HAPTIC INTELLIGENCE
MAX PLANCK INSTITUTC POR INTElllGENT SYSTEMS
Ham ard Katherin# J. Kuch#n6oc1r
Htętoc tnte2Qeoce Dopartmoa!. Ptinck Insatuie for totefcgent Systems
The sottn#s# of human ftnp#r padł f«ndor$ highły ot grasplng o6)2cU
Mow much does the swrat wwtW bom M Kun soft#n the inger pad?
II« not fcasibic to mcchancefty test n v2vo slon to utocrstand the prcperbes ot 2># stratum corneom layer tostead. wo Infer the itratum coc ncuma properttos by Mtnj a tn2e etement model to particulir correct data. We thtn compore the optlmlaed materia! propenie2 for dry. nahrU. and mc2t Inger2
liardMoicaiyaoiro
1 0#1gn tor th# lng#r pr#1ng KmUabom « CO^SOC _«
crr#WAtr#v^K^.»#A2i.1r<t#f«iołr#HM
• Ocr#no»2m#r«m»»rorm10r#<SOACr3«peMU2 b#hv##n r# v>9#c «rd ta {M<u pOA#
• «Y©»«vjc1j Vecorr#c<#r««#ndeer«mieMorce#l#ech •3 pMtoi cr N fcrjM boa
- Th# oo#ao#r##im>c Keten b#h»##n is» lng#r p#d #nd to# 2"
gaMpUA#wMMn»tdior«a#e«#^#o#Kfi(iouehPł W,"-M
2 #iod#<: On# l#cm Ogd#n mod#f
• w-£ur 2/T 2JT-U■2#»•1«#t3-2|4) w#i«d2•2#«<4(?>tyce******
• Botot»««fr4lumco«aw2end2»r#Mof r2 V>Q#r tam# toto2 »«• hyp«
W# Morcfod tor oę&~2x«J y vato#s th2» protode ta mnm«m m#2n tooared #rror |MSC) b#hw#en th# ston/etod and m#etur#d data by r#peaftng the s^rotaion 2eh de«r#m comfcwator2 ot #K and ^.
Th# opamoed v2to#s o2 th# thear modjtoa tor eech bitu# typ# (Kretom corrxvm re#t oł W>g#ł) produce a r#lafconihp txe»e#n the imtatod correct af#a a^J »a normai tero# ctose to th# ttUacmhe fcxnd m th# m#aur#d data.
The e2vaet#d ccmbrnatons (hk-M#/) repreient t» aohosa of th# t#K#d fnp#r to th# dy. raiurai. anj »0d« Kaaea
i Apco-atosltl
Ahghlwf<yłw*lw MW tefff^Buwęr* *rg*S COMttt to*e |S00 Ki)
lorc4pt^fMlw<acfif«a irwearM toai
remarf>ceo"yr«tl>P3
2 Eip#rm#m2
• TM Kt^cl n ^ ITYWM N2 OP h#p lAJ^ ^
no10 to k##p a 1e#r o2 amim !#2•< to#tof #1 ftoftO o P#4>^ «1#2« 0hy2c2#>2c2tt 1•##( 2My«0y fili mitr)
• Th#
i»er
Th# tog2 pr#M#s la ptadpm «p to 1 .$ N «nd ton Mb e1
Th# rao2fcr# Ol re vig#r 4 <n#»ms#e #01r Ahu##n»jeioc» co^)#to3c«t<2ay#4Sim##.p#rto2wQ orewecfcei iStoabtotMcncrfwe^ftiunnoafcseconsfton2
1. Oiy ftoger
2 tafty* ftng*
#»♦<
Mrr
«eai2 ip*
. 1.124 kP#
3 Modllrto#r
• ttlttkP*
• ti,, OMkp#
• Th# (totrcK# mean moćiture vatoes re^jeed dflerent shear modufc to 2eh2v# a good fu to e2penmer2ai data
• Th# modt ftnp#r pad haa higher correct aree than the oTor ino fmger condrtons at the sam# normai lorce.
• The stratom com#um ot re most bng#r o mdced KgnfcantJy sotter o2x - J2.tSSkP2) tfvan th2 ot the sam# finptr v%hen <tsłnanaturalo2K • 6bJi2kpj)ordry bK • eaotoihj) itate.
• Each raw correct image ts transtormed to eetmate the mage rot the camera wc2rfd see it A was posrfcncd perpendcuUr to the gtass pfato
The raw deu consdt ot une-stacfced i^gerpont imago2 and the wrwcntog normai lorce measuremerfci
• The two data soutoes are sync2rori»d «2h one another based cn tm# ttamps
• Otobei threshcfdng and then tocai adaptwe rresho»<$ng on each image prodoce binery mapes. prov2dng t2e reaf łnjor correct ar2a |2J.
• The gros2 correct area is całcutatod by oxoractng the corrze2 hol of the potefts constoered to be in contact