デジタルデザインの未来

Digital

Design

in

Future

FL

)"

tsT

Jv

7'-ve

-1'

>t

oxx

HidekiAOYAMA KeioUniversity

Etheses

tskeept

Sq#

Concept!basicdesignpiaysan importantrole to_produce

industrial_productscoping withdiversjficationof customer:s

preference.

This

paper

describes

digitalsupport systems

for

theconcept/basic design

_processes

which can shorten the

leadtimeofa _productand expand designer'sideas,

One

jsa system to automatjcally

_generate

candidate solutions forthe

designfromconcept which isexpressed bynatural _language so calied KanseiWords.Second one isa system using mixed reality technologytoevaluate designabilityand operabiljty without making

_prototypes,

And other threesystems

for

sup-portingdesignprocessesare also introducedinoutiine,

1.

Introduction

lt

is

a

key

issue

totimelyoffer _productsmeetimg with

diver-sifying

_preference

ofcustomers. Ibtimelymake offerof prod-ucts, it

is

an essentialand importantfactorto shorten _the de-veloping time of a

_product.

On theother hand,as

it

has

been

become hardtomake differentiationof a _productinfunction

design,style designisbecomjng more and more important_to make _product

distinction

[1],

[2],

[3].

Abilityand _performance

for

designactivity mainly

depend

on hislhersensitivity

(Kansei)

and experience. Itneeds long

timetolearnand master

high

sensitivityand rich experience.

Indetajldesign_processes,

CAD

_{Computer

Aided Design)

systems are used asan extremely usefuland _powerfultoolfor

covering hisfher_poorexperience. However,thereisvery few

supporting system tosupport concept and basicdesign

pro-cesses,

ltisan importantrole forindustrjaldesignerstosupply

industrial

_products

which satisfycustomer's demands and

needs. Makingthelong-rangeestimation of customeris

de-mands isso hardso thatshortening

lead

timeforsupplying a productjsan indispensableissueforcoping withthe short-range estimation of customer's them.

In

the_presentstate,to

make _higherefficiencyforthedownstream _processesfrom

detail

design

is

_going

totechnical

difficulties

becausethe effi-ciency

is

reaching tothelimit.Therefore,thetjme

for

upstream

2r'if(#Mxkle

specialissueofjapanesesocietyforthescienceofdesign

vol.15-4 no.60 2DOe

processes

:theconcept and basicdesign

_processes

including designevaluation _processes,should be shortened formaking short leadtime,

Asmentjoned above, systems

for

supporting theconcept and basicdesjgn_proces$es

have

tobedevelopedforthe fol-lowingreasons :

. to_{support and cover} Iower_{sensitivityand}

_poor

_experience

of_young de$igners,and

. to_{shorten the} lead_{timeofan industrial}

product

for

meeting short-range estimation ofcustomer's demands.

inthis_paper,digitaldesignsupport systems which can cope with above i$suesareintroduced.

2.

Basic

3D

Model

Construction

System from

lmage/

Concept

lnthischapter, a method toconstruct themany basic shapes satisfying the_given

impressionlconcept

isintroduced.

Theshapes _playa role as candidate solutjonsforthedesign

and are indicatedtoa

designen

The designercan decidethe solution fromvarious candidates

_generated

bythesupport

system and hislherexperiences. A basicsystem jsthen

de-veloped according tothe_proposedmethod and thevalidity of

thesystem isverified. Inthedevelopment ofa basicsystem,

design

for

car bodyformisadopted asan example-exercise.

2.1

KanseiVVbrds

A

concept and impressionof a _product are usua[ly

ex-pressedina naturallanguage,The author callsthewords

ex-pressingaconcept and impressiont`Kansei

Words",

The

Kan-sei Words expressing theimpressionsforcar bodyformswere

investigated

by

a

_{questionnaire.}

Sixtywords were then

ex-tractedfromspoken words and were categorjzed

by

the cius-teranalysjs using thedataobtained from the_{questionnajre.}

Asthe resuits ofthecategorization, ten

Kansei

Word$

were

determinedtoexpress theimpressionsforcar bodyforms;the

tenKansei

Wbrds

are`'cute'', "sporty", t'casuai", _formalt',"soft", "stable", "powerful", "luxurious", _un-tired",

