Below is the uncorrected machine-read text of this chapter, intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text of each book. Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.
2 Anthropometric Models Human body models come in many forms, including two- dimensional drafting board templates, sizing manikins, three- dimensional physical dummies for biodynamic tests, and com- puter analogs. The discussion of anthropometric models will cen- ter largely on computer analogs. Most computer models were developed with a particular purpose in mind such as biodynamic testing, strength assessment, or human factors evaluations. What- ever their differences, models share a basic need for accurate rep- resentation of body size, shape, and proportion in all of their exasperating permutations. Much of this challenge fails in the domain of physical anthropology and engineering anthropometry. TH1: ANTHROPOM1DTRIC DATA BASE In the United States, formation of an anthropometric data bank was initiated in 1973 by C.E. Clauser of the Harry G. Arm- strong Aerospace Medical Research Laboratory (AAMRL). The data bank was meant not only as a repository for information from a variety of sources but also as a facility in which such data would be processed and cast in a comparable format to permit recall and analysis for design purposes using computer routines. Over the years, the data bank has expanded steadily; today it con- stitutes a unique anthropometric source for designers, engineers, and modelers. The 1985 holdings of the AAMRL's anthropometric data bank 4
s included 51 separate surveys. Most of the data are based on mil- itary rather than civilian populations and on male rather than on both male and female groups. These disproportionalities in coverage are not by design but are due to the limitations of avail- able data. Table 2-1 provides a listing of the current holdings of the data bank for U.S. and foreign military populations. The surveys range in time from 1946 to 1981, with the majority being conducted in the 1960s. In any one survey, as few as 46 body size variables were measured with the largest number of variables (189) being measured in the U.S. Air Force's 1967 survey. In all, more than 300 measured variables, from one or more surveys, are included in the current data bank. A major survey to update the anthropometric data base for the U.S. Army was begun in 1987 and Is scheduled for completion in 1988. Data on foreign mili- tary populations in general include fewer subjects and variables. For an overview, check the NASA Anthropometric Source Book (National Aeronautics and Space Administration, 1978~. The anthropometric data base for the U.S. civilian population as a whole is rather weak. No comprehensive anthropometric study of the civilian population has ever been undertaken. Our knowledge of U.S. civilian body size variability (see Table 2-2) comes primarily from the various health and nutrition surveys, beginning with the first 1962 Health Examination Survey (HES) (Stoudt et al., 1965~. In these surveys, investigators were concerned with health and nutritional assessment and obtained only limited anthropo- metric data (primarily on mass-related dimensions, such as girths and skinfolds). In the 1962 HES survey, 12 workspace and 7 nutrition-related dimensions were measured on a nationwide sam- ple (n = 6,672) of civilians aged 18-79 years. These data, limited as they are in the description of body size variability and the fact that they are 25 years old, are still the best available for the U.S. civilian population. The sample was sufficiently large, however, to provide adequate descriptors for sex, age, and some racial groups. Since the utility of military data for civilian populations has of- ten been challenged, McConville et al. (1981) attempted to match military samples with the HES civilian samples on the basis of height and weight. For the mates, results were good in that almost 98 percent of the civilians from the HES study were matched with U.S. Army subjects from a single survey. By comparing seven dimensions that were similarly measured in the U.S. Army and
6 TABLE 2-1 Military Data Contained in the AAMRL Anthropometric Data Bank Survey D ate Population Sample Size Variables (No.) U.S. Military Males 1950 U.S. Air Force pilots 4,000 146 1950 U.S. Army aviators 500 46 1964 U.S. Nary aviators 1,529 98 1965 U.S. Air Force ground personnel 3,869 161 1966 U.S. Army ground personnel 6,682 73 1966 U.S. Nary enlisted 4,095 73 1966 U.S. Marines enlisted 2 008 73 1967 U.S. Air Force Biers 2 420 189 1970 U.S. Army fliers ~ 1~482 88 Total 26,585 U.S. Militarv Females 1946 U.S. Women's Army Corps 7 563 65 1968 U.S. Air Force women 1 905 139 1977 U.S. Army women 1.331 151 Total 10,799 U.S. military total Foreign Militant Populations (Male) 37,384 1960 Turkish armed forces 912 151 1961 Greek armed forces 1,071 151 1961 Italian armed forces 1,342 151 1961 Korean military fliers 264 132 1964 Vietnamese military forces 2,129 51 1967 German air force 1 466 152 1969 Iranian military 9 414 74 1970 Latin-American armed forces 1,985 76 1970 Royal Air Force aircrew 2,000 64 1972 Royal Air Force head study 500 46 1972 Royal Australian Air Force 482 18 1973 French military fliers 174 118 1974 Royal New Zealand Air Force aircrew 238 63 1974 Canadian military forces 565 33 1977 Australian personnel 2,945 32 1975 British Army surrey 1,537 61 1975 English Guardsmen 100 61 1976 English Transport Corpsmen 161 61 1976 United Kingdom Gurkhas 36 61 1976 Hong Kong Chinese military 73 47 1981 Israeli aircrewmen 360 63 Foreign military total U.S. and foreign military total 27,754 65,138 SOURCE: Harry G. Armstrong Aerospace Medical Research Laboratory anthropometric data bank (1985).
