Professional Documents
Culture Documents
Display With Different Sized Fields-Of-View: Performance and Head Movements Using A Helmet-Mounted
Display With Different Sized Fields-Of-View: Performance and Head Movements Using A Helmet-Mounted
CONTENTS 1. INTRODUCTION
1. Introduction Specifying the size of the field-of-view (FOV) for a helmet-
2. Methods mounted display (HMD) is a complex task. For example, for a
2.1. Description constant sensor size and optical design, and using just three of
2.2. Stimuli and apparatus the many interdependent parameters, the following holds true:
2.3. Experimental design Increasing the FOV size, increasing the image resolution, or
2.4. Subjects decreasing the HMD weight may all be expected to increase
3. Results
3.1. Dependent variables performance. However, increasing the FOV size increases the
3.2. Performance data HMD weight and decreases image resolution. Furthermore, the
3.3. Head movement data relation between FOV size and display weight is complex and
3.4. Target replacement data depends on other variables, such as eye relief, exit pupil size,
4. Discussion and transmissivity. Similar complex relations exist between the
4.1. Magnitude of the effects other parameters.
4.2. Reasons for the effects This complexity calls for a cost/benefit analysis using data
4.2.1. Effect of task complexity on the effects of each parameter on performance. Some data of
4.2.2. Effect of FOV size this nature do exist. For example, increasing the FOV size makes
4.3. Specifying FOV size
5. Conclusions geographic orientation and target acquisition Increasing
6. Acknowledgments the image resolution produces a reduction in target recognition
7. References time and increases the probability of target recognition.2'3 Char-
acter recognition accuracy is also improved with increasing tele-
vision display resolution.4 Increasing helmet weight has been
shown to decrease neck muscle endurance times.5 Although
these data are valuable for providing qualitative design guide-
Invited Paper EO-106 received Dec. 24, 1989; revised manuscript received lines, they are of limited value for quantifying HMD parameters
March 12, 1990; accepted for publication May 10, 1990.
1990 Society of Photo-Optical Instrumentation Engineers. for a number of reasons. First, data on FOV size were collected
0• a 'I
- 40
Fig. 2. View inside VCASS showing a superimposed FOV 450 wide x 42.5° high. Targets (ar-
rowheads) and threats (squares) were visible only when they were inside the superimposed
FOV.
Targets (arrowhead shapes) or threats (squares) 2.2° high and TABLE I. The nine target conditions used in the experiment. The
wide were presented above the horizon, within a gaming area number of targets that had to be replaced was equal to the number
of targets presented, less the number of threats.
1200 left and right and 900 upward from straight ahead. Targets
were positioned pseudo-randomly in a manner that ensured that
'flIREATS
irrespective of the number of targets, they were unclustered,
unpredictable, and distributed over the entire gaming area. 9 6 3 $
A terrain scene was visible in the total 1200 x 600 ambient
FOV. The terrain scene had a horizon at O elevation and rec- NJMBF 9 1 2 3 4
ognizable landmarks. The targets and threats were visible only TAETS 6 5 6 7
when they were within the superimposed FOV, which was po- PRESEN'1D 3 8 9
sitioned centrally within the ambient FOV (Fig. 2). This com-
bination of ambient and superimposed FOV simulated a see-
through HMD used to view the output from a head-steered sen-
sor. A head-stabilized reticle cross (3.4° high and wide) was 3. RESULTS
constantly visible in the center of the superimposed FOV. The results presented in this paper are from three conditions,
referred to in Table I as conditions 8, 6, and 3. These three
2.3. Experimental design conditions represent the instances when subjects were presented
The two factors manipulated in the experiment were target con- with 3, 6, or 9 targets and had to deal with 3 threats. The method
dition (see Table I) and FOV size. The experiment evaluated of describing the conditions refers to the number of targets pre-
five superimposed FOV sizes: 2O x 20°, 45° x 42.5°, sented (P), the number of threats (T), and the number of targets
600 x 50°, 900 x 60°, or 120° x 60°. Subjects were trained replaced (R). Thus, in the 9P:3T:6R condition, the subjects were
with the 45 combinations of 9 target conditions and S FOV sizes presented with 9 targets, 3 of which turned to threats, leaving
during one session, for approximately 1 .5 h, including rest pe- 6 targets to be replaced. The reasons for choosing these con-
nods. During training they were presented with all target con- ditions are (1) they represent the early stages of a trial, in which
ditions in the same order (8, 9, 5, 6, 7, 1 , 2, 3, 4) with suc- it might be expected that the dependent variables are maximally
cessively decreasing FOVs. In the second session, administered sensitive to the independent variables, (2) they provide the wid-
on a different day, subjects were given 5 practice conditions est range of task complexities while still providing comparable
followed by all 45 conditions in a random order. data, and (3) space limitations. The complete data are presented
elsewhere .