NII-Electronic Library Service Patan ParamO (a)Sideview A

(b)

Frontview Fig.1

,

Parametersdetiningcar bodyshape

I,

BLA/Pai'am.6 Param.9 Param.10 mub.i

#

/ps -vauame

me･av

¢

･{

/,,lr:gi'1pm'.tLff.i1

_{nvrb pptup'L-- l.}

,

moa-.-,F-uz'st

ew'r#

vanrv.KntTrmMJ

-.ketFM--y

tas-

-w

lable1. Factorloadingsforside-view$hape impressions FactorlFactor2Factor3Faclor4Factor5 stable

.Q.g62n.o4g0237O.249O.190

spaciousO.Slle.174O.109O.42I-O.102 cute O.ll7'os74n.lgoo.essO.O14 casual-O.371og6g4.012O.099O.222 soft

-o.eo7O,820'-O.067-O.2SS-O.122

lm-tired-O.182O.t62O.884LO.lele.l72 1uxwiouse.o67O.344o7ggO.234-O.210 foIMalO.366n.321e779o.17eO.oo5 powerfu1O.429-O.090e.}22O.S89,O.O07 SPOItY-O.8S30213O.206-O.037O.137

lable2, Factorloadingsforfront-viewshape impressions

;

Fig,2.Develeped system

'･x･･i,eeIM.za.-･intlh,ww1/lu･"-...-...tt,t.

//ti'g,ill･i,'lilu,,ge.t?･

,

iilllllilva111'#.'

illiiitde

'Iliiiill･i,,,l l'i'lt'l'1/'llt'1:'l'l./;-uack /l.;Il'il'i{l･I'cu//'ders

'1'/'ll/'

"

rvr" ''"Ilii

2.2 3D Model Generationfrom KanseiWbrds

Questionnaires,inwhich fiftytwo _{persons gave points}in

from O to5 on each of thetenKanseiWords forthe

typi-caltwentythreeside-view formsand eighteen

_persons

_gave

points

infromO to5on each ofthetenKanseiWords forthe typicaltwelve

front-view

forms,

were executed inorder totie theKanseiWbrds tocarbody forms.From the

_{questionnaires,}

impressionsforthetypicaltwentythreeside-view formsand

the typicaltwelve front-viewformshave been expressed by

theKanseiWords with weights, Astheside-view formsand

the

front-view

forms

are

defined

by

the

form

parametersas

shown inFigure1,theKanseiWords expressing impressions

for

car

body

forms

can

be

relatedtotheparameters

defining

car

body

forms

by

using a neural network model. However,

the neural network model relatingthe

Kansei

Words

tothe

form_parameterscould not _generatethe reasonable formfrom the KanseiWords with the weights representing a concept

becausetheKanseiWords have overlapping meaning ineach other. Forthe_problem,another neura[ network model was constructed asfol[ows.

The common

factors

and the

factor

loadings

were extracted

bythefactoranalysis using the KanseiWords and weights

obtained

from

thequestionnaires,

lables

1

and

2

show the

factor

loadings

for

side-view shape impressionsand front-view shape impressionsderivefromthe

factor

anaiysis.

The extracted factorsdo not have overlapping inthe

meaning.

Consequently,

a neural network model which was

constructed

by

reiatingthecommon

factors

tothecarform

parameterscould generatethereasonable formfromthe com-mon

factors

withweights.

In

theneura[ network model, the

inputlayeristhecommon factorsand the output layeristhe

form _parameters.The$um FPV3,of thefactorIoadingon each common factoris_giventoeach element ofthe inputlayenThe

sum FPVi,,ofthefactorloadingisderivedfromequation

(1

).

r'ifltr#ewsreIg specialissueefjapanEsesocletytorthesclenceofdeslgn

yol.15-4 no,60 2DOB

NII-Electronic Mbra

3

(a}

Soprty_peint:high

(b)

Sableand spacious _points:high

Fig.3._{Resultsgeneratedbydeveloped}system

FP}1,=E(PVleightxF2ictorLoading)

_{1)

ln

theabove equation

(1),

_tr3i",

"maight",

and 'FLrctet'Laading" mean the factornumbec theweight

_given

toaKanseiWord as an impression,and thefactorloadingofa KanseiWord on the common factorsshown

lables

land

2,

respectively.

