AIAA 2009-6323

AIAA Atmospheric Flight Mechanics Conference

10 - 13 August 2009, Chicago, Illinois

Flight Dynamic Requirements for UAVs - Do They Really

Exist?
Richard Colgren1
University of Kansas, Lawrence, KS 66045
Lance Holly2
Viking Aerospace, Lawrence, KS 66049

There exist decades of experience in developing and documenting flight dynamic and
handling qualities requirements for manned aircraft. Several requirements documents with
associated user guides have been generated and are in general use. Such documents only
apply to UAVs in a general and limited sense. Attempts have been made to generate similar
documents for UAVs. These either apply to a particular platform, specify requirements only
based on sensor performance or mission maneuvers, or impose inappropriate constraints on
the vehicles flight dynamics. Requirements imposed on the operator are very different
depending on if the pilot is flying in traditional R/C mode, flying from a simulator,
operating the vehicle in an augmented mode, or acting as a systems supervisor under
completely autonomous operations. Further, previous training on one system or in another
mode of flight operation can adversely prepare the pilot or operator for problems when
flying in another mode or on a different platform. Examples of differences between flight
dynamics requirements and handling qualities, and between different operating modes, are
also discussed within this paper.

Nomenclature
C-H scale
FCS
n
p
q
r
R/C
UAV
V

=
=
=
=
=
=
=
=
=
=
=
=
=

Cooper-Harper scale
Flight Control System
Load Factor
Roll Rate (rad/sec)
Pitch Rate (rad/sec)
Yaw Rate (rad/sec)
Remote Control
Unmanned Air Vehicle
Velocity
Pitch Attitude (rad/sec)
Roll Attitude (rad/sec)
Heading (rad/sec)
Deflection (%)

I. Introduction

here exist several decades of experience in developing flight dynamic and handling qualities requirements for
manned aircraft, both fixed wing and rotorcraft. Several requirements documents with associated user guides
have been generated and are in general use. A key document for use in describing and evaluating flying qualities of
manned fixed wing aircraft is Ref. 1. Several excellent supporting documents are given in its list of references. An
equivalent document for use in describing and evaluating the flying qualities of rotary wing aircraft is given as Ref.
1

Associate Professor, Aerospace Engineering Department, 2120D Learned Hall, Lawrence, KS 66045, AIAA
Associate Fellow.
2
Managing Partner, Avionics and Flight Control Systems, Viking Aerospace, 100 Riverfront Road, Suite B,
Lawrence, KS 66049.
1
American Institute of Aeronautics and Astronautics
Copyright 2009 by Richard Colgren and Lance Holly. Published by the American Institute of Aeronautics and Astronautics, Inc., with permission.

2. Multiple useful supporting documents are also given in its list of references. Both in the case of fixed wing
aircraft and in the case of rotorcraft the C-H scale and pilot evaluation are the gold standard for the evaluation of
flying qualities. Differences are seen between the tasks used for evaluating fixed wing aircraft versus rotorcraft. For
example, coupling is allowed between axes on rotorcraft to a degree not allowed on fixed wing vehicles. As Ref. 1
and Ref. 2 were developed specifically for manned aircraft, such documents only apply to UAVs in a general and
limited sense. Attempts have been made to generate similar documents for UAVs. Some of these apply only to a
particular platform. These were developed by a particular manufacture along with their customer to support their
aerial platform. They also provide some guidance on developing flying qualities based on their specific experiences
for use on other aerial vehicle programs. Other documents specify requirements only based on sensor performance
or on mission maneuvers, or impose inappropriate constraints on the vehicles flight dynamics. Requirements
imposed on the operator are very different depending on if the pilot is flying in traditional R/C mode, flying from a
simulator, operating the vehicle in an augmented mode, or acting as a systems supervisor with the vehicle flying
under completely autonomous operations. Previous training while operating one system can adversely prepare the
pilot or operator for problems encountered while flying a different aerial platform. Further, previous training while
operating in one mode can adversely prepare the pilot or operator for problems encountered while flying in another
mode. Examples of the differences between flight dynamics requirements and handling qualities requirements, and
between the different operating flight modes, are also discussed within this paper.