2.4. Subjects
The experiment used paid volunteer male subjects aged between 3.1. Dependent variables
1 8 and 30 years , with uncorrected or corrected visual acuity of The dependent variables may be categorized into performance
20/30 or better. The experiment used a total of 17 subjects, 8 data, head movement data, and target replacement data. Per-
of whom had experience at the target replacement component formance at the task was measured as the percentage of threats
of the task from a previous experiment. hit and the time threatened. The latter was the time between a
target turning into a threat and either shooting the subject (time
threatened = 5 s) or being shot by the subject (time threat-
ened < 5 s). The total time threatened was divided by the num-
her of threats to give a mean time threatened.
Horizontal and vertical head (helmet) positions were vector
summed to provide head pointing angle (HPA) displacement and
then differentiated to HPA velocity. The standard deviation of w
0
100
90
80
70
60 3-2927:
;
:
the time histories was used as a measure of the magnitude of
displacement or velocity. Standard deviation is the same as the
a. fr,
50
3, 6 OR 9 PRESENTED
root mean square for time histories with a mean of zero. 40 3 ThREATS
Target replacement error was determined by a least error
30
method. A matrix was formed of the absolute angular distances
between each target and each replacement. Average replacement
I U U U U
error was determined by matching nonguessed replacement po- 20 45 60 90 120
sitions with the target positions, in the manner that produced FIELD -OF -VIEW (DEG)
least total absolute error, and then dividing by the number of
(a)
nonguessed replacements.
3.2. Performance data 4.5 3. S OR S PRESENTED
Performance decreased with decreasing FOV. Subjects hit fewer 6
threats and were threatened for longer durations with small FOVs 4.0 9 3 ThREATS
6
Cl)
3.5
are shown in Fig. 3 . Analyses of variance (performed on the z0
data from the 6 conditions in which targets turned to threats)
indicated that there was a significant effect, on both dependent
0
Ui
Cl)
3.0
,
6 9 6
2.0
3
3—3 3
ened: targets [F(5,80) = 39.82 p < 0.01]; FOV [F(4,20) =
18.09 p < 0.01); targets x FOV [F(20,320) = 2.33 p < 0.01].} 0 4560 0 120
Comparisons of the data at each FOV were conducted using FIELD -OF -VIEW (DEG)
''t' , tests, and the probability of any two means being the same (b)
are shown in Table II. The associated t values are shown in
Table III. It should be noted that with multiple comparisons of Fig. 3. Mean performance data for 17 subjects. (a) Threats hit; (b)
means, there is an increase in the probability of finding a dif- time threatened. Values on the abscissa refer to the width of the
ference between conditions when no such difference exists. One superimposed FOVs.
option to deal with this is to accept two means as being different
only when the probability value is less than the usually accepted of the task, in which the subject searched for the targets
phases
0.05 (e.g. , about 0.01). Another option, which has been adopted and then shot the threats. The subject was allowed more time
in this paper, is to present the complete data and allow readers to search for the targets with the smaller FOV, which resulted
to judge the interpretations that have been made. in a longer time history. With the 20° FOV, for the first ap-
In the least complex condition (3 presented in Fig. 3), there proximately 15 s , there is evidence of a systematic attempt to
were no differences between any of the FOVs (at the p < 0.05 scan the gaming area. The amplitude of the displacements sug-
level) with either dependent variable. In the more complex con- gests that the subject made sweeping movements from side to
dition (6 presented), the 20° FOV was significantly different side while elevating his head after each sweep. These displace-
(p < 0.01) from the other FOVs, with both dependent variables, ments are reflected in velocity changes that are large and smooth
suggesting that a minimum FOV size to achieve performance (Fig. 5). From both subjective reports and previous data,7 it
would have to be larger than 20°. The results of the 9 target would appear that these initial sweeps were sufficient to inform
condition are more complex. With both dependent variables, the the subject that there were more than 6 targets, but not where
900 and 120° FOVs were not significantly different from one they all were. After approximately 20 s there was a change in
another. The 20°, 45°, and 60° FOVs were either significantly the nature of the head movements. These can be characterized
different, or close to significantly different, from the 9fO or 120° as large amplitude, low frequency displacements, over which
FOVs, or both. One interpretation of this grouping is that a FOV are superimposed smaller, higher frequency movements. The
greater than 60° is required to achieve optimum performance. resultant velocity time history shows low amplitude, "choppy"
movement. At 60 s the first of the three threats changed shape.
