stature, age, and gender effects on reach motion...

Stature, Age, and Gender Effects on Reach Motion Postures

Don B. Chaffin and Julian J. Faraway, University of Michigan, Ann Arbor, Michigan,Xudong Zhang, University of Illinois, Urbana-Champaign, Illinois, and Charles Woolley,University of Michigan, Ann Arbor, Michigan

The rapid adoption of software to simulate human reach motions in the design ofvehicle interiors and manufacturing and office workstations has required asophisticated understanding of human motions. This paper describes how morethan 3000 right-arm reaching motions of a diverse group of participants werecaptured and statistically modeled. The results demonstrate that stature and agehave a larger effect than does gender on reach motion postures for motions cho-sen by the participants while reaching to targets placed throughout a typical auto-mobile interior We propose that these methods, models, and results can assistthe further development of human motion simulation software for ergonomicpurposes, such as for the design or evaluation of vehicle interiors or industrialworkplaces, to ensure that various population groups are pl-hysically accommo-dated.

INTRODUCTION

Many common tasks require that a personcoordinate the movement of multiple segmnentsof the body to reach successfully to a designat-ed location in space. The task of interest in thispaper is reaching to a variety of locations with-in the interior of an automobile (e.g., movingthe right hand from the steering wheel to thecenter console, overhead console, radio, orpassenger side door). Some of these motionscan require one not only to extend the upperextremity toward the designated target locationbut also to rotate the shoulder and torso toassist in the motion. As described in a recentstudy by Zhang and Chaffin (1997), thesetypes of motions can be modeled based on afour-segment linkage system consisting oftorso, shoulder, arm, and forearm-hand seg-ments. Various empirical and kinematic model-ing approaches have been used to predict thesetypes of complex motions (see Badler, Phillips,& Webber, 1993, for further information onthese models).

The purpose of this paper is to empirically

describe and analyze the effects of size, age,and gender on the postures chosen while per-forming these types of reaching motions. Ifthese three demographic effects can be satis-factorily quantified, the understanding ofhuman motion will be enhanced and futurehuman motion simulation software will moreaccurately reflect these population attributes inthe resulting motion predictions.

There is little debate that the size of an indi-vidual often has an important effect on pos-tures when the reach target is located awayfrom the body. In a static sense, when theupper extremity is extended so that the hand-forearm-arm segments are nearly aligned, thereach boundary is defined by the length ofthese body segments unless the shoulder andtorso are also allowed to move toward the tar-get. Computerized human reach simulations,such as those performed by SAMMIE'TM,Crewchiefm, SAFEWORK"M, and JACKTM,often predict reach boundary conditions with-out shoulder and torso assistance, unless theuser specifically modifies the initial torso andshoulder postures (Karwowski, Genaidy, &

Addrcss correspondence to Don B. Chaffin, Center for Ergonomics. University of Michigan, Ann Arbor. M1 48019-2117;[email protected]. HUMAN FACTORS, Vol. 42, No. 3, Fall 2000. pp. 408-420. Copyright t 2000, Human Factorsand Ergonomics Society. All rights reserved.

STATURE, AGE, AND GENDER

Asfour, 1990). This strategy, though renderinguseful approximations of extreme reach pos-tures, does nothing to assist a workplace orvehicle designer in understanding how different-size people actually choose to move and posi-tion themselves when performing a variety ofreach tasks in their particular reach boundaryconditions. This latter type of information isvaluable inl designing a workspace in which aperson can reach to various areas in an unob-straLcted fashion.

A reason to believe that age could affectcomplex reaching motions is derived from thefollowing studies. The most general evidence isfrom the 1987 National Hlealth InterviewSurvey (Supplement on Aging), in which self-reports of performance declines by older indi-viduals were greater in motor tasks requiringmultisegmental coordination (e.g., reachingoverhead, crouching, lifting objects, andbathing) than in simpler motions (e.g., shakinghands or eating; Lawton, 1990). Also, one lab-oratory study of 61 participants 20 to 80 yearsof age by Potvin, Syndulko, Toutellotte,Lemmon, and Potvin ( 980) disclosed thatdecreases in performance time were greaterwith age when the tasks required multisegmen-tal motions (e.g., cutting with a knife orputting on a shirt), than with simpler motions.Ball-throwing motion studies by Haywood,Williams, and VanSant ( 199 1 ) revealed thatolder adults tended not to rotate their trunkand arm as much as younger adults did. Theyconjectured that these changes could becaused by loss of flexibility in the shoulder gir-dle; by fear of pain (or actual pain), inhibitingtorso and shoulder motions that approachone's range of motion at a joint; or by thedesire to protect against postexertion soreness.Finally a recent study by Cavanaugh et al.(1999) disclosed that in standing forwardreaches, older participants tended not to rotateand flex their torsos as much as younger onesdid, thus demonstrating diminished maximumvolitional reach capability with age.

