1. INTRODUCTION
Spatial
disorientation is still a pilot killer.
Spatial disorientation is associated with the experiencing of an
orientational illusion for pilots [Gillingham 1985]. Loss of situational awareness has been a major concern in
military aviation, where one encounters a wide aircraft aerodynamic envelope
(acceleration and velocity) [Barnum 1968, Cheung 1995]. Requirements for close formation flying
under poor weather conditions is another factor related to this
phenomenon. Civilian aviation has
not placed much emphasis on this pilot-system limitation [Kirkham 1978, Johnson
1989, AOPA 1993]. Is it a problem
only for high-performance fighter pilots?
Recently
spatial disorientation was identified as a link in the causal chain of factors
related to a major US air carrier crash.
A DC-9 jetliner had a collision with the ground because of an encounter
with a microburst and the pilots’ subsequent loss of situational awareness due to somatogravic illusion
[NTSB 1994]. The flight crew became disoriented
during the transition from Visual Meteorological Conditions to Instrumental
Meteorological Conditions.
What
type of computer graphic display, head-up display, or head-mounted display
might facilitate spatial orientation?
What should be the contents of a training syllabus against
disorientation? Is our
understanding of its pathophysiology sufficient to permit the design of truly
better instruments and training?
In
aerospace medicine textbooks, there has been no quantitative description about
natural body and head alignment of pilots other than in a straight and level
flight. Only recently this was
challenged by two ground simulator studies [Patterson 1995A, Smith 1995]. Some flight instructors teach students
that keeping their head and body straight along the body z-axis (longitudinal
axis, Figure 41) is the proper body
alignment. But the simulator study
found that pilots laterally deflect their heads during turning maneuvers.
Positional
alignment of head and body during flight maneuvers is a significant limiting
factor for the design of instruments and cockpit layout. The latest designs of Head-Up Displays
or Head-Mounted Displays are more sensitive to this geometry because of their limited
angle of view. In order to view
some military head-up displays, the eyes must be kept within an 8 x 13 cm (3 x
5 inch) field. These limits are
easily exceeded if pilots roll their head around their head x-axis (deflect the
head laterally).
The
physiological dynamics of pilot head motion during flight has yet to be
investigated.
2. BACKGROUND
Actively
controlled human flight began with hang gliders, under visual flight
condition. The Wright Brothers
from Dayton, Ohio began controlled, powered flight after their extensive
experiments with gliders. The next
breakthrough, was the invention of the aircraft’s attitude display.
Prior
to the emergence of the apparatus, there was an interesting instrument
arrangement used by Charles A. Lindbergh for his solo transatlantic flight in
1927. Since his forward view was
blocked by a fuel tank, he used a periscope to provide a visual reference
[Roscoe 1966]. Although he is said
to have used forward slipping during approach to gain a better view of the
runway, the periscope deprived him of peripheral vision, which is important for
pilots [Kochhar 1978].
Based
upon his design of a gyroscopic stabilizer for ships, Elmer Sperry, Jr. [Laboda
1995], in 1910, extended the technology by developing a gyroscope for use by
pilots for determining aircraft attitude.
The Sperry Artificial Horizon allowed Lt. James H. Doolittle to fly his
NY-2 Navy trainer airplane ‘blind’ on 24 September 1929. This historic apparatus is on display
at the US Air Force Museum, Dayton, Ohio.
Figure 1.-Computer graphics attitude display.
A display of a Fokker F-100 jetliner. It is the latest design in use, but stays with the concept
of a stationary aircraft symbol with a moving horizon.
Figure 2.-Variation in Attitude Indicator
design.
This AI in a 1966 Piper PA30B twin
piston-engine airplane has a bank pointer that moves with the horizon bar,
instead of with the miniature airplane. Many AI’s of this design are still in
use today. This kind of design
variation is also seen in VOR (Very High Frequency Omni Range) indicators.
Figure 3.-View of AI and the earth
horizon.
The Attitude Indicator (arrow) and the
earth horizon seen from cockpit of a CH2000 single engine airplane. The roll angle of the aircraft is 64°
to the left. This picture is aligned
to the page so that it looks natural to the reader, while actually it differs
from the retinal image of the pilots in the cockpit because of the limited
motion of the head in flight.
[ AOPA Pilot, 36(12),1993: 47 ]
There
is a speculation that Lindbergh’s success with the periscope may have affected
the design of the artificial horizon (now called the Attitude Indicator, Figure 1, Figure 2). All operational attitude indicators,
except for those of Russian design, use ‘moving horizon’ symbolic design. The horizon bar (Figure 1) resembles that of a short portion of the earth horizon as
if seen through a periscope (Figure 3).
Peripheral
displays, such as the para-visual director, peripheral command indicator,
HOVERING display, and the Malcolm horizon stimulate peripheral vision to
provide aircraft attitude information [Stokes 1988]. They have proven to be relatively effective despite degraded
visual acuity associated with peripheral vision. Visual acuity in Snellen’s fraction decreases from 1.0 in
the visual center to 0.2 at 10° off center, 0.1 at 20°, and 0.07 at 30°
[Westheimer 1992].
How
can we present attitude information to maximize foveal vision? Although the options are many, modern
Russian fighters use an attitude display whose miniature airplane, instead of a
horizon bar, rolls.