3.3. Head movement data About 1 s later the displacement and velocity records show that
Representative time histories of HPA displacement and velocity the subject made a large amplitude, fast head movement. Using
for two FOVs in the 9 target condition are shown for one subject the time histories of horizontal and vertical head displacement,
in Figs. 4 and 5. The time histories are for the first and second it is possible to interpret what happened. The subject made a
TABLE II. Probabilities of the mean data from two FOVs being the TABLE Ill. The "t" values for the comparisons of means shown in
same. There were no significant differences (probabilities were greater Table II. N = 17.
than 0.08) between FOVs in the 3 target condition. Comparisons
were made using "t" tests.
OW
OW 20 45 60 90
20 45 60 90
45 —3.636 % THREATS HIT
60 —4.386 —1.000 (6 TAIGETS)
45 0.002 % THREATS HIT FOV 90 —5.638 —0.293 0.566
60 0.001 0.332 (6 TA1GETS) 120 —4.927 —2.063 —0.566 —1.074
FOV 90 0.000 0.773 0.580
120 0.000 0.056 0.580 0.299 45 —1.975 THREATS HIT
60 —2.768 —0.643 (9 TA1ETS)
45 0.066 % THRFMS HIT FOV 90 —3.917 —2.864 —2.057
60 0.014 0.529 (9 TA!GETS) 120 —3.631 —2.167 —1.852 0.523
FOV 90 0.001 0.011 0.056
120 0.002 0.046 0.083 0.608
45 3.396 TI THREAD
45 0.004 TI THREATEN
60 4.605 1.081 (6 TAIETS)
FOV 90 6.302 0.852 —0.066
60 0.000 0.296 (6 TAIGETH) 120 3.965 0.788 0.115 0.142
FOST 90 0.000 0.407 0.948
120 0.001 0.442 0.910 0.889
45 1.765 TIME THREAT
60 1.820 531 (9 TA85E!IS)
45 0.097 TI THBEA1 OW 90 3.665
0 .
2.735 1.718
60 0.088 0.603 (9 TA1E'1s) 120 4.358 3.240 3.170 0.625
FOV 90 0.002 0.015 0.105
120 0.001 0.005 0.006 0.541
4. DISCUSSION
' 'ballistic' '
head movement from the extreme right to the left, The questions addressed by this experiment were, how much
which passed across a threat located straight ahead. The move- does reducing the FOV size of an HMD affect performance at
ment continued past the center to the left. The subject then a visual search/monitoring task, and what is the shape of the
brought the threat within the FOV as quickly as possible. The FOV/performance curve. The study also sought to elucidate some
of the mechanisms by which FOV size might affect performance.
approximately 2 s period immediately following the rapid head
movement, where the velocity and displacements were close to Measurements were made of a number of different, but inter-
zero, indicate where the subject aligned the reticle over the threat related, dependent variables, and the experiment manipulated
and shot it (with 2 s to spare). both the FOV size and the complexity of the task. This discussion
The 1200 FOV condition exhibits large amplitude, sweeping will attempt to answer the questions by using the different per-
displacements, but of a lower amplitude than the 200 FOV con- spectives provided by the data.