2.3 System Development and Results

A system

to

automatically

_generate

the

3D

model

from

the KanseiWords expressjng car

images

hasbeen developedby

using theneuralnetwork model as mentioned above, Figure

2

shows the interfacewindow ofthe system bywhich the

Kan-sei Words with weights can

be

input

into

thesystem.

Figure3

(a),

(b),

and

(c)

show the 3D model

_generated

by

thedevelopedsystem. The formshown inFigure

3(a)

shows

theresult from thatthesporty impression_point

[s

hjgh

(5

point)and the_pointsof other impressionsare middle

(3

_point).

The formshown inFigure3(b}shows theresult

from

thatthe

stableand spacious _pointsare high

(5

_point)and the

_points

of other impressionsare middle

(3

_point},The

form

shown

in

Figure3(c}shows theresult fromthatand the luxuriousand

formai

_pointsare high

(5

_point)and the_pointsof other impres-sions are mjddie

{3

_point).

As shown intheseresults, jtwas confirmed thatthedevelopedsystem could _generatetheform

4Twy#ffnk:e

specialissueefiapanesesocietyforthesctenceofdesign

vol.15-4 no.60 20D8

Fig.4.Design_processferdigitalcamera

fromthe KanseiWords expressing concept/impressions ofa car.

3.

Design

Evaluation

System

Using

Mixed

Reality

fechnology

Figure

4

shows an example of

design

processes

for

a

digi-talcamera. Asshown

in

Figure

4,a designerfirstconsiders and ponders his/herideatoconcrete the

_product

forms by drawingmany sketches, The _{productdesigns}represented as sketches are evaluated

by

making rough mock-ups and the 3D digitalmodels areconstructed

by

a

3D

CAD

system under

theevaluation of the sketches and the rough mock-ups. _The constructed 3D digitalmodels arethen converted tothe phys-icalmock-ups bya rapid _prototypingsystem toevaluate the

designabiljtyand operability ofthem.The

final

decision

ofthe designisconducted bythedetailedmock-ups after designing

theuser interfaceincludingthebuttonarrangement. _Inthese

processes,

making mock-ups needs much time and cost. This

problem must be solved tohave highcompetjtive abiljty of

product

development.

Forthis

problem,

a method toevaluate a _productdesignby using virtual models insteadot _physicalmodeis isbeing devel-oped. Sincevirtua[models: 3D digitalmodels, are constructed

in

a virkualspace, itisvery easy and inexpensivetomake and

modify them.And recent computer

_graphics

technologycan

make the

3D

digital

models inhighreality

[4],

[5].

However, virtual models havea serious

_prob]em

tha±itisimpossibleto evaluate _product

design

in

touchingthem

_[6].

Thismeans that virtual models have

difficulty

in

using virtualmodels fordesign

NII-Electronic Library Service

Fig,5.Advantagesand disadvantagesofvirtualmodel, physicalmodel, and hybridmodel

lnthischapter, an evaluation method of designabilityand operability of a _productwhich hastwo advantages: of touch-ingwith _physicalmodels and highefficiency with virtual mod-els isintroduced.Inthemethod, a hybridmodel covering a _physicalmodel with a virtual model isconstructed bythe mixed realitytechnologyinthemixed space. As characteris-ticsofthethreemode[s: a virtual model, a_physicalmodel, and

hybridmodel, are shown inFigure5,thehybridmodel can work forminor model change without making a _physicalmodel

for

design

evaluation.

3.1

Development of System

3.1.1

Foundation

of

The

Method

The introducedmethod evaluates and inspectsthe

design-abilityand operabilityof a

designed

product

by

lookingand touching a physicalmodel covered with a virtual model, The virtual model covering aphyslcalmodel can beeasily changed

by

a

3D

CAD

system,

Virtual

models

having

similar

forms

to

a physicalmodel can

be

evaluated

by

touching the _physical model without an uncomfortable feeling.Humans donot have an uncomfortable feeiingeven ifthetouchingmodel form is sfiightlydifferentfromthelookingmodel form.The reason isas

follows.

