Illusory Sense of Human Touch from a Warm and Soft Artificial Hand

1534-4320 (c) 2013 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission. Seehttp://www.ieee.org/publications_standards/publications/rights/index.html for more information.

This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI10.1109/TNSRE.2014.2360533, IEEE Transactions on Neural Systems and Rehabilitation Engineering

TN

AbdevwhtoupsydemcanfrothiaccAt 28.hasdevhanartsofindpalwitparhanthethatheposare

Ihan

Masupof

Nat JDep DGra YEng M

T

NSRE-2014-00

bstract—To tovelopment, wel

ho have lost thuching becomeychosocial issumonstrate that n be perceived om a human his goal. First, wecording to their

room tempera.4°C at the skins a Shore duroveloped a procend. To compatificial hands, aftness map by rdentation force lmar side of thth skeletal strurticipants’ armnds, but they em. Receiver oat a warm and e touch is frossibilities for pre more socially

Index Terms—nd illusion.

HE human hbones, musc

anuscript received pported in part by Engineering, Qata

tional University oJJ Cabibihan ispartment, Qatar UnD Joshi is with taphic Era UniversiYM Srinivasa andgineering DepartmMA Chan is with

I

John-

T

0086

ouch and be ll-being, and r

heir arms and es a serious ues and socia

the touch fromby another perand. We descre made participr preferred warature, the prefn surface of a someter value ofess to create a

are the skin sa robotic indenrecording the d

of 1 N was aphe hand. Resultucture is as sofms were touchwere preventedperating charasoft artificial

m a human hrosthetic and roacceptable.

Prosthetics, bio

I. INTRO

hand has a cocles, tendons, l

April 2, 2014; accthe Research and ar University and

of Singapore. s with the Mecniversity, Qatar (ethe Electrical andity, India. d A Murugananthament, National Univ

GE Global Resear

IllusorWa

-John CabibihIEEE,

touched are relationships. H

hands due to concern that

al stigma. In m a warm and rson as if the toribe a three-stepants select artirmth and softneferred warmth soft silicone rubf 30 at the OO rubber hand reoftness of a h

nter was emplodisplacement dapplied to 780 dts showed that ft as a human hed with humd to see the h

acteristic curvehand can creathand. These fobotic hands th

omimetics, artif

ODUCTION

omplex anatomligaments, arte

cepted August 25, Graduate Studies

d the Academic R

chanical and Inemail: john.cabibihd Electronics Eng

am are with the Eleversity of Singaporch, Germany.

ry Senarm a

han, Senior MMark Aaron

vital to humHowever, to th

accident or wt often leads

this paper, soft rubber ha

ouch were comep process towaficial skin sampess characterist

was found to bber material tscale. Second,

eplica of a humhuman hand aoyed to producata when constdata points on an artificial hahand. Lastly,

man and artifichand that touch

analysis suggete an illusion t

findings open hat are lifelike a

ficial skin, rubb

my consisting eries, nerves, a

2014. This work Office of the Coll

Research Fund of

dustrial [email protected]). ineering Departm

ectrical and Compore, Singapore

nse of Hand So

Member, IEEEChan, and Ar

man hose war,

to we

and ming

ard ples tics.

be that

we man and ce a tant the

and the cial hed ests that the

and

ber

of and

was lege

f the

ring

ment,

uter

Fig. 1. Aligaments

the protform thoutermothe extlayers bdermis, sweat gis also tactile spalm anside of t The charactestiffens various secondsenough enough and paifear, dcommuncharactefeaturesfingers

Humaoft Art

E, Deepak Joshrrchana Muru

Anatomy of the s, muscles, artery,

tective layer ohe innermost stost protective lternal world [broadly classif

and the hypoglands, and the

densely packsensing [4]. A nd the digits tothe hand.

skin tissue eristics: it eawhen the conshapes of obj

s; it is tough to provide a to soothe som

ins. Through tdisgust, love

unicated to oeristics, it is s for prosthetidue to an accid

an Toutificial

hi, Yeshwin Muganantham, M

human hand. Dand nerves at the v

of the skin [1,tructure of thelayer and acts [3]. It is comfied as the str

odermis. Apart blood vessels

ked with receplayer of subcuo create a cush

of the humasily deforms ntact force becojects, but retur

against extercomforting to

meone else’s phtouch, distinct, gratitude others [5]. G

reasonable tic devices for dent or war.

uch frol Hand

Mysore SrinivMember, IEE

Depiction of the volar side of the ri

, 2] (Figure 1)e hand. The skas the primary

mprised of sevratum corneum

from the papi, the volar sideptors to facilitutaneous fat pahioning effect

man hand hwith slight

omes large; it rns to its origirnal elements,

ouch to a babyhysical and emot emotions sucand sympathGiven these to mimic som

those who lo

om a d

vasa, MemberEE.

1

major tendons, ight hand.

). The bones kin forms the y interface to veral internal m, epidermis, illary ridges, e of the hand tate efficient ds lies in the on the volar

as versatile contact and conforms to

inal shape in yet is soft

y; it is warm otional aches ch as anger, hy can be

remarkable me of those ost hands or

r,

1534-4320 (c) 2013 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission. Seehttp://www.ieee.org/publications_standards/publications/rights/index.html for more information.

This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI10.1109/TNSRE.2014.2360533, IEEE Transactions on Neural Systems and Rehabilitation Engineering

TNSRE-2014-00086

2

To date, significant advances have been made in the recovery of the motor and sensory functions that were lost due to amputation. With a surgical technique called targeted muscle reinnervation, Kuiken et al [6] demonstrated that real-time control of multiple degree of freedom joints for prosthetic arms and hands can be achieved when the nerves from the residual arm are transferred to alternative muscles sites and electromyography signals are recorded by electrodes at the skin surface. Tactile feedback was found to be essential in order to control a prosthetic hand in an intuitive manner. Marasco et al [7] created an artificial sense of touch by coupling a pressure sensor on a prosthetic hand while feedback was achieved through a haptic interface (i.e. tactor), where proportional pressure is applied to stimulate the cutaneous nerves that were redirected to the skin of the residual limb. Self-reported and physiological measures from amputees suggest that a vivid sense of ownership of a prosthetic hand can be created by providing cutaneous tactile feedback. More recently, Raspapovic et al [8] demonstrated that an amputee's motor commands can be simultaneously decoded and sensory feedback can be delivered in real time for an amputee to bidirectionally control a prosthetic hand. From the contact information provided by the tactile sensors at the prosthetic hand, sensory information can be provided to the amputee by stimulating the median and ulnar nerve fascicles using implanted electrodes. For the first time, they showed that a blindfolded amputee can identify stiffness and shape of three different objects only from the tactile information provided by the prosthetic hand. Upper limb loss has been found to dramatically change a person's body image [9-11]. To this end, creating a sensation that a foreign object is a part of the body has been investigated in the so-called Rubber Hand Illusion [12]. This illusion is created by applying synchronized brush strokes to a rubber hand in full view of the participant and to the participant's own hand, which is hidden behind a screen or under a table. Since its discovery, several reports have confirmed that an illusion of touch has been felt by non-amputees [13-15] and amputees [7, 16] alike, with both experimental groups experiencing ownership of the rubber hand. It was explained that the illusion occurs due to the attempt of brain's perceptual systems to interpret visual, tactile, and proprioceptive information resulting into a re-calibration of the location of touch and the felt position of the hand with the result that the touch appears to be felt by the rubber hand [12, 16]. Furthermore, it has been reported that the experience of body ownership applies only to objects that have the same appearance of the body part [17]. It was demonstrated that participants experienced a sense of ownership only for a realistic prosthetic hand and not for a plain wooden block or even a wooden hand. The study suggests that the object being viewed by the participants must fit with a reference model of the body in order to maintain a coherent sense that the object can be a part of the body. Considerable advances have been achieved to make prosthetic hands and fingers indistinguishable from the missing body parts not only in terms of anatomical structure [18, 19] but also in the replication of skin tone, pores, and hair [20, 21].