In
the past, no commercial competition was seen between the ‘moving horizon’ and
‘moving airplane’ display. This is
probably due to the initial success of the Sperry artificial horizon and the
comprehensive analogy of periscopic view (‘cut out’ earth horizon appears in an
Attitude Indicator) [Poppen 1936, Roscoe 1966]. Other equivalent terms, which are often confusing, are
summarized in Table 1. It is worth noting that an integrated
display which changes from ‘moving aircraft’ to ‘moving horizon’ for roll-in
has been devised [Fogel 1959, Roscoe 1975].
Table 1.-Terms equivalent to ‘moving horizon’
and ‘moving aircraft’.
moving
horizon moving
airplane
inside-out outside-in
fly-to fly-from
moving
card or tape moving
pointer
earth
referenced aircraft
referenced
aircraft
stabilized space
stabilized
in
aircraft coordinates in
earth coordinates
modified
from [Johnson 1972]
A
study by the Federal Aviation Administration Civil Aeromedical Institute with a
Beechcraft T-34 [Hasbrook 1973] compared the two types of display and concluded
that:
Data from many of
these earlier studies suggest that the outside-in (moving- aircraft)
indicator provides better pilot performance: but this in-flight study fails,
in the main, to show any such well defined, overall
advantage.
Despite
pioneer efforts, it was not until 1930 that military pilots were taught that
the instruments should be used as source for flight information and that they
should not fly by the “seat of their pants” [Malcolm 1984]. This is not widely understood among
today’s leisure pilots.
When
one questions the dynamics associated in comparing a ‘moving horizon’ with
‘moving airplane’, the answer is ambiguous. Missing is the recognition of natural behavior of the
pilots. What is the natural
position of the head relative to the cockpit?
When
asked if the head moves during flight, most pilots will answer ‘no’. Only a few pilots admit that they roll
their head. Even demonstration
team members of the US Navy [Patterson 1995A] and the US Air Force (Figure 4) were observed to roll their
heads around the head z-axis (tilt their head) in a direction opposite to the
center of the turn in visual flight. Pilots prefer to view a picture as if aligned to the
earth horizon rather than to the cockpit (Figure
3). The only literature about
head motion in flight, which surfaced, gave a general indication that subject
pilots kept their head z-axis normal to the ground [Hasbrook 1973].
Recent
simulator sutdies revived this old but unanswered question [Patterson 1995A,
Smith 1995]. If pilots are rolling
their heads, is a ‘moving horizon’ better adapted to prevent roll reversal
(mistakenly initiate roll to the opposite direction)? What if the display moves with the head instead of being
fixed to the cockpit? What is the
natural motion of the head that should be assumed in a new display design? Which attributes of display design
should be adopted?
Figure
4.-Head roll of F-16
pilot in 9 G turn.
This U.S. Air Force Thunderbirds
demonstration team member is quickly rolling into 9 G turn. The roll (bank) angle of aircraft in
this picture is 64° (ÐBOH’). His head is rolled to the right around head x-axis
(laterally deflected to the right) with an angle of 15° (90° - ÐAOH) relative to the cockpit. Cockpit level is represented by line
HH’. Because the body is leaned to
the right with an angle of 3° (ÐHOS), the angle of head roll relative
to the body z-axis is 12° (15° - 3°).
There is a slight yaw of the head to the left, which does not affect
more than 1° for this head roll angle relationship to the cockpit. A transient maximum roll of the head of
24° relative to the cockpit was observed when roll of the aircraft was at 60°
(just before this picture). After
this picture, roll of aircraft was kept 82° to 85° to keep 9 G turn, which
theoretically requires 83.6° of bank [ cos- 1 (1/9) ]. Head roll angle was kept approximately
6° to the right relative to the cockpit while 9 G 360° turn was continued.
[ International Video Corporation 1990 ]
3. RESEARCH
OBJECTIVES
The
purpose of this research is to determine how pilots (flying and non-flying)
align their heads during actual flights in a general aviation aircraft.
It
is hypothesized that general aviation pilots in visual flight rule (VFR)
conditions align their heads with the horizon of the earth for visual
orientation, which causes head rotations around the x-axis of the head. This study examines this hypothesis by
recording and analyzing pilot head motion during actual VFR flight in a general
aviation aircraft.
To
supplement the flight test, head motion data were collected during the ground
taxing phase of a flight.
4. METHODS
4.1 Subjects
A
total of 10 civilian volunteer pilots holding current FAA medical certificate
(either class 2 or 3) and FAA pilot rating (either private, commercial, or Air
Transport Pilot) participated (Table 2).
The
protocol for participation by
human subjects was approved, prior to flight test, by the Office of
Research and Sponsored Programs, Wright State University.
Table
2.-Profile of pilot
subjects.
subject number (n = 10)
|
(numbered as tested) |
total flight time [hours] (airplane) |
age [years] |
|
8 |
80 |
38 |
|
9 |
110 |
37 |
|
6 |
485 |
41 |
|
7 |
570 |
22 |
|
1 |
710 |
31 |
|
4 |
780 |
24 |
|
3 |
1200 |
26 |
|
5 |
1230 |
30 |
|
10 |
2600 |
34 |
|
2 |
4000 |
47 |
|
mean, sample SD |
1177 ,
1165 |
33.0 , 7.52 |
All subjects held current FAA certificate for the research
airplane. All 10 subjects were
male with an age rage of 22 - 47 years and airplane flight tie of 80 - 4000
hours. None had military flying
experience.