Humans havehighpositionsensibility on vision butlow po-sition sensibility on touching.Thisisthekey

point

tomake the introducedmethod successful. Inother words, even ifthere

areiittledifferencesofthebutton

_positions

orthemodel forms between a _physicalmodel and acovering virtual model,

hu-mans donot feelun-comfortableness inlookingand touching

themodel because

_position

identificationbecomes more

ef-fectiveinvisionthanintouch.

3.1.2

ldentificationMethod of RelativeRelationship

tween

Real

Space

Coordinates

and

Virtual

Spece

CoordinatestoMake Mixed

Space

The

introduced

system offersa

design

tool

having

both

ad-vantages of

_physical

modes and virtualmodels bymerging

therealapace and a virtualspace. The one oftheessential

Fig.6,ldentiticationofcoordinates byARtoolkit

(http://www,hltl,washington.edu/artoolkiV)

(http:11mtd.fh-hagenbeng.aVdepoVgraphicsfartoo[kit/)

technologiestodevelopsuch des[gntoolistoidentifythe co-ordinates of thereal space and avirtuaj space and tonaturally overlap them.

The real space _pictureand avirtual space _pictureare

simul-taneously_giventoan operator byusing a video see-through

head mount display

{HMD)

[7],

[8].

Inthismethod, itis

ex-tremelyimportantto

_precisely

cover arealobjectwithavirtual objectwithno timedelay.Inordertorealizethis,theAR toolkit was utilized as shown

in

Figure

6.

The

AR

toolkitcan continu-ously and

_precisely

identifythe relative relationship between

therealspace coordinates and thescreen coordinates of a video see-through HMD inrealtime.

The method using theAR toolkitrecognizes a marker set on a

_physical

model inthe real space. When a _physicalmodel

havinga marker ismoved and rotated bydesigner'shands

in

theevaluation _process,

the

marker sometimes comes into

a

blind

spot.Inthe case, there[ative rela ±ionshipof the

coor-dinates

must

be

lost,

lnorder tosolve this_problem,a hybrid

identification

method has been developed,When themarker

is

in

aview spot,thecoordinates relationship betweenthereal space and a virtual space isidentifiedbyusing theAR toolkit.

On

theother

hand,

when the maker isina blindspot, the co-T-ifltz7anve"seg specialissueetjapanesesocietyforthescienceefdesign vol.15-4 ne.60 2008 NII-Electronic Mbra

5

y

''RapidProlot>'ping.t"=ttt/t/t/t,.

tt=tt

"/,sw,.,y/..m･,v.1/li'･'.

RealSpace DataGiebe

ttprvl

HMD //tttt//./t///t,t/1/t./////･..//t.,////./.111rl'///11t'/llll/,..././//t,.,//-･

Trackin!Seiisor' Trackin!Senser' TrackingSensot

/tt/t

t/tt./lpt./.t.t./li

1･//.t"ttt,l/ t/,.l/ttl}/,/･,

t･/･//･/.

'

_,lll.

t/11/./,. =i/'

t/t.

3PCAI]Dava FingerJoi"t.tsig]e 3DPosiTienmldPos ¢Data 3DCG

/.t. I.'l/,

.///ilt#.

_it.I

"'

_lt.

t

tttt S F.S･ l. f.IY, 11/･fF. :su/, /1.,,!t =../t eel{'

.n.it.rt.

ttt/.,.ty.'i/'"

･;,lg･',li#,/l･i,11,'t/t/t"f/'//''''ttt//ttt.ttt/ttttt.

l-.

ttt

VirtualHand

tl.ttttX,M

VirttLalX･'iew

.,m/t,n//-///･/･//.

tttt/tt'/x/./t/

_{ttttttttt.tttttt.//ttttt.tt.t.tttt.t/tt/...t/t.//ttttt.}

tttttt:ttttttttttt/:/tt

VirtualSpace

Fig.7.Hardware components ofsystem

(a)Physjcalmode)witha

(b)

Physicalmodelwitha

maiker magneticsensor

Fig.8.Markerand magnetic sensor

ordinates relationship jsidentifiedbythe_positionand the

atti-tude

detected

by

amagnetic trackingsensor set on a_physical model and theHMD. The reason using thehybrididentification

method isthattheAR toolkitcan accurately identifythe

coor-dinatesrelationship thanamagnetic trackingsensor buthas

blindspots.