7 TABLE 2-2 U.S. Civilian Population Data Contained in the AAMRL Anthropometric Data Bank Survey Date Sample Size Variables (No.) Adult Males 1961 Air traffic controllers 1962 Health Examination Surrey (HES) 1962-1981 Matched Health Examination Sunrey (HES) (ages 18-65) 1974 Law enforcement officers 1975 Health and Nutrition Examination Surrey 678 65 3,091 18 2,761 70 2,989 23 (HANES) (ages 18-74) 6,563 11 1980 Health and Nutrition Examination Surrey (HANES II) (ages 18-75) 5,921 13 1981 U.S. rninere 270 44 U.S. civilian males total 22,273 Adult Females 1962 Health Examination Surrey (HES) 3,581 18 1971 Airline stewardesses 423 73 1975 Health and Nutrition Examination Sunrey (HANES) (ages 18-74) 10,123 11 1980 Health and Nutrition Examination Surrey (HANES II) (ages 18-75) 6,598 13 1981 U.S. miners _ 86 44 U.S. civilian females total U.S. civilian total 20,811 43,084 SOURCE: Harry G. Armstrong Aerospace Medical Research Laboratory anthropometric data bank (1961-1981). civilian survey, the authors demonstrated that the matching pro- cess provided representative anthropometry for the civilian male sample that was adequate for some design purposes. Matching proved to be less successful for women. The civilian women were heavier at every increment of stature, on average, than the military women. The matched military sample did not adequately characterize the distribution in the total female civilian population. For those that were successfully matched, however,
8 the correspondence between other body dimensions for civilian and ~rulitary women was quite good. Within limitations, the matching procedure has proven to be a useful technique for estimating the body size variability of a population for whom only limited anthropometric data are avait- able. The procedure is limited, however, to the range of body sizes within the base population from which the matches are drawn. By and large, all these data have been collected by using tra- ditional anthropometric tools and techniques. What is available, then, is a series of univariate descriptors of body size in terms of heights, lengths, breadths, depths, girths, and surface curvatures (Figure 2-1~. The military surveys in particular were designed to satisfy a variety of users, predominantly pattern makers and de- signers of personal protective equipment. Body dimensions for the layout of workspaces have also received attention, but only a few dimensions have been obtained strictly for human body models. The need for personal protective equipment for the head and face has required a large number of dimensions including surface arcs, breadths, and a series of headboard measurements (Figure 2-2) to relate a series of points in three-dimensional space to a common origin. Using these points and assuming bilateral symmetry, it becomes possible to develop face forms of sizing models for de- signers based on anthropometric data and the artistic ingenuity of a sculptor. Such forms are then reproduced and provided to designers who are involved in a specific design problem. This has turned out to be an extremely successful mask which is used in newer aircraft in which ~9 forces are common and 9-9 forces are not unknown. The need for anthropometric data translated into a three- dimensional form has extended into other areas as well. The requirements for body forms of ~ and ~year-old children for crash injury research necessitated the interpretation and integration of data from some six different sources, no one of which could be considered as the principal source. The resultant integrated data were rendered into three-dunensional body forms (Young et al., 1983~. Even when a strong, traditional anthropometric data base exists, it may not be as comprehensive as necessary to develop human body models. The need for sizing of partial pressure suits for U.S. Air Force aircrews led to the translation of the height- weight sizing system into three-dimensional models. The body for
9 ~- -low I~ Act - ~ T l FIGURE 2-1 Typical univariate descriptors of body Ale. SOURCE: Files of Anthropology Research Project, Inc., Yellow Springs, Ohio. each size was characterized as a sequence of body girths at specific levels, each girth having a breadth and a depth, with appropriate segment lengths. The development began with an armature to which mesh was affixed to bring the form roughly up to size. Plaster of Paris was applied to bring the forms to final size and shape (McConville et al., 1963~. Such body forms were designed specifically for sizing of a particular item of personal protective clothing. Each incorporated a specific statistical breakout of the data. Hence, their use is generally limited. (One exception is the "Ion" regulars body form that was used to provide the body size and shape for the biodynamic analog developed by Payne and Band [1971i, called DYNAMIC DAN.) In all of these sizing models, it was necessary to integrate traditional data from a series of independent studies to produce a usable body model. But the end product was most often a result
10 ,/ :~1)-'~ ;~ v' FIGURE 2-2 Head and face measurements. SOURCE: Files of Anthropol- ogy Research Project, Inc., Yellow Springs, Ohio.
