A Methodology for Detecting Pilot Fatigue and Head Tilt Measured Using
Flight Simulation of an Unmanned Aerial Vehicle
by
Judith Ann Bailey
A thesis submitted to the Graduate Faculty of
Auburn University
in partial ful llment of the
requirements for the Degree of
Master of Science
Auburn, Alabama
May 5, 2013
Approved by
John E. Cochran Jr., Department Head and Professor of Aerospace Engineering
Harold Zallen, Distinguished Visiting Research Professor of Aerospace Engineering
Brian S. Thurow, W. Allen and Martha Reed Associate Professor of Aerospace Engineering
George Flowers, Dean of the Graduate School and Professor of Mechanical Engineering
Abstract
The objective of this research project was to determine whether the human factors of
fatigue and head tilt can be detected in real time by measuring the head motion of pilots  ying
simulated unmanned aircraft missions. Test subjects  ew a set of maneuvers using a  xed-
base  ight simulator programmed with a General Atomics MQ-1 Predator UAV simulation,
while an infrared tracking system tracked the six-degree-of-freedom motion of their heads.
Test subjects completed three sessions of data collection. The data collected was date, time,
translational motions of x, y, and z, and rotational motions of yaw, pitch, and roll. The
collected data was then tabulated and plots of the six-degrees-of-freedom versus time were
generated for analysis. Visual observations of the subjects? head motion during the session
were compared to the plotted data. Analysis of the data collected supports the conclusion
that motions in translational y and rotational pitch are the most promising indicators of
fatigue. On the other hand, rotational motions consisting of coupled yaw and roll appear
to be the best indicators of head tilt. Due to the limited amount of data, these conclusions
are preliminary. However, the methodology presented for observing evidence of phenomena,
such as pilot fatigue and head tilt, show considerable promise.
ii
Acknowledgments
The author would like to thank Dr. Harold Zallen for his continued support during the
last four years not only in this research endeavor but in the excitement of life and lessons
that you should never back down and never give up. His expertise and advice has been the
cornerstone of this research and the evolution of the author into a better researcher.
The author would also like to thank committee members Dr. John E. Cochran, Jr. and
Dr. Brian S. Thurow. The author is grateful for Dr. Cochran?s assistance throughout the
research study, and appreciative of his steady focus that has helped the author stay on track.
Dr. Thurow has been a constant mentor throughout the author?s college career. His support
and expectation of the author has helped her grow into a better leader and a better engineer.
Finally, the author would like to thank her parents, John and Melissa Bailey, and her
 ance, Ryan Sherrill, for their continued love and support throughout her college career.
May the Force Be With You.
iii
Table of Contents
Abstract . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . ii
Acknowledgments . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . iii
List of Figures . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . vi
1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1
2 Review of Literature . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4
2.1 Fatigue Detection . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4
2.2 Head Tilt . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5
2.3 Unmanned Aerial Vehicles . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6
2.3.1 General Atomics Predator . . . . . . . . . . . . . . . . . . . . . . . . 7
2.3.2 UAV Human Factors . . . . . . . . . . . . . . . . . . . . . . . . . . . 7
3 Instrumentation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11
3.1 Computer One . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11
3.2 Computer Two . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13
4 Test Subject Flight Details . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 16
4.1 Test Subject Characteristics . . . . . . . . . . . . . . . . . . . . . . . . . . . 17
4.2 Test Subject Calibration . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 17
4.3 Assignment of Tasks . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 18
4.3.1 Simulator Sessions . . . . . . . . . . . . . . . . . . . . . . . . . . . . 19
4.3.2 Modi cations to Microsoft Flight Simulator X . . . . . . . . . . . . . 19
4.4 Data Acquisition . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 20
5 Results and Discussion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 22
5.1 Fatigue . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 22
5.2 Head Tilt . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 26
iv
5.3 Anecdotal Comment . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 26
6 Conclusions and Recommendations . . . . . . . . . . . . . . . . . . . . . . . . . 35
6.1 Further Areas of Study . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 36
6.2 Modi cations to Instrumentation . . . . . . . . . . . . . . . . . . . . . . . . 37
Bibliography . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 38
A Institutional Review Board Protocol, 12-082 EP1203 . . . . . . . . . . . . . . . 42
B Flight Simulator Ensemble . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 45
B.1 MGI Fixed-Base Flight Simulator . . . . . . . . . . . . . . . . . . . . . . . . 45
B.2 Embedded TrackIR Capture System . . . . . . . . . . . . . . . . . . . . . . . 45
C Modi cation of Predator Simulation Instrument Panel . . . . . . . . . . . . . . 48
D Butterworth Filter Code . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 51
v
List of Figures
1.1 The Six Degrees of Freedom[8] . . . . . . . . . . . . . . . . . . . . . . . . . . . 3
2.1 General Atomics MQ-1 Predator[31] . . . . . . . . . . . . . . . . . . . . . . . . 8
2.2 General Atomics RQ-1 Predator[31] . . . . . . . . . . . . . . . . . . . . . . . . . 9
3.1 MGI Fixed-Base Flight Simulator with Infrared Tracking Technology . . . . . . 12
3.2 TrackIR4 and TrackClip[8] . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 14
3.3 Visual of the TrackIR Capture Software[42] . . . . . . . . . . . . . . . . . . . . 14
3.4 TrackIR4 Reading Position and Orientation via TrackClip[8] . . . . . . . . . . . 15
4.1 Cen-Tech Self-Leveling Laser Level Setup . . . . . . . . . . . . . . . . . . . . . 18
4.2 Sample of the Data . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 21
5.1 Y and Pitch Degrees of Freedom[8] . . . . . . . . . . . . . . . . . . . . . . . . . 23
5.2 Flight Session 2, First Complete Standard Landing Pattern for Subject 1 . . . . 24
5.3 Flight Session 2, Final Complete Standard Landing Pattern for Subject 1 . . . . 24
5.4 Flight Session 3, First Complete Standard Landing Pattern for Subject 1 . . . . 25
5.5 Flight Session 3, Final Complete Standard Landing Pattern for Subject 1 . . . . 25
5.6 Flight Session 2, First Complete Standard Landing Pattern for Subject 2 . . . . 27
vi
5.7 Flight Session 2, Final Complete Standard Landing Pattern for Subject 2 . . . . 27
5.8 Flight Session 3, First Complete Standard Landing Pattern for Subject 2 . . . . 28
5.9 Flight Session 3, Final Complete Standard Landing Pattern for Subject 2 . . . . 28
5.10 Yaw and Roll Degrees of Freedom[8] . . . . . . . . . . . . . . . . . . . . . . . . 29
5.11 Flight Session 1, Maneuvers for Subject 3 . . . . . . . . . . . . . . . . . . . . . 29
5.12 Flight Session 2, Maneuvers for Subject 3 . . . . . . . . . . . . . . . . . . . . . 30
5.13 Flight Session 3, Maneuvers for Subject 3 . . . . . . . . . . . . . . . . . . . . . 30
5.14 Flight Session 1, Maneuvers for Subject 4 . . . . . . . . . . . . . . . . . . . . . 31
5.15 Flight Session 2, Maneuvers for Subject 4 . . . . . . . . . . . . . . . . . . . . . 31
5.16 Flight Session 3, Maneuvers for Subject 4 . . . . . . . . . . . . . . . . . . . . . 32
5.17 Flight Session 1, Maneuvers for Subject 5 . . . . . . . . . . . . . . . . . . . . . 33
5.18 Flight Session 2, First Complete Standard Landing Pattern for Subject 5 . . . . 33
5.19 Flight Session 2, Final Complete Standard Landing Pattern for Subject 5 . . . . 34
A.1 IRB Stamped Flyer Approved for Recruitment . . . . . . . . . . . . . . . . . . . 43
A.2 IRB Stamped Handout Approved for Recruitment . . . . . . . . . . . . . . . . . 44
B.1 Original Predator Ground Control Station[49] . . . . . . . . . . . . . . . . . . . 46
B.2 Revised Predator Ground Control Station[50] . . . . . . . . . . . . . . . . . . . 46
C.1 Original Predator Simulation Instrument Panel . . . . . . . . . . . . . . . . . . 49
vii
C.2 Modi ed Predator Simulation Instrument Panel . . . . . . . . . . . . . . . . . . 50
D.1 Data With No Filter . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 52
D.2 Data With Butterworth Filter . . . . . . . . . . . . . . . . . . . . . . . . . . . . 52
viii
Chapter 1
Introduction
With the increased use of unmanned aerial vehicles (UAVs) in the U.S. military and
other federal and state agencies, i.e. Border Patrol, Treasury, Agriculture, Commerce, et
al., as well as the upcoming expectancy of their use in commercial airspace, human factors
a ecting UAV pilots has become an important topic of discussion. Unlike the traditional
aircraft pilot, a UAV pilot is  ying the aircraft from a base station on the ground or in another
vehicle miles away from the location of the actual aircraft. This change in environment is
a new way of operating for the traditional pilot. Generally, the view from the UAV?s nose
camera is portrayed on the screen along with the instrumentation of a conventional aircraft.