3.1

.3

Developed

System

Flgure7shows the

hardware

components ofthedeveloped

system. Anaverage

form

moder of

designed

and constructed

3D digitalmodels jsformedbyarapid

_prototyping

system and

isused as a _physicalmodel.

A

magnetic trackingsensor isset on thebackside ofthe marker attached on a

_physical

model to identifythecoordinates ofthe _physjcalmodel. Thesystem has

a head mount display

{HMD}

of avideo see-through typeon which a magnetic trackingsensor issetto identifythescreen

coordinates of theHMD. The relativerelationship

between

the_physicalmodel coordinates and the screen coordinates is identifiedbytheAR toolkitand

the

two magnetic tracking sen-sors. Therefore,a _physicalmodel is_preciselycovered with a virtual modei according totherelative coordinates

[9],

A designerwith a HMD can toucha _physicalmodel cov-ered with a virtual model ina dynamic stereoscopic vision.

g,o,e;ct,g'geheg,.,of

N

e--

---

(a>Overlapplngby (b)Overlappingby

mafker magneticsensor

Fig,9,Physicatmodel covered with virtualmodel

Data_grovesare used ±odetectthe

joint

angles of an operator' s fingers.And magnetic tracking$ensors areset on thedata grovesto

_detect

_thepositionsand theattitudesof operator" s hands.Thevirtua[ _hands are constructed and simulated in

a virtual space with these

data:

the

joint

angles ofan

opera-tor'sfingersand the _positionsand theattitudesof operator's

hands.The

_joint

angles ofanoperator's

fingers

and the touch-ing_positionson thefingerstoa _physicalmodel enable one to quantitativelyevaluate and inspecttheoperabilityof the

de-signed user interfaceincludingbuttonarrangement.

3.2

Executionof Developed

System

3.2.1

Markerand Magnetic

ftacking

Sensor

Figures8

{a)

and

(b}

show a _physicalmodel with a marker and a magnetic trackingsensor, respectively.

As

mentioned above, themarker isused foridentifyingthe coordinates

rela-tionshipbetween a_physicalmodel and thescreen ofa

HMD

projecting

avirtual space. And when the marker ishiddenin

the

_proces

of designevaluation, the magnetic tracking sensor

mounted on thebackof themarker isused forthe

identifica-tion.

Figures

9

(a)

and

_{b)

show theviews of a _physicalmodel covered withavirtualmodei according totherelatjve

coordi-67ifly\nezakEg

specialissueofjapanesesecjetyforthescienceotdesign

(a}

Virtualmociel oon$tructiom interface

{b>

Censtfuctedvirtualmodel Fig.1O.Virtuaimodel construction

nates identifiedbythemarker and

by

the magnetic tracking sensors, respectively.A$shown inFigure

9

(b},

themagnetic

trackingsensor was working when themarker was hidden.

3.2.2

Virtual

Model

Construction

The

developed

system can construct a virtual model by

converting CAD data

(STL

data)of _parts,assembling them, and _giving

functions

such as adisp[ay, a button,a LED,etc. to them. Figure1O(a}shows theinterfacetoassemble _partsand

givefunctionstotheparts.The interfacecan also designate

thematerial _qualitytoeach _parttomake realistic computer

graphics.And Figure1O(b)shows an example of a construct-ed virtual model.