In addition to functional and aesthetic considerations, Murray [22] argued that the usage of prosthesis plays a social role in the lives of amputees—prosthesis use can ward off social stigmatization. This finding was corroborated by Ritchie et al [23] where they found that a prosthesis can help upper limb amputees cope, to feel normal again, and “not to stand out”. Limb loss and prosthesis usage can alter a person's social life and the quality of social interactions [24-26]. All these findings are of particular importance as they provide evidence on the possibility of satisfying the functional, body image, aesthetic, and social requirements of those who lost a hand. What remains unknown, however, is whether an artificial hand would feel realistic to the person being touched. In the present article, we consider a scenario where the user of a prosthetic hand touches another person. We ask whether experimental participants could feel that the touch from an artificial hand with warm and soft characteristics feels like touch coming from a human hand. We describe a three-step process to investigate this. First, we present the selection process for warm and soft synthetic skins that participants preferred to have on an artificial hand. We then show how we included these features into a replica of a human hand. Finally, we describe an experiment wherein participants were touched with a human hand and artificial hands, but they were prevented from seeing the hand that touched them.

II. PERCEPTION EXPERIMENTS WITH ARTIFICIAL SKIN

SAMPLES

A. Participants

A total of 165 healthy subjects (95 males, 70 females, all 17-28 years old) were recruited from the National University of Singapore. Participation was voluntary.

B. Experimental design

To determine the desired thermal and mechanical characteristics of an artificial hand, we asked participants to touch various skin samples and select a sample that they felt similar to human skin tissue. Participants were free to choose their own exploration strategy. They were not limited on the time to make a decision. The skin samples were arranged to have a gradiation of soft and warm features and were laid out in a 4×4 array (Figure 2a). We designed each row to have a temperature gradient of 5°C at the sample’s surface (i.e. 22, 27, 32 and 37°C). We wanted to determine the temperatures because we do not know which temperature is preferred by the subjects for a lifelike prosthetic hand. This is especially important because the room temperature has an effect on the skin temperature of the participant’s hand [27]. For the columns, we selected 4 different materials of increasing softness. The details of the materials selected are provided in Table I. The lower Shore durometer value corresponds to a softer material. The materials were selected based on their usage in earlier works on prosthetic or robotic skins [28, 29], embedding materials for tactile sensors [30, 31] and for soft robotics [32, 33]. All the skin samples were fabricated using standard moulding techniques. They were identically colored to remove

1534-4320 (c) 2013 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission. Seehttp://www.ieee.org/publications_standards/publications/rights/index.html for more information.

This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI10.1109/TNSRE.2014.2360533, IEEE Transactions on Neural Systems and Rehabilitation Engineering

TN

anypigwitthe

Eco

Pro

Dra

Pol(PD

C.

Figtembaswitwitglumato samtemthebe

Fig4×4andarti

tem

EprotemimsystemtemaccTheman effUSvo

NSRE-2014-00

y visual bias. Tgment (Silc Pigth silicone paie pigment to ad

TABLE I

Material Name

oflex, OO-30

ochima, GLS 40

agon Skin, series 2

lydimethylsiloxaneDMS), Sylgard 184

Design of the

gure 2b showmperature sensse. The polyimth dimensions th maximum cued on extruaterial thermall

its low thermmples were gmperature wasermometer in fabout 50-60%

g. 2. Touch experim4 array with columd rows arranged aificial fingertip sk

mperature sensor m

Each sample oportional comperatures of

mplemented in stem was integmperature of tmperature senscuracy ±0.5°C

he sensor wamploying it on t

on-and-off mafect transistor (SA). A pull-ultages of the

0086

The samples wg, Smooth-On, nt base (Psych

dhere properly.

. DETAILS OF TH

Manufact

Smooth-On,

Prochima, It

20 Smooth-On,

e 4

Dow CorninUSA

e Embedded He

ws the assembsor at the surfamide heater (Kof 6.35×25.4×current rating

uded polystyrely insulated themal conductiv

given 3.9 W os maintained free air. The rel

%.

ments on artificiamns arranged accorccording to increakin sample consi

mounted on the skin

employed a ontroller in o

22, 27, 32 ana microcont

grated to it (Figthe fingertip ssor (LM335, N

C). The sensor as calibrated the circuit. Theanner using me(MOSFET; IRF

up resistor wae microcontro

were colored usUSA). The pig

ho Paint, Smoo.

HE MATERIALS

turer Shore ODurome

Numb, USA 30

taly 55

, USA 70

ng, 86

Heating

bly of a skinace and a heatinKapton heater×0.25 mm3, has

of 3 A. The ene foam (Se samples fromvity of 0.03 Wof power. That 21°C, as mlative humidity

l skin samples. (arding to materials asing warmth. (b) isting of an emb

n surface.

hybrid on-offorder to reand 37°C. The roller and a

gure 3a). The inamples were National Semiwas glued to to room tem

e heater was metal–oxide–semF630, Fairchild

as also used foller and the

sing a flesh-tongment was mixoth-On, USA)

S SELECTED OO eter

ber

MaterialDesignatio

A

B

C

D

n sample withng element at s, Minco, USAs 7.1 Ω resistan16 samples w

Styrofoam). Tm one another dW/m·K. All

he ambient romeasured withy was recorded

a) Samples laid ouin increasing softnCutaway view of

bedded heater an

ff scheme andach the desi

algorithms wdata acquisit

nitial skin surfameasured withconductor, USthe skin surfa

mperature befmade to operatemiconductor fied Semiconductfor matching e MOSFET.

ned xed for

on

h a the A), nce

were This due the om h a d to

ut in ness f an

nd a

d a red

were ion

face h a SA; ace. fore e in eld-tor, the A

microcointegratwere usamplesrepresenthermocwhich wdesign fingertip

Fig. 3. Alalgorithmtemperatuskin surfa

D. The

There wwhen thorder tothe fouanalyzemeasuredensity,data she

. Taspecific

ontroller (Atmeted developmensed to maintas to the desirentative readingcouple-based was attached towas then indp samples.

lgorithm for the emm used for controures of 22, 27, 32 aace temperatures re

ermal Characte

will be a chanhe human fingeo calculate thaur materials wer (TCi, C-Ted thermal co, ρ, of each maeets. Specific

able II gives c heat.

ega 2560, Atmnt board (Ardu

ain the tempered temperaturegs collected fmultimeter (U

o the surface ofdependently im

mbedded heating aolling the heatinand 37°C for eachesulting from the h

erization of the

nge in temperaertip touches that temperature,were required.

Therm Technoonductivity, katerial was obtheat, c, was tthe thermal c

mel, USA) couuino, Smart Prrature of the s. Shown in F

from the data U1252A, Agif each material

mplemented on

and representativeng system to reah of the material saheating system des

e Materials

ature at the conhe skin materia, the thermal p. A thermal ologies, Canak, and effusivained from thethen calculatedconductivity, d

3

upled with an rojects, Italy) 16 fingertip

Figure 3b are logger of a

ilent, USA), l. The circuit n all the 16

e results. (a) The ach the desired ample. (b) Initial sign.

ntact surface al sample. In properties of conductivity da) directly vity, e. The eir respective d as /density, and

1534-4320 (c) 2013 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission. Seehttp://www.ieee.org/publications_standards/publications/rights/index.html for more information.

This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI10.1109/TNSRE.2014.2360533, IEEE Transactions on Neural Systems and Rehabilitation Engineering

TN

A (

B (P

C (D

D (

E.

Thme20 30 thopartemtemthe(mthefortemsuc28andthe

F.

