Effects of Vehicle Interior Geometry and AnthropometricVariables on Automobile Driving Posture

Matthew R Reed, Miriam A. Manary, Carol A. C. Flannagan, and Lawrence W.Schneider, University of Michigan Transportation Research Institute, Ann Arbor, Michigan

The effects of vehicle package, seat, and anthropometric variables on posture werestudied in a laboratory vehicle mockup. Participants (68 men and women) seiect-ed their preferred driving postures in 18 combinations of seat height, fore-aftsteering wheel position, and seat cushion angle. Two seats differing in stiffness andseat back contour were used in testing. Driving postures were recorded using asonic digitizer to measure the 3D locations of body landmarks. All test variableshad significant independent effects on driving posture. Drivers were found toadapt to changes in the vehicle geometry primarily by changes in limb posture,whereas torso posture remained relatively constant. Stature accounts for most ofthe anthropometricalyv related variability in driving posture, and gender differ-ences appear to be explained bv body size variation. Large intersubject differencesin torso posture, which are fairly stable across different seat and package condi-tions, are not closely related to standard anthropometric measures. The findingscan be used to predict the effects of changes in vehicle and seat design on drivingpostures for populations with a wide range of anthropometric characteristics.

INTRODUCTION

Accurate prediction of driving posture isessential for vehicle interior design. Optimal posi-tioning of the controls, displays, and restraint sys-tems depends on a detailed understanding of howand where drivers of widely varying sizes will sit.Early research into the problem of control andseat placement was concerned primarily withimproving the comfort of designs rather than pre-dicting how people would respond to particularvehicle and seat geometries (Lay & Fisher, 1940).Beginning in the late 1 950s, designers began touse planar and three-dimensional (3D) manikins.based on the pioneering work of Dempster1 955), to assess leg room and control reach(Geoffrey, 1961 lKaptur & Mvyal, 1961 ).

Recently. advances in computer technologyhave led to the development of 3D softwarem.odels of the entire human body that areincreasingly used for vehicle design (Porter,Case. Freer, & Bonney, 1993). These human

figure models provide a useful visualization ofvehicle occupant size. shape, and position, butthe success of design evaluations conductedwith these models is strongly dependent on theaccuracy of the model posture.

The current study investigates the effects onwhole-body driving posture of three variablesthat are known to have important effects onseat position and eye location (FIannagan, Manary,Schneider, & Reed, 1998; Flannagan, Schnieder,& Manary, 1996; Manary. Flannagan, Reed, &Schneider, 1998). The analysis is intended toprovide an understanding of the individual andinteractive effects of seat height, steering wheelposition, and seat cushion angle on all of themajor posture characteristics of interest forvehicle interior design.

METHOD

ParticipantsThis study was conducted in three phases,

Address correspondence to Matthew P. Reed, University of Michigan Transportation Research Institute, 2901 Baxter Rd.,Ann Arbor, MI 48109-2150; mreed@umich.edu. HUMAN FACTORS, Vol. 42, No. 4. Winter 2000, pp. 541-552.Copyright @ 2000, Hluman Factors and Ergonomics Societv. All rights reserved.

Winter 2000 - Human Factors

each of which used different participants and adifferent set of conditions. The participantswere 68 licensed adult drivers who were select-ed in gender-stature groups spanning morethan 95% of the stature range in the U.S. pop-ulation (Abraham, Johnson, & Najjar, 1979).Table I summarizes the participants' staturedistribution by phase.

FacilitiesTesting was conducted in a reconfigurable

vehicle mockup that allowed the seat height,fore-aft steering wheel position, and seat cush-ion angle to be varied over a wide range. Theseat and control layout, termed the vehicle pack-age, was specified and measured using stan-dard reference points and dimension definitionsdocumented in Society of Automotive Engi-neers (SAE) Recommended Practice 11 100 andother related practices (described in SAE,1997). Figure 1 illustrates these dimensions ina side view of a generic package. The x axis inthe package coordinate system runs positiverearward, the y axis positive to the driver'sright, and the z axis positive vertically. The ori-gin is defined by a different point on each axis.The origin x coordinate is defined by the ballof foot (BOF) reference point, whereas the ori-gin z coordinate is defined by the acceleratorheel point (AHP).

The seating reference point (SgRP) is ob-tained using the weighted, contoured H-pointmanikin (SAE 1826) to measure a reference

point on the seat. The height of this pointabove the heel rest surface (AHP) defines theseat height and is termed H30, following thedimension definitions in SAE J1100. Seat cush-ion angle (L27) specifies the orientation of thelower part of the seat (seat pan) with respectto horizontal and is measured using the H-point manikin with a procedure described inSAE J826. The steering wheel is characterizedby the coordinates of the center of the frontsurface of the wheel, the angle of the front sur-face of the wheel with respect to vertical, andthe diameter of the wheel. The horizontal dis-tance from the center of the steering wheel toBOF is a key package dimension and is termedSW-BOFX.