11 of the scuIptor's skill in providing the final shape by filling in those areas for which no anthropometric data were available. ANTEROPOM1:TBIC COMPUTER. MODELS The anthropometric data input to the human engineering eva]- uation models ~ far more extensive than the simple lengths, diam- eters, and circumferences used to specify the size of the geometric forms of the early models. Most of the human engineering eval- uation models are based on the simulation concepts of intercon- nected links, originally outlined by Braune and Fischer (1889) in their classic biomechanical analysis of the German infantryman. This approach was refined and expanded by Dempster (1955), who studied the body as a series of interconnected links that he defined as "straight-line distances between adjacent centers of rotation. Early geometric modeling (don Meyer, 1873) reduced the body to a series of ellipsoids and spheres to arrive at estimated mass and centers of gravity of body segments. In 1960, Simons and Gardner developed a man-mode} by approximating the body segments as uniform geometric shapes. They represented the ap- pendages, neck, and torso by cylinders and the head by a sphere. Using Barter's (1957) equations for the ma" of the individual segments, they computed the inertial parameters for the geomet- ric forms and calculated the total-body moments of inertia. This work, elementary in many respects, was the genesis of much of the present biodynam~c modeling activity. In a study of the dynamic response of weightless man, Whit- sett (1962) refined the anthropometric mode} developed by Simons and Gardner (1960) by increasing the number of body segments from 8 to 14 by using additional geometric shapes to approximate more closely the shapes of the various body segments (Figure 2-3~. Whitsett's 14 segments include a head, a torso, two upper arms, two lower arms, two hands, two upper legs, two lower legs, and two feet. The head is modeled as an ellipsoid, the hands are spheres, the upper and lower arms and legs are frustums of circular cones, and the feet are rectangular parallelepipeds. The physical properties incorporated by Whitsett into the mode} included body size data from Hertzberg et al. (1954), mass properties from the regression equations of Barter (1957), and center-of-mass and segment-density data from Dempster (1955~. The equations for the mass moments of inertia were standard for
12 ~` HEAD End ~ NECK Or. lit l!\: u an,, - 61f me UPPER LEG an,, tJ HAI :D LOWER LEG FOOT ELLIPSOID ~ ELLIPTICAL Jo TORSO CYLINDER UPPER ARtd FRUSTUM OF A RIGHT CIFICULAR In LOWER CONE (l ~ l SPHERE is) FRUSTUM OF A RIGHT CIRCULAR CONE RECTANGULAR 00 PAI?Al-LEL£PIPED o FIGURE 2-3 Segmented human and model. SOURCE: Whiteett (1962~. the particular geometric forms used; only the mass moment of inertia equation for the frustum of a right circular cone needed to be derived. In 1963, Gray refined this basic model. In 1964, Hanavan published the results of a study intended to (1) design a personalized mathematical man model, (2) analyze the model, (3) prepare a generalized computer routine for calculating the inertial properties of any subject in any body position, and (4) develop a design handbook for a series of percentile body forms in 31 body positions. The mode} was made up of 15 s~rnple geometric forms hinged at the end of each of the primary segments. While the torso was considered as two linked segments and the head as a third linked segment, they lacked motion. Hanavan, in a manner similar to that used by Gray, defined the body posture by assigning Euler angles to each of the segments and then calculated the inertial dyadic tensor and the center-of-mass locations for a specific body in specific positions. Hanavan used the mash predictive equations described by Barter (1957) as input. This