dition. Head velocity was smoother and of a consistently larger
amplitude than in the 200 FOV condition. 4.1. Magnitude of the effects
The mean head movement data are shown in Fig. 6. For the Reducing the size of the FOV reduced performance by an amount
conditions with 6 and 9 targets presented, there were smaller that was dependent on the complexity of the task (Fig. 3). For
amplitudes of head displacement and greater amplitudes of head the least complex task (3 presented), reducing the FOV from
velocity with a large FOV than with a small FOV. Analyses of 120° to 20° produced about a 10% reduction in the number of
variance, performed on the three conditions in which there were threats hit and an 8% increase in time threatened (data normalized
3 threats, revealed a significant effect of number of targets and to the 120° FOV condition). For the most complex task
FOV but no significant interaction between the two. {HPA dis- (9 presented), reducing the FOV produced a 45% reduction in
placement: targets [F(2,32) = 72.56 p < 0.01]; FOV [F(4,64) = threats hit and a 43% increase in time threatened. Similarly, the
3.9'7p <O.O1};targets x FOV[F(8,128) = l.fi7p = 0.113]. smallest FOV required to produce optimum performance was
HPA velocity: targets [F(2,32) = 21 .19 p < 0.01]; FOV dependent on task complexity. For the least complex task, a 20°
tF(4,64) = 6.03 p < 0.01]; targets x FOV [F(8,128) = 1.60 FOV was sufficient. For a more complex task, a FOV larger
p = 0. 13].} The absence of an interaction precluded the com- than 20° was required. For the most complex task, a FOV larger
parisons of individual FOVs. than 60° was required.
4.2. Reasons for the effects
3.4. Target replacement data 4.2.1. Effect of task complexity
The mean replacement errors for the 6 and 9 target conditions To successfully monitor more targets, the subjects had to move
(9P:3T:6R and 6P:3T:3R) are shown in Fig. 7. In addition, the their heads faster and/or remember more target locations. The
mean replacement errors for the condition 3P:OT:3R are shown head movement data (Fig. 6) show that subjects moved their
for comparison. Analyses of variance (performed on all 6 con- heads slower, not faster, with more targets. Two explanations
ditions in which targets were replaced) showed a significant for this behavior are that the subjects were restricted from making
effect of number of targets [F(5,80) = 4.97 p < 0.01] but no fast head movements (for example, by the helmet mass) or that
significant effects due to FOV [F(4,20) = 2.37 p = 0.087] nor they were not sufficiently certain about the target locations to
any interaction between number of targets and FOV size. make large amplitude, fast, head movements. It is unlikely that
9 PRESENTED 3 THREATS
3 —3
36
3, 6 OR 9 PRESENTED
666
Cl)
Ui 3 ThREATS
Ui
C,
Ui
0 32
Cu
w 30 9
LiJ
ck
0
w 28
a 45O O i!o
FIELD -OF VIEW (DEG)
(a)
0 10 20 30 40 50 60 70 80
0
z0
C)
6O
55 3
/3\ 3
Ui
C,) 9
50
TIME (S) Cl)
Ui
Ui 6 OR 9 PRESENTED
45
Fig. 4. Time histories of vector summed horizontal and vertical head C,
Ui
displacement (head pointing angle) for one subject. O is straight 0 40
3 ThREATS
ahead.
0 4560 0 10
FIELD -OF -VIEW (DEG)
9 PRESENTED 3 THREATS
(b)
200
Fig. 6. Mean head movement data for 17 subjects. Values on the
100 ordinate are the standard deviation of the (a) displacement and (b)
velocity time histories during the search and shoot phases of the
task.
a
z0 —100
w0 —200 6 OR 9 PRESENTED
Cl
—
I
I '
I ' I
' I 20 6
3 ThREATS
(I) 18 9'
w
w 200 U) 16
0
w
Ui
a 100
3 PRESENTED . 0 THREATS
—100 10 N
—200 8
10 20 30 40 50 60 70 20 45 60 90 120
Fig. 5. Time histories of head pointing angle velocity for one subject. Fig. 7. Mean target replacement error for 17 subjects. Targets were
Data were derived by differentiating the displacement time histories replaced in the 6 and 9 target conditions following search and shoot
in Fig. 4. activity, performance at which is shown in Fig. 3.
the subjects' performance was velocity limited. Subjects were The apparently inverse relation between HPA velocity and
capable of making high velocity movements with 3 targets (Fig. 6), replacement error, discussed in the previous section, suggests
and when they were certain about the location of a threat, they that in the 6 and 9 target conditions subjects were less certain
were also capable of doing so with 9 targets (Fig. 5, 200 FOV, about target locations. This may have been because with smaller
shortly after the start of the shoot phase). Therefore, uncertainty FOVs there was a lack of information about the patterns formed
about target location is a more likely cause of slow head move- by the targets. Such information was easier to acquire with a
ment. large FOV, where it was available at a glance, whereas with a
Replacement error may be considered a measure of the cer- small FOV successive samples of the ,gaming area had to be
tainty that subjects felt about target locations, with large errors integrated to achieve the same effect.7' Observations and sub-
being associated with low certainty. If head velocity was me- jective reports support the contention that these were information-
diated by certainty, it might be expected that head velocity and gathering strategies used by the subjects. The implication is that
replacement error would be inversely related. The condition with subjects were less able to code and/or recall nonvisual infor-
highest velocity (3P:3T:OR) had no replacements, so a measure mation about target locations (e.g. , kinesthetic information about
of certainty for this condition had to be estimated. The most head position).