Figure 2 illustrates the test mockup schemat-ically. The seat was mounted on a motorizedplatform allowing unrestricted fore-aft travelalong a path inclined 60 to the horizontal. Seatcushion angle was varied by pivoting the entireseat around a lateral axis. Seat height was setby adjusting the height of the heel surface rela-tive to the seat. When the seat height was in-creased, the angle of the accelerator and brakepedals with respect to the horizontal wasreduced, consistent with the pedal plane angleequation given in SAE 11516. In keeping withtrends documented in vehicle fleet data (Flan-nagan et al. 1996, 1998), the steering wheeland instrument panel height were also loweredslightly with respect to the SgRP at higher seatheights (25 mm lower/90 mm of seat height

TABLE 1: Participant Pool

Participant Stature Gender Phase 1 Phase 2 Phase 3 All

Group Range (mm) (n) (n) (n) (n) (n)

0 <1511 Female 3 3 6

1 1511-1549 Female 5 0 0 5

2 1549-1595 Female 3 3 6

3 1595-1638 Female 5 0 0 5

4 1638-1681 Female 3 3 6

5 1681-1722 Female 3 3 6

6 1636-1679 Male 3 3 6

7 1679-1727 Male 3 3 6

8 1727-1775 Male 5 0 0 5

9 1775-1826 Male 3 3 6

10 1826-1869 Male 5 0 0 5

11 >1869 Male 3 3 6

Total 20 24 24 68

542

INTERIOR GEOMETRY AND DRIVING POSTURE

Steering Wheel Angle

Steering WheelDiametrter

g Steering Wheel to BOF X Seat

(L6) Podi

AcceleratorPectal

BatiRef

<- (BO

Package Origin (0,0)

of Foot I \erence Point \

Seat Cushion Angle (L27)

Accelerator Heel Point (AHP)

x

Seat Height(H30)

IFigure 1. Vehicle package geometrv. Expressions in parentheses are Society ofAutomotive Engineers nomenclature from SAE JI 100 (SAE, 1997).

Pedals Moved Horiz'Steering-Wheel-to-B

Headliner Position Constantfor All Conditions

Steering Wheel Angle \Changed with Seat Height

instrurent Panetand Steering Wheel SgRP Vertical /Vertical Movement PositionWith Seat Height A Constant for

. / ~~~~~~All Conditions Fore-Aft Seat Position Seatback Recline UnderTravel Under Subject Contro. Subject Control

ontaliv To Changeall-ot-Foot Distance

Floor Height Varied to\ Change Seat Height/ \ A

Peda! Plane Angle VariedWith Seat Height I

Seat Cushion Angle Changedby Rotation Around This Axis

Figure 2. Schematic of vehicle mockup showing adjustment axes.

increase), and the steering wheel angle with quired and allowrespect to vertical was increased at higher seat steering wheel toheights (20/90 mm of seat height increase).The fore-aft position of the steering wheel with Test Conditionsrespect to the pedals was varied by moving the Three packagpedals along a motorized horizontal track, manipulated ind(rather than moving the steering wheel. This each variable intereduced the amount of seat track travel re- tial portion of the

ed the instrument panel andremain in a fixed relation.

e and seat variables werelpendently, using a range ofnded to represent a substan-vehicle fleet. These variables

543

-

54 Wite 200 - Hua Fatr

were selected based on the findings of severalstudies of driving posture conducted at theUniversity of Michigan Transportation ResearchInstitute both in laboratory mockups and invehicles (Flannagan et al., 1996, 1998). Table 2lists the 18 configurations used in testing. Seatheight (H30) was set to 180, 270, and 360 mm,corresponding to a wide range of vehicle types,from sporty cars to minivans. Seat cushionangle (L27) was set to 110 and 18°. The horizon-tal distance from the steering wheel center tothe ball of foot reference point (SW-BOFX)was adjusted between 450 and 650 mm.

Steering wheel position with respect to BOFis correlated with seat height in this experi-ment design because it is not possible to manip-ulate both variables over a large range withoutcorrelation. At high seat heights, drivers tendto sit closer to the pedals, necessitating a moreforward steering wheel position. In fact, acrossthe vehicle fleet, a substantial correlation existsbetween these two variables. Figure 3 shows aplot of seat height and SW-BOFX for 158 re-cently produced vehicles, along with the testconditions used in this study. At each seatheight level, the test conditions were selectedto span a substantial fraction of the range of

production vehicles. The correlation betweenseat height and steering wheel position in theexperiment is addressed in the analysis, pri-marily by considering subsets of the test con-figurations for which the two variables areorthogonal. Figure 4 shows seat cushion anglesfor 65 vehicles, along with the two levels usedin testing.

