PowerGlove
Concepts and current results
Veltink P.H. 1, Kortier H.G. 1, Schepers H.M. 3, Sluiter V.I. 1, Brookhuis R.A. 2, Lammerink T.S.J. 2, Wiegerink R.J. 2
1Biomedical Signals and Systems, MIRA Institute; 2Transducer Science and Technology, MESA+ Institute,
University of Twente, Enschede, the Netherlands,
3Xsens Technologies B.V., Enschede, the Netherlands
First author e-mail: p.h.veltink@utwente.nl
Dynamic interaction between the human body and the environment can be assessed by measuring force and estimating velocity at their interface. A sensorized glove with movement and force sensing is developed in which this principles is applied.
Keywords – inertial movement sensing, six degree-of-freedom force / moment sensing, power sensing, load identification, dynamic interaction.
1. INTRODUCTION
Measuring the physical interaction between the human body and the environment is important in medicine, ergonomics and sports. Potential applications include the assessment of motor performance in rehabilitation, measuring whether the human body is loaded within safe limits in ergonomics and training for optimal performance of motor activities in sports.
The physical interaction between the human body and the environment can be assessed by measuring force and velocity at their interface (Fig 1a). Power transfer can be derived from the inner product of force and velocity [1,2], the dynamics of environment can be derived by relating movement to force under conditions in which the human body is the actor and the environment behaves passively and does not change in certain time periods [1,3].
It is the objective of our current PowerSensor project to demonstrate these principles and to develop a sensorized glove that is able to measure hand and finger movements as well as interface forces. The sensor system that is to be applied on the finger tips and selected finger and hand segments will include inertial and magnetic sensors for movement and an accurate 3D force sensor for interface forces (Fig. 1b). Optionally, the 3D force sensor may be extended by 3D moment sensing. This paper provides an overview of our current research achievements in this area, with reference to papers that provide more detail.
F
v
(a) (b)
Figure 1. The PowerGlove concept [1]. (a) Dynamic interaction between the human body and the environment can be assessed by measuring force and estimating velocity at the interface, (b) a sensorized glove with movement and force sensing is developed in which these principles are applied.
3DAHM2012
Figure ground unit an measur the mas 2. MA Feasibi We pro the env Concep The po load. F by inte applyin global c was sub position informa multipl power t Concep A seco estimat estimat Experim Work p at the i compar (Fig. 2) Develop In addit of the s Inertial Hand a of the f estimat hand an of the h extensiv 2. Feasibility d to a table fol nd an ATI-mi red movement ss. Similar tes ATERIALS AND ility demonstr opose two con vironment, bei pt 1: Sensing p wer transfer b irst, relevant m egrating the si ng known orie coordinates us ubtracted and n. Integration ation about ch lying force an transfer. Perfo pt 2: Identifyin ond order load ted acceleratio tion [3]. These mental method performed and interface betw red to the cha ).