However, much of the intuitive feel a pilot gains from sitting in an aircraft during  ight is lost
due to the lack of vestibular and haptic cueing[1]. This is a similar scenario that occurs when
a pilot is \ ying" a  xed-base  ight simulator. As a result of this similarity, it is postulated
that methods for detecting human factors a ecting pilots during a simulator session can be
applied to pilots  ying a UAV where action can be taken to prevent an accident in real time.
By using an infrared head tracking system, the primary interest of this study is two human
factors associated with head motion: fatigue and head tilt.
The  rst human factor of interest, fatigue, is prevalent in UAV  ight and believed to
be the cause of multiple crashes upon landing[2]. Fatigue is de ned by the Federal Aviation
Administration (FAA) as, \a condition characterized by increased discomfort with lessened
capacity for work, reduced e ciency of accomplishment, loss of power or capacity to respond
to stimulation, and is usually accompanied by a feeling of weariness and tiredness[3]." A UAV
pilot?s work schedule di ers markedly from that of a traditional pilot, and awareness of this
challenge is increasing as F-15 pilots are transferred to UAV Training and Operations[4].
1
Studies have also shown that continuous viewing of computer screens puts stress on the eyes
of the operator and cause the operator to become fatigued[1]. These factors coupled with
the stress of  ying a multi-million dollar aircraft that cannot be seen or touched during a
mission, makes a UAV pilot?s situation unique.
Because of its association with head motion, the second human factor of interest is head
tilt. Head tilt is a medical condition similar to torticollis, a twisting of the neck with the
head tipped to one side and the chin to the other[5]. Also, the term head tilt is hypothesized
to be an optokinetic re ex, a condition of the inner ear that causes a subject to misjudge
the direction of line of sight from the subject to a point or the orientation of an observed
object[6]. Thus, motion of the head during  ight may be related to head tilt through subjects
trying to compensate for these conditions. Currently in aviation medicine, there is no formal
medical description for the head motion associated with head tilt, but previous research has
shown that it is indicated by a yawing and rolling motion of the head[7].
In this work, the objective was to determine the head motions indicative of fatigue and
head tilt. Infrared tracking technology was used as the method for data collection to measure
the six-degree-of-freedom (6DOF) motion of the test subject?s head, see Figure 1.1. Since
this information could provide useful for unmanned aviation systems operators, ground based
operations of the General Atomics MQ-1 Predator UAV were simulated. The simulation was
 own by volunteer test subjects who held at least a Federal Aviation Administration Private
Pilot Certi cate. The Master Set of maneuvers  own by the test subjects represent a typical
UAV mission along with particular maneuvers that would allow further observation of fatigue
and head tilt. Three sessions for each test subject were conducted to provide replicate data.
In Chapter 2, background information is provided on past studies of detection and measure-
ment of fatigue and head tilt as well as information on UAVs and their associated human
factors. In Chapter 3, the instrumentation used in the study is presented, while Chapter 4
presents the protocol used during the simulator sessions. In Chapter 5, the results of the
study are summarized, and in Chapter 6, conclusions are discussed and recommendations
2
are made for future study. The human subject protocol for this research was approved by
the Auburn University Institutional Review Board (12-082 EP1203), see Appendix A.
Figure 1.1: The Six Degrees of Freedom[8]
3
Chapter 2
Review of Literature
According to Wiegmann and Shappell[9], \pilot fatigue is di cult to observe directly,
much less quantify." As a result, the aeromedical perspective in their \Human Factors Anal-
ysis and Classi cation System[9]" de nes pilot and  ight crew fatigue as a cause of aircraft
accidents not just a contributing factor. The FAA agrees with this view and describes fatigue
as a result of many factors pertaining to sleep including: the amount of recent sleep a crew
member has received; the number of continuous hours awake; a crew member?s accumulated
sleep debt and work load; the time of day of the crew member?s  ight; and the number of
time zones to be crossed[10]. The FAA has developed speci c criteria for airline pilots and
crew members in order to reduce the risk of fatigue occurring during aircraft operations. A
summary of the factors associated with pilot fatigue is o ered by Caldwell[11], and some of
the methods used to study these factors and determine the FAA?s fatigue prevention crite-
ria are provided by Rosekind[12], Van Dongen[13], Frakes[14], and David[15]. The FAA is
focused on using this information to prevent fatigue indirectly rather than trying to detect
fatigue and provide a remedy in real time before an accident occurs.
2.1 Fatigue Detection
In medicine, fatigue is de ned as \a non-pathologic state resulting in decreased abil-
ity to maintain function or workload due to mental or physical stress[16]." Currently, de-
tecting fatigue in commercial truck drivers in the transportation industry is an ongoing
activity[17][18][19][20]. There are many driver fatigue detection techniques that can be di-
vided into two broad categories: techniques that monitor the driver directly and techniques
that monitor the driver?s operation of the vehicle[21][22]. Techniques that monitor the driver
4
directly include intrusive measuring of brain waves, heart rate, and/or pulse. However, intru-
sive measures require sensors to be attached to the driver that cause discomfort. Therefore,
more research is focusing on non-intrusive measures such as: eye tracking, eyelid movement,
gaze tracking, head movement, and facial movement. Techniques that monitor the vehi-
cle include measuring the vehicle?s speed, lateral position, turning angle, and/or moving
course. The most established driver fatigue detection technique used is the monitoring of
eye movement.
Preliminary research using head tracking to detect fatigue in commercial truck drivers
has been conducted by Xie[23], Smith[24], and Ji[25]. The most recent study by Ji proposes
that nodding of the head indicates the onset of fatigue. By using an established eye tracking
technique, the frequency of nods a subject experiences over time can establish a given fatigue
level. This technique uses the distance between the eyes and the eyes? locations as inputs
to a Kalman  lter algorithm in order to predict where the face is located in the image. A
conclusion drawn from this research is that the given algorithm could estimate the orientation
of the head and resulting number of nods accurately. However, a criterion for indicating levels
of fatigue that correspond to the number of nods experienced has not been established.