ThChsof32surwhpar= 327tem(Ecprepro heaUptraconchaandcalto

NSRE-2014-00

TABLE II. TH

Material C

Ecoflex)

Prochima)

Dragon Skin)

PDMS)

Initial Human

he initial skineasured as they

and 30 minutsubjects (15 m

ose who particrticipants weremperature was mperatures at te participant’s

model 62 Miermometer wasr all measuremmperature at thcceeding mean.4°C (SD = 1.d 30 minutes, e temperatures

Results

he participants hance for selecftness, 46% of% preferred mrface temperathile 31% selecrticipants’ pref36), material A°C temperatur

mperature resucoflex OO-30)eferred temperoportion of higThe majority’sat transfer propon contact, ansferred to tnduction. The anged dependid artificial filculated the su[34, 35]:

0086

HERMAL PROPE

Thermal Conductivity (k)

(W/m·K)

0.15

0.23

0.16

0.12

n Skin Temper

n temperaturey entered the etes after they emales, 15 femcipated in the e made to sit i

maintained atthe papillary w right hand uini, Fluke, Us maintained a

ments. Upon enhe fingertip wn temperatures9°C) and 27.8respectively. Fat the 10 minu

selected the cting a samplef the participamaterial B. Foture was selectcted the 32°Cferences clusteA with 32°C res (n = 28). ults to 28.4°) was the morature can be

ghly selected tes choice was ccess that occuthe heat fromthe surface osurface tempering on the initngertips and

urface temperat

ERTIES OF THE M

Density (ρ) (kg/m3)

1,065

1,123

1,081

1,040

ratures

es of the pexperiment rooentered. We ra

males, all 18-25skin sample en a relaxed pot 21°C. We mwhorl of the inusing an infraUSA; accuracat a constant dntrance to the

was 28.9°C (SDs were 29.0°C

8°C (SD = 1.8°For the subsequute mark were c

samples showe is 6.25% (1

ants preferred mor temperatureted by 45% o

C temperature.ered on materia(n = 26), andBy proportio

C. In summst preferred so

e set to 28.4°emperatures. consistent withurs when one tom the humanof the artificrature at the intial temperaturtheir thermal

ture of each m

MATERIALS

Specific Heat (c(J/kg·K)

1,558

1,244

1,435

1,610

participants wom, and after andomly selec5 years old) frexperiments. Tosition. The romeasured the s

ndex fingertipared thermomecy ±1°C). T

distance of 22 room, the meD = 2.2°C). TC (SD = 2.0°°C), after 10, uent calculatioconsidered.

wn in Figure out of 16). Fmaterial A whe, the 27°C sf the participa Notice that

al A with 27°Cd material B won, the preferary, material oft material. T°C based on

h the result of ouches a surfan fingertip wial fingertip

nterface of contres of the huml properties. W

material accord

c)

were 10,

cted om

The om kin of

eter The cm ean The C), 20,

ons,

4a. For hile kin

ants the

C (n with

red A

The the

the ace. was

by tact

man We ing

(1)

where the humtemperatemperaconductinitial t(§II.E). fingerpavalues fwere exvalid, tcriterionThe Fou

where characteby the thermal

/to be lethe usetemperaFor the in Figur28.3 an27°C, temperaof the pthe pro28.4°C

Fig. 4. PrThe selecclusteringtemperatutemperatuskin samp

is the resultman and artifature of the hature of the tivity, ρ is dentemperature of

The thermal cad was estimafor k, ρ, and cxperimentally the Fourier nun to be satisfieurier number is

contact time,eristic length, contact area,

l diffusivity o. The Four

ess than 2.1×10e of Eqn. (atures at the comost preferred

re 4a, the calcund 29.9°C for

and materialatures are also participants’ fioportioned tem(§II.F).

referred warm andcted material type ag at material A ure, and materialures at the contacples.

ting surface teficial fingertip

human fingertiartificial fing

nsity, and c is f the participancontact coefficated in [36] toc for each of obtained (Tabumber, , ofed to assume as

, t, was set , defined as

, was calculatof the materirier number fo0-4. The cri(1) is valid. ontact interfacd material andulated skin surmaterial A w

l A with 32closest to the

ingertips of 29mperature was

d soft skin characand initial surface with 27°C temp

l B with 27°C tct interface when

mperature upops, isip, igertip, k is specific heat. nts’ fingertips cient, , oo be 1,181 J/mthe artificial s

ble II). For Eqf less than 5a semi-infinite

to 5 secons material voluted to be 4×1ial, α, was cr each materiaiterion was sat

The calculace are shown id temperature cface temperatu

with 27°C, mat2°C, respectivaverage initial

9.0°C (§II.E). s earlier calcu

cteristics of the sktemperature of th

perature, material temperature. (b) the human finger

4

on contact of s the initial is the initial the thermal The average was 29.0°C

of the human m s½K. The skin material qn. (1) to be ×10-2 is the model [37].

(2)

nds and the ume divided 10-3 m. The alculated as

al was found tisfied. Thus, ated surface in Figure 4b. combinations ures are 28.4, terial B with vely. These l temperature Incidentally,

ulated to be

kin samples. (a) e artificial skins

A with 32°C The calculated

rtip touches the

1534-4320 (c) 2013 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission. Seehttp://www.ieee.org/publications_standards/publications/rights/index.html for more information.

This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI10.1109/TNSRE.2014.2360533, IEEE Transactions on Neural Systems and Rehabilitation Engineering

TN

I

A.

Adatsili9 cbonto histhetheUn

B.

comfroperof hansubstruinvsub

Figsub3D mouthe prinhum

FtheTo

NSRE-2014-00

III. LIFELIKE A

Participant

A 35-year-oldta from his dicone rubber recm in breadth anes, and 19 cmthe tip of the m

s hand’s measue anthropometre indentation niversity of Sin

Experimental

Two designs ompare the soft

om silicone rrceptual correldeformation cnds were madbject’s right hucture while volved in fabrbject’s hand ar

g. 5. The fabricatibject’s hand. (b) CT

surface reconstruuld and the bonesmould. The liquid

nted model of theman subject’s hand

Figure 5a showe subject’s hanoshiba Medica

0086

ARTIFICIAL HA

AND EXP

d male subject dominant righteplica of a humas measured acm in length as middle finger. Turements wereric data in [38]experiment w

ngapore’s Instit

l Design

of artificial hantness of the hurubber. Here, late of skin coaused by an ap

de from the dand. The first the second onricating the syre shown in Fig

ion process to repT scan image show

uction of the humas to construct a syd silicone rubber i mould and the bd obtained after the

ws the Compund. A helical al Systems,

AND: DESIGN, CPERIMENT

participated it hand were uman hand. The cross the ends measured fromThe subject wae close to the . The experim

was approved tutional Review

nds were fabriuman hand to t

softness is ompliance, whipplied force [3data collected

one made usene did not. Tynthetic replicgure 5.

plicate a human hwing the bone andan hand. (d) The mynthetic hand. (e) is poured through ones. (g) The come silicone rubber h

uted TomograpCT scanner (Japan) used

CONSTRUCTION

in the study. Tused to createsubject’s hand

of the metacarm the wrist creas chosen becau50th percentileental protocol by the Natio

w Board.

icated in orderthe replicas ma

defined as ich is the amou9]. Both artificfrom the hume of the skelet

The various steca of the hum

hand. (a) The humd the skin contoursmodel of the two-Assembled modethe large hole. (f)

mpleted replica ofhas cured.

phy (CT) scan(AquilionTM

d the follow

N,

The e a d is rpal ase use

e of for

onal

r to ade the unt cial

man ton eps

man

man . (c) part

el of ) 3D f the

n of 64, ing

paramet500 ms were pImages,reconstrwere poBelgiumstereolitto desigthe bonprovisio5e). A USA) Geometthe bonresolutiand 160 The mas the smixed ainto theof 21°Chuman fabricathand wskeletonouter diskeletonartificiabones (N

C. Exp

The expin FiguSwedenrepeatabsensor resolutithe z ditip of 2robot’s the resuMateria

Fig. 6. Eforce on displacema stable p

ters during theexposure time

processed in t, USA) to proruction of a huost-processed m) to design athography (STgn the model ones at accurate on for pouring

3D printing with Fullcure

tries, USA) wanes and the moions are 600 do00 dpi in the z material selectsynthetic skin and degassed ine mould. After C and relative hand with the

tion process wawith the same m

n. The use of imensions of thn and the one al hand with bNB), respectiv

periment setup

perimental setuure 6. The ron) has six bility of 0.03

(ATI Industions of 0.06 N irection. The in2 mm diametwrist to apply ulting displaceal for the video

Experimental setupthe hand. A f

ment. The inset shoposition with respe

e scan: 120 kV,e and 0.5 mm sthe Vitrea 2 oduce a threeuman hand (Fiwith the 3-M

a replica of thTL) format. Thof a two-part mdepths from th

g liquid silicosystem (modeeVeroWhite ras used for fabroulds (Figure 5ots per inch (dpaxis. The accuted earlier (i.e. material. The n a vacuum chr 24 hours of c

humidity of 5e skeleton wasas repeated to omaterial, but wthe same two-he two artifici without, are

bones (WB) anvely.

up used for theobotic arm (Idegrees-of-fremm. It is intetrial Automatin the x and yndenter, a cylier, was attacha constant forc

ement is meao of the experim

p. The robotic armforce/torque sensoows the hand as it

ect to the robot.