Figure
5.-Research
airplane. Cessna C172L Skyhawk,
fixed-gear, 4-place.
Figure
6.-Instrument
layout of the research airplane.
Attitude Indicator type is
EDO-AIRE 5000F-6. The miniature
airplane of the AI moves with the bank pointer, which is the standard in U.S.
at present. This picture was taken
from pilot eyepoint with a comparable field (35 mm film, f = 35 mm).
Figure.
7.-Visual
Flight Rule 1:500,000 sectional chart for the flight test air space. The area marked by arrows was used for
the research flights. The meteorological
condition was better than VFR minimum; Dayton visibility was 10 statute miles
or better for 9 subjects and 7 miles for subject #9. Dayton wind was 6-15 knots. Flight altitude was 1500-3500 feet Above Ground Level, more
than 1000 feet below the clouds.
All flights were completed within daylight hours.
4.2 Aircraft
The
aircraft used for this study was a 150 horse power, single-engine land,
fixed-gear, 4-place general aviation utility airplane (Figure 5, Cessna C172L, N7116Q, based at Brookville Air Park, airport code I62). All subjects were volunteers
familiar with this airplane type.
The
research airplane was equipped with a set of standard Instrument Flight Rule
instruments, which includes an attitude indicator (attitude gyro). Figure
6 shows instrument panel arrangement.
This picture corresponds to the pilot’s view in flight. Attitude Indicator is located at top
center of the ‘T’ layout for cardinal Instrument Flight Rule instruments.
4.3 Airspace
and weather
The
airplane took off from Brookville airport and returned. Flight maneuvers took place in a
practice area approximately 15 km southwest of the airport (Figure 7). Because of Dayton jet traffic, flight altitude was kept
below 4,000 feet Above Ground Level.
The altitude flown was between 1,500 and 3,500 feet AGL, and was at
least 1,000 feet below the clouds.
Visibility
reported at Dayton International Airport at the time of flights was better or
equal to 10 statute miles for 9 subjects; 7 statute miles for subject #9. Surface wind at Dayton was reported at
6 to 15 knots, generally from southwest.
There was a slight turbulence for subject #2 due to thermals. Other subjects had no turbulence. Surface temperature at Dayton
International Airport (308 m above Mean Sea Level) was between -4 to +11
°C. All flights were completed
within daylight hours under Visual Meteorological Condition. Weather was at all times above federal
Visual Flight Rule minimums.
4.4 Head tilt
recording
The
subjects were seated in the left seat of the airplane. They wore a small, light-weight
(70 g) CCD camera(Sony CCD-MC1, f=3.6 mm) on a headset (Figure 8 ). Video
recording was made during taxi and flight with a NTSC-VHS portable video
cassette recorder (Panasonic AG-170).
The center of the videocamera field was
aimed at the center of a white bar, 530 mm in length, located on the top of the instrument panel. The horizon of the earth was visible in
the camera field (Figure 10). Alignment of the camera relative to the
earth horizon was calibrated with a grid pattern and a protractor while on the
ground (Figure 11). Angular readings were entered into
computer by keyboard (Figure 9).
Subjects’ head motion relative to the
cockpit were analyzed from the recorded video picture. Head roll angle around the head x-axis
was measured on a TV monitor screen with a protractor, frame by frame
(Panasonic AG-1970 video cassette recorder and Sony SSM-2010 monitor) (Figure 12). In-flight pictures were
compared to a calibrated video picture recorded during the preflight phase on
the ground.
Figure
8.-Videocamera
mounted on a headset.
SONY CCD-MC1 lightweight
videocamera was mounted onto David Clark H10-13.4 aviation headset. The head pad of the headset was removed
to minimize height. Cables were
not interfering with pilot head motion.
Line-of-sight of the camera had an angle range of 10 - 15° relative to
horizon when a pilot is in the seat (Figure
10).
Figure
9.-Data acquisition system diagram.
The video tape pictured with a portable videorecorder was manually
analyzed and the result was entered to computer through keyboard.
Figure
10.-Dimension of videocamera setting in the research airplane.
Figures are shown in mm. The height of the subjects ranged from
172 cm to 190 cm, which varied the line-of-sight angle of the videocamera from 10˚
to 15˚ relative to horizon. This
range of camera angle did not cause rotational skew of the cockpit image more
than 1˚.
B: reference bar at the top
of instrument panel.
C: optical center of CCD
videocamera attached to headset.
E: eyepoint of subject.
Figure
11.-Calibration of videocamera angle.
Angle of the head of a pilot
and reference white bar on instrument panel is in calibration with a protractor
mounted on the head, a grid, strings and weights.
Figure
12.-Cockpit picture in analysis.
Video images were measured
manually frame by frame with a 20-inch video monitor and a protractor. Both aircraft roll and head roll were
measured in this example frame.
|
|
Figure 13.-Head alignment of a pilot in 45° bank
turn to the left.
Posterior view of a pilot flying Cessna
172P. Actual roll (bank) angle of
the airplane was 48° to the left at the moment this video frame was
captured. The head is rolled to
the right around head x-axis with an angle of 11° relative to the cockpit. This student pilot was trying to keep
45° bank turn. It is common to
observe small amount of jerky transient head roll component superimposed onto
steady bias. Note that this
picture was taken by a videocamera different from that used for data
collection.