2.2 Head Tilt
In 1960, Greenberg[36] published a short article theorizing that there was a correlation
between eye dominance and head tilt. In the 1990s, research was performed by Dr. Fred
H. Previc at Brooks Air Force Base to determine this relationship between eye dominance
and head tilt. He sought to prove Greenberg?s hypothesis that vertical misalignment of
the eyes are caused by a lateral head tilt resulting from ocular dominance. Previc?s research
focused on measuring a subject?s sight dominance and vertical misalignment of the eyes then
comparing it to two-dimensional photographic measurements of the subject?s head tilt. His
study demonstrated that sighting dominance is related to ocular misalignment produced by
lateral head tilt and therefore predictive of head tilt in most subjects[37][38]. Since Previc?s
5
research in the mid 90?s, most head tilt research has focused on determining the cause of the
medical condition known as torticollis, which is associated with a twisting of the neck to one
side accompanied by painful muscle spasms.
In 2008, the precursor to this research study was conducted to determine motions of the
head that indicate head tilt using  ight simulation. After an initial round of testing to  nd
quali ed subjects without vestibular issues, non-pilots viewed a  ight simulation and pilots
viewed, then  ew the same simulation. Data of the 6DOF head motion was collected via
infrared tracking and analyzed to graphically demonstrate the head tilt observed during the
 ight simulation testing. As a result, Zallen et al.[7] determined head tilt to be indicated
by not only a two-dimensional lateral rotation of the head, as shown by Previc[37], but also
by a three-dimensional yawing rotation of the head. Another result of this study was the
preliminary  nding that fatigue was indicated by a translational y motion of the head. As
a result, the current research study was established to determine further information on the
indicators of fatigue and to give further conclusion to the  ndings that head tilt is indicated
by rolling and yawing motions of the head.
2.3 Unmanned Aerial Vehicles
With the invention of the modern  ight simulator,  ight simulation was used extensively
in the aviation industry to train pilots and  ight crew along with assisting in aircraft design
and development as well as human factor and ergonomics issues. Besides the precursor to
this study[7], other head-tracking research using  ight simulation has focused on spatial
awareness during  ight. Pilots have to interact with a complex system through a human-
machine interface that must be monitored frequently and systematically. By monitoring a
trainee?s eye and head motion during a simulation exercise, an instructor can give feedback
on the level of the trainee?s awareness and where he needs to improve his monitoring[26].
Currently, the concept of  ight simulation has been remodeled into a ground control station
for remotely  ying UAVs.
6
2.3.1 General Atomics Predator
The General Atomics Predator has two designations, MQ-1 and RQ-1. The General
Atomics MQ-1 Predator, shown in Figure 2.1, is weaponized to conduct o ensive operations,
irregular warfare, and high value target and individual prosecution. Typically, a laser-
guided AGM-114 Hell re missile will be mounted under each wing, however, armament varies
upon the nature of the target and could include a Singer air-to-air missile, Brilliant Anti-
Tank (BAT) ammunition, or the Small Diameter Bomb (SDB). The General Atomics RQ-1
Predator, shown in Figure 2.2, is not weaponized, but it is tasked with long endurance,
medium altitude surveillance and urban reconnaissance missions. It is equipped with a
Northrop Grumman AN/ZPQ-1 Tactical Endurance Synthetic Aperture Radar (TESAR)
and a Wescam Versatron 14TS Infra Red/Electro-Optical (IR/EO) sensor turret[27]. The
Predator has a Rotax 914F four cylinder engine with 115 horsepower. It has a wingspan of
55 feet (16.8 meters), length of 27 feet (8.22 meters), and height of 6.9 feet (2.1 meters).
The Predator has an empty weight of 1,130 pounds (512 kilograms) with a maximum takeo 
weight of 2,250 pounds (1,020 kilograms) corresponding to 665 pounds (100 gallons) of fuel
and 450 pounds (204 kilograms) of payload. Cruise speed is around 84 mph (70 knots) with
speeds up to 135 mph (117 knots), range of 770 miles (675 nautical miles), and a ceiling
of 50,000 feet (15,240 meters)[28]. The Predator unit cost was approximately $4 million in
 scal year 2010 dollars[29]. However, the Predator is part of an Unmanned Aircraft System
(UAS) that consists of four sensor/weapon equipped aircraft, a ground control station (GCS),
a Predator Primary Satellite Link (PPSL), and spare equipment along with operations and
maintenance crews for deployed 24-hour operations, which cost approximately $20 million
in  scal year 2009 dollars[30].
2.3.2 UAV Human Factors
With the increased use of UAVs, attention has been turned to the human factors associ-
ated with being a UAV pilot[32][33][34]. There are two main human factors issues associated
7
Figure
2.1:
General
Atomics
MQ-1
Predator[
31
]
8
Figure
2.2:
General
Atomics
RQ
-1
Predator[
31
]
9
with UAVs that can lead to fatigue and be ampli ed by a pilot?s head tilt, ergonomic inter-
face issues and control issues. The ergonomics design of the operation station is the main
point of complaint from UAV pilots including the heads-up-display (HUD), seat design, and
control layout. Two medical issues attributed to bad ergonomic design that UAV pilots
experience are lumbar kyphosis and computer vision syndrome (CVS). Lumbar kyphosis is a
problem that occurs at the musculoskeletal level when one sits for an extended period of time
and the lumbar region of the back  attens out or bends outward which results in discomfort
and a lack of productivity[1]. CVS is de ned by the American Optometric Association as a
\complex of eye and vision problems related to near work which are experienced during or
related to computer use[35]." Speci c problems with the original Predator setup included the
HUDs, one was at eye-level and one was mounted higher up, which forced the pilot to crane
his neck in order to view the screen; and the joysticks, which were elevated above the arm
rest of the chair and required the pilot to lift his arms constantly in order to maneuver the
aircraft. The HUD also used red graphics on a blue background resulting in further stress on
the pilot?s eyes. Although the Predator setup has been improved to a three monitor setup
in a horizontal con guration at eye level with one joystick and keyboard at arm level with
an easier to read HUD, ergonomic issues still play a role in the human factors of Predator
pilots.
Besides the ergonomic interface, the loss of control due to a mediated visual  eld with
no vestibular and haptic cueing result in a stressful environment for the a UAV pilot. The
loss of vestibular and haptic cueing can cause delay in the recognition of an engine failure,
ice accumulation, or approach to stall, which can result in an accident. The mediated  eld
of view related by the UAV?s nose camera can hamper the pilot?s control of the aircraft.
This is especially true during landing where peripheral cues are valuable, and is a potential
variable in many UAV crashes that occur upon landing[2].
10
Chapter 3
Instrumentation
A complete system setup of the study is shown in Figure 3.1. The MGI Fixed-Base
Flight Simulator is operated using Computer One. It has a three screen monitor setup
for a better viewing experience, and it is equipped with Precision Pro pedals and yoke for
increased control of the aircraft simulation. The infrared tracking receiver, used to measure
the head motion of the subjects, can be seen mounted on the top of the middle monitor of
the  ight simulator?s three screen con guration. The infrared tracking receiver is plugged
into the rear of Computer Two and operated separately from the  ight simulator software
to prevent an overload of Computer One?s system.