pment of the P tion to demon sensorized glo l sensing of ha and finger mov
fingers and th tion method w nd finger mov hand and 3D a ve distributed (a) y demonstrati llowing arbitr ni45 6 DoF f t and force in sts were perfor DMETHODS ration of conce ncepts concern
ing the estima power transfer between the bo movement qu ignals of the entations at th sing the estim the resulting
drift was redu hange of heig nd velocity, m ormed work w ng load dynam d model, incl on, velocity a e movement qu ds applied in t d load dynamic ween a handle ange of potent PowerGlove nstrating the p ove, including
and and finger vements are e he back of the was initially de vements. It is n accelerometers d movement se ion of power ary trajectorie force/moment formation and rmed using sp epts
ning the assess tion of power r between the ody and the en uantities were 3D angular v he start and e mated orientati acceleration i uced by apply ght was not us multiplying m was estimated b mics luding mass, nd displacem uantities were the feasibility cs were estim and the load, ial energy and
rinciples of dy movement an r movements estimated by fu e hand with a eveloped using now evaluated s and magneto ensing system, sensing and es (b) using a t sensor (a). W d compared to pring loads [2, sment of the d r transferred an
body and the environment w
estimated: the velocity senso
end of the mo ion at any mom integrated to ying known ze
sed. Power tra moment and a by integrating damping and ment to measur e derived as de demonstration mated from iner while displac d the load dyn
ynamic intera nd force sensin fusing inertial kinematic an g a model that d using a com ometers on the , including 3D (b) load identific sensorized ha Work perform o changes of p ,3]. dynamic intera nd dynamics o environment was estimated e sensor orien rs and correc ovement. The ment. Subsequ velocity, and ero velocity at ansfer was es angular veloci power transfe d stiffness par red force as f escribed above n rtial movemen cing the load. namics with th action assessm ng. and magnetic natomical mod t generates sen mplete inertial/m e segments of D gyroscopes i ation. A mass andle with an med and mass potential ener
action between of the environ
for the displac tation was est ting the resul 3D accelerat uently, the gra
velocity subs t the start and stimated at eac ty and adding er over time [2 rameters, was function of tim e. nt sensing and The estimated he known cha ment, we are de movement se del of the hand
nsor signals w magnetic sens f index finger a s currently be s was moved n Xsens inertia s were estima rgy and actual
n the human b nment. acement of an
timated at any lting estimate tion was trans avitational acc sequently to c d end of the m ach moment in g both compo 2]. s identified by me using leas d force/momen ed work perfor aracteristics of eveloping a pr ensing on the d. The applied when performin sor module on and thumb [4] eing developed from the al sensing ated from l value of body and unknown y moment for drift, sferred to celeration change of movement, n time by onents of y relating st squares nt sensing rmed was f the load rototype segments d optimal ng certain n the back ]. A more d. 3DAHM2012 43
Figure (c) mom Six deg The six by man An app Capacit compon part, fo determi shear fo part. An [5]. 3. RE Feasibi The est accurat The ma and dam due to m Develop Inertial The firs movem magnet descrip measur Figure prototy (a) 3. Working p ment sensing, gree-of-freedo x degree-of-fre ny thin silicon plied load to t tive read-out nents simultan our triangular ined by measu orces and a m n applied shea ESULTS ility demonstr timated work te within 4 % ass parameter mping. The st mass and dam
pment of the P l sensing of ha st version of t ment sensing m tometers on th ptions of the se ring hand and
4. (a) first ver ype kinematic principle of the (d) shear forc om fingertip fo eedom fingert pillars. These the sensor resu
is used to det neously. For d r electrodes a uring the chan moment around ar force cause ration of conce performed on for varying m of the mass lo iffness param mping. Varianc PowerGlove and and finger the inertial and module on the he segments o ensor fusion a finger movem (a) rsion of a mov glove with pa top electrode (b) e six degree-o e sensing orce/moment s ip force/mome e pillars are the ults in a smal ect this displa determining fo are realized in nge in capacita d the normal a es a differentia epts n a 9.4 kg mas movements wit oad was identi eter of the spr ces accounted r movements d magnetic mo
back of the ha f the index fin algorithms, thi ments are prov
vement sensor artial instrume Fz of-freedom for sensor ment sensor con