II. Flying Qualities and their Assessment

In evaluating the handling qualities of piloted aerial vehicles great reliance is made on the Cooper-Harper (C-H)
scale3 and on pilot subjective assessment of their ability (including the required workload) to accomplish specific
tasks.1, 2 A wide variety of closed-loop high-gain (high-bandwidth) tasks, many requiring precise tracking, have been
developed for the evaluation of aircraft flying qualities. The recommended tasks for the evaluation of these flying
qualities are based on the Flight Phase Category. The achieved performance is stated in terms of Levels. The
familiar C-H scale used is shown in Figure 13. It is necessary to use equivalent definitions between the CooperHarper scale shown in Figure 13 and the Level definitions. Level 1 is Satisfactory, Level 2 is Acceptable, and Level
3 is Controllable. Typically, a Cooper-Harper pilot rating of 1 to 3 defines Level 1, a C-H rating from 4 through 6
defines Level 2, and a C-H rating from 7 through 9 defines Level 3. Note that Level 3 is not necessarily defined as
safe. This is consistent with the C-H scale. Cooper-Harper ratings of 8 and 9 are defined as cases where
controllability is in question. In a few cases the boundaries are further modified for specific considerations. The
bounds are sensitive to pilot workload. Examples of recommended tasks for evaluating flying qualities for Flight
Phase Category A include air-to-air tracking, air-to-ground tracking, aerial refueling, close formation flying, and
captures. No recommended tasks are given for Flight Phase Category B because this category consists of low-gain
(low-bandwidth) tasks. Recommended tasks for the evaluation of flying qualities for Flight Phase Category C are
again tracking tasks, including close formation flying, precision landings (with and without vertical and lateral
offsets), takeoffs, and captures. Note that, in many cases, the queues the authors of Ref. 1 and Ref. 2 expected to be
experienced by the pilot are not experienced by the UAV pilots or operators.
There are no recommended tasks for Flight Phase Category B because this Flight Phase Category generally
consists of low-gain (low-bandwidth) tasks. It is expected that flying qualities problems encountered while operating
within this Flight Phase Category will normally be exposed in the more demanding tasks used for evaluating Flight
Phase Categories A and C, or by normal operations during the flight test program. Thus, for piloted air vehicles,
special tasks for this Flight Phase Category are normally not considered necessary. However, Flight Phase Category
B tasks are normally of much longer duration than the tasks in the other Categories. It can also be argued that UAVs
will spend a greater percent of their flight time in the Flight Phase Category B than piloted aircraft do. Often the
UAV pilot or operator is not as tightly within the control loop as is the pilot of the manned aircraft. The UAV pilot
also does not get the variety of queues that the manned aircraft pilot does. This causes concerns on the
appropriateness and timeliness of the pilots reaction. Pilot fatigue may become a significant factor in certain
mission critical Category B tasks, even in the case of the UAV operator. In this case an evaluation of this kind of
task might be required.
Flight envelopes are described by a speed (V) and load factor (n) graph as shown in Figure 21. It is used to define
the Operational and Service Envelopes. Also, it represents the conditions where the requirements apply at a given
altitude. However, load factor is rarely a flying qualities queue for UAV operations. In all cases the load factor
denotes air vehicle maneuverability without regard to the available thrust. Note that flying qualities specifications
2
American Institute of Aeronautics and Astronautics

place no requirements on load-factor capability in constant-speed level flight. These flight envelopes are defined at
various altitudes corresponding to the Flight Phases.1, 2