, 150mA, 75 mslice thickness.Imaging Soft

e-dimensional igure 5b and 5atic software

he human handhe same softwamould that wohe skin surfacene rubber (Fig

el 350, Objet resin (Code ricating the ST

5f). The machipi) in both the

uracy is up to 0 Ecoflex OO-3liquid silicone

hamber before bcuring at room 50-60%, the res formed (Figuobtain a second

without the pre-part mould enal hands. The henceforth dennd artificial h

e indentation teIRB-140, ABBeedom with egrated with a tion, Apex,

y directions, aninder with a hhed to the grice to the skin ssured (see Su

ments).

m applied a constor then measuredt rested on a clay m

5

mA exposure, . The images

ftware (Vital (3D) digital

5c). The data (Materialise, d’s bones in are was used ould position e and have a gure 5d and Geometries, 830, Objet

TL models of ine's printing x and y axes .1 mm. 30) was used e rubber was being poured temperature

eplica of the ure 5g). The d design of a

esence of the nsured equal one with the noted as the

hand with no

ests is shown B Robotics, a position force/torque USA) with d 0.125 N in emispherical ipper on the surface while upplementary

tant indentation d the resultant mould to ensure

1534-4320 (c) 2013 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission. Seehttp://www.ieee.org/publications_standards/publications/rights/index.html for more information.

This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI10.1109/TNSRE.2014.2360533, IEEE Transactions on Neural Systems and Rehabilitation Engineering

TN

poon ordregusePQthedef

Figandwer

D.

A tABdonrobinddefgrithetesto divwatesand disx

whdisdisdef

E.

StaIBM

NSRE-2014-00

The hand beinsition (Fig. 6 predefined re

der to compagions are displed to identify

QRS. We used e hand, like thfine the bound

g.7. Standardized td artificial hands. re programmed to

Software Des

top-level flowcBB robot contrne using the Rbot system. Todenter was manfined in Figurid of test pointe index, middlst region into atheir shorter l

vided into a 25as divided intost points were d were sufficieThe final skinsplacement eq, , in thr

here d is tsplacement sesplacement (i.formation prod

Data process

atistical analysM Corp., USA

0086

ng tested in theinset). The indgions of the ha

are the humanayed in Figure

the test bounthe natural la

he papillary waries.

testing areas for iPoints ABCD anthe robotic arm

sign

chart of the soroller is shownRAPID progra

o record the refnually jogged tre 7. The robots within these le, and ring fina 30×5 grid forlength, the thu5×5 grid with o an 8×10 grid

available for ent to create a dn displacemen

quation betweeree dimensiona

the displacemen on the x e. z axis) m

duced by apply

sing

ses were perfoA). For all analy

experiment wdentation testsands to create n and artificie 7. A standardndaries, shown

andmarks on thwhorls and va

indentation experind PQRS form the

ftware code prn in Figure 8. Pamming languference positioto the corners

ot was programpredefined tes

ngers, the progr a total of 150umb and the li125 test pointsof 80 test poi

the human andetailed softnesnt was measuren two pointsal space, given

ment. There or y directio

measured the ying a force of

ormed using Syses, statistical

was set in a sups were perform

softness mapsal hands. Tho

dized method wn as ABCD ahe palmar sidearious creases,

iments on the hume test boundaries

rogrammed in Programming wuage of the ABons, the tip of of the boundar

mmed to creatst boundaries. Fgram divided 0 test points. Dittle fingers ws each. The paints. Overall, 7d artificial hanss map. red by using s , , aby:

was negligions. The verti

amount of s1 N.

PSS (version l significance w

pine med s in ose was and e of

to

man that

the was BB the ries e a For the

Due were alm 780 nds

the and

(3)

ible ical kin

21, was

set at a of the performpost-hoTukey-Kmethod

Fig. 8. A setting thhands.

F. Skin

The sofderived using cMathWfrom thhand (Faveragecontour9a indicmm. Rewere obFigure 9with anartificia2.5 to 7WB concloser to

1Data are http://doi.

probability vaskin displac

med using one-c multiple comKramer Hon

d.

top-level descripthe indentation bou

n softness map

ftness maps od from the skicontour plots

Works, USA). Ae measured da

Figure 9). The e filter to smoor plots. The socated that the egions around bserved to hav9b and 9c shownd without thal hand with NB7.5 mm, the dindition were wo the human ha

available from the.org/10.5061/dryad

alue of p < 0.0cement data o-way analysis mparisons test

nestly Signifi

tion of the program

undaries, and for in

of the human in displaceme in a softw

A 2D colouredata and was sup

raw data wereothen the suddftness map of displacement

the proximal ive the least diw the softness mhe presence ofB condition hasplacements ofwithin 0.5 to and’s skin com

e Dryad Digital Red.vm4m4

5. A statisticalof the three of variance (At was performeicant Differen

m flow for calibrandenting the huma

and synthetic ent data and ware package

d contour plot perimposed on e processed widen gradient ch

the human has were withinnterphalangealisplacement onmaps of the artf the skeletonas displacemenf the artificial 6 mm and app

mpliance.

epository:

6

l comparison hands was

ANOVA)1. A ed using the nce (HSD)

ating the sensor, an and artificial

hands were were plotted (MATLAB, was created each type of

ith a moving hanges in the and in Figure n 0.2 to 5.75 l (PIP) joints n all fingers. tificial hands

n. While the nts within the hand having peared to be

1534-4320 (c) 2013 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission. Seehttp://www.ieee.org/publications_standards/publications/rights/index.html for more information.

This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI10.1109/TNSRE.2014.2360533, IEEE Transactions on Neural Systems and Rehabilitation Engineering

TN

Fighumhan

G.

Figthehasartdiswitsigindp <F(2p <forNBavebonwh14Tasofthethe

Fighumbonthe

NSRE-2014-00

g. 9. Skin softnessman hand; (b) artind with no bone str

Comparisons

gure 10 compae artificial hands the least ditificial hands. splacements ofthin each fin

gnificant differdex: F(2,447) =< 0.001; ring:2,372) = 375.7< 0.001. Post-r human vs. WB hands diffeerage displacemnes differed f

hile the artifici3% against the

aken together, ftest among the bony structue artificial han

g. 10. Skin displacman hand, artificines (NB). Post-hocgroups being com

0086

s maps from the inificial hand with bructure (NB).

s: human vs. ar

ares the displacds. It consistenisplacement, fA one-way A

f the three grounger and the rences: thumb:= 196.4, p < 0.: F(2,447) = 37, p < 0.001; a-hoc analyses

WB hands, humered significanments were cofrom 14% to ial hand withoe human hand

the artificial he three types re. However, d with bones w

cements at 1 N coial hand with bonc analyses show si

mpared for each of

ndentation tests wbone structure (W

rtificial hands

cements of the ntly shows thatfollowed by th

ANOVA was pups (human, W

palm. The a: F(2,372) = 2.001; middle: F399.8, p < 0.0and the palm: Fshowed that th

man vs. NB hantly at p < 0mpared, the ar47% against t

out bones diffefor every finghand without

of hands due tthe displacemewere closer to

onstant force (menes (WB) and artignificant differencthe fingers and the

with 1 N force forWB); and (c) artifi

human hand at the human hahe WB and Nperformed on

WB and NB hananalyses yield

222.8, p < 0.0F(2,447) = 344001; little fingF(2,237) = 128he displaceme

ands, and WB 0.001. When rtificial hand wthe human ha

ered from 54%ger and the palt bones was to the absenceents recorded the human ha

an and SEM) for tificial hand withces (p < 0.001) fore palm.

r (a) icial

and and NB the

nds) ded 01; 4.1, ger: 8.1, ents vs. the

with and,

% to lm. the

e of for

and.

the

h no r all

H. Reg

The getopogracomparhuman displace30×5 grthe firstthe palmcompareach setest hanhand bdisplaceregions pair for Figurthe indiindicatedifferenand the of the thand, wFigure data betshows fmissing

Fig. 11. Rdisplacemartificial suggestin

human ha

IV. PE

A. Par

A total years oSingapo

B. Exp

The pahand thhand. Wthe hair

gion-wise analy

eometrical struaphy of the srison was doneand artificial fement comparrid in the humt row of the artm, we dividedrison. A two-taet of human annds in pairs as based on the ement. That isof human and

r the tests. re 11 consolidividual test ree a noticeable nces were foune human hand. thick volar lig

which can be 1). In compartween the humfew regions ofg bone structure

Region-wise statistment data. (a) Hum

hand without bong no significant

and.