When
the head of pilot rolled to the right, it was reflected by the left roll of the
white bar on the top of the instrument panel. Actual roll angle of the aircraft was also measured from the
angle between the earth horizon and the white bar on the instrument panel (Figure 13). The initiation of aircraft rolling motion, the return to
level and straight flight, and the angle of the head relative to the cockpit
(the white bar) were all evident in video picture. Frame-by-frame analysis of the picture was made necessary to
determine the roll angle of the head because there were small-angle motions
superimposed on relatively large-angle static components, the result of
establishing in aircraft turn (constant bank angle).
Time
resolution of the measurement was 1/30 seconds (video frame interval). The angular resolution limit of
rotation around head x-axis was unmeasurable because the resolution limit was
much better than the precision of mechanical rotatory positioning of
videocamera on a tripod. A
protractor used for reading had 1° step in scale.
There
was an artifact in the picture rotation in the head coronal plane when subjects
rolled their head around the head z-axis (roll in the horizontal plane). The amplitude of this (around x-axis)
angular artifact was 1.0˚ per 10˚ of head roll (around z-axis). That is, when subjects looked 10˚ to
the left (horizontally), it seemed as if they tilted their head 1° to the left. It was revealed in post-flight video
analysis that all necessary scenes were recorded with less than 10° of the head
roll around the head z-axis. Thus
artifact effect due to this phenomenon was less than 1° in this study.
There
was no measurable artifact in the picture rotation due to head roll around the
head y-axis (head pitch change).
Overall
precision and systemic error estimation for angular measurement was 1˚. This was considered a degrade for this
study. No positional drift of the
videocamera, due to headset slippage relative to the skull, was observed in the
post-flight video analysis.
4.5 Flight task
The
subject sat in the left seat of the airplane. The investigator took the right seat. The flight task was categorized into
three flight modes; AI (Attitude indicator) mode, visual mode, and non-flying
mode.
•
Attitude Indicator Subject
flew the aircraft with attitude indicator and mode visual
reference. This mode examined
subject’s response in combined visual/instrument environment.
• Visual mode Subject
flew the aircraft with visual reference only (with
no instrument reference). This mode examined subject’s response with pure
outside visual
reference.
• Non-flying mode Right
seat pilot (the investigator) flew the aircraft; subject
observed outside of aircraft. This
mode examined subject’s response in alert, motivated but non-controlling
state.
For
each flight mode, 6 turns were flown.
Bank angle of the aircraft for every turn was assigned to a subject
randomly by the investigator. The
bank angles used were: right and left 45 ˚, right and left 30 ˚, right and left
15 ˚. These bank angles are also expressed with + sign indicating right bank
and - sign indicating left bank, i.e. -45, -30, -15, +15, +30, and +45
degrees. Head motion recording was
continued in level and straight flight (bank angle 0 ˚) and ground taxi.
The
first run of flight maneuvers was started from AI mode, then to visual mode,
and ended in non-flying mode, with 6 x 3 = 18 turns in a run. Within each
flight mode, 6 bank angles were arranged randomly.
Each
subject had 3 runs, 18 x 3 = 54 turns in total. Order of flight mode was kept the same (AI, visual, then
non-flying mode), in order to avoid flying successively in a same mode. The order of bank angles within each
mode was changed randomly.
An
example of the order of flight maneuvers with raw head angle data are shown in Table 15 (in Appendix).
4.6 Ground
task
No
specific assignment was given to subjects for ground maneuvers (turns on runway
or taxiway). Video images for
ground turn analysis were selected from continuous video record, based on
picture quality (lighting condition was a major factor).
4.7 Pilot
briefing
Before
a flight, the subject pilot was briefed orally as follows:
“The purpose of this research is to
measure your neck motion during the flight. Since some amount of natural cervical motion is anticipated,
DO NOT restrict your body motion.
Fly and look out as you always do.
But do not keep looking into the side windows during record turns (to
keep the angle of the videocamera within a favorable range). Make clearing
turns before record turns if necessary.”
“Count to three out loud when flying
level and straight (counting was used to time straight and level portion of the
flight and clarify subject’s intention).
Roll in and out at your usual rate. Count to ten out loud after you have established the specified
bank angle.”
For
the Attitude Indicator mode, subjects were orally briefed as follows:
“Use the attitude indicator to
establish and keep the specified bank angle.”
For
visual mode, subjects were orally briefed as follows:
“Do not refer to the instrument panel
for this turn.”
For
non-flying mode, subjects were orally briefed as follows:
“I have the airplane controls. You will look out for other traffic.”
No
specific instruction was given to the subjects for ground taxi.
4.8 Experimental
design
The
ground simulator data (Patterson 1995A) showed that variance of head roll angle
is relatively small when aircraft roll angle is within 45˚. From this result, it was estimated that
head roll angles in flight test would have a standard deviation of 8 degrees or
less (Table 3).
Table
3.-Ground simulator
study data.
Aircraft roll angle Head roll
angle [deg] Sample mean (n=14) variance
45˚ left 13.75 32.36
30˚ left 11.32 11.75
15˚ left 5.31
3.74
15˚ right -4.79 5.53
30˚ right -10.39 15.31
45˚ right
-12.87
39.30
The largest Standard Deviation was 6.27° (square root of
39.30) for 45° right aircraft roll.
Arbitrary assumption that flight data will have 30 % larger SD than
simulator study yielded SD estimation of about 8° for flight test data.