similar condition was one in which there were 3 targets and no The data from this experiment do not support the hypothesis
threats (3P:OT:3R). In this condition, shown in Fig. 7, the re- that the need to point the eyes in the same direction as the head
placement error was the lowest. Furthermore, neither HPA ye- affected head velocity. However, in view ofthe well documented
locity nor replacement error show a consistent effect of FOV evidence of the links between the head and the eyes' ' and the
size in the 3 target condition. The other two conditions in Fig. 7 implications of such an effect on performance, it is reasonable
show some evidence (although not statistically significant) of to call attention to this possibility as an area for future research.
decreasing replacement error with increasing FOV, whereas the For example, one of the ways suggested to alleviate the perfor-
data in Fig. 6 show increasing velocity with increasing FOV. mance decrement of a small FOV is to cue the user about target
Because of this apparently inverse relation between replacement locations. It is hypothesized that this would increase the certainty
error and head velocity, it seems possible that head velocity was about target positions and so improve performance. However,
mediated by how certain the subjects were about target location. if one of the ways a small FOV affects performance is by re-
ducing head velocity via the head/eye interaction, a target cue
may not produce the increase in head velocity required to com-
4.2.2. Effect ofFOV size
pletely restore performance.
A large FOV allowed subjects to use both head and eye move-
ments to search large portions of the gaming area. Eye move-
ments are faster than head movements, and coordination of the 4.3. Specifying FOV size
head and eyes to acquire visual targets is a much-practiced and The decrement in performance caused by FOV size is task-
well-coordinated In addition, a large FOV afforded dependent. Therefore, to make design recommendations, more
subjects the use of their peripheral vision. The combination of must be known about the intended task. It can be assumed that
peripheral vision and eye movements gave subjects forewarning for tasks requiring the integration of information from a large
about the location of targets, which may have reduced the oc- gaming area with many objects, a larger FOV is required than
currence of overshoots during the visual acquisition of targets. for a more restricted gaming area with few objects. Performance
In contrast, the head and eyes had to point in the same direction appeared to depend on how well the subjects could integrate
for subjects to extract information with the smaller FOVs, and information. The task of information integration was aided by
peripheral vision provided less information. Therefore, the syn- head movements, but in this experiment, it was not limited by
ergistic effect of eye movement was lost, the opportunity to how much or how quickly the subjects could move their heads.
make overshoots increased, and subjects had to move their heads Analysis of head and eye movements while performing candidate
more. tasks may provide useful data about the size of the likely gaming
In the more complex target conditions, subjects made faster area. More traditional task analysis may reveal the number of
head movements with a large FOV than with a small FOV objects about which location-information is needed. It should
(Fig. 6). This may seem counter to intuition. It might be ex- be possible to generate a set of equal performance curves, using
pected that in order to cover the same area, head movements experimentally derived data, that specify the range of sizes of
must be faster with a small FOV than with a large FOV. There FOV and gaming area required to achieve a specified perfor-
may be three reasons for the observed results. Subjects may have mance for a fixed number of targets. Such information would
reduced their head velocity with small FOVs so as not to over- aid in the specification of FOV sizes.
shoot the targets (there was less forewarning). With a small
FOV, subjects were less certain about the target positions, and
5. CONCLUSIONS
so reduced their head velocity . The need to restrict eye move-
ments with the small FOVs affected the manner in which subjects Reducing the size of the field-of-view of a helmet-mounted dis-
moved their heads, which was reflected in reduced velocities. play reduced performance at a search and shoot task, as measured
If fear of overshooting was a problem, it should be expected by the number of threats hit and the time for which the subject
to decrease HPA velocity at the smaller FOVs in the 3 target was threatened.