3.1 Computer One
Computer One operates the hardware and software needed to support the MGI Fixed-
Base Flight Simulator, see Appendix B. It is a standard Dell Optiplex 740 Desktop with 4 GB
of 1.9GHz DDR2 DRAM with an ATI Radeon 6500 graphics card. The other components
of Computer One related to the study are as follows:
1. CH Flight Simulator Yoke (Model 99914) [39]
2. CH Precision Pro Pedals (Model 88785) [39]
3. Three, 19" Dell LCD Monitors (Model EI98WFPv)
4. Matrox TripleHead2Go (Model KFE5S204) [40]
5. Microsoft Windows XP Professional Version 2002 Service Pack 3 Operating System
11
Figure
3.1:
MGI
Fixed-Base
Fligh
tSim
ulator
with
Infrared
Trac
king
Te
chnology
12
6. Microsoft Flight Simulator X: Gold Edition
7. First Class Simulation?s \UAV Predator" Add-on [41]
First Class Simulation?s \UAV Predator" software uses the MQ-1 version of the General
Atomics Predator. It is equipped with a Hell re missile under each wing. The missiles cannot
be  red during the simulation which requires the test subjects to take o and land with a
missile under each wing. This is a scenario seen by UAV Predator pilots when a mission is
aborted and returned to base.
3.2 Computer Two
Computer Two is a standard Hewlett Packard Compaq 8200 Elite Convertible Minitower
PC with 16 GB of PC3-10600 Memory operating Microsoft Windows XP Professional Version
2002 Service Pack 3 with a NVIDEA NVS 300 PCIe graphics card. This computer operates
the hardware and software needed to support the Embedded Track IR Capture System[42],
see Appendix B. The TrackIR4[8], shown in Figure 3.2, is an infrared receiver, originally
designed for use with gaming software, that tracks the 6DOF motion of the head. It tracks
a head-mounted TrackClip[8], also shown in Figure 3.2, within a  eld of view of 46 degrees,
a sample rate of 120 frames per second, a raw sensor resolution of 358 x 288, a sub-pixel
resolution of 1/20th pixel, reporting resolution of 7100 x 5760, and a horizontal resolution
of 154 sub-pixels per degree.
The TrackIR4 is mounted atop the center of the three LCD monitor con guration of
Computer One and connected to a USB 2.0 port on the rear of Computer Two. Data is
recorded on Computer Two using the TrackIR Capture software, shown in Figure 3.3, as
the test subject  ies the UAV Predator simulation on Computer One. The triangular shape
of the three re ective phosphor markers on the TrackClip is key to having vectors for the
TrackIR4 to measure the di erence in position and orientation over time via infrared signals,
see Figure 3.4. The distances read by the TrackIR4 to form this vector are the three sides
13
of the triangle formed by the phosphor markings along with the distance from the top most
marker to the plane of the bottom two markers.
Figure 3.2: TrackIR4 and TrackClip[8]
Figure 3.3: Visual of the TrackIR Capture Software[42]
14
Figure 3.4: TrackIR4 Reading Position and Orientation via TrackClip[8]
15
Chapter 4
Test Subject Flight Details
The Master Flight containing eighteen maneuvers was performed by Kay[43], a Federal
Aviation Administration Certi ed Flight Instructor. The Master Flight has a duration of 76
minutes and 24 seconds. The Master Set of eighteen maneuvers in the Master Flight to be
performed by each test subject follows:
1. Take o and ascend to 20,000 feet.
2. Initiate a standard rate turn to the left for 180 degrees of heading change.
3. Level  ight at 20,000 feet for 7 minutes.
4. Descend to 3,000 feet and level o .
5. Initiate a standard rate turn to the right for 180 degrees of heading change.
6. Ascend to approximately 4,000 feet.
7. Descend back to 3,000 feet.
8. Complete a 360 degree level turn to the right at 40 degrees of bank.
9. Complete a 360 degree level turn to the left at 40 degrees of bank.
10. Complete a 360 degree ascending turn to the right at 35 degrees of bank.
11. Complete a 360 degree ascending turn to the left at 35 degrees of bank.
12. Complete a 360 degree descending turn to the right at 35 degrees of bank.
13. Complete a 360 degree descending turn to the left at 35 degrees of bank.
16
14. Begin a standard landing pattern and land the aircraft.
15. Taxi down the runway for  ve seconds and then take o .
16. Climb to approximately 1,800 feet.
17. Repeat steps 14, 15, and 16, until  ve landings have been completed.
18. After the  fth landing, the simulation is complete.
4.1 Test Subject Characteristics
Test subjects were recruited from the student population at Auburn University as well
as from the general population in the surrounding area. It is noted for the study, that the
student/lesser experienced pilots have more current simulator experience than the older/more
experienced pilots recruited from the general population. There were a total of thirteen test
subjects who  ew for the study. Nine of the test subjects were between the ages of 20-27
and the other four between the ages of 55-70. All test subjects were male as a result of a
lack of interest in the study from the female pilots recruited.
4.2 Test Subject Calibration
Each test subject?s head was zeroed before a  ight simulator session began. This pro-
cedure was accomplished with a horizontal and vertical Cen-Tech Self-Leveling Laser Level
Item 92703[44] located immediately behind the  ight simulator seat. The laser has maxi-
mum power of 5 mW, Class IIIA Protection, and has a wavelength of 650 nm +/- 10 nm.
Calibration establishes standardized tracking between the TrackIR4 and the TrackClip by
performing the  ve steps that follow:
1. The participant is seated at the  ight simulator with the Cen-Tech Laser Level mounted
37.5 inches to the rear of the seat. The Cen-Tech Laser Level is calibrated using the
blue LED light located dead center on the middle monitor of the  ight simulator.
17
2. The calibrated laser controlled with a level and rotating bezel is focused on the test
subject?s position to establish zero motion, as shown in Figure 4.1.
3. On Computer Two, the TrackIR Capture Program shows a Control Menu with com-
mands and recording slots for 6DOF.
4. All recording slots are reset to zero at the beginning of every session.
5. The TrackIR4 receiver has three colored status LED?s: red - no signal is received and
paused; orange - signal received and paused; green - signal received and running. In
order to begin recording, the green LED must be illuminated to receive and record all
data.
Figure 4.1: Cen-Tech Self-Leveling Laser Level Setup
4.3 Assignment of Tasks
The test subjects meet for three sessions. The duration of the  rst session is two hours,
and the duration of the last two sessions is one and a half hours. All three sessions require
18
the test subjects to  y the Master Set of maneuvers described previously. However, the test
subject?s orientation to the  ight simulator in the  rst session requires a slightly di erent
process than the  nal two sessions.
4.3.1 Simulator Sessions
At the beginning of the  rst session, each test subject is shown a sheet listing the Master
Set of maneuvers they will be asked to perform. Next, they are seated at the simulator and
calibrated, as described in Section 4.2. The  rst task is viewing a 15 minute and 53 second
video splice of the Master Flight that includes: initial take o , one ascending 360 degree turn,
one descending 360 degree turn, and the  nal two landings. This allows the test subject to
become familiar with the Predator simulation interface without having to worry about  ying
the aircraft. After the video, the test subject can ask questions about the setup and then
practice  ying the UAV Predator simulation until they are con dent in their handling of
the aircraft. Once con dent, the test subject will  y the Master Set of maneuvers with
data collection. The  nal two sessions simply consist of practice time and the  ying of the
Master Set of maneuvers with data collection. Temperatures at the MGI Flight Simulator
were recorded for every session with an infrared precision digital thermometer. The range of
temperatures during the session was 79:1 F (26:2 C) - 85:6 F (29:8 C) for all test subjects.