Phases I and 2 were conducted using a seatfrom a Ford Taurus, a typical midsize sedanseat that has minimal contouring or bolstering.The Dodge Neon seat used for Phase 3 testingwas a sportier, bucket seat with a firm, promi-nent lumbar support. The objective in switch-ing seats was to evaluate whether the effects ofseat height, steering wheel position, or seatcushion angle differed for the two seats.

ProceduresParticipants were recruited by advertise-

ment and by word of mouth. All were licenseddrivers with at least four years' driving experi-ence. The study objectives were explained toeach participant, and written consent was ob-tained. The participant changed from streetclothes into test garb, consisting of loose-fittingshorts and a short-sleeve shirt with a slit in the

TABLE 2: Test Conditions

PhaseConfiguration Seat Cushion Seat Height SW-BOFXNumber N 1 2 3 Angle (L27, °) (H30, mmn) (mm)

1 44 x x 11 270 450

2 68 x x x 11 270 500

3 68 x x x 11 270 5504 68 x x x 11 270 6005 44 x x 11 270 650

6 44 x x 18 270 4507 68 x x x 18 270 500

8 68 x x x 18 270 550

9 68 x x x 18 270 600

10 44 x x 18 270 650

12 48 x x 11 180 550

13 48 x x 1 1 180 650

14 48 x x 1 1 360 450

15 48 x x 11 360 550

16 24 x 18 180 550

17 24 x 18 180 650

18 24 x 18 360 450

19 24 x 18 360 550

Note: Condition 11 included a modification to the seat. Data from Condition 1.1 are excluded from this analysis.

Winter 2000 - Human FaCtors5A4l

INTERIOR GEOMETRY AND DRIVING POSTURE

E

co

I

a)0f)

buu n=158

402

300

43v S ~ ~ ~ ~ ~ 4

200

400 500 600 700

Steering-Wheel-to-BOF Distance (mm)

Figure 3. SWV-BOFX versus seat height in 158 pro-duction vehicles. Large dots are test conditions usedin the current study.

back to allow access to posterior spine land-marks. The participants wore their own com-fortable driving shoes.

The test conditions were presented to theparticipants in random order. Table 2 lists theset of test conditions presented in each phase.In each trial, the participant was screened fromthe vehicle mockup while the test conditionswere set by the experimenter. The seat trackposition was adjusted to the estimated meanpopulation seat position, and the seat backangle was set to 23G, using the SAE J826 man-ikin torso angle measure. The participantentered the mockup and adjusted the fore-aftseat position using a motorized control andadjusted the seat back recline angle using amanual adjuster. The participant was instruct-ed to operate the pedals and steering wheeland to continue to adjust until a "normal, com-fortable driving posture" was obtained. A stat-ic road scene was displayed on a large screenin front of the drivers to provide consistentvisual cues. In Phase 1, the participants werefree to choose any hand position on the steer-ing wheel. In Phases 2 and 3. participants wereinstructed to place their hands on the steeringwheel at the 10 o'clock and 2 o'clock posi-tions. Although many other hand positions arepossible in driving, the 10 o'clock and 2o'clock positions are reasonable, and standard-izing the hand locations gives greater meaning

n = 65

1 1 degrees 18 degrees

7 10 13 16Seat Cushion Angle (deg)

- 15

- 10

-5

-1l 9

Figure 4. Seat cushion angle distribution for 65 pro-duction vehicles. Test conditions are shown withvertical lines. Vertical axis is count.

to the elbow angles. There was, however, nosignificant difference in elbow angle for similarconditions in Phases I and 2.

After the participant obtained a comfortabledriving posture, the experimenter recorded bodylandmark locations using the sonic digitizerprobe. Table 3 lists the body landmarks. Thepubic symphysis landmark was palpated andrecorded by the participant after training bythe experimenter. All other landmark locationswere recorded by the experimenter. The entiremeasurement required approximately 60 s,after which the participant exited the mockupto prepare for the next trial. Testing with eachparticipant lasted approximately 2 h.

