35

Flight Dynamics of a Flapping-Wing

Air Vehicle
Roman Y. Krashanitsa,1 Dmitro Silin,2 and Sergey V. Shkarayev3
University of Arizona, Tucson, Arizona, 85721, USA
Gregg Abate4
Air Force Research Laboratory, Munitions Directorate, Eglin AFB, FL 32542-6810, USA
ABSTRACT
The research and development efforts presented in this paper address the flight dynamics
of a flapping-wing air vehicle (ornithopter). The 74-cm-wing-span ornithopter was
equipped with an automatic flight control system that provides stability augmentation and
navigation of the vehicle and flight data acquisition. Wind tunnel tests were conducted
with the control surfaces fixed in the trimmed position and flapping motion of the wings
activated by a motor at a constant throttle setting. Coefficients of vertical and horizontal
force, and pitching moment were determined at a free stream velocity of 7.25 m/sec, and
the angle of the stroke plane varied from 0 to 40 degrees. A series of flight tests were
conducted with fixed controls, demonstrating ornithopter stability in all axes. Proportional
control laws were programmed into the autopilot for the closed-loop controls. A number
of flights of the autonomous ornithopter were conducted with the telemetry acquisition.
During the autonomous flights, the ornithopter performed waypoint and altitude
navigation, demonstrating stable performance.
Keywords: flapping, flight, ornithopter, dynamics, wind tunnel, stability, experiments.

NOMENCLATURE
sp
angle of stroke plane
2
CX
horizontal force coefficient C X = X / 0.5 V S

CY

2
vertical force coefficient CY = L / 0.5 V S

CM

2
pitching moment coefficient C M = M / 0.5 V Sc

C X , CY , C M
c
X
f
Y
M
S
TS
V

horizontal force, vertical force, and pitching moment coefficient derivatives with
respect to the angle of the stroke plane
wing mean geometric chord
horizontal component of aerodynamic force
flapping frequency
vertical component of aerodynamic force
pitching moment
wing planform area
throttle setting
free stream velocity

I. INTRODUCTION
In recent years, a number of research studies have been conducted on the important aspects of flapping
flight in nature and on the aerodynamic design of flapping-wing micro air vehicles (ornithopters). The
ornithopters generate lift and thrust for forward motion using their flapping wings, emulating birds and
1Research

Assistant Professor, Dept. of Aerospace and Mechanical Engineering, 1130 N. Mountain Ave.

2Graduate

Research Assistant, Dept. of Aerospace and Mechanical Engineering, 1130 N. Mountain Ave.

3Associate
4Team

Professor, Dept. of Aerospace and Mechanical Engineering, 1130 N. Mountain Ave., Member, AIAA