13 technique was then applied by Tieber and Lindemuth (1965) and Bobbins et al. (1971) and is still in use. However, with better mass distribution data (McConville et al., 1980, and Young et al., 1983) and the availability of powerful large-scale computers, this approach has become increasingly outmoded. Most current human body models, such as those incorporated in BOEMAN (Ryan, 1971), SAMMIE (Bonney and Case, 1976), and COMBIMAN (Kroemer, 1973; McDaniel, 1976), begin with a link system, which is simplified from the human skeleton. Model- ers assign ranges of joint motion to the primary joints, and finish with an enfleshment procedure to give the mode} its final physical form. Depending on the intended use of the model, additional refinements such as mass distribution properties (Articulated To- tal Body Model; Kaleps, 1978) or visibility plots (COMBIMAN; Kikta et al., 1982) are added. In every case, an adequate an- thropometric data base is required for the construction of these models. T]IRE~D~IENSIONAL ANTHROPOMET1lY The existing anthropometric data base does not contain three- dimensional anthropometric data. It has been possible to use the existing univariate anthropometric descriptors to develop three- dimensional models, but such approaches have been compromises at best that are dependent on a series of approximations and assumptions regarding the relationship of individual dimensions. The traditional anthropometric data base lacks a common origin point to which the individual measurements can be related. In a recent series of studies, stereophotometric techniques were used to obtain mass distribution estimates for a sample of 31 male (McConville et al., 1980) and 46 female (Young et al., 1983) subjects and to relate these mass distribution properties to the anthropometry of the individuals. This procedure, similar to aerial photography, requires paired cameras in front and back of the subject (Figure 2-4) to obtain the stereoplates (McConville et al., 1980~. The plates are read, resulting in a "terrain maps of the body (Figure 2-5) from which contours, volumes, and mass distribution estimates can be obtained. The 31 male subjects were measured for some 75 body dimen- sions, and the 46 female subjects were measured for a comparable
14 MOP 1~ ~ - ~71\ d! An I FIGURE 2-4 Stereo camera array. SOURCE: McConville et al. (1980~. but expanded set of 92 body dimensions. After the anthropome- try was obtained, some 77 targets were affixed to the body land- marks to facilitate their location cluring the stereophotometric assessment. Volume, center of volume, and principal volumetric moments and axes of inertia were calculated. The prunary body segments used In these studies were defined by using planes of segmentation similar to those used in previous cadaver studies (Chandler et al., 1975; CIauser et al., 1969~. The use of stereophotogrammetry made possible the comparable ana- lytic segmentation of live subjects and facilitated the delineation of additional segments, such as the thorax, abdomen, and pelvis. An anatomical axis system was established for the total body and for each segment. These were right-hand orthogonal systems based on palpable, largely bony landmarks and were used to pro- vide a consistent reference for the principal axes of inertia for each segment regardless of body and segment position. The axis sys- tems were defined using a minunum of three noncolinear points on each segment located as far apart as was feasible. The anatomical axis system shown in Figure 2-6 for the head segment was estate fished using the right and left tragion landmarks and the right
15 . 1 i ~ . 8 X PXI3 X tO 'I -.K~ 0.508 1.000 1.~0 2.060 3. 1 3.080 too N. - , _! _ - _ _ _s FIGURE 2-5 aTerrain maps of the hump body. SOURCE: McConville et al. (1980~. infraorbital landmark. A fourth landmark, sellion, was used to translate the origin of the axis system to the m~dsagittal plane. Anthropometric techniques developed for the mass distribu- tion studies may have considerable merit for developing an an- thropometric data base for human body modem. The anatomical axis system for each segment and for the total body help to define postural orientation in three-dimensional space. Segmental land- marks are related to the segmental axes and to the total body axes, with body mass distribution characteristics predicted through re- gression equations based on the anthropometry of the model.
