Anificial Hand Gripper Controller via Smart Glovefor Reh abilitation Proces s

A. M. Mohd Ali, R. Ambar, M. M. Abdul JamilDepartment of Electronic Engineering, Faculty of Electrical and

Electronic Engineering, Modeling and Simulation Research

Laboratory. University Tun Hussein Onn MalaysiaParit Raja, Batu Pahat, 86400 Johor.Malaysia

Email : mahadi@uthm.edu.my

Abstracl-Most parts in this research are dedicated to control ofmulti finger grippers with emphasis on the finger tips or fingerjoints. By controlling a multi finger gripper, vye enable thegripper to handle an object; in another words, controlling a multifinger gripper can be viewed in terms of controlling an object'spose and the forces between the object and its environment.Hence, an object pose controller with feedback from an objectpose sensor suits multi finger gripper control. Also due to thenon-linear dynamic system behavior in the joints of most multifinger grippers, an effective, easily-adaptable joint controller isemployed. The paper discusses the object pose controller withgreat detail in a new joint controller. Since the joint controller is

based on microcontroller thus, we do not use an exact analyticalmodel for this case.

Keywords- Rehabilitation; Flex Sensor; Assistive Device; ForceSensorl Monitoring Device.

I. INTnopUCTION

The technology developments since 70's era until now are

rapidly changing the robotic hand engineering history t U.Currently, existing robotic hand can be divided into four typeswhich is; Robot hands of 80's, Commercial hands, Researchhands and Prosthetics. Development of robot hands early inthe 80's start with, Soft gripper Hirose Soft Gripper by ShigeoHirose from Tokyo Inst. Technology. This development beganin the late 70's with I DOF when it utilized pulleys at jointsand create evenly distributed forces t2l. Then, in 80's,Tomovic and Bekey pioneering effort in development of firstrobotic hand prototype called Belgrade / USC hand afterWorld War II which consists of four DOF ( I for each pair offingers and two for thumb) t3l.

In the same era, more development and researches done inthis field to upgrade the prototype technology. For exampleStanford/JPl hand prototype with nine DOF design. Featuresuch as four tendons and finger also were designed forfingertip manipulation which is combined with strain gaugefingertip sensors [4]. In another study, a new prosthetic handis being tested at the Orthopedic University Hospital inHeidelberg Grip which functions almost like a natural hand. Itcan hold a credit card, use a keyboard with the index finger,

20 I 2 International Conference on Biomedical Engineering (ICoBE),Penan g,Malaysia,21 -28 February 201 2

A. J. Mohd Wahi, S. SalimUniversity Kuala Lumpur-British Malaysia Institute

Kolej Kemahiran Mara Ledang.Batu 8, Jln Sg Pusu, 53100, Selangor.

Email : jee-5363@)zahoo.co.uk

and lift a bag weighing up to 20kg. It's the world's firstcommercially available prosthetic hand that can move eachfinger separately and has an outstanding range of gripconfigurations. The construction artificial hand gripper'sindividual finger can be quickly removed by simply removingone screw. Thus, the developed prosthetics can easily swapout fingers which require servicing and therefore patients canreturn to their everyday lives after a short visit to the clinic [5-6]. Banett hand from Barrett Technology Incorporated used 4

motors, one motor per finger for three finger and plus anotherspread motor for palm. The breakaway technology allowsfingers to adapt to object geometry. It's also including theoptical encoder for position sensing. Fischer et. al. t7l andHong et. al. t8] presented three fingered, multi jointed robotgripper for experimental use. Results of the control of thegripper on joint level, the Cartesian behavior of the fingers andsome experiences with the grasping and manipulationexperiments using the presented system were reported.Basically, prosthetic hand gripper can be used to develop and

evaluate different approaches of stable grasping and objectmanipulation. The gripper system provides the basic means interms of position and force control to perform experimentsabout grasping and object motion in a useful way.