Even though most normal reaching motionsin daily activities do not require maximumeffort, muscle strength does become importantas the motion of a particular segment approach-es the end of a joint's volitional range of motion.Because muscle strengths vary by as tnuch as

15:1 in the normal healthy population(Chaffin, Andersson, & Martin, 1999), andbecause women's shoulder and arm strengthsaverage about 55% of men's (Laubach, 1978),extended reaches in which the shoulder ishighly stressed may result in a woman'smotions being different from a man's. A com-parison study (Chaffin, Georgi, Baker, &Nussbaum, 1998) of six women and six menperforming reaches to a passenger side doorindeed disclosed that the shoulder strengthrequirements, when the upper extremity wasextended toward the passenger door, wereequivalent to 34% of the average woman'sstrength and 26% of the average man'sstrength. Others (Gallagher, Zuckerman,Cuomo, & Ortiz, 1996; Kumar, Chaffin, &Redfern, 1988; Murray, Gore, Gardner, &Mollinger, 1985; and Yates, Kamon, Rodgers,& Champney, 1980) have shown that age (par-ticularly after about age 50) is associated witha general decrease in population strengths.Collectively, these studies support the hypothe-sis that older women demonstrate less armabduction/flexion angles than do younger menwhen reaching away from the body and, thus,may use more torso compensation (motions)to complete such a reach.

To test these hypotheses, we asked a groupof men and women of varied statures and agesto sit in a driving simulator and reach to vari-ous targets while being carefully monitoredwith a three-dimensional (3D) motion capturesystem. The resulting motion data were used toderive joint angle changes for each reach,which were then analyzed by a new functionalregression analysis method to determine thepotential elfects of stature, gendei- and age.

METHODS

Participants

Thirty-eight participants (20 men and 18women), demographically representative interms of stature and age of drivers, volun-teered to serve in the experiment and werepaid for their participation. Table I presentsthe mean and standard deviation values of ageand gross anthropometry of these 38 partici-pants. These values compare reasonably well

409

Fall 2000 - Human Factors

TABLE 1: Mean and Standard Deviation Values of Participants' Age, Stature, and Weight

Gender Age (years) Stature (cm) Weight (kg)

Female (n = 18) 34.2 ± 14.1 162.2 ± 8.8 60.9 ± 10.7

Male (n = 20) 36.3 ± 16.8 175.2 ± 9.4 79.3 ± 15.6

Overall (N = 38) 35.3 + 15.4 171.3 ± 11.1 70.6 ± 16.2

with those published for the U.S. population ingeneral (NCHS, 1976).

Experimental Procedures

The experiment was conducted in a drivingsimulator developed in the Center for Er-gonomics at the University of Michigan. Thedriving simulator consisted of a mock-up of avehicle with a driving scene graphically ren-dered on a big-screen monitor in front of thevehicle, coupled to the primary controls (e.g.,steering wheel, accelerator and brake pedals) ofthe vehicle. As illustrated in Figure 1, 38 tar-gets, grouped into five areas emulating typicalzones to which a driver would normally reach,were presented in the simulated vehicle, Thesefive target areas included a center console,instrument panel (radio and climate control),overhead console, glove box, and passenger-

OVERHEAD CONSOLE

side door. The later statistical analysis combinesreaches to the glove box and passenger doorand refers to these as far reaches toward thepassenger-door area.

During the driving task. participants wereintermittently signaled, by a graphic on thevideo monitor and a tone, to reach toward oneof the 38 targets in a natural manner and totouch a smiall switch at the target. When theextended finger touched the target switch, theywould remain in that posture for about 3 s, andthen a tone would signal them to return to theoriginal driving posture. Each reach was repeat-ed once, and the order of a total of 76 reachingmovements was randomized. A practice sessionwas provided to allow participants to gain suffi-cient familiarity with the tasks. The fore and aftseat position was adjusted as preferred by eachparticipant at the beginning of the experimeent.

DISPLAY

MOVEABLETARGET

g<0 GLOVEBOX / FRAME

FR

SNEAR_ OR-~~~ "PASSENGER DOH:ORI"

Figure 1. Rear view of driving simulator an-d rcach targets (not to scale).

410


but up-down and tilting motions were fixed.The seat pan had a 90 angle from horizontal,and the back rest had a 22° angle from vertical.The seat had medium-sized side bolsters on theseat pan and seat back.