ERCEPTION EXP

rticipants

of 28 healthy old) were recrore. Participatio

perimental Des

articipants werhat touches theWe compared thry part of the

lysis

ucture of theskin tissue aree between specfingers to inverisons. For exa

man index fingetificial index fi

d the 8×10 gridailed paired t-tend artificial hahuman vs. WBregion of m

s, displacementd artificial han

dates the pairedegions. The sh

pattern of regnd (p > 0.05) This could be

gaments at the2 to 3 mm in

rison, the analman and NB arf similar skin ce.

tical comparison oman vs. artificial haones. The shadedifference betwe

PERIMENTS WIT

HANDS

subjects (20 mruited from thon was volunta

sign

re instructed tem is from a hheir responses forearm by thr

e bone and te not uniformcific regions inestigate their eample, the firser data was cofinger, and so od into 2×2 regest was performand data. We B hand, and humeasurement t data from eq

nds were consid

d t-test results haded areas ingions where nfor the WB ar

e attributed to e PIP joints ofn thickness [4lysis of skin drtificial hands compliance be

of human and artifand with bones and areas have p-

een the artificial

TH HUMAN AND

males, 8 femalhe National Uary.

to determine human hand or after they werree types of h

7

the external m. Hence, a n each of the ffects on the

st row of the mpared with

on. In case of gions for this med between matched the

uman vs. NB of the skin

quivalent test dered as one

by showing n Figure 11a o significant rtificial hand the presence f the human

40] (also see displacement (Figure 11b)

ecause of the

ficial hands skin nd (b) human vs. values > 0.05, hand from the

D ARTIFICIAL

les, all 17-26 University of

whether the r an artificial re touched at ands (Figure

1534-4320 (c) 2013 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission. Seehttp://www.ieee.org/publications_standards/publications/rights/index.html for more information.

This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI10.1109/TNSRE.2014.2360533, IEEE Transactions on Neural Systems and Rehabilitation Engineering

TN

12andartcoltemandsurnaïhanthe12b

Figbonsofttouof t

soft

Dtheexpon(i.eneiRatason thecermopoSyhelThequtriaeffwethachacon

NSRE-2014-00

a): a human had warm artiftificial hands hld artificial hanmperature tookd warm artificrface temperatïve to the hyponds that they we hands were kb; also see Sup

g. 12. Perception nes. (a) From left tft and warm artificched the participathe soft and cold a

ft and warm artifici

Detection theoe participant’speriments maye of the two ste. yes or no tither stimulus

ating experimensk. For the cur

a 5-point scae touch is fromrtain that the toouse, participanint scale. A

ystem, USA) wlp the experim

he applied forcually presentedals. The first 6fects. A total oere obtained. Tat at least 100 aracteristics (Rnclusion [41].

0086

and, a soft andficial hand whad the bone snd had no heatk the ambient ial hand was pture of 28.4°Cothesis of the will be touche

kept out of the pplementary Vi

experiments withto right: the humancial hand. (b) Exp

ants. (c) Receivingartificial hand and

ial hand and the hu

ory [41] states s ability to

y be distinguishtimulus classestask). The secis null (i.e. r

nts give more rrent paper, wele where 1 de

m an artificial houch is from ants entered theforce sensor

was positioned menter regulatece was about 1d in a pseudor

6 trials were diof 588 respons

This data set waobservations aROC) curve a

d cold artificialith the heatinstructure (§III.ting system andtemperature o

programmed wC (§II.F). The p

experiment and with. Duringview of the paideo).

h the human and n hand, soft and coperimental set-up g operating charac

the human hand.

uman hand.

that for expertell two stim

hed. The first is contains onlycond one is rerating experiminformation th

e asked particienotes that theyhand and 5 dena human hand. eir responses a(FingerTPS, at the particip

e the amount N. The three random order iscarded to accses (28 subjectas sufficient asare needed in ranalysis to dra

l hand, and a sng design. BH). The soft ad the skin surfa

of 21°C. The swith an initial sk

participants wnd to the typesg the experimearticipants (Figu

artificial hands wold artificial hand, where the test hateristics (ROC) cu(d) ROC curve of

riments involvmuli apart, ts detection whthe null stimu

ecognition whments on a scal

han in a “yes-nipants to respoy are certain t

notes that they With a compu

according to aPressure Prof

pant’s forearmof contact forhand types wfor a total of

count for practts, 21 trials eas it was suggesreceiver operataw a meaning

soft oth and face soft kin

were s of ent, ure

with and

ands urve f the

ing two here ulus here le). no” ond that are

uter a 5-file

m to rce.

were 27

tice ch)

sted ing

gful

C. Dat

We useof correhand (trtouch frbetweencomparmethodobserveis 0.5 anThe pro

D. Res

For eacTable I(very chuman curves. and micurve thAUC othat theas a toutouch frfrom the In codifficultthe humthe coldAUC iswarm a0.05 supass as

TABLE

Hand type

Human Cold Warm

For ovebasic desimple hook-tyamputee Contiare nowintelligedexterouamputeeobject's and thralso bee

ta processing

ed the ROC cuectly recognizerue positive) tofrom the human the two Ared with the med, the p-value wed sample AUCnd that the nul

obability select

sults

ch type of hanIII shows howertain to be phand). These A result that

istakenly recohat is coincidf 0.5. In other

e touch from anuch from a hrom the soft ane human hand ontrast, Figurety in recognizi

man hand (AUd hand was fous significantly

artificial hand, uggesting that t

if it were touch

III FREQUENCY(T

Very Certain to be

Prosthetic (1)

C

Pro

21 49 20

V. DIS

er a century, tesign. It has begrasping task

ype prosthesis hes [44]. inuing advancw convergingent prosthetic us grasping (ees according t

s size or shaperough experimen demonstrate

urve analysis [4ed touches by o the proportionan hand (false Areas Under tethod of Hanlewas determinedC is found whell hypothesis ated as significa

nd that was uw the subjectsprosthetic hand

data were furt cannot discriognized touch dent to a diagor words, an An artificial han

human hand. Fnd cold artifici(AUC = 0.697

e 12d suggestsing the soft andC = 0.562, p =

und to be less thdifferent from

the p-value wathe touch fromh from a huma

Y OF EACH RESPTOUCH RECOGN

Certain to be osthetic

(2)

Neutral

(3) 35 24 49 33 34 52

SCUSSION AND C

the prosthetic een a robust anks. However, has been high

cements from g towards th

devices. Prose.g. [21, 45, 46to their motor e can now be

ments on the rued that an artif

41] to relate ththe warm or cn of mistakenlypositive). The

the Curves (ey and McNeilld as the probaben the true popussumes that th

ant was p < 0.0

used to touch t recognized th

d) to 5 (very crther analyzediminate betwecorresponds

onal line and AUC closer to nd could not beFigure 12c shoial hand can b7, p < 0.0001). s that the partd warm artifici= 0.0611). Thehan 0.05 indicam an area of as found to be

m this artificialan hand.

PONSE FOR EACNITION)

Certain to be

Human (4)

Very t

Hu

65 46 45

CONCLUSION

hook [43] hasnd reliable termthe rejection at 50% among

several researhe developmensthetic hands 6]); it can be ccommands [6felt by an ampubber-hand illuficial hand can

8

he proportion cold artificial y recognized e differences (AUC) were l [42]. In this bility that the ulation AUC e area is 0.5. 5.

the subjects, hem from 1 certain to be

d using ROC een correctly to an ROC occupies an 0.5 suggests e recognized ows that the e recognized

ticipants had al hand from e p-value for ating that the 0.5. For the greater than l hand could

CH STIMULUS

y Certainto be uman (5)

Sum

51 196 19 196 45 196

s retained its minal tool for

rate for the g upper limb

rch domains nt of more can perform

controlled by , 47, 48]; an putee [7, 8]; usion, it has

n be felt as if

1534-4320 (c) 2013 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission. Seehttp://www.ieee.org/publications_standards/publications/rights/index.html for more information.