Data AnalysisThe body landmark data were used to calcu-

late the locations of joints defining a kinematic-linkage representation of the body, illustratedin Figure 5. These landmark acquisition proce-dures and methods for estimating joint loca-tions are described in detail in Reed, Manary,and Schneider (1999). For example, hip jointlocations were estimated using the measuredlocations of the anterior-superior iliac spineand pubic symphysis landmarks. The resultingbody segment positions and orientations wereanalyzed to determine the effects of the experi-mental variables on driving posture. Six vari-ables of primary interest are defined in Table 4

545

*.AL_.

Winter 2000 - Human Factors

TABLE 3: Definitions of Body Landmarks

Landmark Definition

Glabella

lnfraorbitale

Tragion

Occiput

Corner of eye

C7, T8a, Tl 2a

Suprasternale (manubrium)

Substernale (xyphoid process)

Anterior-superior iliac spine(ASIS; right and left)

Posterior-superior iliac spinea(PSIS; right and left)

Pubic symphysis

Lateral femoral condyle

Wrist

Acromion

Lateral humeral condyle

Lateral malleolus

Undepressed skin surface point obtained by palpating the mostforward projection of the forehead in the midline at the level of thebrow ridges.

Undepressed skin surface point obtained by palpating the mostinferior margin of the eye orbit (eye socket).

Undepressed skin surface point obtained by palpating the mostanterior margin of the cartilaginous notch just superior to the tragusof the ear (located at the upper edge of the external auditory meatus).

Undepressed skin surface point at the posterior inferior occipitalprominence. Hair is lightly compressed.

Undepressed skin surface point at the lateral junction of the upperand lower eyelids.

Depressed skin surface point at the most posterior aspect of thespinous process.

Undepressed skin surface point at the superior margin of the jugularnotch of the manubrium on the midline of the sternum.

Undepressed skin surface point at the inferior margin of the sternumon the midline.

Depressed skin surface point at the anterior-superior iliac spine,located by palpating proximally on the midline of the anterior thighsurface until the anterior prominence of the iliac spine is reached.

Depressed skin surface point at the posterior-superior iliac spine,located by palpating posteriorly along the margin of the iliac spineuntil the most posterior prrominence is located, adjacent to the sacrum.

Depressed skin surface point at the anterior margin of pubicsymphysis, located by the participant by palpating inferiorly on themidline of the abdomen until reaching the pubis. The participant isinstructed to rock his or her fingers around the lower margin of thesymphysis to locate the most anterior point.

Undepressed skin surface point at the most lateral aspect of thelateral femoral condyle, measured on the skin surface or throughthin clothing.

Undepressed skin surface point on the dorsal surface of the wristmidway between the radial and ulnar styloid processes.

Undepressed skin surface point obtained by palpating the mostanterior portion of the lateral margin of the acromial process of thescapula.

Undepressed skin surface point at the most lateral aspect of thehumeral condyle.

Undepressed skin surface point at the most lateral aspect of themalleolus of the fibula.

(cont'nued next page)

546

INTERIOR GEOMETRY AND DRIVING POSTURE

TABLE 3 (continued)

Landmark Definition

Medial shoe point Point on the media! aspect of the right shoe medial to the firstmetatarsal-phalangeal joint (approximately the ball of the foot).

Shoe heel contact point Point on the floor at the center of the right shoe heel contact areawith the foot in normal driving position contacting the acceleratorpedal.

'These points are not accessible when the participant is sitting in a conventional vehicle seat but are recorded in other sitting andstanding experimental situations to characterize the participant's torso geometry. See text for details.

and illustrated in Figure 6. HipX is the hori-zontal location of the hip joint (average ofright and left) aft of BOF. Ilip-to-eye angle isthe side-view angle of a vector from mean hipto the center-eye point with respect to verticaland is a measure of overall torso recline. Thecenter-eye point (hereafter called the eye point)is an eye location estimate on the body (

LegFoot /

line with the fore-aft coordinate of the infraor-bitale landmark, the lateral coordinate of theglabella landmark, and the vertical coordinateof the corner-eye landmark.

RESULTS

center- Data from Phases I and 2 for Conditions Ithrough 1 0, representing five steering wheelpositions at two seat cushion angles, wereextracted for initial analysis. Steering wheelpositions were normalized by subtracting the

leck midrange value at the 270-mm seat height(550 mm). Within-subject analysis of variance(ANOVA) identified highly significant effects

Thorax of both test variables on most of the posturevariables of interest. Table 5 summarizes effectsthat were significant with p < .01. In no case

lomen was the interaction between seat cushion angleand steering wheel position significant.

Steering wheel position had a strong effecton fore-aft hip joint location (HipX). Moving

Eye

C7.'T1

HeadAngle

Pelvis Angle

Figure 5. Kinematic linkage representation of driv- Figu.re 6. Illustration of posture variables.ing posture.