e spring eleme ll displacemen acement, deter orces/moments in the bottom ance between axes, comb-sh al change in c ss and a spring th durations b ified within 5 ring load was d for were abov
ovement sens and and small nger and the th is movement s vided in anoth r PCB with 3D entation of ind (c) rce/moment se nsists of a silic ents of the for nt of the top e rmining all ax s which cause m wafer (Fig. these electrod aped electrode capacitance wi g with a sprin etween 3 and % error, with identified wit ve 99 %. ing system, in l distributed P humb, is illust sensing system her paper in thi
D acceleromet dex finger and
Q3 Q1 Q2 Q4 ensor [5] (a,b)
con top electro rce sensor for electrode relat xial force com out of plane d
3a,b,c). Eac des and the top
es are realized ith respect to g constant of 5 s [2]. h negligible fo thin 3 % error ncluding a com PCBs with 3D trated in figur m and the initi is proceedings
(b) ter and magne d thumb [4] Mx (d) ) normal force ode which is s all degrees of tive to the bot mponents and
displacement ch force comp
p wafer. For m d in the top an
the top wafer
88 N/m appea orces due to st r with negligib mplete Xsens acceleromete re 4. Further ial evaluation s [4]. etometer, (b) t C1 C e sensing, supported f freedom. ttom part. all torque of the top ponent is measuring nd bottom (Fig. 3d) ared to be iffness ble forces ers and of the first 2 Fx 3DAHM2012 44
Figure measur Six deg The six up to 50 around directio 4. DI It shoul perturb estimat are perf daily-li In the c demons the glov In addit glove is control 5. AC This re the Tec 6. RE [1] P.H Ap [2] P.H env [3] H.G Int 30 [4] H.G of thi [5] R.A six Co acc 5. (a) SEM pi ring forces and gree-of-freedo x degree-of-fre 0 N in normal each axis. Fo ons, moments ISCUSSION ld be noted th bing a constant ted dynamics formed in a cl ife situations, coming period strate the pow ve. tion to assessi s expected to l, as well as in CKNOWLEDGM esearch is supp chnology Prog EFERENCES H. Veltink, D pplication 200 H. Veltink, H vironment, IE G. Kortier, P ternational Co August – 3 S G. Kortier, H the XII Intern is proceedings A. Brookhuis x-axis force-to onference on cepted. (a) icture showing d moments, (b om fingertip fo eedom force-m l direction and orce measurem showed a full hat load identif t passive load are the combi losed dynamic
possibly by id d, we plan to c wer sensing an
ing the dynam be useful in b n animation an MENT ported by the gram of the Mi Device and m 090056445, M H.G. Kortier, H EEE Transactio .H. Veltink, L onference of t September 201 H.M. Schepers national Symp s. s, R.J. Wieger orque sensor Micro Mech
g the top elect b) Six degrees orce/moment s moment senso d 10 N in shea ments showed l scale error o fication was p d. When the lo ned effective c loop. It is, th dentifying the combine inerti d load identifi mic interaction biomechanical nd serious gam Dutch Techn inistry of Econ method for me March 5, 2009. H.M. Scheper ons on Biome Load identific the IEEE Eng 11, 3 pp. , P.H. Veltink posium on 3D rink, T.S.J. La with a large hanical System trode with the s of freedom f
sensor or has been rea ar direction, th a full scale er f 2.8% around performed und oad is varying dynamics of b herefore, impo performed ac ial and force s fication princip n between the l analysis of h ming. nology Founda nomic Affairs easuring the rs, Sensing p edical Enginee cation during gineering in M k, inertial sen D Analysis of ammerink, M
range for bio ms, MEMS 2
e silicon pillar force-moment
alized and test he moment ran rror of 0.9 % i d the normal a der the conditi and/or genera both human b ortant to asses ctivity. This is sensors in the ples when inte
human body a and function a ation STW, ap s. dynamic inter power transfer ering, vol. 56, object handli Medicine and sing of hand f Human Mov M.J. de Boer, K omechanical 2012, Paris F (b) s and the com
sensor mount ted [5] (Fig. 5 nge is between n normal and and 1.1 % arou on that the hu ates forces on ody and load, s the measure s to be further prototype sen eracting with t and the enviro and assessmen pplied science raction betwe r between the 2009, pp. 171 ng, Proceedin Biology Socie and finger mo ement, Bolog K. Ma, M.C. applications, F, 29 January mb-structure fo ted on a test P 5). The force r n 0 and 25 N of 0.4% in sh und the other uman body is a the human bo , since measur ement conditio investigated. nsorized glove the environme onment, the se nt of human m e division of N een bodies, U e human body 11-1718. ngs of the 33r ety, Boston M ovements, Pro gna I, 18-20 Ju Elwenspoek, The 25th inte y – 2 Februa or PCB [5] range is mm hear axes. actively ody, the rements on in e and ent using ensorized motor NWO and US Patent y and the rd Annual MA USA, oceedings uly 2012, Scalable ernational ary 2012, 3DAHM2012 45