Figure 1. Cooper-Harper Flying Qualities Scale

Figure 2. Flight Envelope for Fixed Wing Aircraft

3
American Institute of Aeronautics and Astronautics

Proof of compliance when accomplishing these demonstration tasks consists of pilot comments and C-H ratings.
For Level 1, pilot comments must indicate satisfaction with the aircrafts flying qualities. This includes no worse
than mildly unpleasant deficiencies. The median C-H ratings must be no worse than 3.5 in calm air or in light
atmospheric disturbances. Note again that many of the queues used in the evaluation of piloted aircraft are different
or not even available to the UAV pilot or operator. For Level 2, pilot comments must indicate that whatever
deficiencies may exist, aircraft flying qualities are still acceptable. Median C-H ratings must be no worse than 6.5 in
calm air or light atmospheric disturbances. For Level 3, pilot comments must indicate that the aircraft is at least
controllable despite the deficiencies in flying qualities. The median C-H ratings must be no worse than 9.5 in calm
air or light atmospheric disturbances. In moderate to severe atmospheric disturbances pilot comments and C-H
ratings must comply with requirements on the relationship between Levels and qualitative degrees of suitability. The
degrees of suitability are defined in Table 1.
Table 1. Suitability of Flying Qualities
Satisfactory

Flying qualities clearly adequate for the mission Flight Phase. Desired performance is
achievable with no more than minimal pilot compensation.

Acceptable

Flying qualities adequate to accomplish the mission Flight Phase, but some increase in
pilot workload or degradation in mission effectiveness, or both, exist.

Controllable

Flying qualities such that the aircraft can be controlled in the context of the mission
Flight Phase, even though pilot workload is excessive or mission effectiveness is
inadequate, or both. The pilot can transition from Category A Flight Phase tasks to
Category B or C Flight Phases, and Category B and C Flight Phase tasks can be
completed.

Actual task performance is not recommended for use as proof of compliance because it is even more subject to
pilot variability than pilot comments and C-H ratings. The performance objectives suggested in the evaluation tasks
are developed for use with the C-H scale, which was not developed with UAVs in mind. Specific definitions of
desired and adequate performance objectives attempt to reduce pilot variability by insuring that all of the evaluation
pilots attempt to achieve the same level of performance. In these performance objectives, adequate performance is
set at a level sufficient to successfully perform similar tasks within the operational service. Desired performance is
set at a more demanding level to insure that system deficiencies are exposed. Although task performance is not
recommended as proof of compliance, task performance should be recorded and analyzed by the flight test engineers
to insure that pilot ratings are reasonably consistent with the level of performance achieved and that all pilots are, to
a close approximation, achieving the same level of performance.
The evaluation of aircraft flying qualities is basically a subjective science, and human variability makes analysis
of the results a difficult proposition. The decreased number of queues and the increased delays experienced by the
UAV pilot or operator experiencing these queues makes this an even more difficult proposition. Nevertheless, there
are steps that can be taken to reduce the variability in the results. It is absolutely necessary that multiple evaluations
be conducted at each test condition. Studies of C-H rating variability have indicated that three pilots is the minimum
number of pilots for an adequate evaluation.4, 5 However, Ref. 5 further demonstrated that the point of diminishing
returns was reached after including about six pilots within an evaluation. Therefore, the recommended number of
pilots per test condition is three to six. Careful selection of the evaluation pilots is extremely important in reducing
the variability in the results.
In order to insure that all of the pilots attempt to achieve the same level of performance, it is extremely important
to explicitly define the desired and adequate levels of task performance. This insures consistency and reduces the
effects of pilot variability.1 Best results are achieved with task performance defined in terms of quantifiable
objectives which the pilot can readily observe in real time. This can pose a great difficulty when evaluating UAV
flying qualities, where the pilot or operator can be spatially and temporally removed from immediate motion
queues.6, 7 Furthermore, consideration must be given to defining objectives that can be adequately recorded so that a
flight test engineer can confirm that pilot ratings are reasonably consistent with task performance. Some UAV
systems, especially those for smaller vehicles, can provide few if any recorded flight parameters. Defining
4
American Institute of Aeronautics and Astronautics