[Patterson 1995A]
The
number of subjects necessary was estimated to be at least 10 with the
assumptions below [Odeh 1975]:
• randomized block experimental design
with subjects as blocks
• five degrees difference in head roll
angle is to be detected among flight modes (with Attitude
Indicator, visual, and non-flying status.)
• average variance of the groups is 3/4
that of random variation
• error rate 
• estimated standard deviation of head
roll angle is less than 8 degrees
The
statistical design for this study is a factorial design, 10 subjects 
3 flight modes 
7 aircraft bank angles. The dependent variable was head roll
angle. Subjects were treated as a
random factor.
5. RESULTS
5.1 Flight
conditions
The
10 subjects completed all assigned tasks.
Since the flight maneuvers were simple turns, only one subject was
requested to re-try two of the turns due to shallower bank angles than specified. All turns were coordinated to within
half ball deviation on the inclinometer.
Flight
time (block to block) for subjects ranged from 30 minutes to 1 hour 18 minutes,
average was 50 minutes. Recording
maneuvers in the air took approximately 20 minutes for 54 turns. The remainder of the flight was devoted
to taxi, ascent and approach, horizontal repositioning in the airspace (mostly
for wind correction), and test equipment operation.
5.2 Quality
of recorded image
Spot
measurement of head position in flight pictures was not difficult n = 10). However, transient blockage of camera
view was not infrequent with the pilot’s hand or the sun in view. The quality of video images were
satisfactory in lighting and in camera angle for 26 flight maneuvers with 3
subjects (#8,9,10) for frame-by-frame time series analysis.
For
ground phase analysis (n =10), 8
ground maneuver images from 5 subjects (#4,5,6,9,10) were satisfactory for spot
measurement; only one image was subjected to frame-by-frame time series
analysis (subject #10).
By
comparison to the earth horizon, it was found that the attitude indicator had a
systemic error of 5˚. The attitude
gyro indicator onboard, EDO-AIRE 5000F-6 Serial No.29005F, indicated correct
bank angle to the right, but 5˚ more than actual to the left. That is, 45˚ bank to the left was
actually at 40˚. It was still in
compliance with federal regulations, including TSO-C4C for Attitude instruments
[Kelly Instruments]. Correcting
the instrument was technically difficult.
Since there was no redundant equipment onboard the aircraft for
real-time reference, data were not correctable for this instrument error. This did not affect the precision or
systemic error of head angle measurement.
However, careful analysis for the difference in the right and the left
turn was required.
5.3 Body leans
Simultaneous
and independent recording of the head motion and the torso motion was not
feasible because only one videocamera was onboard the research airplane.
Specific attention was paid to the torso lean of subjects, by the investigator
in the right seat. To evaluate the
leans of subject body, the lateral deviation of the shoulders relative to a
seatback was visually monitored.
Only the first 6 turns by subject #9 was observed to have had leans,
with shoulder lateral deviation in excess of 5 cm. None of other subjects presented deviation more than 3
cm. Forward leaning was
frequently monitored when subjects tried to check upward traffic in non-flying
flight mode.
Head
roll angle data with noticeable body lean from subject #8 is presented in Table 4. The data were included in the data set for this study,
because there was no meaningful trend in these values.
Table
4.-Head roll with body
lean, subject #9.
aircraft roll
angle run #1* run #2 run #3 mean
left
45° 4
2
3 0.3
left
30°
7 2 10 5
left
15°
0
1
3 1.3
right 15° 5 5
0 3.3
right 30° 4 1 1 -2
right 45° 5 5 -11 -7
*Body lean was
observed in the 6 maneuvers of run #1.
All values are in degrees.
Negative sign means head roll to the left.
5.4 Pilot
head roll angle
The
head roll angles of subjects, around the head x-axis, during in-flight turns
showed a distinct relationship to the bank angle of airplane (Figure 14, Figure 15, Figure 16). Neutral point of the head around the
head x-axis, where head roll angle is 0˚, was defined as the position of the
head immediately before the aircraft began a move to roll-in. Raw in-flight data are shown in Table 16 (in Appendix).
The
magnitude of head roll response of each subject varied. However, each subject’s response was
generally symmetrical in the right and the left aircraft roll. Sample mean head roll and group
standard deviation are shown in Table 5,
Figure 17, Figure 18, and Figure 19. As seen from mean and Standard Deviation values, all data
showed the pilots’ behavior to tilt (roll around the head x-axis) their head
toward outside of the turn (in the direction that eyes follow the earth
horizon).
|
|
Figure 14.-Head roll angle around
head x-axis, in Attitude Indicator
mode of flight. All 10 subjects. There is a considerable individual
variation. However, each
subject’s head roll curve is symmetrical
about direction of aircraft
roll.
Figure 15.-Head roll angle around
head x-axis, in visual flight
mode. All 10 subjects. There
is a considerable individual
variation. However, each
subject’s head roll curve is symmetrical
about direction of aircraft
roll.
Figure 16.-Head roll angle around
head x-axis, in non- flying
flight mode. All 10 subjects. There is a considerable individual
variation. However, each
subject’s head roll curve is symmetrical
about direction of aircraft
roll.
Positive values for head or aircraft
roll mean roll to the right; negative values mean roll to the left (-45° in
aircraft roll angle means left 45° bank).