Leader, Airframe Dynamics & Robust Control, AFRL/RWGN, Associate Fellow AIAA

Volume 1 Number 1 2009

36

Flight Dynamics of a Flapping-Wing Air Vehicle

insects. Furthermore, a successful ornithopter design provides a verifiable physical model of flight in
nature. The evolution of birds and flying insects has led to physiques that continue to surpass
engineered designs of the same scale for important performance criteria such as maneuverability and
controllability. Biomimetic flight principles might have significant positive influences on the design of
flight control systems. However, just mimicking the flight of birds and insects is insufficient for
designing flapping-wing vehicles,1 and without accurate empirical data and a clear understanding of
flapping flight dynamics, further progress in this technology will be difficult.
Analogous to flight dynamics of conventional aircraft, flapping flight dynamics involves three main
topics: maneuver performance, stability of motion, and control mechanisms. This section provides a
review of the results of previous investigations of flapping flight dynamics including flight of birds,
insects, and manmade vehicles and devices.
Ellington2 summarized the flight kinematics and dynamics of insects. This overview provides a
useful framework for prospective insect-size micro air vehicle (MAV) designs. The motion of the
flapping wing is described with respect to a flapping plane, also called a stroke plane. Insects have been
observed to perform quick maneuvers by tilting the stroke plane of their wings. Lateral direction
changes can be accomplished by a roll of the stroke plane (often by the simultaneous increases of
flapping amplitude and the pitch angle of the outside wing). Tilting the stroke plane is similar to the use
of cyclic control to tilt the rotor disk of a helicopter in a particular direction, resulting in the helicopter
moving in that direction. For slower flight and hovering, the body hangs below the wing base, and the
insect benefits from a passive pendulum-like stability.
Taylor3 reviewed neurophysiology and biomechanics literature concerning the mechanisms of
flight stability and controls in insects and found that it has been difficult to provide a formal analysis
of insect flight controls. The main obstacle for such analysis is in obtaining accurate empirical data on
the aerodynamic forces and inertia properties in insects. Even analysis of the existence of passive
stability mechanisms remains a difficult problem. Insects use different mechanisms of controls and a
number of control inputs. Flapping frequency and amplitude are main control inputs for longitudinal
control in flies. On the other hand, the longitudinal flight control in locusts is more complicated and
includes flow interactions between fore- and hindwings affecting a force balance between them. Most
of the studies addressed longitudinal motion of insects, but there was a lack of empirical data for
lateral motion.
Taylor and Thomas4 provided a detailed study of the dynamic stability of the desert locust in
longitudinal motion. The study is based on the linearized equations of rigid-body motion, which is a
classical approach5 for stability of uncontrolled motions of an airplane. The application of this approach
to a flapping apparatus implies averaging through a flapping cycle of the position of the center of mass,
moments of inertia, and aerodynamic forces. Gyroscopic terms are assumed to be negligible. Another
issue with this approach is that in the animal closed-loop controls, it is impossible to isolate the
muscles inputs from the dynamics of the animals body. In the study,4 both aerodynamic stability and
control mechanisms are studied as one coupled phenomenon. Rate derivatives, such as pitching
moment derivatives with respect to the angle of attack and to the pitch rate, were not included in the
model. These derivatives could play an important role in damping out oscillatory modes in the locust.
The remaining aerodynamic derivatives were measured on locusts rigidly tethered to the force balance
in the wind tunnel. The eigenvalue analysis showed one stable oscillatory mode and two non-oscillatory
modes, one of which is unstable. In real life, locusts are stable in flight and, therefore, the obtained
unstable modes reflect aforementioned shortcomings and limitations of the approach.
Sun and Xiong6 examined the dynamic stability of a bumblebee in hovering flight. In order to
overcome difficulties in the aerodynamic measurements on tethered insects and in the determination of
aerodynamic derivatives, computational fluid dynamics simulations were used instead. Similar to the
previous study,4 the insect was modeled as a single rigid body and the equations of motion were
linearized about an equilibrium point. Eigenvalue and eigenvector analysis indicated an unstable
oscillatory mode caused by coupled horizontal motions and pitching. It was shown that pitching
moment derivatives are important for providing effective damping. It was noted that the period of
unstable oscillations is 50 times the wingbeat period. Thus, the insect has enough time to make
adjustments to the flapping wing kinematics and to alter the predicted unstable motion. Lack of
feedback controls and use of a linear, time-invariant dynamic model in the previous studies4,6 did not
allow a proper modeling of flight stability in insects. In general, a linear dynamic model is not adequate
for highly maneuverable insects, birds, and micro air vehicles.
International Journal of Micro Air Vehicles