This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI10.1109/TNSRE.2014.2360533, IEEE Transactions on Neural Systems and Rehabilitation Engineering

TNSRE-2014-00086

9

it were a part of the body [12, 16, 49]. All these are giving upper limb amputees hope to regain what has been lost. Considering the technological trends above, it can be expected that prosthetic hands will touch and be touched by others during social interactions. To our knowledge, the current paper is the first to demonstrate that a warm and soft silicone artificial hand can be used to create an illusion of the sense of human touch. This was accomplished through a three-step process. First, participants were asked to select the samples that they felt were similar to human skin from artificial skin samples that were laid out in a 4×4 array. Majority selected a silicone material with a Shore durometer value of 30 at the OO scale and the selected temperatures clustered at 27 and 32°C in an ambient room temperature of 21°C. When proportioned, the selected temperature was calculated to be 28.4°C. Our results suggest that a soft rubber material with a warm skin surface temperature is critical for the artificial skin to be perceived as lifelike. Rubber materials are known to be poor conductors of heat. The thermal conductivities of the 4 materials tested herein were low with values from 0.12 to 0.23 W/m·K. However, it would be possible to mimic the warmth of the human hand’s skin tissue by controlling the embedded heater’s power supply by adjusting the heat that will be transmitted to the surface of a rubber material [27, 50]. A more lifelike artificial skin is the one having a surface temperature that is similar to the initial surface temperature of the human skin upon contact. This means that when a person touches a sample, it tends to feel more lifelike when the person does not experience a significant difference on the warming or cooling sensation on his/her finger at the moment of contact. The contact temperature will naturally rise and approach the human body temperature due to the elimination of the natural air convection from the exposed finger. Second, we described a method to replicate the geometric features of a human subject’s hand and the softness of the hand's skin tissue. Through CT scan, computer-aided design, 3D printing and silicone moulding technologies, we found that the softness of the human hand’s skin tissues can be mimicked by replicating the surface topology of the human hand and the geometries of its skeleton. The soft material selected in the 4×4 array experiment was used as a substitute for the human skin tissue. Due to the absence of ligaments and tendons like those in human hands, there were regions on the fingers and the palm that were softer than the human hand. Overall, the DIP and PIP regions were not significantly different from the human hand (p> 0.05). We used these regions of the artificial hand to touch the forearm of the participants. To integrate other mechatronic components for a more intelligent prosthetic hand, there is still substantial space that is available for mechanisms and embedded electronics at the skeleton and at the dorsal part of the hand. From our initial design, only 5-7 mm of artificial skin thickness are needed to be allocated for embedded heaters and for the soft material to replicate the human hand’s skin compliance. Finally, we combined the previous results to design the artificial hands for the perception experiments. Subjects who

were naïve about the experimental objectives were touched with a human hand and two artificial hands. One artificial hand was soft but it felt cold because the skin surface temperature was similar to that of the room temperature of 21°C. The other artificial hand felt warm due to the embedded heaters, which raised the skin surface temperature to 28.4°C. With the participants’ field of view being restricted, they were asked to recognize whether they were touched by a human or an artificial hand. The subjects were touched at the hairy part of the forearm where a subclass of unmyelinated afferents (C-tactile, CT) are known to be present [51]. The CT afferent has been implicated in the coding of pleasant tactile sensations. ROC curve analysis suggests that an illusion of human touch can be created from an artificial hand with embedded heaters on soft synthetic skin and a skeleton structure (AUC = 0.562; p = 0.0611). An early study suggested that friction and the surface topography of a material are significant factors when the human finger pad makes a sideward movement on a surface, which can further influence the sensation of pleasant touch [52]. The current paper only investigated simple touches of 1 N normal force and not the sliding touch movements that are similar to a caress. Such movements would entail controlled lateral movements of 1-10 cm/s [53]. Likewise, the fingerprint ridges and the sweat from the skin pores have been shown to contribute to the contact mechanics of sliding movements [54]. In future studies, it would be interesting to investigate the effects of artificial fingerprint ridges on the pleasantness of the touch through controlled lateral movements similar to a caress. More advanced touching movements, like caress and handshakes, have social, emotional, and cultural ramifications. The present work focused only on simple mechanical touch. Despite its seeming simplicity, social touching for artificial hands has been a neglected research area. From the experiments described herein, we found that softness and warmth as important features. Future work can look into the other design variables that can make social touches (e.g. handshake or caress on the face, the hands, and arms) non-discriminable even through an artificial hand. Interpersonal touch induces strong and reliable changes in autonomic activity (e.g. skin conductance, pulse and respiration) between the interacting partners [55]. A possible extension to the current work is to investigate the emotional valence of the touch from a lifelike artificial hand. That is, we might be able to see effects of this novel type of prosthetic hand for affective touches. There have been related works on determining the emotional state of participants by touch from haptic interfaces. Some of the earlier works have explored that idea with a haptic jacket [56] or a haptic sleeve like in our earlier work [57]. However, there have been none for prosthetic or robotic hands. In the Affective Teletouch Technology that we proposed [57], we asked participants to watch a sad movie while the spouse touched the arm of the participant. In the other experimental condition, we used a haptic sleeve to provide vibratory and warmth stimuli on the subject’s arm. A similar experimental design could be used for

1534-4320 (c) 2013 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission. Seehttp://www.ieee.org/publications_standards/publications/rights/index.html for more information.

This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI10.1109/TNSRE.2014.2360533, IEEE Transactions on Neural Systems and Rehabilitation Engineering

TNSRE-2014-00086

10

the soft and warm artificial hand as described herein. For practical implementation later on, it would be helpful to consider a layered skin structure that addresses the multiple requirements to have soft features for social touching [28], for skin compliance and skin conformance for tactile sensing purposes [58] as well as the need to protect the underlying structures through a tough skin layer [50]. In an earlier work [29], we investigated how to design a soft skin for social touching interactions, which also considers other requirements for wear, puncture, and tear. For the condition of having a softer internal material and a stiffer external material with a 0.8 mm thin layer, we found that the results of a layered structure will only have a difference in displacement of about 7% as compared to a structure that is homogeneous. These previous findings suggest that it would be possible to achieve both skin softness as well as toughness by varying the properties of the skin layers. The present results are important because they provide an early evidence that an illusion of human touch can be created by a warm and soft silicone artificial hand to the person being touched. When applied to prosthetic hands, these findings have a potential to help prosthesis users cope with the functional and psychosocial effects of losing a part of their body.

ACKNOWLEDGMENT

The authors are grateful to Jaclyn Ting Lim for the illustrations.

REFERENCES [1] C. A. Moran, "Anatomy of the hand," Phys Ther, vol. 69, pp. 1007-13,

1989. [2] C. L. Taylor and R. J. Schwarz, "The anatomy and mechanics of the

human hand," Artif Limbs, vol. 2, pp. 22-35, 1955. [3] Y. Lanir, "Skin Mechanics," in Handbook of Bioengineering, R. Skalak

and S. Chien, Eds., ed New York: McGraw-Hill, 1987, pp. 1-25. [4] R. S. Johansson and A. B. Vallbo, "Tactile sensibility in the human

hand: Relative and absolute densities of four types of mechanoreceptive units in glabrous skin," Journal of Physiology, vol. Vol. 286, pp. 283-300, 1979.

[5] M. J. Hertenstein, D. Keltner, B. App, B. A. Bulleit, and A. R. Jaskolka, "Touch communicates distinct emotions," Emotion, vol. 6, pp. 528-533, 2006.

[6] T. A. Kuiken, P. D. Marasco, B. A. Lock, R. N. Harden, and J. P. A. Dewald, "Redirection of cutaneous sensation from the hand to the chest skin of human amputees with targeted reinnervation," Proceedings of the National Academy of Sciences of the United States of America, vol. 104, pp. 20061-20066, 2007.

[7] P. D. Marasco, K. Kim, J. E. Colgate, M. A. Peshkin, and T. A. Kuiken, "Robotic touch shifts perception of embodiment to a prosthesis in targeted reinnervation amputees," Brain, vol. 134, pp. 747-758, 2011.