Table
5.-Result of pilot
head roll angle around head x-axis during aircraft turn.
aircraft
roll angle
[deg] attitude
indicator flight mode visual
flight mode non-flying
flight mode
mean
[deg]
group
S.D.
mean
[deg]
group
S.D.
mean
[deg]
group
S.D.
left
turn 45 5.5 5.2 9.0 5.8 8.3 6.3
30 4.9 4.3 6.8 4.4 6.0 3.8
15 2.2 2.3 4.9 3.9 3.0 2.7
right
turn +15 -2.4 1.5 -2.0 2.0 -3.8 4.3
+30 -4.1 3.7 -3.3 2.0 -5.8 5.8
+45 -4.8 3.6 -4.8 4.8 -7.1 6.9
Negative values of aircraft roll angle means bank to the
left; positive values mean bank to the right (-45° means turn to the left with
a bank angle of 45°.) Negative
values of head roll means left lateral deflection of the head (-2.4° means that
the head was deflected 2.4° to the left.); positive values mean right lateral
deflection of the head. Opposite
sign of head roll to aircraft roll means that the subject rolled the head in
the direction that retinal image is stationary. The maximum Standard Deviation value of 6.9° (at right 45°
turn) was close to the pre-test estimation of 8°. Data from this table is used in Figures D, E, and F. In Attitude Indicator flight mode,
subject pilot used both visual cue and Attitude Indicator to fly; in visual
flight mode, subject used only outside visual cue to fly; in non-flying mode,
subject acted as non-flying pilot, looking outside for other air traffic
without controlling the airplane.
Figure 17.-Pilot head roll
around head x-axis during aircraft turn, in Attitude Indicator mode of
flight.
Sample mean and sample Standard Deviation plot of 10
subjects. Each vertical bar
represents +/- 1 SD. Data are from
Table 5. Negative sign for head or aircraft roll means roll to the
left (-45° aircraft roll means left turn with 45° bank angle.) Slope of the curve is less than those
of visual mode (Figure 18) or
non-flying mode (Figure 19). Head deflection is symmetrical for both
direction of aircraft roll. Head
roll response is linear to aircraft roll within +/- 30° of aircraft roll(bank)
angle. Restricted head motion
response was observed at +/- 45° of aircraft roll, which was not seen in visual
or non-flying mode.
Figure 18.-Pilot head roll
around head x-axis during aircraft turn, in visual mode of flight.
Sample mean and sample Standard Deviation plot of 10
subjects. Each vertical bar
represents +/- 1 SD. Data are from
Table L. Negative sign for head or aircraft roll means roll to the
left (-45° aircraft roll means left turn with 45° bank angle.) Slope of the curve is more than that of
Attitude Indicator mode (Figure 17),
and similar to that of non-flying mode (Figure
F). Head deflection is larger
in left turn of the aircraft. Head
roll response is linear to aircraft roll within range of this experiment (+/-
45° of aircraft bank).
Figure 19.-Pilot head roll
around head x-axis during aircraft turn, in non-flying mode of flight.
Sample mean and sample Standard Deviation plot of 10
subjects. Each vertical bar
represents +/- 1 SD. Data are from
Table 5. Negative sign for head or aircraft roll means roll to the
left (-45° aircraft roll means left turn with 45° bank angle.) Slope of the curve is more than that of
Attitude Indicator mode (Figure 17),
and similar to that of visual mode (Figure
18). Slope of the curve is
symmetrical to both direction of aircraft turn. Magnitude of head roll response is linear to aircraft roll
within range of this experiment (+/- 45° of aircraft bank).
5.5 Comparison
of three flight modes (AI, visual, and non-flying)
Head
tilt (roll around the head x-axis) response due to aircraft bank angle was
measured in Attitude Indicator mode (subject pilot uses both outside visual
cues and attitude gyro indicator for spatial orientation), visual mode (only
outside visual cue is used for orientation), and non-flying mode (subject
observes outside without controlling airplane).
The
head roll angle data were statistically tested with regression, ANOVA, and
t-test.
5.6 Regression
analysis
The
relationship between aircraft bank and head roll angle, depicted in Figure 17, Figure 18, and Figure 19,
can be approximated with a linear model.
When we assume their relationship as,
[head
roll angle] = B 
[aircraft bank angle]
where B is a regression
coefficient, the value of B for Attitude Indicator mode was -0.127; visual
mode, -0.163; non-flying mode, -0.183.
95% t-test confidence intervals for B value were calculated as in Figure 20. This figure shows that when subject pilots were flying in AI
mode, they had less head roll relations to an aircraft roll than they did when
they were flying in either visual or non-flying mode.
regression
coefficient
Figure 20.-Regression coefficient
(ordinate) for aircraft bank angle
vs. head roll
angle.
Each
bar shows 95 % t-test confidence limits for coefficient (B) for Attitude
Indicator mode, visual mode, and non-flying mode.
Subjects
have significantly less head roll in Attitude Indicator mode.
This
graph corresponds to Figure 21. The ANOVA analysis presented in Table 7 also dictates that Attitude Indicator and non-flying mode
are most
separated among three flight modes.
The values for (upper limit, B,
lower limit) are:
AI
mode (-0.114,
-0.127, -0.140)
visual
mode (-0.147,
-0.163, -0.179)
non-flying
mode (-0.164,
-0.183, -0.202).
Figure
21.-Regression analysis for head roll angle vs. aircraft roll angle.