quantifiable and recordable task performance objectives and setting appropriate levels of desired and adequate
performance is the most difficult part of planning flying qualities evaluations.
A method used to reduce the effects of variability is the long-look evaluation technique. In this method, the
pilot continues or repeats the task until becoming confident in the evaluation and assigning a C-H rating. The longlook approach allows the pilot a more extensive appraisal of the test condition. It eliminates the effects of unique
events encountered during a single run. It allows the pilot to get over the learning curve. The problem is that in
repeating the task the pilot or operator can get overly familiar with the task and give it an improved C-H. The
concern is that with decreased task queues under UAV operations this improvement in C-H rating (with familiarity)
is magnified. The normal evaluation procedure is to specify a minimum number of runs to be performed before a
rating can be given, and then allow the pilot to make additional runs as necessary.
Pilot comments should be considered the most important data. A C-H rating is only a summary of observed
flying qualities characteristics reduced into a single number. Pilot comments identity the specific deficiencies, if
any, that must be corrected. However, these comments are flavored by the interface with the UAV. Furthermore,
the long-look technique filters the effects of deficiencies on the C-H rating because, over several runs, the pilot
learns to compensate for some deficiencies. When pilot comments are given for every run, the comments will
hopefully identity all observed deficiencies, even those which can be compensated for. Comments on succeeding
runs provide guidance on the pilot's ability to compensate for the deficiencies. The final C-H rating should indicate
the relative significance of these deficiencies. Therefore, pilot comments must be recorded and analyzed for every
test run.
Time and cost constraints prohibit piloted evaluation of every task in every possible aircraft configuration at
every possible point within the flight envelope. The conditions that must be evaluated are the most common
operating conditions, operating conditions critical to the mission of the aircraft, and the worst case conditions. They
emphasize those where the quantitative, open-loop flying qualities requirements are violated. As most air vehicles
have multiple flight control modes, all mode transitions should be evaluated at common, mission critical, and worst
case conditions. Of specific concern are mode switches which are done automatically. Furthermore, the degradation
due to atmospheric disturbances should be demonstrated by evaluation within the different disturbance levels.
Evaluation of the effect of severe atmospheric disturbances may be performed in ground simulation. When using a
simulation to predict the degradation of flying qualities due to severe atmospheric disturbances, it is necessary to
correlate C-H ratings gathered from the simulation sessions in light to moderate turbulence with C-H ratings
obtained from flight test in light to moderate turbulence for the same task. Simulations for UAVs operating under
remote control often provide poor visual queues, making the evaluations less certain.
Any closed-loop task, performed aggressively, offers the possibility of being used to evaluate an air vehicle's
handling qualities and PIO characteristics.1 When developing a specification for a particular program, Ref. 1 and
Ref. 2 provide guidance in constructing these tasks. However, the appropriateness of these tasks and the queues that
the pilots and operators use to evaluate the vehicles performance at accomplishing these tasks can be significantly
deferent for UAVs. In fact, they may differ significantly between UAV programs. This due to the various pilot to
UAV interfaces, the variety of UAV mission profiles, and the different levels of autonomy. These differences, along
with the greater uncertainty with respect to appropriate evaluation queues, have prevented the development and use
of general flying qualities guidelines as have been developed for piloted air vehicles, as demonstrated by Ref. 1 and
Ref. 2.

III. UAV Flight Testing Examples

As a published example of UAV flying qualities analysis, Ref. 7 documents the flight test of a half-scale radiocontrolled model of the Pioneer unmanned air vehicle (UAV). In this testing, static longitudinal and lateraldirectional stability-and-control characteristics were measured. The air vehicle was instrumented to measure controlsurface deflections, angle of attack, sideslip angle, and airspeed. Telemetry was used to downlink the data to a
ground recorder. A wind-tunnel test of a 0.4-scale Pioneer vehicle model at full-scale Reynolds numbers was also
carried out. Finally, a CFD numerical analysis was conducted using a low-order panel method. The flight-test
determination of the neutral point was used to insure that the static margin was sufficient to conduct tests throughout
the flight envelope. Crosswind limits were determined from steady-sideslip maneuvers. The flight data was shown to
compare very favorably with other data sources. This study was interesting in the use of a variety of different scaled
5
American Institute of Aeronautics and Astronautics

vehicles and test articles. Other examples of the use of scaled UAVs for the prediction of the flight dynamics of
piloted vehicles also exist and are of interest.6
The flight testing of multiple fixed wing and rotary wing UAVs have been conducted by the authors. The
following portion of this section provides a couple recent examples of issues that can arise with such flying qualities
assessments. This includes Case 1 where confusion over which operating mode the UAV is in and previous training
when operating under different modes causes difficulties. It also includes Case 2 where adverse training due to
flying other vehicles and previous operations under fully manually modes causes problems. These cases pertain to a
small 50 V electric UAV, the Aggressor II, shown in Figure 3.