547

548 Winter 2000 - Human Factors

TABLE 4: Posture Variable Definitions

Variable Definition

HipX Fore-aft distance from the mean hip joint location to the ball-of-footreference point

Hip-to-eye angle Angle in the side view (x,z) plane of the vector from the mean hipjoint to the center eye point with respect to vertical

Center eye point An eye location estimate on the body centerline with the fore-aftcoordinate of the infraorbitale landmark, the lateral coordinate of theglabella landmark, and the vertical coordinate of the corner-eyelandmark

Pelvis anale, thorax angle,head angle x,z (side view) plane angle of the respective segment with respect

to vertical

Lumbar flexion Pelvis angle minus thorax angle

Cervical flexion Head angle minus thorax angle

Elbow angle Angle between the arm and forearm segments in the plane of thesegments; smaller values indicate greater flexion

Knee angle Angle between the thigh and leg segments in the plane of thesegments; smaller values indicate greater flexion

TABLE 5: Effects of Steering Wheel Position and Seat Cushion Angie:Configurations 1-10 in Phases 1 and 2

Normalized SteeringWheel Position Seat Cushion Angle

Variable (-100 to +100 mm) (1 1°-18°)

HipX (mm) 89.6 -6.0Hip-to-eye angle 3.1 0.59Lumbar flexion n.s. 2.0Cervical flexion n.s. n.s.Elbow angle -26.5 n.s.Knee angle 16.3 -3.6

Note: Listed values are mean differences between conditions. Positive values indicate the indicated changein the independent variabie resulted in a rearward, more reclined, or more flexed dependent variable value.N.s. indicates effect was not significant p > 01),

the steering wheel rearward 200 mm from themost forward position resulted in an averagerearward movement of the hip joint of about90 mm. The rearward movement of the steer-ing wheel reduced elbow angle by an averageof 26.5° while increasing knee angle by 16,3°.Overall torso recline (hip-to-eye angle) increas-ed about 3° with the 200-mm increase in steer-ing-wheel-to-pedal distance. The significanteffects of steering wheel position were largelylinear. Figure 7 illustrates the steering wheeland seat cushion angle effects on FlipX. Theeffect of steering wheel position on hip location

is approximately linear over a 200-mm range,and the steering wheel effect is very similar forthe two seat cushion angles.

Changing the seat cushion angle from 1°' to180 had significant but smaller average effects onposture. The higher cushion angle resulted in hipjoint locations an average of 6 mm further for-ward and increased overall torso recline by lessthan 1°. Net lumbar spine flexion increased by 20,but cervical flexion was not significantly affect-ed. Elbow angles were not significantly affected,but: knee angles were reduced an average of 3.60.

In Phase 1, the participants were free to posi-

INTERIOR GEOMETRY AND DRIVING POSTURE

900 -

E 850-E

aI 800-

750 --10Q -50 0 50

SWtoBOFX (norm) (mm)100

Figure 7. Steering-wheel-position and seat-cushion-angle effects on HipX in Phases I and 2, Config-urations I-1.0.

tion their hands at any location on the steeringwheel, whereas in Phase 2 the hand locationswere restricted to 10 o'clock and 2 o'clock.There were no significant differences in torsoposture between the phases, indicating that thehand location restrictions did not have impor-tant effects on torso posture.

Seat Effects and InteractionsIn Phase 3, six of the conditions tested in

both Phase I and Phase 2 were repeated withdifferent participants and with a different seat.Data from these conditions (2, 3, 4, 7. 8, 9)were extracted to assess differences in posturebetween seats and potential interactions be-tween the seat type and the steering wheelposition or seat cushion angle.

There were two significant differences inaverage posture between the two seats, using abetween-subjects comparison. The participantsin the Neon seat sat an average of 15 mm fur-ther forward (HipX) and with 3,4G less lumbarflexion than the participants in the Taurus. Thedifference in HipX is attributable partly to thefact that the Neon participants were about 7mm shorter, on average, than the Taurus parti-cipants. Shorter-stature drivers are known toselect more forward seat positions, on average(Flannagan et al., 1998). Torso recline was notsignificantly different between the seats.

More important. however, the effects of thesteering wheel position and seat cushion angleon the posture variables did not differ signifi-cantly between the seats (p > .05 for the inter-action terms). This suggests that the effects of

these two variables on posture do not dependon the seat geometry.