Spherical reflective markers were placed onpalpable body landmarks identifying the rightwrist, right elbow, right shoulder, supraster-nale, and right and left anterior-superior iliacspine (ASIS), as well as on a head-mountedframe (see Figure 2). An optoelectronic motionanalysis system (MacReflexTmi, Qualisys Inc.,Glaston, CT) with four cameras captured themotions of these markers at a sampling fre-quency of 25 liz. This system has the ability totrack the captured motions as a succession oftwo-dimensional images (four images corre-sponding to four cameras for each time frame),and then delivering the 3D Cartesian coordi-nates of the markers as a time series.

A four-segment biomechanical linkage thatrepresents the torso and upper extremity(Figure 2) was constructed by connecting thewrist joint, elbow joint, shoulder (glenohumer-al) joint, suprasternale, and the bottom of the

Anterior sagittalplane of head(helmet mounted

Ulnar notchof radius

torso. A separate neck-lhead link was included,but the analysis of motions of this segment isnot included here. In addition, for graphic pur-poses, the torso link included a pel ic triangu-lar linkage, which was scaled and positioned inthe sagittaL plane as described in Chaffin,Faraway, and Zhang (1999).

The 3D coordinates of the surface markersfor the wrist, elbow, and shoulder were trans-lated into the corresponding internal jointcenters using a procedure developed by Nuss-baum, Zhang, Chaffin, Stump, and Raschke(1996). Implicit in this linkage representationare three assumptions: (a) the four body seg-ments contained in the system are rigid links;(b) the right hand is considered as a rigidextension of the forearm (i.e., there is no rela-tive motion between the hand and forearmand the distal tip of the extended index fingeris the end of this link); and (c) the bottom ofthe torso link is estimated as an axis connect-ing the two ASIS markers. Based on this link-age, the 18 joint angles defined in Table 2were computed for every time frame. The pro-files of these joint angles described the

Top of head(helmet mounted)

Acromlon

Interior jugular notch-..... of the manubrium

Anterior superiorili a ellac spine (ASIS)

Rod across the base /of the metacarpal bones

Figure 2. Depiction of linkage system and body surface markers.

411


TABLE 2: Angles Used to Define Postures and Motions

Global Angles Local Included Angles

1. Trunk forward flexion 1. Neck-head flexion2. Trunk lateral flexion 2. Neck-head rotation3. Trunk axial rotation 3. Neck-head lateral4. Arm vertical 4. Shoulder vertical (relative to torso)5. Arm horizontal 5. Shoulder horizontal (relative to torso)6. Forearm vertical 6. Humeral rotation7. Forearm horizontal 7. Elbow included8. Hand vertical 8. Forearm rotation (supination/pronation)9. Hand horizontal

10. Hand rotation

time-varying joint kinematics during eachmeasured reaching movement. These jointangle data then served as the input for thesubsequent statistical analysis to allow us todetermine the significant effects of stature,age, and gender.

Statistical Analysis

A more detailed description of the statisticalmethods used in this paper may be found inFaraway (1997), and a general exposition ofth~e analysis of functional data appearing inRamsay and Silverman (1997).

As an example of the data, the elbow-included angle when reaching to the glove boxarea is depicted in Figure 3. As shown, theang,ular changes for each reach are relativelysmooth. Because the motion times when reach-ing to the 38 targets varied among people andtarget locations, the data were rescaled so thatthe motion times represent the proportion of

ElbowIncluded

a(t)(degrees)

Time(normalized)

ReachConcluded

Figure 3. lypical elbow-included angle data with asmoothing spline function.

time required to complete a motion. Thisallowed the angle values for each joint to bedirectly compared for each of the experimentalconditions of interest throughout the move-ment without the confounding effects of slowand fast motions. This nonnalizing procedurewas also justified by an earlier study (Zhang &Chaffin, 1997), which concluded that instanta-neous postures were not significantly affectedby the speed of reaching movements when per-formed no faster than about 20% above theirchosen normal speed.

Though both global and local joint angleswere computed for the reach motions, only thefollowing six angles are reported here: trunkforward flexion (global), trunk lateral flexion(global), trunk rotation (global), shoulder ver-tical deviation of arm from torso (local), shoul-der horizontal deviation of arm from torso(local), and elbow flexion included angle(local). We believe that these six angles are themost intuitive angles and, generally, are inde-pendent of one another (e.g., if one laterallyflexes the torso with the arm held in a constantposture relative to the torso, the includedshoulder vertical abduction angle values willnot change, given that they are defined with alocal coordinate system that rotates with thetorso).

The statistical model used to determine thepotential effect of stature, gender, and age onthese six angles is

a,,,ft) = OJt) = PJ(t)x + j(t)yj + 3 (t)zj +,(t)height, + t(t)age, + ft(t)sexi + ei,(t),

where heighti is standing height in centimeters

412

.2 .4 .6 .8 0


of individual i; agei is 0 for those younger than50 years and I for 50 years or older; sexi is 0for female and I for male; (x1,y1,zj) are thecoordinates of the jth target; and aij(k() is theangle curve over time for individual i, target i,and replicate k (k = 1,2). More precisely, the(xj,yj,z1 ) are measured relative to an origin,defined as the right ASIS point for the parti-cipant.

