logica fuzzy (Master'sthesis) . DIES, University of Bologna, Bologna, Italy, 1996.

10. Lovely D, Stocker D, Scott R. A computer aided myoelectrictraining system for young upper limb amputees . JMicrocomput Appl 1990 ;13 :245—59 .

Klir GJ, Yuan B . Fuzzy sets and fuzzy logic : theory andapplications. Englewood Cliffs, NJ : Prentice-Hall ; 1995.Posthoff C, Sonntag P. Fuzzy methods in expert systems forconfiguration and control . Proceedings of the Third IEEEInternational Conference on Fuzzy Systems . Orlando, FL;1994.Fathi M, Lambrecht M. Ebflatsy : a fuzzy logic system tocalculate and optimize parameters for an electron beamwelding machine . Fuzzy Sets Syst 1995 ;69 :3—13.Wei W, Mendel JM . A fuzzy classifier that uses both crispsamples and linguistic knowledge . Proceedings of the ThirdIEEE International Conference on Fuzzy Systems . Orlando,FL ; 1994.Davalli A. MCA3 version 3 .1-microprocessor controlledarm—user manual . Bologna: INAIL Centro Protesi Ricerca eDocumentazione; 1996.Kelly M, Parker PA, Scott RN, et al . The application of neuralnetworks to myoelectric signal analysis : a preliminary study.IEEE Trans Biomed Eng 1990;BME-37 :(Mar) :221-30.Hudgins B, Parker PA, Scott RN, et al. A new strategy formultifunction myoelectric control . IEEE Trans Biomed Eng1993 ;BME-40 : (Jan) :82-94.Farry K, Walker I, Baraniuk R . Myoelectric teleoperation of acomplex robotic hand . IEEE Trans Robotics Autom 1996.Bonivento C, Melchiorri C. Towards dexterous manipulationwith the UB Hand II . 12th IFAC Congress; Sydney, Australia.Invited paper, session Fr-A-7 ; July, 1993.

Submitted for publication March 3, 1997 . Accepted in revisedform October 29, 1997.

11.

12.

13.

14.

1.5.

16.

17.

18.

19 .

Automatic tuning of myoelectric prostheses

Documents