4.3.2 Modi cations to Microsoft Flight Simulator X
After six sessions, it became apparent that there was an issue with the interface between
the test subject and the simulation. Analysis of the data showed that yawing to the right
was occurring with a higher frequency than assumed normal. It was determined that the
original placement of the instrument panel in the simulation was not easily visible because
of its small size and location at the bottom of the right most screen in the three monitor
setup. As a result, the panel was enlarged and moved to the center[45], see Appendix C.
Another change made during the study resulted from the observation that the manner in
19
which some test subjects were landing the aircraft would have resulted in a crash in the real
world. However, the simulation was not detecting and depicting a crash. Subsequently, the
crash function in Microsoft Flight Simulator X was enabled to increase the reality of the
simulation.
4.4 Data Acquisition
After the test subject has been zeroed, the test examiner will simultaneously start the
TrackIR Capture Program on Computer Two and a stopwatch, then instruct the test subject
to start the engines and begin the  rst maneuver. The test subject will indicate when that
maneuver is complete, and the test examiner will mark the time on the stopwatch and
instruct the test subject to begin the next maneuver until the test subject has completed
the entire Master Set of maneuvers. The Embedded TrackIR Capture System converts the
test subject?s head motion collected by the TrackIR4 into: the date, time in one-hundredth
of a second, translational parameters of 6DOF x, y, and z in one-thousandth of a millimeter,
and rotational parameters of 6DOF yaw, pitch, and roll in one-thousandth of a degree. A
 nal representation of the data is shown in Figure 4.2. Shakedown runs were performed to
establish a procedure that would be safe and viable for the human test subjects.
20
Figure 4.2: Sample of the Data
21
Chapter 5
Results and Discussion
The data for thirteen test subjects was collected and tabulated using Microsoft Excel
2010. A Butterworth  lter, see Appendix D, was used to remove noise from the data and
two-dimensional plots were constructed for analysis. Ten of the test subjects  ew all three
sessions, one subject  ew two sessions, and two subjects  ew one session. Because of com-
pleteness, only the data of the ten subjects who  ew all three sessions was analyzed. Based
on the observations of the test examiner, the author of this paper, the  nal sequence of
landings were evaluated for fatigue by looking at the y and pitch motions of the head, while
the four 360 degree ascending and descending turns were evaluated for head tilt by looking
at the yaw and roll motions of the head. Because of the small population of subjects, the
other degree-of-freedom head motions could not be determined relevant to fatigue or head
tilt.
5.1 Fatigue
The most obvious place to search for evidence of fatigue in the data was in the sequence
of landings at the end of the Master Set of maneuvers. During the landings, the test observer
noted the test subjects changing their posture in the  ight chair, which resulted in a change
in the y and pitch motions of the head, see Figure 5.1. The test observer interpreted these
shifts as visual signs of fatigue. There were two di erent scenarios of shifts noted by the test
observer that were evenly distributed among the test subjects. The  rst scenario included a
vertical upright shift with high frequency pitching oscillations of the head, while the second
scenario included a downward shift of the head accompanied by fewer pitching oscillations.
22
To study the data obtained by the TrackIR4 for these two scenarios, the y and pitch degrees-
of-freedom were plotted.
Figure 5.1: Y and Pitch Degrees of Freedom[8]
In Figures 5.2 - 5.5, the y and pitch degrees-of-freedom for the  rst and  nal complete
standard landing patterns of a test subject, denoted as Subject 1, are plotted for the  nal
two sessions, respectively, as a representation of the  rst scenario. As Subject 1 settled into
 ying the landing pattern, he began to recline in the  ight chair. As a result, his head moved
down, which is represented by the decrease in the y motion of the head. Also during this
time, a high frequency oscillation in the pitching motion of the head is occurring, and is
associated with the subject continually glancing from the simulation view to the instrument
panel in order to keep the aircraft on course. Finally, as Subject 1 begins his turn from base
leg to  nal approach, he begins to straighten back up in the  ight chair, which correlates
with an increase in the y motion of the head as he readies himself for landing the aircraft.
In Figures 5.6 - 5.9, the y and pitch degrees-of-freedom for the  rst and  nal complete
standard landing patterns of a test subject, denoted as Subject 2, are plotted for the  nal
two sessions, respectively, as a representation of the second scenario. As Subject 2 settles
into  ying the landing pattern, he begins to lean toward the computer screen. As a result,
23
Figure 5.2: Flight Session 2, First Complete Standard Landing Pattern for Subject 1
Figure 5.3: Flight Session 2, Final Complete Standard Landing Pattern for Subject 1
24
Figure 5.4: Flight Session 3, First Complete Standard Landing Pattern for Subject 1
Figure 5.5: Flight Session 3, Final Complete Standard Landing Pattern for Subject 1
25
his head moves down, which is represented by the decrease in the y motion of the head. The
pitching motion of the head is less frequent than seen in scenario one, because Subject 2 is
moving his head up and down to look between the the simulation screen and the instrument
panel, which is represented by the higher amplitude y motion.
5.2 Head Tilt
After observing a tilting of the head during the four 360 degree ascending and descending
turns, analysis of these maneuvers revealed head tilt results similar to Zallen et al.[7], and
provide additional evidence that yawing and rolling motions of the head are indicators of
head tilt, see Figure 5.10. In Figures 5.11 - 5.13, the yaw and roll degrees-of-freedom for
the 360 degree ascending and descending turns of all three sessions of a subject, noted
as Subject 3, are depicted, respectively. An interesting observation is that the yaw and
roll motion become more pronounced during portions of the two descending turns than in
the two ascending turns. This correlates with the higher degree of di culty when  ying a
descending turn.
Another interesting observation is the symmetry between yaw and roll. The TrackIR
Capture Program measures yaw to the right and the left as positive and negative angles,
respectively, while roll is measured to the right and left as negative and positive angles,
respectively. This coupling is similar to the motion of an aircraft as described by the subject
of  ight dynamics. In the lateral dynamics of an airplane, a roll rotation produces both a
yawing and rolling moment while subsequently a yaw displacement produces a yawing and
rolling moment[47]. These observations are consistent as shown in Figures 5.14 - 5.16 that
depict the head motion of another test subject, denoted as Subject 4.
5.3 Anecdotal Comment
There were two distinct groups of test subjects as described earlier, younger/lesser
experienced pilots and older/more experienced pilots. The older test subjects had more
26
Figure 5.6: Flight Session 2, First Complete Standard Landing Pattern for Subject 2
Figure 5.7: Flight Session 2, Final Complete Standard Landing Pattern for Subject 2
27
Figure 5.8: Flight Session 3, First Complete Standard Landing Pattern for Subject 2
Figure 5.9: Flight Session 3, Final Complete Standard Landing Pattern for Subject 2
28
Figure 5.10: Yaw and Roll Degrees of Freedom[8]
Figure 5.11: Flight Session 1, Maneuvers for Subject 3
29
Figure 5.12: Flight Session 2, Maneuvers for Subject 3
Figure 5.13: Flight Session 3, Maneuvers for Subject 3
30
Figure 5.14: Flight Session 1, Maneuvers for Subject 4
Figure 5.15: Flight Session 2, Maneuvers for Subject 4
31
Figure 5.16: Flight Session 3, Maneuvers for Subject 4
trouble adjusting to the  ight simulator than the younger test subjects even though they
had an average of more than 5000 hours and 40 years of experience compared to the younger
test subjects. This can be attributed to the familiarity the younger test subjects have with
simulators compared to the older test subjects. The patterns indicating fatigue and head
tilt described above which represent  ights of the younger test subjects also represent the
patterns of the older test subjects. The di erence can be seen in the increased frequency of
motion exhibited that corresponds with the uncertainty visibly shown and expressed by the
older test subjects. Figures 5.17 - 5.19, denoted as  own by Subject 5, are representative of
an older test subject?s maneuvers and landings, respectively.