The 95 % confidence band for
Attitude Indicator, and linear regression lines for visual/non-flying mode are
shown. Negative values for angle
mean roll to the left. Linear regression
line for Attitude Indicator mode is significantly different from that of visual
mode or non-flying mode. This
graph corresponds to Figure 20.
5.7 ANOVA
analysis
Analysis
of Variance (ANOVA) test was applied to examine the effects of subject or
flight mode difference. Absolute
(sign-less) values of head roll angle data were used for calculating ANOVA
table because mean of head roll for a subject is close to zero due to the
symmetrical value distribution over aircraft bank angle (Table
6). There was a
difference in mean head roll angle among flight modes (p = 0.0860).
As
a post-hoc test for the effect of flight mode on head roll angle, Scheffé’s
procedure was performed. There was
a difference between the Attitude Indicator flight mode and the non-flying
flight mode (p = 0.0920), i.e. when subjects were working as non-flying pilot,
their head roll around the head x-axis due to aircraft bank was larger than
when they were using an Attitude Indicator as one of spatial orientation cues.
Table
6.-ANOVA
table for subject and flight mode.
|
effect |
df effect |
MS effect |
df error |
MS error |
F |
p-level |
|
subject |
9 |
169 |
180 |
13.8 |
12.2 |
0.0000 |
|
flight mode |
2 |
34.3 |
180 |
13.8 |
2.49 |
0.0860 * |
|
subject x flight mode |
18 |
10.7 |
180 |
13.8 |
0.776 |
0.727 |
Number of subjects was 10. Flight modes were Attitude Indicator, visual, and
non-flying. Absolute (sign-less)
value of the head roll angle was used.
Subject and flight mode both were found to be influential on the
magnitude of head roll angle.
There was no significant combined interaction effect of the two factors.
Table 7.-Post-hoc test
for head roll angle difference among flight modes.
Scheffé’s test
|
p-level |
Attitude Indicator mean = 3.53 |
visual mean = 4.43 |
non-flying mean = 4.91 |
|
Attitude Indicator |
• |
0.366 |
0.0920 * |
|
visual |
• |
• |
0.740 |
|
non-flying |
• |
• |
• |
Absolute (sign-less) values
of the head roll angle were used for the mean head roll angle calculation. There was a considerable (*p = 9.2 %)
statistical difference between effects of Attitude Indicator and non-flying
mode on head roll angle. This
result is the same as regression analysis as shown in Figure 20, which means
that Attitude Indicator and non-flying mode are separated farthest among the
three flight modes.
5.8 t-test
As
the third statistical measure used for comparison among three modes of flight,
t-test matrix was calculated (Table 8). As found by regression analysis, mean
head roll angle for each aircraft roll (bank) angle was significantly larger in
visual mode than that in Attitude Indicator mode in left turn. However, in right turn this difference was not found
significant by t-test.
In the right turn, the t-test indicated that mean head roll angle is
larger in non-flying mode than that in visual mode. This was not seen in either the
regression or ANOVA analyses.
There was no significant difference between these modes in left turn as
indicated by t-test.
In
right turn, there was a possible
difference in mean head roll, as indicated by regression and ANOVA analysis,
between non-flying and AI mode (p = 0.11, 0.11, 0.28).
Because
of a notable discrepancy between the direction of the turn (roll) of the
aircraft in comparison among flight modes, t-test by itself was not sufficient
to derive a conclusion. A
reference to the directional variation shown in Table 11 in view, these directional effects needs careful
evaluation.
Table 8.-A matrix of t-test for difference in head roll among flight modes
at each aircraft roll (bank) angle.
aircraft
|
bank angle |
two-tail t-test for head roll angle
(n = 10 ) p level (* p <
0.05) |
|
||||
|
|
flight mode |
|
AI |
visual |
non-flying |
|
|
left turn |
45 deg |
Attitude Indicator |
• |
0.0026 * |
0.048 * |
|
|
visual |
• |
• |
0.67 |
|
||
|
non-flying |
• |
• |
• |
|
||
|
30 deg |
Attitude Indicator |
• |
0.029 * |
0.39 |
|
|
|
visual |
• |
• |
0.57 |
|
||
|
non-flying |
• |
• |
• |
|
||
|
15 deg |
Attitude Indicator |
• |
0.0077 * |
0.27 |
|
|
|
visual |
• |
• |
0.12 |
|
||
|
non-flying |
• |
• |
• |
|
||
|
right turn |
45 deg |
Attitude Indicator |
• |
0.96 |
0.11 |
|
|
visual |
• |
• |
0.040 * |
|
||
|
non-flying |
• |
• |
• |
|
||
|
30 deg |
Attitude Indicator |
• |
0.29 |
0.15 |
|
|
|
visual |
• |
• |
0.11 |
|
||
|
non-flying |
• |
• |
• |
|
||
|
15 deg |
Attitude Indicator |
• |
0.51 |
0.28 |
|
|
|
visual |
• |
• |
0.11 |
|
||
|
non-flying |
• |
• |
• |
|
||
Absolute (sign-less) values of the head roll angle were
used for the mean head roll angle calculation. The most
significant difference was left turn visual vs. AI mode (the former the
larger, p 
0.029). Then right turn non-flying vs. visual mode (p 
0.11) and right turn non-flying vs.