3.2.3

Executionof Developed System and Evaluationof

OperabilityOfButtonHandling

Thedesignabilityof designedmodels

(virtual

models) which

havesimilar formstoa _physicalmodel can beevaluated

by

lookingand touchingthe_physicalmodel covered withavirtual model indesigner'ssensibility, The operability of button

han-dlingis

_{quantitatlvely}

evaluated by distributionof thecontact

pointsbetween thethumblfingersand buttonsand the

joint's

Fig.11.Magnetic sensors todetect_positiensand directions ot thumb

and hand

Fig,12,festingview

ang[es of theoperator:s fingers,ltisunderstandable thatthe operability must behigherinthecase thatthe

distribution

of

thecontact

points

betweenthethumblfingersand buttonsis concentrated and the

joint's

angles oftheoperator's fingers aresmaH inhandling

buttons,

Figure

11

shows a

data-glove

to

detect

the

joint

angles of theoperator's

fingers.

And positionsand attitudes of the

thumb and thehand are alsodetectedbythe magnetic

track-ingsensors. Figure12shows a view of testingof button

han-dlingusingthevirtualthumb.

Figure

l3

(a)

and

(b}

show the results of thedistributions ofcontact

_points

for

an easy taskand adifficulttask,

respec-tively.

[n

theeasy task,onlyone buttonishandled.And inthe

difficulttask,pluralbuttonsare handled.As shown inFigure

13

(a)

and

(b),

it

is

quitenatural thatthedistributionof the contact pointsisconcentrated inan easy taskthana difficult task.Therefore,

it

may

be

understandable thatdesignabilityof

button

arrangement isquantitative]yevaluated and in$pected bythedistributionof the contact _pointsbetween thethumb

and

buttons.

7･fft>#anstkfi･E･ specialissueofiapanesesocietyforthesclenceofdesign

(a)

Easytask

Fig,13,Distributionofcontacting point

(b)

Djthculttask

4.

0ther

Systems

to

Support

ConceptiBasic

design

Threesupporting systems toconstruc ±a 3D digitalmodel and sophisticate

it

are

introduced

jnthischapter.

4.1 3D Model Construction

System

fromSketches

Adesignerdraws many sketches toconcrete hisfherideain starting _processofcar styledesign.Sometimes,thesketches are a side view, afrontview, a back view, and atop view as shown

in

Figure

l4.

In

another case, a birdview sketch is ad-opted. Inthe currentlyused method, a clay model iscreated

byreferring tosketches, and thedigital3D model is construct-ed by reverse engineering fortheclaymodel. Itneeds much

timeand cost tocreate aclay model and toadapt reverse

en-gineering

foritso thatthe

_processes

should berestrictedfora

fewmodels.

The author hasdeve[opeda system toautomatically con-struct a digitalmodel froma sketch drawn on a screen

pen-tablet.The system can

be

applied fora side view, a set of a side view, a frontview, a

back

view,and a topview, or abird view, Figure15shows an example ofa

3D

model constructed

bythesystem.

4.2

3D

Model ConstructionSystem

based

on

CharacteF

isticLines

As mentioned jnthechapter 2and thesession

4.1

,a basic

3D

digital

model is_generatedfromaconcept or sketches. The basjcmodel should be sophisticated with characteristiclines.

A

system toconsider characteristic

lines

byconstructing them

on a3D digitalmodel hasbeendevelopedbytheauthon

The

system can construct characteristic linesbythesimple operation:

just

drawingcharacteristic

lines

on a 3D model

indicatedon a screen with _pen-tabletfunctionsas shown in Figure16(a).Figure16(b)shows an example of a 3D digital model on which thecharacteristjc linesareconstructed.

4.3

3D

Model

Construction

System

based

on

Highlight

Lines

ln

the

final

processof car style

design,

surfaces composing a model should bebeautifiedfromtheview _pointof highlight

ljnes.

Hjghiight

linesare derivedas a setot _pointssatisfying

Equation

_{2}

forarbitraryconstant values: arbitraryangle op be-tween _thereflectedIightand thelookingdirection

(Figure

l7).

m2(a'n)(b･n)+(a'b)=conset

(2)

Fig,14._Anexample otcaFsketches _{Fig.15.Constructed3D model}

8T'tf(tz\mxk#e

speciaHssueofjapanesesoeietyforthescienceofdesign

NII-Electronic Library Service

(a)

Drawingofcharacteristicllnes

Fig.I6.Modelingusing characterist]c lines

st

i

(b)

Constructionof characteristic Iines

"Eh r'u.