[8] S. Raspopovic, M. Capogrosso, F. M. Petrini, M. Bonizzato, J. Rigosa, G. Di Pino, et al., "Restoring Natural Sensory Feedback in Real-Time Bidirectional Hand Prostheses," Science Translational Medicine, vol. 6, p. 222ra19, February 5, 2014 2014.

[9] D. M. Desmond, "Coping, affective distress, and psychosocial adjustment among people with traumatic upper limb amputations," Journal of Psychosomatic Research, vol. 62, pp. 15-21, 2007.

[10] I. Dudkiewicz, R. Gabrielov, I. Seiv-Ner, G. Zelig, and M. Heim, "Evaluation of prosthetic usage in upper limb amputees," Disability and Rehabilitation, vol. 26, pp. 60-63, 2004.

[11] A. Saradjian, A. R. Thompson, and D. Datta, "The experience of men using an upper limb prosthesis following amputation: Positive coping and minimizing feeling different," Disability and Rehabilitation, vol. 30, pp. 871-883, 2008.

[12] M. Botvinick and J. Cohen, "Rubber hands 'feel' touch that eyes see," Nature, vol. 391, p. 756, 1998.

[13] H. H. Ehrsson, C. Spence, and R. E. Passingham, "That's my hand! Activity in premotor cortex reflects feeling of ownership of a limb," Science, vol. 305, pp. 875-877, 2004.

[14] H. H. Ehrsson, K. Wiech, N. Weiskopf, R. J. Dolan, and R. E. Passingham, "Threatening a rubber hand that you feel is yours elicits a cortical anxiety response," Proceedings of the National Academy of Sciences of the United States of America, vol. 104, pp. 9828-9833, 2007.

[15] M. Tsakiris and P. Haggard, "The rubber hand illusion revisited: Visuotactile integration and self-attribution," Journal of Experimental Psychology: Human Perception and Performance, vol. 31, pp. 80-91, 2005.

[16] H. H. Ehrsson, B. Rosén, A. Stockselius, C. Ragnö, P. Köhler, and G. Lundborg, "Upper limb amputees can be induced to experience a rubber hand as their own," Brain, vol. 131, pp. 3443-3452, 2008.

[17] M. Tsakiris, L. Carpenter, D. James, and A. Fotopoulou, "Hands only illusion: Multisensory integration elicits sense of ownership for body parts but not for non-corporeal objects," Experimental Brain Research, vol. 204, pp. 343-352, 2010.

[18] J. J. Cabibihan, "Patient-specific prosthetic fingers by remote collaboration-A case study," PLoS ONE, vol. 6, 2011.

[19] A. D. Deshpande, X. Zhe, M. J. V. Weghe, B. H. Brown, J. Ko, L. Y. Chang, et al., "Mechanisms of the Anatomically Correct Testbed Hand," Mechatronics, IEEE/ASME Transactions on, vol. 18, pp. 238-250, 2013.

[20] E. L. Leow, B. P. Pereira, A. K. Kour, and R. W. H. Pho, "Lifelikeness in multilayered digital prostheses," Prosthetics and Orthotics International, vol. 21, pp. 40-51, 1997.

[21] C. Connolly, "Prosthetic hands from Touch Bionics," Industrial Robot, vol. 35, pp. 290-293, 2008.

[22] C. D. Murray, "The social meanings of prosthesis use," J Health Psychol, vol. 10, pp. 425-41, 2005.

[23] S. Ritchie, S. Wiggins, and A. Sanford, "Perceptions of cosmesis and function in adults with upper limb prostheses: A systematic literature review," Prosthetics and Orthotics International, vol. 35, pp. 332-341, 2011.

[24] J. Davidson, "A survey of the satisfaction of upper limb amputees with their prostheses, their lifestyles, and their abilities," Journal of Hand Therapy, vol. 15, pp. 62-70, 2002.

[25] A. Esquenazi, "Amputation rehabilitation and prosthetic restoration. From surgery to community reintegration," Disability and Rehabilitation, vol. 26, pp. 831-836, 2004.

[26] P. Gallagher and M. MacLachlan, "Adjustment to an artificial limb: A qualitative perspective," Journal of Health Psychology, vol. 6, pp. 85-100, 2001.

[27] J.-J. Cabibihan, R. Jegadeesan, S. Salehi, and S. Ge, "Synthetic Skins with Humanlike Warmth," in Social Robotics. vol. 6414, S. S. Ge, H. Li, J.-J. Cabibihan, and Y. Tan, Eds., ed: Springer Berlin Heidelberg, 2010, pp. 362-371.

[28] J. J. Cabibihan, S. Pattofatto, M. Jomaa, A. Benallal, and M. C. Carrozza, "Towards Humanlike Social Touch for Sociable Robotics and Prosthetics: Comparisons on the Compliance, Conformance and Hysteresis of Synthetic and Human Fingertip Skins," International Journal of Social Robotics vol. 1, pp. 29-40, 2009.

[29] J. J. Cabibihan, R. Pradipta, and S. S. Ge, "Prosthetic finger phalanges with lifelike skin compliance for low-force social touching interactions," Journal of Neuroengineering and Rehabilitation, vol. 8, p. 16, 2011.

[30] D. H. Kim, J. Viventi, J. J. Amsden, J. Xiao, L. Vigeland, Y. S. Kim, et al., "Dissolvable films of silk fibroin for ultrathin conformal bio-integrated electronics," Nature materials, vol. 9, pp. 511-517, 2010.

[31] S. C. B. Mannsfeld, B. C. K. Tee, R. M. Stoltenberg, C. V. H. H. Chen, S. Barman, B. V. O. Muir, et al., "Highly sensitive flexible pressure sensors with microstructured rubber dielectric layers," Nature Materials, vol. 9, pp. 859-864, 2010.

[32] R. F. Shepherd, F. Ilievski, W. Choi, S. A. Morin, A. A. Stokes, A. D. Mazzeo, et al., "Multigait soft robot," Proceedings of the National Academy of Sciences, vol. 108, pp. 20400-20403, 2011.

[33] F. Tramacere, A. Kovalev, T. Kleinteich, S. N. Gorb, and B. Mazzolai, "Structure and mechanical properties of Octopus vulgaris suckers," Journal of the Royal Society, Interface, vol. 11, 2014.

[34] J. H. Lienhard, IV and J. H. Lienhard, V, A heat transfer textbook. Cambridge, MA: Phlogiston Press, 2008.

[35] L. A. Jones and H. N. Ho, "Warm or cool, large or small? The challenge of thermal displays," IEEE Transactions on Haptics, vol. 1, pp. 53-70, 2008.

1534-4320 (c) 2013 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission. Seehttp://www.ieee.org/publications_standards/publications/rights/index.html for more information.

This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI10.1109/TNSRE.2014.2360533, IEEE Transactions on Neural Systems and Rehabilitation Engineering

TN

[36

[37

[38

[39

[40

[41

[42

[43[44

[45

[46

[47

[48

[49

[50

[51

[52

[53

[54

[55

[56

[57

[58

NSRE-2014-00

6] H. N. Ho and Lsurface during Engineering, vo

7] H. N. Ho and discrimination app. 118-128, 20

] Department of Washington, DC

9] R. Friedman, estimation of so142, 2008.

0] A. Lluch, "IntriJoint stiffness ofMosby, Dunitz,

] N. A. MacmillGuide: Taylor &

2] J. A. Hanley anreceiver operati29-36, 1982.

] D. W. Dorrance4] E. Biddiss, D.

upper limb pTechnology, vo

] S. Roccella, M.Dario, et al., "humanoid emHumanoid Robo

6] C. Cipriani, Mtransradial prosvol. 8, 2011.

7] M. Powell, R.Recognition-BaMovement ConRehabilitation 2013.

] N. V. ThakorTranslational M

9] H. H. Ehrsson, hand: Feeling multisensory br10573, 2005.

0] J. J. CabibihanHakkim, "ApReplacement MDuplication, PC

] H. Olausson, YStarck, et al., "Uinsular cortex,"

2] A. Klöcker, MThonnard, "Phyexploration," PL

] L. S. Löken, J."Coding of pleaNeuroscience, v

4] M. J. Adams, SAndré, et al., "Fof the Royal Soc

] J. Chatel-Goldmincreases autonBehavioral Neu

6] D. TsetserukoCommunicationPerceiving TangW. Bergmann THeidelberg, 201

7] J.-J. Cabibihan,Robotics. vol. 7M.-A. Williams356.