AI mode (p 
0.28) followed. It is notable that there was
considerable difference between left and right turn. In right turn, there was no difference (0.96 
p 
0.29) in visual vs. AI mode. Conversely, there was no difference
(0.67 
p 
0.12) in non-flying vs. visual mode in
left turn. Bank (roll) angle of
the aircraft did not have much influence in these trends. Consult Table 5 for mean values.
5.9 Comparison
between the direction of aircraft roll (left or right)
The
difference in mean head roll angle by the direction of the aircraft roll (turn)
was statistically analyzed by two-tail t-test. Table 9 shows the
complied data from all three flight modes. Except at bank angle of 15°, there was a significant
difference in mean head roll angle value between the left and the right
direction of aircraft turn. The
difference in the mean head roll was 1.5° and 2.1°. These angular values are approximately 30 % of the amplitude
of head roll during turns.
The
experimental condition differs, in terms of right or left, in aircraft seating
(subjects took a left seat of the two in the frontal row in the airplane). It brought an asymmetric pilot view
through windshields for outside visual cues (Figure 22). Hence it
affected the pattern of visual vigilance in all three flight modes.
Although
the window view was asymmetrical in the right and the left, it did not present
equal effect on three modes of flight.
The left-right directional difference in mean head roll angle found in
visual mode (Table 11) was not
observed in the other two modes (Figure
19, Figure 28).
The
difference found in compiled data in Table
9 is thought to have resulted from visual flight mode part of data.
Figure
22.-Field of view for Cessna 172L.
Angular dimensions measured
from the eyes (nasion) of a pilot 172 cm in height in Cessna 172L research
airplane. Drawing is not to
scale. Side windows are weakly
affecting peripheral vision because of its rear position. Field of view in pitch (vertical) is
limited in this airplane type.
Field of view data has not been measured by the manufacturer of the
airplane.
Table 9.-Comparison of head roll
angle between left and right aircraft roll, all flight modes.
all flight modes
|
n = 30 |
head roll mean [deg] |
head roll S.D. [deg] |
t-test (two-tail) p level |
|
|
15 deg bank |
left |
3.4 |
3.1 |
p = 0.35 |
|
right |
2.8 |
2.9 |
|
|
|
30 deg bank |
left |
5.9 |
4.1 |
* left > right p = 0.049 |
|
right |
4.4 |
4.1 |
|
|
|
45 deg bank |
left |
7.6 |
5.8 |
* left > right p = 0.010 |
|
right |
5.5 |
5.2 |
|
|
Absolute (sign-less) values of the head roll angle were
used for the mean head roll angle calculation. The amount of head roll was significantly greater in left
turns, by 1.5° (5.9° - 4.4°) at 30° bank and by 2.1° (7.6° - 5.5°) at 45°
bank. There are no significant
difference in the variance of head roll.
Table 10, Table 11, and Table 12
are presented for each mode of flight.
Table 10.-Comparison of
head roll angle between left and right aircraft roll, Attitude Indicator flight
mode.
|
Attitude Indicator mode n = 10 |
head roll mean [deg] |
head roll S.D. [deg] |
t-test (two-tail) p level |
|
|
15 deg bank |
left |
2.2 |
2.3 |
p = 0.74 |
|
right |
2.4 |
1.5 |
|
|
|
30 deg bank |
left |
4.9 |
4.3 |
p = 0.34 |
|
right |
4.1 |
3.7 |
|
|
|
45 deg bank |
left |
5.5 |
5.2 |
p = 0.56 |
|
right |
4.8 |
3.6 |
|
|
Absolute (sign-less) values of the head roll angle were
used for the mean head roll angle calculation. Although there was a bias in
Attitude Indicator (pilots rolled the aircraft shallower to the left by 5°
because of bias error in AI), there was no significant difference in the amount
of head roll by the direction of the aircraft roll (turn).
Table 11.-Comparison of
head roll angle between left and right aircraft roll, visual flight mode.
visual mode
|
n = 10 |
head roll mean [deg] |
head roll S.D. [deg] |
t-test (two-tail) p level |
|
|
15 deg bank |
left |
4.9 |
3.9 |
* left > right p = 0.040 |
|
right |
2.0 |
2.0 |
|
|
|
30 deg bank |
left |
6.8 |
4.4 |
* left > right p = 0.0074 |
|
right |
3.3 |
2.0 |
|
|
|
45 deg bank |
left |
9.0 |
5.8 |
* left > right p = 0.016 |
|
right |
4.8 |
4.8 |
|
|
Absolute (sign-less) values of the head roll angle were
used for the mean head roll angle calculation. The amount of head roll was significantly greater in left
turns, by 2.9° (4.9° - 2.0°) at 15° bank; by 3.5° (6.8° - 3.3°) at 30° bank; by
4.2° (9.0° - 4.8°) at 45° bank.
The variance of head roll seems to be greater in left aircraft roll than
in right roll, although not to be tested statistically.
Table 12.-Comparison of
head roll angle between left and right aircraft roll, non-flying flight mode.
non-flying mode
|
n = 10 |
head roll mean [deg] |
head roll S.D. [deg] |
t-test (two-tail) p level |
|
|
15 deg bank |
left |
3.0 |
2.7 |
p = 0.45 |
|
right |
3.8 |
4.3 |
|
|
|
30 deg bank |
left |
6.0 |
3.8 |
p = 0.91 |
|
right |
5.8 |
5.8 |
|
|
|
45 deg bank |
left |
|||