Hence, the research and development in this disciplineincreased and more prototypes being commercialized due to

l::1" IJTTj "H,i:l;lffi:,1l J, fi ffi: T'" :','fl;'i; ffi'Jllproj ect I I 0- l2l. Further proj ect were done to develop themechanical parts of the project based on a master-slave system

[4]. This artificial limb looks and acts like areal human handand consists of a Smart Glove and an artificial hand gripper.We also utilized this prototype for mirror visual feedback(MVF) therapy in assisting paralyzed patients rehabilitationprocess U 31. This will be followed in this study for thedevelopment of an enhanced version of artificial hand gripper,where a capability of detecting pressure on each finger ispresented. The aim of this research is to assist handicapindividual in providing them with an enhanced version ofprosthetics hand that can be used in rehabilitation process,which is economical and affordable.

97 8-4577 -1991-21 12l$.26.00 @20 I I IEEE

%||tnrnn f,ryttrp posrtlons

II. MnrHon

The development of this project based on master-slavesystem which involves integrating a Smart Glove and an

artificial multi-finger hand gripper as shown on Figure I Thehand gripper will mimic the actual finger movement made bya human operator through the Smart Glove.

There are two different sensors used in this project, i.e.flex sensors and flexi-force sensors which were attached to theglove. The former is meant for resembling the movement offinger's joint, while the latter is for controlling the pressure ofthe gripper. These sensors and the gripper are then wired to an

Arduino [INO; a microcontroller board based on theATmega3z8. The data from the Smart Glove will be fed intothe UNO, processed, and then channeled to the hand gripper.

From the experiment done in [0],[l] and [3], when flexsensor is bend inward, resistance value increased significantlyas the angle of flex sensor is bend further. However, when it isbent outward, the resistance value decreased gradually. Thesepreliminary finding suggest that flex sensor is clearly suitableto detect finger bending angle by utili zing inward bend of theflex sensor.

Figure 2. Figures showing how flex sensors assembled on the back of SmartGlove.

Figure 3. Figure showing flexi-force sensors attached on the Smart Glove.

B. Flexi-force Sensor

In this paper, flexi-force sensors were attached on each ofthe glove fingertips as shown in Figure 3. Flexi-force sensor is

suitable to measure pressure force between body and externalsurfaces. By attaching the sensor to glove's fingertips, theanalysis of force distribution applied on the fingertips can be

done.However, preliminary experiments on Flexi-force sensor

have been carried out to determine the suitability of the sensoron detecting small forces. Studies shows that force sensitivesensors can be used to detect and measures small amount offorces including individual muscles.

l"--^-'-^'^".-*t.:L'

, oostltsn ii

i ,l^o^ i. i':-'*t ti.

tI

''

i Cottitditt

i l,lrctupocermt

j Af;ttuiti0"iJti0 T,**-ryh+l

&t

porrxtopcintrmpprrq

i

IIt

tr

b*dng dh',rnan I'nnd

Figure l. Block diagram showing the overall system of proposed Smart Glovecontroller for artificial hand gripper.

The most important part after assembling the hardware isto program the microcontroller. The program was written in C.

It needs to be tuned accordingly so that the gripper will mimicthe finger movement. The methodology flowchart of thisproject is shown as in figure l. The followings are theexplanation regarding the flex sensors and flexi-force sensors.

A. Flex Sensor

In this project we use two type of sensor to measure thecapabilities of the smart glove (Master). It is flex sensor andforce sensor. The flex sensors were attached on back of theSmart glove, as show in figure 2. The Smart Gloveincorporates a sensory system which can detect finger flexion,hence the name Smart Glove was given to this hand glove.The sensor is connected to Arduino microcontroller for analogsignal detection.

To read the sensor, its variable resistance is converted to avariable voltage and amplified with an op-amp. Then, theanalog signal is transmitted to the l0 bits ND converter in themicrocontroller side for data processing. The data will be usedto control servo motors which move the artificial handgripper's fingers. Each of the hand gripper's fingers will movesimilarly to the flexion of flex sensor attached on the SmartGlove. The angle is measured between the two tangent lines atthe ends of the flex sensor's body. The flex sensor has atypical electrical resistance variation when flexed or bent.

Drspiay ftatelorm

F,l t]Stt{

fivpertffrntnai

*-'^lbHable am!0rce

i

i mappillg

III. RgsuIr & DIScUSSIoN

A. Flex sensor

This paper consists of several experiments for flex sensor.