Each participant was allowed to adjust theseat position in the fore and aft directions; thustargets requiring forward reaches (e.g., towardthe radio) might not be as susceptible to theparticipants' anthropometry as those to theside (toward the passenger door). Because it isa normal condition of driving that drivers canadjust their seat fore and aft position, allowingthis adjustment in this study appeared to bewarranted. This would have the potential forreducing the effects of anthropometric varia-tions when reaching to targets near the sagittalplane, but it would not have as much effect onlateral target reaches. The results are presentedfor each target area separately to better under-stand this potential difference.

The coefficient curves are :(t), which areestimated by fixing t for each proportional timeperiod. In essence, the model becomes theusual multiple linear regression model, theestimated coefficients of which give the valuesof J3(t) at each chosen t interval for a designat-ed target location. To derive a comparisoncurve, we simplv repeated the process for anevenly spaced grid of 20 values and for eachtarget and demographic condition. The stan-dard errors and other common regression sta-tistics are computed pointwise for the 20periods and for each experimental condition.Coefficient of determination r2 values disclosethat about 70% of the variance in the angledata are accounted for with this type of model(Chaffir, Faraway, & Zhang, 1999).

Faraway (1997) provided additional detailson the construction of the model. One of theadvantages of such a model is that residualcurves may be calculated and examined.Outliers and influential observations may bedetected in a manner analogous to standardregression. Adequacy of fit may also bechecked by comparing the residual variation inour models with the variation found in repli-

cated reaches. We performed these checks forall the models used here and verified theiradequacy.

The inclusion of only one anthropometricvariable (i.e., stature) in the statistical modelwas based on a stepwise procedure in whichbody weight, arm length, seated height, andseveral other anthropom-ietric variables weresystematically added to the regression model.None of these had a statistically significanteffect beyond that provided by stature alone,

RESULTS

The statistical results are presented graphi-cally; the j3(t) effects (solid line) and 95%pointwise confidence bands (dotted lines) thatwere found to be most significant are depictedfor the six major included body angles.Because the direction of the reach motionaffects the particular angles of interest, resultsare presented by the four reach zones (i.e.,reaches to the radio, center console, overheadconsole, and far right reaches toward the glovebox and passenger door).

Only those angles for which stature, gender,or age had both a statistically significant (t >2.0) effect and a practical effect (greater thanan estimated 1 difference in an angle) aredepicted graphically. To illustrate the practicaleffects of the angle differences on chosen pos-tures, the results are also depicted with a 3Dstick figure in several terrminal postures requiredto reach some of the targets.

Stature Effects on Reaching Postures

Because the participants were allowed toadjust the fore and aft position of the seatprior to beginning each experimental session,it was possible that stature (which varied fromabout 144 cm to 195 cm) would have a signif-icant effect on som.e of the far side and over-head reaches but less effect on the forwardreaches to the radio and center console.Indeed, this appears to be the case; Table 3depicts the significant angle effects (per 10 cmof stature) at the destination of the reaches tothe four target areas.

Inspection of Table 3 indicates that whenreaching to the far targets located to the side(i.e., glove box and passenger door handle),

413


TABLE 3: Significant Mean Changes in Six Major Body Angles for Each 10 cm of Stature when theHand Reached to the Four Target Areas

Target Trunk Trunk Trunk Shoulder Shoulder RightArea Forward Lateral Rotation Vertical Horizontal Elbow

Console ns -0.5° ns - 2.1° 3.0° - 7.1°Radio 3.3° -1.0° 2.00 - 7.4° ns - 4.10Overhead 1.10 -0.4' 1.80 - 9.9° ns -10.3'Far Right 6.10 6.20 2.90 -10.0° 3.10 - 2.00

Note: ns = no significant difference could be found.

stature had its greatest effect: Taller peoplehad to use far less trunk lateral bending (about60 less/lO cm of height) and shoulder vertical(abduction) angle (about 10i less/I10 cm) thandid shorter individuals. Increased stature alsohad a profound effect on the amount of shoul-der vertical abduction and elbow flexion angle(about 10°/10 cm) when reaching to the over-head console.

The angle effects and 95% confidence inter-vals for these four major angle changes through-out the reach motions are depicted in Figure 4.These graphs depict how the angles change foreach 10 cm of additional stature. As can beseen, the greatest effect of increased stature istoward the end of the reaclh motions.