Roman Y. Krashanitsa, Dmitro Silin, Sergey V. Shkarayev, and Gregg Abate

37

Taylor and Zbikowski7 presented experimental data on instantaneous forces and moments produced
by flapping wings of tethered desert locusts. These data were approximated using Fourier series and
embedded in terms of obtained periodic functions into rigid-body equations of motion. The change of
the position of the center of mass was not included in the modeling, and pitch rate derivatives were
dropped from the equations of motion as in the work by Taylor and Thomas.4 This resulted in a
nonlinear time-periodic model. The initial value problem for the obtained system of nonlinear ordinary
differential equations was solved numerically using MATLABTM routines. Obtained numerical results
indicate the instability of the solutions, which is consistent with instabilities of linear, time-invariant
formulation of the previous study.4
Some limitations of the referenced studies4,7 were overcome by Taylor et al.8 through measurements
of the pitching moment derivative with respect to the pitch rate. Incorporation of the pitch rate damping
effect into both linear time-invariant and nonlinear time-periodic models remained the unstable
character of the solution, but increased the doubling time of the positive real root. Although the
dynamic models of the locust flight presented in Refs. [4, 6, 7, 8] failed to explain its flight stability,
they provided a useful analysis of the flapping flight dynamics.
Historically, there have been mixed views as to whether a birds tail provides a stabilizing
mechanism and control for the bird in flight or not. Parga et al.9 compiled data for various bird species,
as well as aircraft types, to compare tail volume coefficients. For birds, the tail volume coefficients
were determined using plan views of bird tails presented by Rayner.10 Although, the birds have smaller
tail volume coefficients when compared to conventional aircraft, their tails by virtue of existence
provide some level of passive aerodynamic stabilization.
An adaptive wing is not a very new concept in conventional airplanes. The Wright brothers utilized
control of a deformable-wing by a symmetric warping of the trailing edge geometry for pitch, and an
asymmetric warping for roll. Significant attention has been paid in recent years to the concept of inflight adaptive wing or even morphing of the airplane with applications to large airplanes.
A design case11 presents the conventional MAV with a variable camber wing. The wing design
features a change in its cross-sectional geometry through a synchronous increase/decrease of
camber and reflex. This control technique redistributes aerodynamic pressure spanwise and
chordwise, creating a desirable combination of aerodynamic forces and moments. The design
constraint on the maximum dimension in MAVs dictates the selection of flying-wing
configurations.12 The biggest factor driving the design of the flying-wing aircraft is pitch stability
and control. A negative, nose-down pitching moment for cambered, low aspect-ratio wings at a low
Reynolds number is compensated for by introducing a wing with inverse camber. For this reason,
the adaptive airfoil designed by Aki et al.11 keeps the ratio of inverse camber to maximum camber
constant at 1:3.
A lack of accurate aerodynamic models for instant aerodynamic forces on flapping wings remains
the main obstacle for a formal mathematical analysis of flight dynamics and controls. In a previous
work,13 a quasi-steady approximation14 of aerodynamic forces generated by flapping and rotating
wings was employed in flight simulations and stability analysis of an ornithopter. Rigid-body equations
of motion in the longitudinal direction were combined with approximate expressions for periodic
aerodynamic forces. Stability analysis of the resulting nonlinear system was conducted. The nonlinear
system was assumed to exhibit a limit-cycle behavior in the state space, and the limit- cycle trajectory
was determined numerically. The nonlinear system was linearized about the limit-cycle locus and
further analysis was conducted on the linear system. No explanations were given on how the
linearization procedure affects the topology of the solution. A numerical example was constructed for
the ornithopter and showed the presence of unstable modes.
The dynamic model of a flapping-wing apparatus developed by Malladi et al.15 relies on the direct
application of Newton's second law of motion. This analysis requires a complete specification of all
aerodynamic forces acting on each component of the mechanical system, including two flapping wings,
body, and tail.
A Lagrangian method for the formulation of equations of motion of a rigid flapping-wing air vehicle
was outlined in Buler et al.16 The obtained system of nonlinear ordinary differential equations was
linearized and a standard feedback control was designed for the computational modeling of attitude
maneuvering. Unfortunately, the study did not discuss specifics of the constraints imposed on the wing
motion and did not explain the determination of aerodynamic loads on flapping wings in the numerical
example.
Volume 1 Number 1 2009

38

Flight Dynamics of a Flapping-Wing Air Vehicle

AeroVironment pioneered the design of a radio-controlled micro ornithopter called Microbat.17