/f

Fig.17._{Highlightline}ca[cuiation

t"

t

'kx･..

(a}Originalsurtaces

A

surface issophisticated on thebasisoftherelationship

between

evolute and

involute.

The

detail

algorithm to re-con-structa surface according totherelationshipbetween evolute and

involute

should

be

referredtoanother

_paper

[1

O]

because

the explanation ofitisomitted due topaperspace limitations. A system tosophisticateand re-construct a 3D digitalmodel

fromthe view _pointof highlightlineshas

been

developed

by

theauthor.

Figure1

8(c)

shows an example ofcar shape sophisticated totheoriginal shape shown inFigurel8(a)bythesystem ac-cording totherequired a

highlight

line

shown

in

Figure

l

8(b).

5.

Conclusions

This_papermainly introducedthe two supporting systems

forconcept designand operability evaluation. The firstsystem could automatical[y construct a3D digitalmodel formthekey words

(Kansei

words) which express theconcept and imageof adeveloping

_product.

Thesecond system could evafiuateand

inspectthedesignabilityand operability of a _productwithout making _prototypes.

1naddition,threesystems toautomaticaHy construct the3D

digitalmodel fromsketches, sophisticate a 3D digitalmodel

byconstructing characteristiclineson themodel with simple

(b)

lnputofrequired highHghtline

Fig,18.

(c>Sophistieatedsuriaces

Resultformultiple surfaces

7'ff(y#nvRresce

specLalissueofjapanesesocietyferthescienceefdesign vel.ls-4 no.6e 2ees

NII-Electronic Mbra

9

operation, and sophisticate thesurfaces composing a model

according tothe

highlight

lines

required

by

a

designer

were

describedinoutline,

References

[1]

Mdes

M

T.:Sketch

mapping:Concept

tionforstylists,Proceedingsof lnternational

sium on Automotivelechnoiogy&Automation,

₍₁

998},

341-348.

[2]

Hummels C.C.M, :two-Handed Gedture-Based Car

StylinginaVirtualEnvironment,Proceedingsof

tjonal

Symposium

on Automotivelechnology&

tion,

(1

997),227-234,

[3]

Hummels C.C M. iExploitingtheExpressive:Rapid

Entryof Car Designer's ConceptuaiSketches jntoa

CAD Environment,Proceedingsof lnternational

sium on Automotivefechnology& Automation,

_{l

997),

243-250.

[4]

Drews

R

WeyrichM :lnteractiveFunctionalEvaluationin

VirtualPrototypingillustratedbyan Example of a

struction Machine Design,Proc.of IEEE IECON, Vol.4

(1998),2143-2145.

[5]

Zorriasatine,E et al. :A survey of virtua] _prototyping

techniquesformechanical _productdevelopment,Proc.

InstnMech.Engrs,Vol,2l7,No.B4

(2003},

513-530.

[6]

foshikazu,O.et al.:A MixedRea[itySystemwith Visual

and 1langibleln±eraction Capability,TVRSJ,Vol.9,No.I

(2004),

79-88.

[7]

KutulakosK N,VallinoJR :Calibration-FreeAugmented

Reality,

IEEE

fransaction

on Visualizationand Computer

Graphics,

Vbl.4,No.1

(1

998),

1-20.

[8]

Park H,Park J:lnvisibleMarker Based Augmented

Reality

System,

Proc.

SPIE,

Vol.5960, Pt.1

{2005),

501-508.

[9]

Kiyokawa

K,

Ohno

H,and KurataY :AnOptical

Through

Display

for

Mutua[

Occlusion

witha Real-Time Stereovision

System,

Computers

&

Graphics,

Vol.25,

No.5

(200l),

765-779.

[1

O]

lerukazu

Ogoshi,

Hideki

Aoyama,

Masakazu

Ohnishi

:

Studyon DigjtalStyleDesign,

Journal

of the_Japan

ciety forPrecisionEngineering,_Vol,70,No.4,

495-501

.

{in

Japanese)

10fw

y#an.ptukle

specialis$ueQfjapanesesecietytorthescieneeetdesign