A stick figure of the postures predicted bythe functional regression model for two

Far Right Reaches towards Glove Box and Passenger Door

Trunk Lateral

C),

0

C)p0.

a)

ao

Cr

Shoulder Abduction

Reaches to Overhead Console

Elbow

E

aC)aa)

Q0)

0a)x,C

0.0 0.2 0C4 0.6 0.8 1.0

Proportion of time

5.0

0.0

-10.0

0.0 0.2 0.4 0.6 0.8

Proportion of time

Figure 4. Mea n and 959,%O confidence lines for four angles attributable to stature (angle differences in degreesper 10 cn of stature).

E °c-)0

,, -2.00.-ac;,

O) -4.0*a):a

c -6.0

Shoulder Abduction

E -2.0

4i -4.00-

a -6.00)

a)a -6.0.2 -8.

C -10.0

1.0

I ... ...... -

.- -1 - - - - - - - -

... .......

- -i i � - .- -6- -i - " -4 -- --- � �� - �� " __-_ 4 " __ �" - i -",_ _ � � ; � - � ; 7 � ; � �7 � -__ - ' ' � w !� � ..I, � 4 4 � .. .....

4@14


extreme-stature men is depicted in Figure 5.This demonstrates the average effect of staturealone when performing a lateral far right reach,independent of gender and age. The additiveeffect of gender and age is presented next.

Gender Effects on Reaching Postures

Because anthropometry (other than stature)and gender may have a covariate effect, caremust be taken when analyzing the effect ofgender alone on reaching postures. The resultsassume that stature accounts for all of theanthropometric effects regardless of gender.This may not be the case (e.g., on average theshoulder breadth for women is generally nar-rower than that for men, regardless of stature).Other gender-related attributes, such asstrength, may also affect the differences, as dis-cussed earlier.

Once again, using the postures at the end ofthe reaches to the four target reach areas forcomparison, we found that some men's pos-tures are significantly different from those ofwomen, as depicted in Table 4. In general,however, the gender effects in these data aresmall (less than 30) Two of the largest jointangle differences during the reach motions aredepicted in Figure 6 for shoulder verticalabduction (during the far-right-side glove boxand passenger-door reaches) and for elbowflexion (for overhead console reaches). Theseindicate that men tended to have slightly lessshoulder angle abduction and elbow flexionthan did women for these reaches, especiallytoward the latter phase of the reach motions.

Age Effects on Reaching Postures

To evaluate the potential effect of age, inde-

- Large 95%tile stature Male, Age 40

- Small 5%tile stature Male, Age 40

Figure 5. Stick figure representations of twoextreme-size men reaching to far-right passengerdoor location.

pendent of gender and stature, an analysis wasperformed to determine the potential effects ofbeing younger or older than 50 years. The thresh-old age of 50 was chosen because many individ-uals are very active and drive vehicles in their50s and 60s, and yet concern is often expressedabout the possibility of decliniing perceptive-motor capabilities begirnning around this age, asdiscussed in the introduction. Of the 38 partici-pants, 14 were 50+ years old (the oldest was 74years old), which provided enough statisticalpower to test the effect of age. In the resultingstatistical regression model, age is treated as acontinuous variable with a linear effect on jointangles because the sample size was not largeenough to test nonlinear models.

Table 5 depicts the mean effects for the

TABLE 4: Summary of Significant Mean Changes in Six Major Body Angles (in degrees) for MalesCompared with Females when the Hand Reached the Four Final Target Areas


Console ns 1.00 1.0° 2.00 ns -2.8°Radio ns ns 1.2° ns -1.0° -2.2'Overhead ns 1.30 ns -1.0° ns -2.4°

Far Right 1.90 -1.2° ns -2.5° 1.50 -1.5

Note: ns = no significant difference could be found.

415


Far Right Reaches towards Glove Boxand Passenger Door

Shoulder Abduction

Reaches to Overhead Console

Elbow

0'

03a)

1D ,S. -2-

< 3

0.0 0.2 0.4 0.6 0.8 1.0Proportion of time


Figure 6. Mean angle differences (and 95% confidence lines) for the two angles found to have the largest dif-ferences attributed to being male compared with fe'male anigle predictions.

older group compared with the youngergroup. It reveals that the largest consistenteffect exists in the shoulder horizontal andvertical angles; the older participants tendednot to elevate (abduct) their arm as much anddid not rotate their shoulder forward as muchas did the younger participants. In essence,the older participants tended to keep theirelbows closer to their torsos during the vari-ous reaches. Inspection of the reach graphsdepicted in Figure 7 shows that this "conserv-ative" arm posture is consistent throughoutthe reach motions, not merely at the terminalpostures analyzed in Table 5.