Seat Height Effects and Interactions

In Phases 2 and 3, trials were conducted atthree seat heights (180, 270, and 360 mm)and with two steering wheel positions at eachseat height. Taking only the data from thePhase 3 550-mm steering-wheel conditions(Configurations 3, 8, 12, 15, 16, and 19) givesan orthogonal 3 x 2 design with seat heightand seat cushion angle (24 participants). Inthese data, there are no significant effects ofseat height on lumbar flexion. cervical flexion.or hip-to-eye angle, nor are there any signifi-cant interactions between seat height and seatcushion angle. However, there is a significant,apparently nonlinear effect of seat height onHipX. The average hip location moved for-ward 9 mm as the seat height was raised from180 to 270 mm but moved forward 25 mmwhen the seat height was raised from 270 to360 mm.

This analysis was expanded to examinepotential interactions between steering wheelposition and seat height. To eliminate the numer-ical correlation of the variables, steering wheelposition was recoded relative to the presumedneutral position at each seat height (600, 550,and 500 mm for the 180-. 270-, and 360-mmseat heights, respectively). In the data fromPhase 3, Conditions 2, 4, 7, 9, and 12 through19 form a 3 x 2 x 2 design in seat height, SW-BOFX, and seat cushion angle. Using a within-subject ANOVA, the main effects and inter-actions were tested for each of the dependentvariables. None of the potential two-way inter-actions was significant. nor was the three-wayinteraction (p > .10 in all cases). The maineffect estimates for the three variables weresimilar to those obtained using other analyses.

Gender Differences

Participant Groups 4, 5, 6, and 7 containedmen and women of similar stature. Using onlythese groups (24 participants) and data fromPhases 2 and 3, an ANOVA was conducted todetermine whether or not the effects of the testvariables differed between men and women.No significant interactions with gender wereobserved, indicating that the test variables

x11:18

549

Winter 2000 - Human Factors

affect the driving postures of men and womensimilarly.

DISCUSSION

A three-phase laboratory study was con-ducted to: determine the effects of seat height,fore-aft steering wheel position, and seat cush-ion angle on driving posture. Analyses focusedon three measures of torso posture and threemeasures of limb posture. The principal obser-vations are as follows.

* Seat height, steering wheel position, and seatcushion angle each have significant, largely inde-pendent effects on posture.

* The effects of these three variables are indepen-dent of body size, proportion, and gender.

* Overall body size (stature) is the primary deter-minant of fore-aft hip position with respect tothe pedals, but seat height, steering wheel posi-tion, and seat cushion angle all have significanteffects.

* The ratio of sitting height to stature is an impor-tant predictor of hip-to-eye angle and elbowangle.

* Knee and elbow angles, the primary measures oflimb posture. are strongly influenced by seatheight and steering wheel position. Over therange studied, steering wheel position has thestronger effect.

* Seat cushion angle has a highly significant effecton both lumbar flexion and overall torso recline,but the importance of the effect is diminishedby the restricted range of this variable in vehicledesigns.

The most important observation from thisstudy is that postural adaptations to changes inthe layout of the driving task are accomplishedprimarily by changes in limb posture, whereastorso posture remains largely unaffected. In-stead, torso posture appears to be determinedprimarily bv intersubject differences that arenot closely related to overall anthropometricvariables. A typical sitter's spine flexion doesnot vary substantially across different vehiclelayouts but may differ considerably from thatof other sitters.

Of the three package and seat variables stud-ied, the fore-aft steering wheel position is themost important. if the steering wheel is movedforward 100 mm, drivers respond by movingtheir hips about 45 mm closer to the pedals,accounting for about half the change in steering

wheel position. The change in hip location isassociated with an average reduction in kneeangle of about 80. The more-forward steeringwheel position (65 mm farther in front of thehips) results in elbow angles that are larger byan average of } 3°. In contrast to the relativelylarge changes in limb posture, hip-to-eye angleis reduced by less than 20.

The two seats that were tested produced dif-ferent average postures, but the differences aredifficult to interpret because different partici-pants were used. The seat with the subjectivelymore prominent lumbar support, the Neon,produced postures with less average spine flex-ion. This finding is consistent with previousstudies (Reed & Schneider, 1996; Reed,Schnieder, & Eby. 1995). in which prominentlumbar supports have been found to cause sta-tistically significant but small reductions in thelumbar spine flexion of drivers. More impor-tant, however, the effects of seat height. steer-ing wheel position, and seat cushion angle werenot significantly different for the two seats, sug-gesting that these effects can be considered tobe independent of seat type. A small nonlineari-ty in the effect of seat height on fore-aft hiplocation did not contribute substantially to theoverall regression prediction. Similar findingswere reported by Flannagan et al. (1996) withregard to driver-selected seat position.