condition as well as the more complex 6 and 9 target conditions. The magnitude of the decrement was dependent on the num-
The evidence for a FOV effect on HPA velocity (Fig. 6) for the ber of targets presented, such that the minumum FOV required
3 target condition is less consistent than for the other conditions. to achieve optimum performance increased with increasing num-
This would suggest that caution about overshooting was not the ber of targets. For the least complex task (3 targets) a FOV 200
only cause of the FOV effect. high x 20° wide was sufficient, for a more complex task (6 targets)
a FOV larger than 200 x 200 was required, and for the most 8. M. Venturino and R. J. Kunze, ' 'Spatial awareness with a helmet-mounted
complex task (9 targets) a FOV larger than 600 x 600 was display," inProc. 33rdAnnualMeeting oftheHumanFactorsSoc., 1388—1391
(1989).
needed. 9. M. J. Wells, M. Ventunno, and R. K. Osgood, "Effect of field-of-view
With the more complex tasks, subjects moved their heads size on performance at a simple simulated air-to-air mission,' ' in Helmet
less, but faster, while performing the same task with a large MountedDisplays, J. T. Carollo, ed., Proc. SPIE 1116, 126—137 (1989).
10. M. A. Gresty, ' 'Coordination of head and eye movements to fixate contin-
FOV than with a small FOV. uous and intermittent targets," Vision Research 14, 395—403 (1974).
It is hypothesized that head velocity may have been mediated 11. 0. R. Barnes, ' 'Visual-vestibular interaction in the coordination of vol-
untary eye and head movements,' ' in Progress in Oculomotor Research,
by how sure subjects were about target locations. This hypothesis Fuchs and Becher, eds. , pp. 299—308, Elsevier, Amsterdam (1981).
is supported by an apparent (but not statistically significant)
inverse relation between head velocity and error at recalling
target locations.
There is evidence that helmet mass was not the limiting factor
for performance of the task used in this experiment. Maxwell J. Wells received a B.Sc. degree in
biology and psychology in 1978 from Stirling
6. ACKNOWLEDGMENTS University in Scotland and a Ph.D. degree in
engineering and applied science from South-
This work would not have been undertaken without the inspi- ampton University in England in 1983. His
ration and leadership of Tom Furness. The authors also wish to doctoral dissertation was on the develop-
ment and assessment of a technique for re-
acknowledge the contributions of programmers Jenny Huang, ducing the effects of whole-body vibration on
Ken Aldrich, and Jim Shaw and experimenter Pat Brown. Chuck performance with helmet-mounted displays.
Goodyear and Gary Beckler helped with the statistical analysis Since 1987 he has been employed as a prin-
and preparation of figures. cipal investigator, supporting the Visual Dis-
play Systems Branch of the Armstrong Aerospace Medical Research
Laboratory. He is a member of the Applied Vision Association, the
7. REFERENCES Underwater Research Association, the Human Factors Society, and
1.R. A. Erickson, ''Line criteria in target acquisition with television,' ' Human SPIE.
Factors 20, 573—588 (1978).
2. L. M. Biberman, Perception ofDisplayedlnformation, Plenum Press, New
York (1973). Michael Venturino received his Ph.D. degree
3. B. L. Silbernagel, ''Using realistic sensor, target, and scene characteristics
to develop a target acquisition model," Human Factors 24, 321—328 (1982). in experimental psychology from the Uni-
4. C. H. Baker and R. Nicholson, ' 'Raster scan parameters and target iden- versity of Maine in 1983. He is currently an
tification,' ' in Proc. IEEE 19th Annual National Aerospace Electronics engineering research psychologist in the Vis-
Conf. (1967). ual Display Systems Branch ofthe Armstrong
5. C. A. Phillips and J. S. Petrofsky, "Neck muscle loading and fatigue: Aerospace Medical Research Laboratory at
systematic variation of headgear weight and center-of-gravity,' ' Aviat. Space Wright-Patterson Air Force Base. His re-
Environ. Med. 54(10), 901—905 (1983). search interests include visual cognition,
6. K. R. Boff and J. E. Lincoln, Engineering Data Compendium, Human
Perception and Performance, AAMRL, Wright-Patterson AFB, Ohio (1988).
spatial information processing, and princi-
7. M. J. Wells, M. Venturino, and R. K. Osgood, "Using target replacement pIes of human information processing as ap-
performance to measure spatial awareness in a helmet-mounted simulator," plied to display design. He is a member of
in Proc. 32nd Annual Meeting of the Human Factors Soc., 1429—1433 the editorial board of the journal Human Factors and is also an as-
(1988). sociate member of IEEE Systems, Man, and Cybernetics Society.