The most successful vehicle of this type has a half-ellipse wing planform with a 20-cm wingspan
flapping at 22 Hz. The project proved to be challenging because of the limited knowledge on
unsteady aerodynamics of flexible flapping wings of this small size and a lack of enabling micro
technologies.
DeLaurier and his group developed a 35-cm radio-controlled ornithopter capable of hovering.18 The
kinematics of the four wings (X-wing) mimics the cling-flip mechanism employed by some insects
and birds. Hovering flights in excess of one minute were achieved with a flapping frequency of 28 Hz.
It was noted that transition to forward flight remains a problem, but that it may be overcome by an
intelligent flight stabilization system.
The University of Arizona has developed 28-cm,19 20-cm,20 and 15-cm21 ornithopters. The current
ornithopter designs are partially similar to those of conventional aircraft, in that a fixed tail with an elevator
and a rudder are used to provide longitudinal and lateral control and stability. The smallest (15-cm, 8-gram)
ornithopter can fly for more than 3 minutes at speeds from 1 m/sec to 5 m/sec. With a flapping frequency
of 20 Hz, it is capable of a wide range of maneuvers and it looks deceptively like a large insect, even from
a short distance. In order to perfect this technology, more studies are needed on the flight dynamics and
controls of these vehicles.
Unlike a biological system, flapping-wing aerodynamics and control mechanisms can be studied
explicitly in aerial vehicles since the input parameters specifying flight behavior can be placed under
full control of the experimenter. Therefore, the objective of this study is to investigate essential stability
and control parameters of flapping flight through extensive laboratory and flight measurements on an
autonomous ornithopter. In this study, the experimental ornithopter shown in Fig. 1 was equipped with
an automatic flight control system and utilized in studies of flapping flight dynamics. Stability
coefficients were determined for the ornithopter through wind tunnel measurements. Flight
experiments were conducted with in-flight telemetry data acquisition. Stability of the aircraft via
passive methods was investigated in flight experiments with fixed controls. The flight state parameters
of a flapping-wing apparatus were also measured during autonomous flights and characteristics of the
flight motion were discussed.

Figure 1. Autonomous ornithopter on the ground (left) and in flight (right).

II. ORNITHOPTER DESIGN AND SPECIFICATIONS

The 74-cm ornithopter was built and equipped with an automatic control system providing
stabilization and navigation of the vehicle and in-flight data acquisition and transmission to the
ground station. An off-the-shelf Cybird P2 ornithopter was selected for the study. This ornithopter
underwent a thorough flight testing. Utilizing this flight experience, the ornithopter was modified in
order to improve its payload capability. The original airframe had a tail consisting of a single surface,
with adjustable elevation and controllable rotation, thus allowing directional control. This tail was
replaced with a V-tail with twice the area with the opening angle of 105. The tail boom is made 2.5
times longer. With these modifications, more effective pitch and roll controls were achieved. After
necessary adjustments of the center of gravity position, the ornithopter was able to fly at a maximum
International Journal of Micro Air Vehicles

Roman Y. Krashanitsa, Dmitro Silin, Sergey V. Shkarayev, and Gregg Abate

39

speed of 10 m/sec, withstand wind up to 3 m/sec, and perform a series of sharp turns while
maintaining altitude. Its flight endurance at a moderate throttle setting was in excess of 7 minutes
with a payload of 50 grams. The ornithopter geometry, components, and mass data are listed in
Tables 1 and 2. The moment of inertia about the pitch axis presented in Table 1 is determined with
respect to the center of gravity.
Table 1. Ornithopter specifications

Parameter
Wingspan (cm)
Length (cm)
Height (cm)
Wing area (cm2)
Tail area (cm2)
Tilt angle of the tail (deg)
CG location from wing pivot (cm)
CG location from root chord (cm)
Flapping angle amplitude (deg)
Dihedral (deg)
Moment of inertia (kg m2)

Value
74
53
14
991
231
16
7.3
4.1
55
7.5
1.4 103

Table 2. Ornithopter components

Component
Wing
Tail
Fuselage with gearbox and nose
Motor
RC receiver
Speed controller
2 micro servos
3-cell lithium-polymer battery
Autopilot
Attitude sensor
Radio modem
Misc
Total

Description
Carbon rods, nylon cloth
Carbon rods, nylon cloth
Fiberglass, EPP foam
Speed 370
PENTA 5
Electryfly 20A
Blue Arrow 3.6g
Thunder Power 730 mAh
Paparazzi Tiny
IR sensor board
XB Pro 2.4 GHz
Wires, foam pads, etc