The characteristics of the sensor will be recorded by usingHyperTerminal software. The capability of flex sensor tofollow the bending of hand finger is very good. Flex sensors

were attached on the back of the Smart Glove as shown onfigure 4 to detect human operator's finger bending activity.The sensor is connected to a microcontroller for analog signaldetection. Two types of data were recorded, the first is tomeasure a flex sensor manually using a multi-meter and the

second is though the data recorded from flex sensors attached

on the Smart Glove through HyperTerminal. Flex sensorpresents lOK Ohm resistance value on zero degrees bendingand about 35K Ohm on 90o bending. The experiment belowshows four type of bending for a finger. They have fourposition of bending 0o, 30", 90o and 120".

00

c) 90" d) 12goFigure 4. A Flex sensors attached on the back of Smart Glove and the activity

of bending at different angle for Arrow finger.

Graph I and graph 2 shows the results of experiment done

using flex sensor on thumb, arrow finger and middle finger.We applied bending activity, and the graph shows increase ofresistance value when the sensor was bent which suggest thatthe sensor can be used to detect finger bending activity when itis flexed. By attaching this sensor to the back of the glove, theflex sensor will act as a detector which sends data to themicrocontroller to inform about the gripper is grasping an

object.

B. Flexi-Force Sensor

By pressing the active round surface of the flexi-forcesensor, we recorded the analog raw data. By mapping the

analog data (0-1023 unit), we converted it to voltage value(0-5 volt). As shown on figure 5, we did a simple experimentby attaching flexi-force sensors on the thumb and arrowfinger. Then, subject will press the thumb and arrow finger on

a flat surface (example a table surface) to record the data fromforce sensors.

Graph 3 and graph 4 shows the results of experiment doneon the flexi-force sensor. We applied force on the flexi-forcesensor active surface and graph shows increase of resistance

value when the sensors were pressed, which suggest that the

sensor can be used to detect force when pressure applied to itsactive surface. By attaching this sensor to the gripperfingertips, the force sensor will act as a detector which sends

data to the microcontroller to inform about the gripper is

grasping an object. We expected to be able to control the

amount of force generated on the grasped object based on thevoltage value generated from the flexi-force sensors.

n)

Finger Bendi ng Activity

^50E40g30Y20cE, 10'Fotr

-Arrow finger

h*F*i' M i dd I e fi nger

-;."* -ftipg

finser

a$ ffi'*,*Pinky\'

Graph l. Bending activity for thumb, middle finger and arrow finger.

Fr

E

oI

;#*.g€*,,9ul(l,'

e,

50

40

30

?0

10

0

-Thumb

-{y16w finger

* M idd le finger

F0Ee€c{: +tC)c)()c)c)c)ooc)1r{fntnF*s)FlfqLnF*

F{ :{ F{ Fl

Graph 2. Bending activity for thumb and arrow finger.

b) 3oo

{h}Figure 5. The activity of Force sensor: (a) Thumb and (b) Anow, finger

pressing the flex force sensor on the table.

Thumb Finger Force50

45

40

Ess9soV

7,75Ezort!

&1s10

5

0

1 15 7. 3 3.5 4 4.5 5

Graph 3. Picture above showing the activity of thumb finger for the SmartGlove.

Arrow Finger Force

-fi1.19s7

fing,er

3.s L r.5 2 1.5 3 35 4

Graph 4. Picture above showing the activitv br arrow tff#:)f", the SmartGlove.

IV. CONCLUSION

The different kinematic structure of human and robothand requires the implementation of appropriate force andposition. On this account forward kinematics of the humanhand and inverse kinematics of the Hand gtipper were derivedand a position mapping algorithm based on a projection of thehuman fingertip position on the gripper trajectory has beenproposed. The evaluation in first real hardware experimentshowed a good and promising performance of the positionmapping as a variety of different grasp types ranging fromprecision to power grasps can be performed. The quality of theforce feedback is strongly affected by the ma:<imum torquemeasurable by the Hand gripper and the performance of theforce controller.

AcTNowLEDGMENT

The authors would like to take this opportunity to expresshis heartfelt appreciation to his respectful supervisor Dr. M.Mahadi Abdul Jamil for his supervision, encouragement,contradictive ideas, patience, guidance and invaluable advice.