ze zp ~ //~:\ f l: \ \ 16 4'- ~ 3X / \N ; - up D,~Y. Cg~x. ~ I FIGURE 2-6 Anatomical axis system for the head segment. SOURCE: McConville et al. (1980~. DISCUSSION While there exists a wealth of anthropometric data for a num- ber of populations, and there are methods of extrapolating the data base to other populations, the current data base is deficient for effective human engineering body modeling. Current link sys- tems are largely based on studies by Cotter and Gleser (1958), Dempster (1955), and Snyder et al. (1972~. When data bases from several sources are combined with different study samples, inter- polations and approximations are required to integrate the data into a functional link system. The traditional anthropometric data base is not as helpful in developing a link system as we would like. Anthropometric landmarks lie on the surface of the body and are often removed from the actual joint centers of rotation by various layers of tissue. Thus, the link length that is sought can only be approximated. In addition, joint centers that define the link lengths are often difficult to locate accurately on living subjects and are even more difficult to locate from photographs. A system- atic investigation of a human body link system that incorporates
17 three-dimensional anthropometry developed specifically for com- puter simulation is required. The current anthropometric data base is a collection of uni- variate body size descriptors that lack a unifying origin to which they may be related in a three-~unensional space. It ~ desirable to develop a procedure that can supplement and integrate the exist- ing data base to provide the anthropometry necessary for effective three-~nnensional models. Reynolds (1977) has coined the term system anthropometry, wherein the traditional heights, lengths, and breadths are replaced by thre - dimensional coordinates for comparable point locations from a common origin, and the static anthropometric postures of standing and sitting are replaced with postures relating to work and movement. Before the envisioned system anthropometry can be developed and an effective anthropometric data base created, two basic in- terrelated problems must be resolved. The first is the selection of an effective data collection system which should be accurate (within required limits) and reproducible, be sparing of subject and observer tune, produce immediate digital output, permit rapid transfer to storage for analysis, and be relatively inexpensive. A wide variety of techniques that can describe points and point relationships in three-~nnensional space have been developed over the years. These range Tom rather simple electromechanical digitizers through stereophotogramrnetry to complex systems such as laser imaging. So far none of the existing systems have proven wholly satisfactory. The second problem L8 that even if a suitable system were at hand, we wouic! need to clevelop a method of analysis of the three- dimensional data that the system would generate. In the analysis of traditional anthropometry, we have the solid statistical mode} of the normal distribution. No comparable analytical mode} has yet been suggested for summarizing three-dimensional size and shape data for our application. Even with a complete and realistic anthropometric data base, various "real-life" work factors (e.g., posture, body restraints, clothing) can drastically change the accuracy and validity of the standard data base for many applications, since actual anthropo- metric characteristics may be quite different from those measured under standardized (laboratory) conditions. Garrett and Kennedy (1971), Roebuck et al. (1975), and Van Cott et al. (1978) compared measuring techniques and anthropometric data from 48 sources
18 and noted a lack of standardization in definitions and procedures across different studies. Data comparability was also noted as a po- tential problem in standardization when different instrumentation was used. No systematic study has been attempted to determine whether a number of measurements taken on a large number of participants by different measurement techniques and by different measurers yield equivalent data. The problem is probably most pronounced for measures involving compression of soft body tissue and those requiring a reference to internal skeletal landmarks. Other limitations of the current anthropometric data base, and hence of modem, are the following: Data on U.S. civilians are seriously deficient, particularly for females. Health Ex~nination Survey (HES) and Health and Nu- trition Examination Survey (HANES) data show that the population is taller and heavier than estimated from mili- tary data. There ~ insufficient information on special populations that collectively consist of a large portion of the total pop- ulation, including those over age 65 (about 12 percent of population), those under age 18 (about 26 percent of the population), population extremes (i.e., the tallest, heavi- est, shortest, lightest), and disabled persons. Most anthropometric data are univariate, which limits their application. Neither two- nor three-dimensional data are commonly referenced to a defined reference system. There is no standard procedure other than "artistic scuba turingn for arriving at three-dimensional body shape based on the classical anthropometric data. Various measurement definitions, measurement techniques, and data processing methods have been used in the differ- ent classical anthropometric surveys that constitute the available data base. Therefore, in many cases data are nei- ther interchangeable nor compatible. furthermore, they cannot be relied on to have the same degree of accuracy. Advanced procedures for data collection such as stereophm togrammetry or laser imaging are needed, but they are still in the experimental stages.