DISCUSSION

We believe this study illustrates a newmethod of studying and characterizing how

various population attributes can affect humanmotions and preferred reach postures. Such amethod provides a new means to understandcomplex human motions and, perhaps, to re-solve the 75-year-old Bernstein problem: Howdoes the human body coordinate a large num-ber of muscles and joint rotations so that thehand trajectory is "well behaved"(Latash, 1998)?

We believe this approach also indicates howstature, a dominating factor in movement char-acterization, can be removed from an analysisto reveal more subtle effects arising from. otherpopulation attributes (e.g.. age and gender). Itis interesting to note that when this was done,we found that the older participants preferredto keep the elbow closer to the torso than didyounger participants. One reason for this maybe related to the work described by Hughesand An (1997), which showed that small

TABLE 5: Summary of Significant Mean Changes in Six Major Body Angles (in degrees) for SubjectsOver 50 Years Compared with Those Under 50 Years when the Hand Reached the Four Final TargetAreas


Console ns ns 1.50 -2.4° -3.4° nsRadio 1.1° ns 1.30 -2.3° -7.19 nsOverhead 1.20 ns ns -3.0° -5.6° nsFar Right 1.80 -1.2° 1.50 -4.40 -4.6° ns

Note: ns - no significant difference could be found.

2-

1 -

0.C,

0)

a)o)

':

-3 -

416

- - - 7-�--

_- - " -, --�......................

------

...

`- .1 .

0


Reaches to Radio

Shoulder Horizontal0.0 1

ID -i. o

Zc -1.0

CDa0C0Cv -2,0 :a0)nc< _-o.0

-4 .0

0.2 0.4 0.6 0.8 1.0

Proportion ot time

Far Right Reaches towards Glove Box and Passenger Door

Shoulder Horizontal

0.0

-1 .0

a)a -2.00)

E -3.0

6 -4.0

-5.0


ShouIder Abduction

0.0 0.2 0.4 0.6 0.8Proportion of time

IFigure 7. Mean angle differences (and 95% confidence lines) for the shoulder angles with differences attrib-utable to being older than 50 years compared with being younger than 50 years.

changes in humeral rotation during reachingmotions can have large effects on glenohumer-al joint contact forces and rotator tendonforces, which often can cause irritable tissuesin older shoulder structures. One simulationstudy by Chaffin et al. (1998), using a shoulderbiomechanical model developed by Hogfors,Karlsson, and Peterson (1995), showed thatreaches to the glove box area by a large youngman resulted in about 10% less glenohumeralcontact force than did those by a small youngwoman, who had to abduct her arm more.

Despite the positive findings of this study,limitations are evident. For instance, the reach-es were performed without any weight in thehand. Strength limitations are known to affectat least static postures (Chaffin, Faraway, &Zhang, 1999) so it is reasonable to expect dif-ferences in motions as the dynamic momentsat the shoulder and other joints increasebecause of moving objects in the hand. This

may be particularly relevant to older individu-als with diminished shoulder strength andrange of motion when lifting objects in and outof a vehicle. The fact that the small, olderwomen in this study had to abduct their shoul-ders to an extreme angle when reaching to thepassenger door or glove box raises questions ofhow best to locate and design inside door han-dles, window controls, and other manuallyoperated devices for such a group. In thisstudy the participants simply had to push smallbuttons, not grasp handles or apply forces indifferent directions. The latter requirementwould certainly alter the terminal reach pos-tures depicted in this study.

Another limitation in the study is the type ofvehicle and seat used. Though it is believedthat the vehicle seat used is reasonably indica-tive of a typical passenger car seat, it certainlydoes not represent all types of vehicles (e.g.,bus, truck, van, and sport utility vehicle seats).

0.0

C -2.0CDap -4.0

*'a -6.0

-8.0

0

Shoulder Abduction


0.0

-a -2.0a)aa) -4.0:a

-6.0

-8.0

1.0

......... I............ -_'

--'1_.1..1_ _- ............

o- ... -

4117

------------

.......... ........ .......-......... ...._

.......... -.. , ............... - ........ ............

-... I.......

.................... .......... ...........

4

....... ......................... I.... 11 ......... .... I.....

418 ~~~~~~~~~~~~~Fall 2000 - Human Factors

Figure 8. TFermiinal postures for reaches to the four target areas (A) far r-ight toward glove box/passengerdoor, (B) center console, (CI) radio, and (D) overhead console, depicting conmbined multivariate effects ofstature, age, and gender. Solid lines indicate large (953th percentile) 21-ye'ar-old m-ale predicted postures, anddashed lines indicate smiall (5th percentile) 65-year-old femnale predicted postures. The two upper panels arefront views and the bottom two are side views.

The use of less-contoured and higher seatsprobably would result in larger torso motions.