Mass (g)
15
11
78
28
3
4
7
46
22
7
3
16
248

In the present study, the ornithopter was equipped with a Paparazzi autopilot22 utilizing previous
experience with this autopilots integration12,23 into micro air vehicles. The autopilot includes a
Phillips ARM7 microprocessor, a built-in U-Blox GPS receiver with an 18-mm patch antenna
mounted on the autopilot board, and an infrared sensor board to determine the attitude of the
vehicle. The autopilot board was installed on the top of the airframe right underneath the wing (Fig. 2).
In order to protect the board from vibrations, it was mounted using a block of foam. All the radio
control components were located right under the board in the frame cutouts, thus the weight of the
wiring was also minimized. An on-board antenna and a pair of X-Bee Pro modems by Maxstream,
Inc. provided wireless communication between the vehicle and the ground station. The flight
control software consisted of the autopilot software on-board the aircraft and the ground station
software. In flight, the autopilot sent telemetry data back to the ground station. Currently, the
telemetry data include: GPS-based location data; speed, altitude, and climb rate of the airplane;
pitch and roll angles of the aircraft provided by infrared sensors; autopilot status data; and position
of the control surfaces.
Longitudinal control of the ornithopter was accomplished by proportional control for the altitude
hold, with an inner pitch attitude loop. Similarly, the lateral-directional control was accomplished by
an outer heading hold loop and an inner bank angle control loop, both using proportional control.
Volume 1 Number 1 2009

40

Flight Dynamics of a Flapping-Wing Air Vehicle

Figure 2. Components of the ornithopter.

III. WIND TUNNEL MEASUREMENTS

Wind tunnel testing of the 74-cm ornithopter was performed in the Low Speed Wind Tunnel at the
Aerospace and Mechanical Engineering Department of the University of Arizona (Fig. 3). This opencontour wind tunnel has a 3 4 feet test section and a free stream velocity range from 2 to 50 m/sec.
The flow is laminarized in a settling chamber to less than 0.3% turbulence in the axial direction. The
wind tunnel is equipped with a six-component balance. Force measurements are done using precision
strain gages. Data from these strain gages are logged using three National Instruments SCXI-1321
terminal blocks in a low-noise SCXI-1000 chassis capable of sampling at 330,000 Hz.

Up-stroke
Stroke Plane

Flapping
Wing

Down-stroke

Figure 3. The ornithopter in the wind tunnel (a) and definition of the stroke plane angle (b).

Preliminary flights of the ornithopter were conducted, including horizontal dashes at a constant
speed. During these flights, the pilot controlled the ornithopter via a radio transmitter, keeping the
throttle setting, TS, constant at 65%. The ornithopter trim in the level flight was achieved with the
elevons set at 3 deg for the right and 15 deg for the left. From the analysis of the telemetry data
acquired during trimmed flights, the ornithopter was cruising at an average speed of 7.25 m/sec at the
average pitch angle of 20 deg.
The ornithopter was tested in the wind tunnel in a fully assembled configuration. Thus, measured
aerodynamic forces, coefficients, and their derivatives include the contributions of all components of
International Journal of Micro Air Vehicles

Roman Y. Krashanitsa, Dmitro Silin, Sergey V. Shkarayev, and Gregg Abate

41

12

12

11

11

10

10

9
f, Hz

f, Hz

the ornithopter: flapping wings, V-tail, and fuselage. The elevons were held in the trimmed positions.
Thus, measured aerodynamic forces, coefficients, and their derivatives include the contributions of all
components of the ornithopter: flapping wings, V-tail, and fuselage. The stroke plane angle, sp, is
defined as the angle between the free stream velocity and the normal to the stroke plane of the flapping
wing, as show in Fig. 3b. The free stream speed, throttle setting, and stroke plane angle were varied
during wind tunnel experiments.
It was observed during preliminary wind tunnel tests that turning on the ornithopter caused a
significant change in the free stream velocity. For example, for initial settings V = 7.25 m/sec and Ts =
65% the test section velocity increases by about 12%. As the stroke plane angle increases, the velocity
decreases, reaching about 10% decrement at sp = 40. Similar effects have been observed during
testing of propellers with a disk area comparable to a wind tunnel test section area.24 The suggested
solution to this problem24 is to correct the air speed by using the value of thrust force generated by the
propeller, geometries of propeller and test section. In the present study on an ornithopter, the wind
tunnel speed was manually adjusted and held constant.
The wing flapping frequency, f, was measured using the digital tachometer HCAP0401 at the
constant throttle setting. Variations of the flapping frequency at sp = 0 are presented in Fig. 4a for
different throttle settings and free stream speeds. The flapping frequency increases with both throttle
and airflow speed increase. As can be seen in Fig. 4b, the flapping frequency decreases, when the stroke
plane angle is increased from 0 to 40 deg for constant free stream speed of 7.25 m/sec.