Figure 3. Aggressor II Rotorcraft UAV

Figures 4 through 6 are plots of flight data taken from the Aggressor II UAV helicopters flight control system
(FCS). Figure 4 gives the Euler angle attitudes for the air vehicle in radians. Time is the number of seconds into the
flight. Figure 5 gives the body fixed Euler axis angular rates in rad/sec. Figure 6 gives the pilot inputs as a fraction
of the maximum, with the maximum in one direction plotted as +1 and the maximum in the opposing direction as -1.
The pilot is not a test pilot. The pilot is a seasoned R/C helicopter pilot who is very capable of flying this aircraft in a
purely manual mode, with direct control over the actuators. In this particular case, the ground control station
operator has caused the autopilot and its augmentation system to be disabled. The ground control station operator
then tells the pilot that the autopilot is engaged and requests that the pilot perform a takeoff in Assisted Mode. The
pilot does not confirm that the Assisted Mode is engaged and thus does not recognize which mode of operation the
UAV is in. The three flight control system modes are shown in Figure 7. Assisted Mode, the middle level of
automation, works like a three dimensional cruise control. The Assisted Mode display on the laptop ground control
display is shown in Figure 8. For operations in the Assisted Mode, the pilot has been trained to provide velocity
6
American Institute of Aeronautics and Astronautics

commands to the FCS, while the FCS is to perform stability augmentation, attitude control, and reference velocity
tracking. Without the autopilot engaged the vehicle remains in a purely manual R/C mode, without stability
augmentation and with the commands fed directly to the control servos. The result is that the aircraft does not
respond to the pilots commands in the way that the pilot expects. Even though the pilot should be able to easily
control this helicopter in this mode, because he does not adapt quickly enough to the actual behavior of the
helicopter the vehicle goes into a divergent oscillation.

Figure 4. Helicopter Attitude - Autopilot Off

Figure 5. Helicopter Angular Rates- Autopilot Off

7
American Institute of Aeronautics and Astronautics

Figure 6. Pilot Commands- Autopilot Off

Figure 7. Autopilot Modes

8
American Institute of Aeronautics and Astronautics

Figure 8. Assisted Mode

The second Aggressor II flight example (Case 2) shows a portion of a flight in the Augmented Mode. In this
mode the R/C controller is operating as a velocity controller. The pilot is not a test pilot, but is an R/C helicopter
pilot who is capable of flying this aircraft in a purely manual mode. The pilot is commanding velocity, but is used to
using visual attitude feedback to generate R/C commands. Using attitude queues to generate velocity commands
creates inappropriate pilot actions which might even lead to a PIO. From visual flight observation and from
inspection of the Euler angle time trace plots, the operator reports an approximately 0.5 Hz oscillation. Upon further
examination of the flight data, it is seen that these oscillations are actually commanded responses based upon the
velocity commands given by the external pilot. The first command starts at about 410 seconds. The pilot sees the
attitude change and pulls back on the command from 412 to 413 seconds. The pilot then repeats this process,
commanding velocity from 413 to 415 seconds then opposing this command after the 415 seconds point based on
the increase in vehicle attitude. Note that the Aggressor II is a flybar-less helicopter UAV. Because this vehicle has a
nearly rigid rotor system with no flybar, it requires a larger change in pitch attitude angle to accelerate and jerk a
given amount than does a helicopter UAV with a Bell-Hiller flybar. In this case, the R/C trained pilot, who is used to
visual control based on vehicle attitudes, tends to pay more attention to the attitude changes than the velocity
changes. When in visual range helicopter UAVs change attitude more quickly and in a more noticeable way than
velocity changes can be recognized. Because of this, pilots tend to let attitude changes effect the velocity command
they give to the helicopter. This is more apparent in transition training for UAV helicopters without a flybar than in
transition training for UAV helicopters with a flybar.
Two FCS modifications could be used to correct the oscillation seen in this case. One possibility is that the
acceleration and jerk limiters could be reduced for the case of external pilot reference commands. This option is
disliked by many pilots as this can make the pilot feel disconnected from the vehicle. It effectively introducing more
lag into the system. The advantage to this approach is that the pilot, once familiar with the delay and vehicle
response, only has to learn one mode or feel for the UAV. A second option is to introduce a dual rate switch on
the external pilot controller which allows the pilot to select between high and low speed settings. The effect is that
the pilot has effectively no tendency to PIO the system when in the low setting. The pilot has finer control over the
speed reference command and any oscillations due to a pilot command will be much more apparent. The high setting
can then be used only for fast forward flight in which the pilot command is often a more discrete command, such
as a step increase in speed. The low setting is used again when finely positioning the vehicle. Such switching does
need to take into account flying qualities during transitions, and pilot mode awareness.1
9
American Institute of Aeronautics and Astronautics