32
Figure 5.17: Flight Session 1, Maneuvers for Subject 5
Figure 5.18: Flight Session 2, First Complete Standard Landing Pattern for Subject 5
33
Figure 5.19: Flight Session 2, Final Complete Standard Landing Pattern for Subject 5
34
Chapter 6
Conclusions and Recommendations
The limited data generated during this study indicate that fatigue is evidenced by the y
and pitching motions of the head, while head tilt is evidenced by a coupled yawing and rolling
motion of the head. Although the remaining degrees-of-freedom for each human factor were
analyzed, no correlation with the test observations could be drawn. Also, as a result of the
small population of subjects, a statistical correlation could not be performed to determine
if the additional degrees-of-freedom indicated either fatigue or head tilt.
There is strong evidence that the y motion of the head is a good indicator of fatigue. This
agrees with the research done by Zallen, et al. Additional observation suggests that pitch is
also a motion of the head that indicates fatigue. Certainly, there is more to be determined.
It is recommended that additional research be done on y and pitch to determine subtle
di erences with a statistical number of test subjects. For head tilt, this research supports
conclusions similar to Previc?s research in the 1990s[37] that head tilt is associated with a
lateral rolling motion of the head. It also supports the conclusion similar to those based on
the research done by Zallen, et al.[7], that head tilt can be indicated by not just a lateral
rolling motion of the head but also by a three-dimensional yawing motion of the head, with
the addition that yaw and roll are coupled.
In order to determine de nite indicators of fatigue and head tilt, a statistical number
of test subjects, no less than 30 recommended for results with a statistical error margin less
than 10 percent, need to be evaluated in order to perform correlation statistics between the
6DOF. The sessions need to last a su cient length of time so that there is no doubt the test
subjects are exhibiting fatigue at the end of the session so subsequent analysis can determine
indicators. By also utilizing the motion of the aircraft in conjunction with the subject?s head
35
motion, a fourier analysis can be used to determine the di erence in the subject?s motion
relative to the motion of the aircraft and the subject?s motion relative to the physical e ects
on the body. Finally, instead of looking at the subject?s head motion over time from take-o 
to landing, analysis of the head motion from landing back through the landing pattern in
terms of distance  own could provide a speci c location in the pattern where head anomalies
more frequently occur. This area of interest can be isolated in order to look more speci cally
at the human factors associated with UAV crashes occurring at landing.
It is the author?s conclusion that every pilot will have a unique pattern of head motion
for aircraft maneuvers  own in a simulator-like environment. As a result, the methods used
in this study are recommended as a basis for current application to the monitoring of pilot
head motion during  ight in a simulator-like environment. The pilot?s speci c pattern of
head motion can then serve as a baseline to evaluate the pilot?s head motion in real time
during missions. If anomalies are present, replacing the pilot with a fresher backup pilot can
help avert potential accidents. The pilot can then be evaluated to determine the cause of
the anomaly and preventive measures can be taken to reduce the risk of a reoccurrence.
6.1 Further Areas of Study
Note in Figures 5.11 and 5.13 how the  rst maneuver, an ascending turn to the right for
360 degrees, corresponds with a yawing motion to the right and rolling motion to the left.
However in Figure 5.12, the  rst maneuver corresponds with a yawing motion to the left
and rolling motion to the right. This variation in the direction of Subject 3?s head motion
during the same maneuver is a conundrum that warrants further study. A similar scenario
can be seen in Figures 5.14 - 5.16 with Subject 4.
Another observation includes the e ect of enabling the crash option in Microsoft Flight
Simulator X. Although as a result of this only one test subject crashed and the simulation
had to be restarted, the angle of approach when landing was observed by the test examiner
to be the main cause of the crash. This was also similar to the cases when test subjects did
36
not crash but had hard landings that caused the aircraft to \bounce" down the runway until
they could regain control. This is believed to be an e ect of the loss of control discussed
in Section 2.2.2 and is recommended for future study as another reason for multiple crashes
during landing of UAVs.
6.2 Modi cations to Instrumentation
Two modi cations to the MGI Flight Simulator are recommended before further study
is conducted. During the simulation sessions, some shifts, resulting from fatigue, were so
dramatic that the  ight chair was pushed away from the simulator. Also, the majority of
the test subjects noted that they would have liked to move the chair closer to or further
away from the simulator in order to make it easier to reach the pedals, which would better
accommodate their speci c height during the sessions. In order to prevent these dramatic
shifts and increase the ergonomics of the design, the  oor of the simulator could be built
out where the seat can be bolted to a track that allows the seat to be adjusted forward or
backward to account for the varying heights of the test subjects. Also, the majority of test
subjects noted that the instrument panel for the Predator UAV simulation di ers signi cantly
from a standard aircraft instrument panel. This required a higher degree of attention from
the test subject on the instrument panel during the simulation than normally would have
been required if the simulation instrument panel had been formatted like a standard aircraft
instrument panel. Therefore, it is recommended that the Predator instrument panel be
modi ed to resemble that of a standard aircraft instrument panel.
37
Bibliography
[1] Salvendy, G., Handbook of Human Factors and Ergnomics, John Wiley & Sons Inc.,
Section 61, 2012, 4th Ed.
[2] Drone Crash Database, Drone Wars UK, http://dronewarsuk.wordpress.com/drone-
crash-database/, January 2, 2013.
[3] Salazar, G.J., \Fatigue in Aviation," FAA Civil Aerospace Medical Institute, Publica-
tion OK-07-193, Oklahoma City, Oklahoma, USA.
[4] Etheridge, T., Personal Communication, Florida Institue for Machine and Human Cog-
nition, Pensacola, FL, USA, December 2011.
[5] Torticollis, A.D.A.M. Medical Encyclopedia, http://www.ncbi.nlm.nih.gov/pubmedhealth/
PMH0001757/, December 19, 2012.
[6] Pettorossi, V.E., et al., \Head Position Modulates Optokinetic Nystagmus," Experi-
mental Brain Research, Volume 213, Issue 1, 2011.
[7] Zallen, H., Cochran, J.E., and Bailey, J.A., \Head-Tilt and Pilot Fatigue Measured by
Flight Simulation," Aircraft Engineering and Aerospace Technology: An International
Journal, Volume 84, Issue 1, 2012.
[8] Natural Point, Inc., TrackIR4 and TrackClip, Corvallis, Oregon, USA, 2013.
[9] Wiegmann, D.A., and Shappell, S.A., A Human Error Approach to Aviation Accident
Analysis: The Human Factors Analysis and Classi cation System, Ashgate Publishing
Company, Chapter 1, 2003.
[10] Federal Aviation Administration, \Flight Member Duty and Rest Requirements,"
November 18, 2011.