CONCLUSIONS

With the experimental and statistical miod-cling methods described in this paper, it is

possible to capture and predict even small vari-ations in reach motions. In the reach motionsdepicted in this study, it is clear t1hat anthiro-pometry (as represented by stature) is a veryimportant demiog-raphic variable. Age and, to amuch lesser degree, gender also account forsomie consistent differences. Their com-binedmultivariate effects are graphically depicted inFigure 8, in which the model predicts the pos-tures of a small (about 5th percentile stature ofU'.S. population) 65-year-old woman comparedwith a large (about 95th percentile stature ofU.S. population) 21-year-old man reaching tothe four different target areas.

As depicted in the terminal postures inFigure 8, the combined effects of stature-, age,

and gender are profound. WhI-en armi abductionshoulder angles approach .1 350, as is the casedepicted in Figure 8a for a small older womanreaching to the far right passenger-side door,one must be concerned. Such a posture wasfound by Chaffin et al. (1 998) to be very limnit-ing for many women and some men because ofthe shoulder strength required in such a reach.This would be further aggravated if the personin such a posture had to exert a significant ha-ndforce in a direction niormal to the armn posture(e.g., operating a manual window or door han-dle). Even the less extreme reaclt to the centerconsole was found to be associated with dif'fer-ent teirminal postures and motions, dependingon stature, age, and gender, as depicted in thepostures associated with reachin-g to the center(shift) console depicted in Figure 8b. Here thereach to the center console shows that for asmall older woman peerforming such a reach, theelbow is kept closer to the body than for a largeyoung man. In contrast, when reaching forwardto the radio or overhead (Figures 8c and 8d,

~~~~b';__; .. . .... . .... ...... ~~~~~~~~~~~~~~~~~~~~~~~~~~~~I

A -Far Right BCneConsol

-~~~~~~~~~"-

C - Radio D - Overhead Console

41~~~~~~~~~~~~~~~~~~~~~~~.

418


respectively) the small older women were foundto use a more erect torso posture and moreshoulder abduction, mostly related to theirsmaller size. If size (stature) is equal, then olderparticipants tended to keep the arm closer tothe body during these latter two reaches thandid the younger group, as noted earlier.

It also should be noted that only a simplefour-link human kinematic model was used toderive these conclusions regarding the effects ofstature, age, and gender. In particular, addingseparate pelvic and clavicle motions to thetorso motions should provide a much morerobust method of depicting complex motionsfor both seated and standing reaches. Such amodel is now being developed in the Universityof Michigan Human Motion SimulationLaboratory. With this revised model and withadditional reach studies that include the effectsof hand loads and handle orientations, weshould be able to depict and simulate an evenlarger array of realistic reach motions for futurehuman simulations and computer-aided designapplications described by Badler et al. (1993).We expect that this technology will greatlyimprove the use of ergonomics principles infuture comiiputer-assisted vehicle design systemsas defined by Peacock and Karwowski (1993).These human simulation systems will providemuch more precise analysis of reach capability,object avoidance, and joint biomechanicalinjurv risk than currently exists.

ACKNOWLEDGMENTS

The authors acknowledge the partial supportprovided for this project by a Chrysler Chal-lenge Fund Grant and the helpful advice andeditorial assistance of Deborah Thompson. Weare also indebted to the University of MichiganClaude D. Pepper Older Americans Inde-pendence Center (NIA Grant AGO8808) forassistance in locating appropriate participants.

REFERENCES

Badles; N. L., Phillips, C. B.. & Webber, B. L. (1993). Simulatinghumuans. New York: Oxford University Press.

Cavanaugh, J. T., Shinberg, M.. Rtay, L., Shipp, K. M., Kuchibhatla,M., & Schenkman, M. (1999). Kinematic characterization ofstanding reach: Comparison of younger vs. older subjects.Clinical Biomechanics, 14, 271- 279.

Chaffin, D). B., Andersson, G. B. I., & Martin, B. J. (1999).

Occupational biomechanics (3rd ed.). New York: Wiley.Chaffin, I). B., Faraway, I., & Zhang, X. (1999). Simulating reach

motions. Presented at the SAE Human Modeling for Designand Engineering Conference, The llague, Netherlands.Pittsburgh: Society of Automotive Engineers.