The findings of this study are directly applic-able to vehicle interior design, for which theaccurate prediction of occupant posture is ofconsiderable importance. Although seat heightis often set early in the design process and sub-stantially constrained by exterior styling con-siderations, the fore-aft steering-wheel positionand seat-cushion angle can be manipulated morereadily to accomplish design goals. The findingsfrom this study will help designers to predictmore accurately the effects of these changes ondriver posture.

This study was conducted in a laboratorymockup, without many of the spatial and visu-al cues that are present in an actual vehicle andthat could affect posture, such as mirrors, doors,and seat belts. However, other studies conduct-ed in conjunction with this work have shownthat these factors may not substantially reducethe generalizability of the findings. For exam-ple, changes in forward vision restrictions

---550.

INTERIOR GEOMETRY AND DRIVING POSTURE

resulting from varying instrument panel heightdo not affect driving posture in important ways(Reed, 1998). Further, comparison of posturepredictions derived from these laboratory datawith in-vehicle driver postures indicate that thepostures measured in this study can reliably beused to predict actual driving postures (Flanna-gan et al., 1996; Reed, 1998).

One vehicle geometry variable not exam-ined in this study is vertical steering wheelposition. This variable was excluded because,in practice, there is little flexibility for verticalpositioning with respect to the seating refer-ence point. Examination of interior geometrydata from dozens of vehicles has shown thatsteering wheel height above the seating refer-ence point is a nearly constant value, becausethe uppermost possible position is constrainedby the vision requirements of people with loweye heights and the lowermost position is con-strained by knee clearance and ingress-egressrequirements. In subsequent studies with avertically adjustable seat position, the verticalsteering wheel position will be examined as apotential important factor relating to driver-selected seat height.

The two most important restrictions onthese findings pertain to the use of a two-way(one-degree-of-freedom) seat track. Althoughmost vehicles are still designed initially using atwo-way track. an increasing number of vehi-cles are being designed and sold with a largerrange of adjustment, particularly so-called six-way seat tracks that allow the seat height, seatcushion angle, and fore-aft seat position to beadjusted. Although research is presently underway to quantify driver behavior when theseadditional adjustments are available, the datafrom the current study should be understoodto apply only to two-way tracks. Further, theseat track in this study allowed all participantsto select their preferred seat position withoutrestriction. In many vehicles, the seat tracklength is insufficient to accommodate everydriver at his or her preferred location, resultingin censoring of the seat position distributionand potentially changing posture in ways thatare not encompassed by the findings of thecurrent study.

ln addition to seat-adjustment limitations.the postures measured in this study were ob-

tained after the driver had been seated for onlyabout I min. Postures over a longer driving ses-sion could be different, although Reed et al.(1995) found only small changes in postureduring long-term driving simulations. In dy-namic, on-road driving, there appear to be smallbut important changes in eye location associat-ed with driving duration, probably attributableto gradual compression of the seat foam (Flan-nagan et al., 1996; Manary et al., 1998).

These findings are also limited by the healthand age of the participants. Although the par-ticipants ranged in age from 21 to 75 years(average 35 years), the behavior of a geriatricpopulation might differ from the relatively fitparticipants used in this testing.

With respect to anthropometry., the partici-pants spanned a wide range of stature relativeto the U.S. population. As noted. the effects ofanthropometry on posture were largely linearand generally did not interact with the effectsof package geometry. Hlowever, the small num-ber of participants in each stature category mayinfluence the generalizability of the findings.

This research does not purport to identifyideal or optimal locations for vehicle compo-nents but, rather, examines the effects of par-ticular component locations on driving posture.Although vehicle design is aided by the identi-fication of desirable component locations, par-ticular positions are ultimatelv selected. At thatpoint, accurate predictions of the resultingdriving postures are necessary to complete theinterior design (e.g., to locate secondary con-trols and displays) and to assess the potentialdistribution of postural comfort in the targetpopulation. Thus accurate posture predictionis required to begin the process of identifyingoptimal control locations.

CONCLUSIONS

Drivers adapt to changes in the vehicle andseat geometry primarily through changes inlimb posture, whereas torso posture remainsfairly constant. Large differences in torso pos-ture between participants are not well predict-ed by anthropometric differences. Seat height,steering wheel position, and seat cushion anglehave effects on driving posture that are largelyindependent of body size and gender.

551

Winter 2000 - Human Factors

- ACKNOWLEDGMENTS

This research was supported in part by grantsto the Automotive Seat and Package Evaluationand Comparison Tools (ASPECT) program, aconsortium of auto industry companies. ASPECTparticipants include BMW, DaimlerChrysler,Ford, General Motors. Johnson Controls, Lear,Magna Interior Systems, PSA-Peugeot-Citroen,Toyota, Volkswagen, and Volvo.