[11] Caldwell, J.A., \The Impact of Fatigue in Air Medical and Other Types of Operations: A
Review of Fatigue Facts and Potential Countermeasures," Air Medical Journal, Volume
20:1, October 2000.
[12] Rosekind, M.R., et al., \Pilot Fatigue, Sleep, and Circadian Rhythms: NASA Fatigue
Countermeasures Program," FAA Aviation Safety Journal, pgs. 20-25, 1993.
[13] Van Dongen, H.P., Dinges, D.F., \Circadian Rhythms in Fatigue, Alertness, and Per-
formance," Principles and Practice of Sleep Medicine, Saunders, pgs. 391-399, 2000.
38
[14] Frakes, M.A., et al., \Sleep Debt and Outside Employment Patterns in Helicopter Air
Medical Sta Working 24-Hour Shifts," Air Medical Journal, Volume 26:1, January-
February 2007.
[15] David, M.C., et al., \Fatigue in Airline Pilots After an Additional Days Layover Period,"
Aviation, Space, and Environmental Medicine, Volume 81, Number 11, November 2010.
[16] Strauss, S., \Pilot Fatigue," www.pilotfriend.com/aeromed/medical/fatigue1.htm,
February 18, 2013.
[17] Lal, S.K.L. and Craig, A., \Electroencephalography Activity Associated with Driver Fa-
tigue: Implications for a Fatigue Countermeasure Device," Journal of Psychophysiology,
Volume 15, Number 3, 2001.
[18] Zhang, Z. and Zhang, J., \Driver Fatigue Detection Based Intelligent Vehicle Control,"
The 18th International Conference on Pattern Recognition, 2006, Hong Kong.
[19] Albu, A.B., et al., \A Computer Vision-Based System for Real-Time Detection of Sleep
Onset in Fatigued Drivers," 2008 IEEE Intelligent Vehicles Symposium, Eindhoven
University of Technology, Eindhoven, The Netherlands.
[20] Noor, H.A.M. and Ibrahim, R., \Image Processing Using Eyelid Blinking And Mouth
Yawning to Measure Humans Fatigue Level," Third Asia International Conference on
Modelling and Simulation, 2009, Bandung, Bali, Indonesia.
[21] Wang, Q., et al., \Driver Fatigue Detection: A Survey," Proceedings of the 6th World
Congress on Intelligent Control and Automation, 2006, Dalian, China.
[22] Coetzer, R.C. and Hancke, G.P., \Driver Fatigue Detection: A Survey," IEEE Africon
2009, Nairobi, Kenya.
[23] Xie, X., Sudhakar, R., and Zhuang, H., \A Cascaded Scheme for Eye Tracking and Head
Movement Compensation," IEEE Transactions on Systems, Man, and Cybernetics Part
A: Systems and Humans, Volume 28, Number 4, July 1998.
[24] Smith, P., et al., \Monitoring Head/Eye Motion for Driver Alertness with One Camera,"
15th International Conference on Pattern Recognition, Volume 4, pgs. 636-642, 2000,
Los Alamitos, California, US.
[25] Ji, Q., Zhu, Z., and Lan, P., \Real-Time Nonintrusive Monitoring and Prediction of
Driver Fatigue," IEEE Transactions on Vehicular Technology, Volume 53, Number 4,
July 2004.
[26] Anders, G., \Pilot?s Attention Allocation During Aproach and Landing Eye- and Head-
Tracking Research in an A330 Full Flight Simulator," Institute for Aeronautics and
Astronautics, Technische Universitat, Berlin, Germany.
[27] General Atomics Predator, http:spy ight.co.uk/Predator.htm, January 12, 2013.
39
[28] Unmanned Aerial Vehicles (UAV): Predator, Air Force HSI,
http://www.wpafb.af.mil/shared/media/document/AFD-090121-039.pdf, Novem-
ber 17, 2012.
[29] US Air Force FY 2011 Budget Estimates, US Air Force Aircraft Procurement,
http://www.sa m.hq.af.mil/shared/media/document/AFD-100128-072.pdf, February
22, 2013.
[30] MQ-1B Fact Sheet, Air Combat Command, Public A airs O ce, Langley Air Force
Base, Virginia, USA, 2012.
[31] General Atomics and A liated Companies, http://www.ga.com/image-
library/unmanned-aircraft-systems, March 22, 2013.
[32] Spravka, J.J., Moisio, D.A., and Payton, M.G., \Unmanned Air Vehicles: A New Age
in Human Factors Evaluations," No. AFFTC-PA-05069. Air Force Flight Test Center,
Edwards Air Force Base, California, 2003.
[33] Williams, K. W., \A Summary of Unmanned Aircraft Accident/Incident Data: Hu-
man Factors Implications," No. DOT/FAA/AM-04/24, Federal Aviation Administra-
tion, Civil Aeromedical Institute, 2004.
[34] Cooke, N.J., et al., Advances in Human Performance and Cognitive Engineering Re-
search, Volume 7: Human Factors of Remotely Operated Vehicles, Elsevier Ltd., Chap-
ter 2, 2006, 1st Ed.
[35] American Optometric Association, Computer Vision Syndrome (CVS),
http://www.aoa.org/x5374.xml, February 15, 2013.
[36] Greenberg, G., \Eye-Dominance and Head Tilt," American Journal of Psychology, Vol-
ume 73, pgs. 149-151, 1960.
[37] Previc, F.H., \The Relationship Between Eye Dominance and Head Tilt in Humans,"
Neuropsychologia, Issue 32, pgs. 1297-1303, 1994.
[38] Putnam, S., et al., \Direction and Rliability of Head Tilt in Humans," Laterality, Volume
1, Issue 2, pgs. 153-160, 1996.
[39] CH Products, LLC., CH Flight Simulator Yoke and Precision CH Pedals, Vista, Cali-
fornia, USA, 2010.
[40] Matrox Graphics Inc., Matrox TripleHead2Go, Dorval, Quebec, Canada, 2012.
[41] First Class Simulations, UAV Predator, Banbury, Oxforshire, UK, 2008.
[42] Zallen, H., Embedded TrackIR Capture System, Malone Group International, Auburn,
AL, USA, December 2007.
[43] Kay, M., Certi ed U.S. Federal Aviation Administration Flight Instructor, unpublished
communication, 2012.
40
[44] Harbor Freight Tools, Cen-Tech Self-Leveling Laser Level, Calabasas, California, USA,
2012.
[45] Howe, A., and Slaughter, S., A Fun Way to Enjoy FSX: Tips, Tricks, and Trips, Abacus,
pgs. 187-202, 2007.
[46] Arctic Tern Equipment, http://www.interstateaircraft.com/Equipment.html, February
27, 2013.
[47] Etkin, B. and Reid, L.D, Dynamics of Flight: Stability and Control, John Wiley & Sons
Inc., Chapter 3.8, 1996, 3rd Ed.
[48] AU Institutional Review Board for the Protection of Human Subjects in Research,
http://www.auburn.edu/research/vpr/ohs/, March 2, 2012.
[49] Air Force World, http://airforceworld.com/heli/eng/predator.htm, February 27, 2013.
[50] UAV Predator / Repear Target Destruction Ground Control Station Operations,
http://www.frequency.com/video/uav-predator-reaper-target-destruction/3851824/-
/5-1226454, February 26, 2012.
[51] NaturalPoint, OptiTrack API, Version: 1.82, For SDK Version: 1.1.035, Date
07/25/2008.