] J. Cabibihan, SSkin's Thicknes

0086

L. A. Jones, "Modhand-object inte

ol. 130, 2008. L. A. Jones, "Co

and localization," 006. f Defense, "HumaC2000. K. Hester, B. G

oftness," Experime

insic causes of stiof the upper limb, S 1997, pp. 259-26lan and C. D. C& Francis, 2005. d B. J. McNeil, "Ting characteristic (

e, "Artificial hand,Beaton, and T. C

prosthetics," Disal. 2, pp. 346-357, 2. C. Carroza, G. C"Design and deveotion expressionotics, vol. 4, pp. 18

M. Controzzi, andthesis," Journal of

. Kaliki, and N. ased Myoelectric nsistency and DiEngineering, IEE

, "Translating thMedicine, vol. 5, p.

N. P. Holmes, andof body owner

rain areas," Journa

n, S. S. Ge, S. Sparatuses, Syste

Manufacturing, TemCT/SG2011/00025Y. Lamarre, H. BUnmyelinated tactNature Neuroscie

M. Wiertlewski, ysical factors inflLoS ONE, vol. 8, 2 Wessberg, I. Mo

asant touch by unmvol. 12, pp. 547-54S. A. Johnson, P. Finger pad frictionciety, Interface, voman, M. Congedo,nomic coupling beuroscience, vol. 8, ou, "HaptiHug: n of Hug over a Dgible Sensations. vTiest, and F. T. va10, pp. 340-347. L. Zheng, and C.

7621, S. Ge, O. Khs, Eds., ed: Sprin

S. S. Chauhan, anss on the Subsurf

deling the thermal eractions," Journa

ontribution of therPerception and Ps

an Engineering D

Green, and R. Lental Brain Resear

ffness of the interS. Copeland, Ed., 4.

Creelman, Detectio

The meaning and u(ROC) curve," Ra

" 1912. Chau, "Consumer ability and Reha2007.

Cappiello, J. J. Cabelopment of five-n robot," Intern81-206, 2007. d M. C. Carrozzf NeuroEngineerin

Thakor, "User Prostheses: Improstinguishability,"

EE Transactions o

he Brain-Machine. 210ps17, Novemd R. E. Passinghamrship is associatal of Neuroscience

Salehi, R. Jegadeems and Methomperature Regulat5," 2011.

Backlund, C. Mortile afferents signaence, vol. 5, pp. 90V. Théate, V. Hluencing pleasant 2013. orrison, F. McGlomyelinated afferent48, 2009. Lefèvre, V. Léves

n and its role in griol. 10, p. 20120467, C. Jutten, and J.

etween romantic p2014-March-27 20

A Novel HDistance," in Hapvol. 6191, A. L. Kan der Helm, Eds

Cher, "Affective Thatib, J.-J. Cabibih

nger Berlin Heidel

nd S. Suresh, "Effface Pressure Prof

responses of the sal of Biomechan

rmal cues to matesychophysics, vol.

Design Data Dige

LaMotte, "Magnitrch, vol. 191, pp. 1

rphalangeal joints,ed St. Louis, Lond

on Theory: A Us

use of the area undadiology, vol. 143,

design priorities abilitation: Assis

bibihan, C. Laschifingered hands fo

national Journal

za, "The SmartHng and Rehabilitat

Training for Patoving Phantom LNeural Systems

on, vol. PP, pp.

e Interface," Sciember 6, 2013 2013.m, "Touching a rubted with activitye, vol. 25, pp. 105

eesan, and H. Abods for Prosthtion and Tactile Se

rin, B. G. Wallin,al touch and projec00-904, 2002. Hayward, and J.

touch during tac

ne, and H. Olausts in humans," Na

sque, V. Haywardip and touch," Jour7, 2013. -L. Schwartz, "To

partners," Frontier014.

Haptic Display ptics: Generating Kappers, J. F. van E

., ed: Springer Be

Tele-touch," in Sohan, R. Simmons, lberg, 2012, pp. 3

fects of the Artififiles of Flat, Curv

skin nical

erial 68,

est,"

tude 133-

," in don:

ser's

der a pp.

for stive

i, P. or a

of

Hand tion,

ttern imb and 1-1,

ence

bber y in 564-

bdul hetic ense

, G. ct to

L. ctile

son, ture

d, T. rnal

ouch rs in

for and Erp, erlin

ocial and

348-

icial ved,

and 201

2013, heEngineerialso serveAffiliate (SiNAPSEJournal oBionics aa past ChChapter organizatiConferencInternatioChair of Intelligentechnologrobotics.

India. Curesearch machine processin

ComputatComputat

Braille Surfaces,4.

John-JohbiomedicPisa, Italan InternSupérieuLaboratocurrentlyEngineer

e was an Assisting Department oed as the Deputy Faculty Member E). Currently, he

of Social Roboticand the Internationhair of the IEEE S(terms: 2011 an

tions: he was thce on Social Ro

onal Conference onf the 2013 IEEEnt Systems, Mangies towards lifeli

Deepak and M. from IndPhD degof TechnInstitute research (NUS), b

urrently, he is on leat University of interface, machin

ng.

Yeshwindegree UniversiAutomatworkingMakino industria

Mark AaResearchNational the Cryospostdoctoheat trans

ArrchanaElectricaUniversihas beenOptimizaDepartmthe same

tional Intelligenction, in particular.

" Sensors Journa

hn Cabibihan wcal robotics by thly in 2007. Duringnational Scholarshure de Cachan, Foire de Mécaniquy affiliated with thring Department otant Professor at

of the National UnDirector of the Soat the Singaporeserves at the Edi

s, Computational nal Journal of AdvSystems, Man andnd 2012). He h

he Program Co-Cobotics, Singaporn Social Robotics

E International Cnila, Philippines. ike touch and ge

Joshi did B. TechTech in Instrume

dia in 2004 and gree in biomedical nology (IIT) Delhof Neuroscience engineer at na

before joining Greave from the uniOregon, USA. Hie learning, biome

n Mysore Srinivasin Electrical E

ity of Singaportion and Control

g as a Research Asia Pte Ltd.

al robotics and hig

aron Chan is a R, Germany. He rUniversity of Sinspheric Sciences Loral fellow. His msfer and thermal m

a Murugananthamal Engineering wity of Singapore, Sn pursuing her Phation using Evo

ment of Electrical e university. Her cce and Robotics,

al, IEEE, vol. 14,

was conferred wihe Scuola Superi

g his PhD studies, hhip grant by the France, which h

ue et Technologiehe the Mechanicaof Qatar Universityt the Electrical niversity of Singaocial Robotics Labe Institute of Neuitorial Board of thCognitive Scienc

vanced Robotics Sd Cybernetics Sochas been active Chair of the 201re; Program Chais at Chengdu, ChinConference on C

He is workingestures for prosthe

h in Instrumentatientation and cont2006, respectivelengineering from

hi. He was a vis(ION), Newcastle

ational university raphic Era Univeriversity to pursue his research area i

edical instrumentat

sa received his MaEngineering from

re, with a spl Engineering. H

and DevelopmenHis research in

gh-speed motion co

Research Engineerreceived his PhD.ngapore. SubsequeLaboratory at NA

main research intermanagement of elec

m received the Bwith Honors fromSingapore in 2012.hD on Dynamic

olutionary Algoritand Computer Encurrent research i, in general and

11

pp. 2118-2128,

ith a PhD in iore Sant’Anna, he was awarded Ecole Normale e spent at the

e in 2004.He is al and Industrial y. From 2008 to and Computer

apore, where he boratory and an urotechnologies he International ce, Frontiers in ystems. He was

ciety, Singapore in conference

10 International ir of the 2012 na; and General

Cybernetics and g on the core etics and social

ion engineering trol engineering ly. He received

m Indian Institute iting scholar at

e University and y of Singapore rsity, Dehradun his postdoctoral includes neural-tion, and signal

aster of Science the National

ecialization in He is presently nt Engineer at

nterests include ontrol systems.

r at GE Global . in 2010 from ently, he joined

ASA GSFC as a rests are boiling ctronics.

.Eng degree in m the National Since then, she Multi-objective

thms with the ngineering from nterests include d Evolutionary