REFERENCES

Abraham, S., Johnson, C. L., & Najjar, M. F. (1979). Weight byheight and age for adults 18-74 years: U[nited States, 1971-74,

V'ital and health statistics. Series 11, No. 208 (DHEVW Pub-lication No. 79-1656). Hlyattsville, MD: U.S. Department ofH ealth. Education, and Welfare.

Dempster, W. T. (1955). Space requiremnents of the seated opera-tor- Geometrical, ki,,emnatic, and mechanicaf aspectcs of the

body with special reference to the limbs (WADC Tech. ReportNo. 55-159). Wright-Patterson Air Force Base, Oll: Wright AirDevelopment Center

Flannagan. C. A. C.. Manary, M. A., Schneider, L. W.V & Reed, M. P.(1998). Anr improved seatig accommodation model withapplications to different user populationis (SAE Tech. PaperNo. 980651). Warrendale, PA: Society of Automotive Engineers.

Flannagan, C. A. C.. Schneider. L. W., & Manary, M. A. (1996).Development of a new seating accommodation mode' (SAETech. Paper 960479). In Automotive design advancements inhuman factors. Improving drivers' comfort and perfonrmance(SP-I 155, pp. 29-36). Warrendale, PA: Society of AutomotiveEngineers.

Geoffrey. S. P, (1961). 4 2-D mannikin - The inside story: X-raysused to determine a new standard for a basic design tool tSAETech. Paper No. 267A). New York: Society of AutomotiveE,ngineers.

Kaptur, V., & Myal, M. 11961). The General Motors comfortdimenisioning system (SAE Tech. Paper No. 267B). New York:Society of Automotive Engineers.

ILay, Vs E.. & Fisher, L. C. (1940). Riding comfort and cushions.SAEJournal. 47. 482-496.

Manary, M. A., Flannagan, C. A. C., Reed, M. F.. & Schneider, L. W(1998), Development of an imnproved driver eye position model

(SAE Tech. Paper No. 980012). Warrendale, PA: Society ofAutomotive Engineers.

Porter. J. M., Case. K., Freer. M. T., & Bonney, M. C. (1993).Computer-aided ergonomics design of automobiles. In B.

Peacock & W. Karwouwski (Eds.), Automotive ergonomics (pp.43-77 ) London: Taylot & Francis.

Reed, M. P. (1998). Statistical and biomechanical prediction ofautomobile driving posture. Unpublished doctoral dissertation,University of Michigan, Ann Arbor. Ml.

Reed. M. P., Manarv, M. A., & Schneider. L. WV. (1 999). Methodsfor measunng and representing automobile occupant posture(SAE Tech. Paper 990959). Warrendale, PA: Society of Auto-motive Engineers.

Reed, M. P., & Schneider, L. W. 1(996). Lumbar support in autoseats: Conclusions from a study of preferred driving posture(SAE Tech. Paper No. 9604781. In Automotive designadvancements in human factors: Improving drivers' comfbrtand performance (SP-I 155, pp. 19-28). WN'arrendale, PA: Societyof Automotive Engineers.

Reed, M. P., Schneider, L. W., & Eby, B. A. H. (1995). Someeffects of lumbar support on driver seated posture (SAE Tech.Paper No. 950141). In Human factors in vehicle design:Lighting, seating and advanced electronics (SP 1088, pp.9-20). Warrendale, PA: Society of Automotive Engineers.

Society of Automotive Engineers. 19971. Automotive engineeringhandbook. Warrendale, PA: Author.

Matthew P. Reed is an assistant research scientist inthe Biosciences Division of the University of Michi-gan Transportation Research Institutc. Ite rcceiveda Ph.D. in industrial and operations cngineeringfromn the University of Michigan in 1998.

Miriam A. Manary is a senior research associate inthe Biosciences Division of the University of Mich-igan Transportation Research Institute. She receivedan M.S.F.. in bioengineering from the University ofMichigan in 1994.

Carol A. C. Flannagan in an assistant research scien-tist in the Biosciences Division of the University ofMichigan Transportation Research Institute. Shereceived a Ph.D. in psychology from the Universityof Michigan in 1993.

Lawrence v Schneider is a senior research scientistand head of the Biosciences Division of the Uni-versity of Michigan Transportation ResearchInstitute. Hie received a Ph.D. in bioengineeringfrom the University of Michigan in 1973.

Dale received: November 22, 1999D)ate accepted.: Mlarch 3 1 2000

552

COPYRIGHT INFORMATION

TITLE: Effects of vehicle interior geometry and anthropometricvariable on automobile driving posture

SOURCE: Human Factors 42 no4 Wint 2000WN: 0035002939002

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.