[52] Butterworth, S., \On the Theory of Filter Ampli ers," Wireless Engineer, Volume 7,
pgs. 536-541, 1930.
[53] Cochran, J.E., Department Head of Aerospace Engineering at Auburn University, per-
sonal communication, February 25, 2013.
41
Appendix A
Institutional Review Board Protocol, 12-082 EP1203
In order to do research involving human testing, an Institutional Review Board (IRB)
Protocol must be obtained. The IRB insures that the rights and welfare of human test sub-
jects are not violated by reviewing all research activities that involve human test subjects for
compliance with all applicable federal, state, local, and institutional regulations, guidelines,
and ethical research principles[48]. The Auburn University IRB Protocol, 12-082 EP1203,
for this study was approved on March 3, 2011 to allow human testing as described in Chapter
4. Test subjects were recruited from the student population of Auburn University and from
the general population in the surrounding area with the requirement that the test subject
have a Federal Aviation Administration Private Pilot Certi cate. The IRB approved  yer
and handout used for recruitment are shown in Figures A.1 and A.2
42
Figure A.1: IRB Stamped Flyer Approved for Recruitment
43
Figure A.2: IRB Stamped Handout Approved for Recruitment
44
Appendix B
Flight Simulator Ensemble
B.1 MGI Fixed-Base Flight Simulator
The MGI Fixed-Base Flight Simulator was designed by Malone Group International?s
Dr. Harold Zallen, President and CEO, in January 2008. The MGI Fixed-Base Flight
Simulator was fabricated at Baird Corp by James J. Baird, Jr., Architect and CEO, and Dr.
Harold Zallen. It was completed December 2008 and has been e ectively used for research. In
November 2011, the MGI Fixed-Base Flight Simulator was donated to the Auburn University
Samuel Ginn College of Engineering Department of Aerospace Engineering by Malone Group
International for permanent retention by the Department of Aerospace Engineering.
The MGI Fixed-Base Flight Simulator was designed to be similar to the original General
Atomics MQ-1 and RQ-1 Predator UAV ground control station at Creech Air Force Base,
Ironstone, Nevada, USA. Subsequent to the design of the MGI Fixed Base Flight Simulator,
the USAF changed from a one operator setup, see Figure B.1, to a three-member team with
separate LCD screen, a view of two of the members is shown in Figure B.2. The three-
member crew for the Predator consists of a pilot to control the aircraft, an enlisted aircrew
member to operate sensors and weapons, and a mission coordinator who assumes the role of
navigator[30].
B.2 Embedded TrackIR Capture System
The Embedded TrackIR Capture System is a composite of programs that converts the
test subject?s head motion into a text  le that can be analyzed with various programs[42].
45
Figure B.1: Original Predator Ground Control Station[49]
Figure B.2: Revised Predator Ground Control Station[50]
46
It includes the TrackIR Software, Optitrack API Software, and two subprograms, trackir-
win32dll.dll and ocx.exe, that allow the TrackIR Capture Program to operate. The TrackIR
Software, NaturalPoint Version 4.1.030.Final, allows the TrackIR4 to communicate with the
computer. The OptiTrack API Software[51] converts the analog signals of the TrackIR Soft-
ware to digital data. The trackirwin32dll.dll is a dynamic link library  le that is directly
embedded into the Microsoft Operating System. It works with the ocx.exe  le to link with
the Optitrack API Software and produce a text  le of the data for analysis.
47
Appendix C
Modi cation of Predator Simulation Instrument Panel
The instrument panel of the Predator UAV simulation was originally very small and lo-
cated at the bottom of the right most screen in the three monitor setup, shown in Figure C.1.
In order to make it easier for viewing, the panel code for the Predator was modi ed[45]. The
steps used to modify the panel code are as following:
1. Make sure Microsoft Flight Simulator X is shut down.
2. Place the mouse over the Microsoft Flight Simulator X icon, right click, and select the
Properties option.
3. Click the Find Target button on the General Tab.
4. Double-click the SimObject folder, then the Airplane folder, the FCS MQ-1 Predator UAV
folder, and  nally the Panel folder.
5. Right-click the panel.cfg  le and select the Open option.
6. Open the  le using the program Notepad.
7. Scroll down in the document to the section in brackets denoted [window00].
8. Modify the line in the code that starts with gauge00 by replacing the string of numbers,
\553,482,302,198," with the string of numbers, \325,400,450,375."
9. Modify the line in the code that starts with gauge01 by replacing the string of number,
\172,482,302,198," with the string of numbers, \0,482,302,198."
10. Save the  le and close all folders.
48
11. Start Microsoft Flight Simulator X and visually con rm that the change has been made
to the panel, see Figure C.2.
The  rst two numbers in the string of code denote the starting position of the upper-left
most corner of the instrument panel. By changing the  rst two numbers of gauge00 from
\553,482" to \325,400," the panel now begins 325 pixels to the right and 400 pixels down
from the upper-left corner of the screen opposed to its previous location of 553 pixels to the
right and 482 down from the upper-left corner of the screen. The  nal two numbers denote
the size of the panel. The panel denoted gauge00 starts o 302 pixels wide and 198 pixels
tall, but it is enlarged to 450 pixels wide and 375 pixels tall. This change resulted in easier
viewing of the instrument panel for all test subjects.
Figure C.1: Original Predator Simulation Instrument Panel
49
Figure C.2: Modi ed Predator Simulation Instrument Panel
50
Appendix D
Butterworth Filter Code
A Butterworth  lter is a signal processing  lter that removes certain frequencies of data
in order to reduce interfering signals and background noise resulting in analysis of purer
data[52]. After collecting and tabulating the 6DOF data in Microsoft Excel, a butterworth
 lter was used in Matlab Version R2010a to clear the data of noise so that the underly-
ing patterns associated with fatigue and head tilt could be easily viewed. A copy of the
Butterworth  lter code used[53] can be seen below.
function Y = butterworth(X,T,WCO,NP)
Y(1,1) = X(1,1);
Y(2,1) = X(2,1);
Y(NP,1) = X(NP,1);
Y(NP - 1,1) = X(NP - 1,1);
CFC = WCO;
wd = 2.*pi*CFC*2.0775;
wa = sin(wd*T/2)/cos(wd*T/2);
denom = 1. + sqrt(2.)*wa + wa^2;
a0 = wa^2/denom;
a1 = 2.*a0;
a2 = a0;
b1 = -2.*(wa^2 - 1)/denom;
b2 = (-1. + sqrt(2)*wa - wa^2)/denom;
for I = 3:1:NP-2
Y(I,1) = a0*X(I,1) + a1*X(I - 1,1) + a2*X(I - 2,1) + b1*Y(
I - 1,1) + b2*Y(I - 2,1);
end
for J = NP - 2:-1:3
Y(J,1) = a0*X(J,1) + a1*X(J + 1,1) + a2*X(J + 2,1) + b1*Y(
J + 1,1) + b2*Y(J + 2,1);
end
51
There are four inputs to the function  le: X, T, WCO, and NP. X is the array of data;
WCO is the frequency being  ltered; NP is the number of data points in X; and T is a
function of NP, T = 1/(NP-1).
In Figures D.1 and D.2, plots of the same data can be seen without the Butterworth
 lter and with the Butterworth  lter, respectively.
Figure D.1: Data With No Filter
Figure D.2: Data With Butterworth Filter
52