Chaffin, D. B., Georgi, C., Baker, G., & Nussbaum, M. l1998).Dynamic shoulder loads in reaching and materials handlingtasks. In Proceedings of the Human Factors anid ErgonomicsSociety 42nd Annual Meeting (pp. 917-920). Santa Monica,CA: Human Factors and Ergonotnics Society.

Faraway, 1. (1 997). Regression analysis for a functional response.'echnometrics, 39, 254-262.

Gallagher, M. A., Zuckerman, J. D., Cuoono, F, & Ortiz, J. (1996).The effect of age, speed, and ann dominance on shoulder func-tions in untrained men. Journal of Shoulder and Elbow Surgery,5(1), 3541.

Haywood, K. M., Williams, K., & VatsSanit, A. (1991). Qualitativeassessment of the backswing in older adult throwing. ResearchQuarterly for Exercise and Sport, 62, 340-343.

Hogfors, C., Karlsson, D)., & Peterson, B. (1995). Structure andinternal consistency of a shoulder model. Journal of Bio-mechanics, 28, 767-777.

Hughes, R. E., & An, K. N. (1997, September). Monte Carlo simu-lation of a planar shoulder model. Medical and BiologicalE'nigineering and Comnpuing, pp. 544-548.

Karwowski. W., Genaidy, A. M., & Asfour, S. S. (Eds.). (1990).C'omputer-aided ergonomics (pp. 138-156). London: Taylor &Francis.

Kumar, S., Chaffin, D. B., & Redfern, M. (1988). Isometric andisokdnetic back and arsi lifting strenglths: D)evice and measure-ment. Journal of Biosiechanics, 21, 35-44.

tatash, M. L. (1998). Progress in mrotor control: Bernstein'sTraditions irn Movement Studies, Vol. 1. Champaign, IL1:Human Kinetics.

Laubach, L. L. (1978). Hunman muscular strength. In WVebbAssociates (Eds.), Antlcropometric source book: IVol. 1 (NASARef. 1024). Ifouston: NASA Scientif'ic and Technical Infor-mation Office.

Lawton, M. P. (1990). Aging and performnance of home tasks.HIuman Factors. 32, 527-536.

Murray, M. P., Gore, D. R., Gardner, G. M., & Mollinger. L. A.(1985). Shoulder motion and muscle strength of normal tnenand woimen in two age groups. Clinical Orthiopedics, 198,268-273.

National Cetnter for Health Statistics. (NCI IS). (1976). Height andweight of adults 18-74 years old in the United States(Advanced Data PlIANES Survey series 10 and 11, No. 3, pp.1-8). Hyattsville, NiD.

Nussbaunm, M. A., Zhang, X., Chaffin, D. B., Stump, B. S., &Raschke, U. (1 996). A reduced surface snarker set for upperlimb kinensatics: I leuristics and optiniization. In Plroceedingsoft he 20th Annual Meeting of Americani Society of Bio-nmechanics (pp. 251-252). Ann Arbor, MI: American Societyof Bionsechanics.

Peacock, B. L ., & Karwowski, W. (1993). Automotive ergonomics.Bristol, PA: Taylor & Francis.

Potvin, A. R., Syndulko, K., Toutellotte, W W., Lemmson, 1. A., &Potvin, J. HI. (1980). IEuman neurologic function) and tise agingprocess. Journal of the American Geriatrics Society. 28 1 -9.

Ransay, J., & Silverman, B. 11997). Functional data anaiysis. NewYork: Springer-Verlag.

Yates, 1. NV., Kaenon, E., Rodgers, S. H., & Champney, P. C.(1980). Static lifting strength and maximial isometric voluntarycontractions of back, arm, and shoulder uscusles. Ergonomics,23, 37 47.

Zhang, X., & Chaffin, D. B. (1997). Task effects on three-dimen-sional dynamic postures during seated reaching movements:An investigative schetne end illustraiion. I-Human Factors, 39,659-671.

Don B. Chaffin is a professor in the Departments ofIndustrial and Operations Engineering and Biomed-ical Engineering at the University of Michigan. Hereceived his Ph.D. in industrial engineering from theUniversity of Michigan in 1968.

419


Julian J. Faraway is a professor in the Departmrent ofStatistics at the University of Michigan. He receivedhis Ph.D. in statistics from the University ofCalifornia, Berkeley in 1987.

Xudong Zhang is an assistant professor in theDepartnent of Mechanical and Industrial Engi-neering at the University of Illinois-Champaign. Hereceived his Ph.D. in industrial and operations engi-neering from the University of Michigan in 1997.

Charles Woolley is a research engineer in the Centerof Ergonomics at the University of Michigan. Ilereceived his masters degree in bioengineering fromthe University of Michigan in 1980.

Date received: July 30, 1999Date accepted: Februay 15, 2000

420

COPYRIGHT INFORMATION

TITLE: Stature, age, and gender effects on reach motion posturesSOURCE: Human Factors 42 no3 Fall 2000

WN: 0028902939006

The magazine publisher is the copyright holder of this article and itis reproduced with permission. Further reproduction of this article inviolation of the copyright is prohibited..

Copyright 1982-2002 The H.W. Wilson Company. All rights reserved.

stature, age, and gender effects on reach motion...

Documents