Figure 9. Helicopter Velocities Autopilot in Assisted Mode

Figure 10. Helicopter Attitude - Autopilot in Assisted Mode

IV. Summary and Conclusions

There exist years of experience in developing and documenting flight dynamic and handling qualities
requirements for manned aircraft. Several requirements documents with associated user guides are in general use.
However, the appropriateness of some of these piloted vehicle tasks and their associated motion queues for use in
UAV requirements is questionable. The tasks and queues that the pilots and operators use to evaluate a UAVs
performance can be significantly different for other UAVs. Such piloted vehicle requirements documents only apply
to UAVs in a general and limited sense. Any closed-loop task, performed aggressively, does offer the possibility of
being used to evaluate an air vehicle's handling qualities and its PIO tendencies. In fact, the tasks and requirements
may differ significantly between UAV programs. This due to the various pilot to UAV interfaces, the variety of
UAV mission profiles, and the different levels of autonomy. These differences, along with the greater uncertainty
with respect to appropriate evaluation queues, have prevented the development and use of general UAV flying
qualities guidelines. Further, previous training on one system or within another mode of operation can adversely
prepare the pilot/operator for problems when flying in another mode or on a different platform. Examples of the
differences between flight dynamics requirements and handling qualities, and between these different operating
modes, were covered within this paper.
10
American Institute of Aeronautics and Astronautics

References
1

anon, Flying Qualities of Piloted Aircraft, MIL-STD-1797A, 30 Jan. 1990.

2
Baskett, Barry J., et. al., Handling Qualities Requirements for Military Rotorcraft, ADS-33E, US Army Aviation
Directorate, Redstone Arsenal, AL, 21 Mar. 2000.
3
Cooper, G. E. and Harper, R. P., Jr., The Use of Pilot Rating in the Evaluation of Aircraft Handling Qualities, NASA-TND-5153, Apr. 1969.
4
Kidd, E. A. and Bull, G. Handling Qualities Requirements as Influenced by Prior Evaluation Time and Sample Size,
Cornell Aero Lab TB-1444-F-I, Feb. 1963.
5
Dukes, Theodore A., Guidelines for Designing Flying Qualities Experiments, NADC-85130-60, Jun. 1985.
6
Foster, Tyler and Bowman, Jerry, Dynamic Stability and Handling Qualities of Small Unmanned Aerial Vehicles, AIAA2005-1023, 43rd AIAA Aerospace Sciences Meeting and Exhibit, Reno, Nevada, 10-13 Jan. 2005.
7
R. M., Howard, R. M., Bray, D. F., Lyons, Flying-qualities Analysis of an Unmanned Air Vehicle, Journal of Aircraft,
Vol. 33, No. 2, pp. 331-336, Mar.-Apr. 1996.

11
American Institute of Aeronautics and Astronautics