RBTengNCES

t I I A. Wright and M. Stanisic, "Kinematic Mapping between the EXOSHandmaster Exoskeleton and the Utah-MIT Dextrous Hand," in Proc. Ofthe IEEE International Conference on Systems Engineering, 1990, pp.

809-81 l.

l2l Hands Overview Slideshow Slide (2010),: Retreived August 23 ,2010;from : http : I I gr aphics. cs. cmu. e du/n sp/course/ I 6 -89 9 I .

t3l Paetsch,W., Kaneko,M.; Intelligent Robots and Stems'90.'Towards a NewFrontier of Applications', Proceedings. IROS '90. IEEE Internationalworkshop on 3-6 Julai 1990,pp,853-858 vol2.

t4l L. Pao and T. Speeter, "Transformation of Human Hand Positions forRobotic Hand Control," in Proc. of the IEEE International Conference onRobotics and Automation, vol. 3, May 1989, pp. 1758-1763.

t5] S. Ekvall and D. Kragic, "Interactive Grasp Leaming Based on HumanDemonstration," in Proc. of the IEEE International Conference onRobotics and Automation, vol. 4, April, May 2004, pp. 35191524.

t6l J. Aleotti and S. Caselli, "Grasp Recognition in Virtual Reality for RobotPregrasp Planning by Demonstration," in Proc. of IEEE InternationalConference on Robotics and Automation, May 15-19, 2006, pp. 2801-2806.

t7l M. Fischer, P. van der Smagt, and G. Hirzinger, "Learning Techniques ina Dataglove Based Telemanipulation System for the DLR Hand," inProc. of the IEEE International Conference on Robotics and Automation,vol.2, May 1998, pp. 1603-1608.

t8] J. Hong and X. Tan, "Calibrating a VPL DataGlove for Teleoperating theUtahA4IT Hand," in Proc. of the IEEE International Conference onRobotics and Automation, vol. 3, May 1989, pp. 1752-1757.

t9] S. Berman, J. Friedman, and T. Flash, "Object-action Abstraction forTeleoperation," in Proc. of the IEEE International Con/brence onSystems, Man and Cyber netics, vol. 3, October 2005, pp.26311636.

[0] A. M. Mohd Ali, M. Y. Ismail, and M. M. Abdul Jamil, "Development ofArtificial Hand Gripper for Rehabilitation Process", IFMBE proceedings:vol. 35, pp. 785-788, 5th Kuala Lumpur International Conference onBiomedical Engineering (Biomed), Kuala Lumpur, Malaysia, inconjuction with the 8th asian Pacific Conference on Medical and

-J|.,11ry1[1

(time)

III

t

t

50

45

40

3'.tIroEru"f;ro$rt

10

5

o

{n}

fr

Biological Engineering (APCMBE 20ll) 20-23 June 2011, Springer-Verlag Berlin Heidelberg.

[UR. Arnbar, M. S. Ahmad and M. M. AMul Jarnil, "Design andDevelopment of Arm Rehabilitation Monitoring Device", IFMBEproceedings:vo1.35, pp. 781-784, 50' Kuala Lumpur InternationalConference on Biomedical Engineering @iomed) Kuala Lumpur,Malaysia, in conjunction with the 8tl' Asian pacific Conference onMedical and Biological Engineering (APCMBE 201l) 20-23 June 2011,Springer-Verl4g Berlin.

[2]A. Malik Mohd Ali, M. Yusof Ismail, M. Matradi " Development ofArtificial Hand Gripper" ASQED 20ll (Numhr 5l) .The 4th AsiaSymposium on Quality Elecffonic Design (ASQED 2012) Kuala Lumpur,Malaysia. (unpublished).

[3]A. Malik .Mohd Ali, Radzi Bin Ambil, A. Jalaludin M. Wahi, M. MalradiAbdul Jamil Journal of Engineering Packaging , Development ofMaster-Slave Robotic Hand for Minor Visual Feedback (lvfvF)Therapy, Copyright- 20ll by UNIKL BMI - November 20ll,Vol.l,ppl -6.

[4] A. Malik .Mohd Ali, AJalaludin M.Wahi, Radzi Bin Ambar, M. Matradi.AMul Jamil, Journal - International Journal of Integrated Engineering(UE) Development of Artificial Hand Gripper by usingMicroconfioller. Volume 3 No 3, December 2011.