8
7

8
7

V=0

V=6
V=7.25

V=8

Ts=40%

Ts=50%

Ts=60%

Ts=70%

Ts=80%

Ts=100%

4
40

50

60

70
80
TS , %
(a)

90

100

10

15

25

20

30

35

40

(b)

Figure 4. Flapping frequency variation with throttle (a) and angle of stroke plane (b).

Horizontal and vertical force components and pitching moment were measured using the wind
tunnel-based coordinate system. In this coordinate system, the positive horizontal axis is directed
downstream, and the positive vertical axis is directed upward. The horizontal force is defined as a
positive force if the drag is greater than the thrust generated by flapping wings. The pitching moment
was determined about the center of gravity of the ornithopter.
The aerodynamic coefficients CX, CY, and CM were determined at a free stream velocity V = 7.25 m/sec.
The stroke plane angle was varied from 0 to 40 degrees. The throttle setting for flapping wings was chosen
at 45%, 55%, 65%, and 75%. At V = 7.25 m/sec, these settings correspond to a flapping frequency of 7.2,
8.2, 9.2, and 10.2 Hz, respectively.
As can be seen in Fig. 5, the vertical force coefficient increases with throttle increase. Note that no
abrupt stall is observed in Fig. 5 for all throttle settings tested. For the throttle setting of 65%, free stream
velocity V = 7.25 m/sec, and stroke plane angle sp = 20, it was found CY = 0.84 and CY = 2.09 rad1
Volume 1 Number 1 2009

42

Flight Dynamics of a Flapping-Wing Air Vehicle

1.6
1.4
1.2

CY

1
0.8
0.6
0.4

Ts=75%
Ts=65%

0.2

Ts=55%

Ts=45%
0

10

15

20

25

30

35

40

Figure 5. Vertical force coefficient variation with stroke plane angle.

(Fig. 5). Reminder that these parameters correspond to trimmed level flight conditions. Using these
parameters, the vertical component of the aerodynamic force was determined to be only 0.3% greater
than the ornithopters weight.
The horizontal force coefficient curves presented in Fig. 6 are convex. In these plots, a negative
force should be interpreted as a forward thrust. At a moderate stroke plane angle, an increase of the
throttle setting provides a higher thrust generated by the flapping wings. As the angle increases, the
curves converge for all tested throttle settings. For Ts = 65% and sp = 20, the horizontal force
coefficient is 0.00017, which is very close to zero. The curve slope at this point is C X = 0.84 rad1.
The pitching moment coefficients presented in Fig. 7 are positive at sp = 0 and have negative
slopes for all throttle settings tested. At a throttle setting of 65% and sp = 20, the coefficients are
0.5

0.4

CX

0.3

0.2

0.1
Ts=75%

Ts=65%
Ts=55%
Ts=45%

10

15

20

25

30

35

40

Figure 6. Horizontal force coefficient variation with stroke plane angle.

International Journal of Micro Air Vehicles

Roman Y. Krashanitsa, Dmitro Silin, Sergey V. Shkarayev, and Gregg Abate

43

0.4
Ts=75%

0.3

Ts=65%

0.2

Ts=55%

CM

0.1

Ts=45%

10

15

20

25

30

35

40

Figure 7. Pitching moment coefficient variation with stroke plane angle.

CM = 0.0087 and C M = 1.1 rad1. Note that pitching moment is close to zero for conditions
observed for the real balanced level flight. In the present study, the weight polar was subtracted from
the pitching moment data. This correction was not performed in the previous work25 resulting in
shifts in pitching moment plots.
A positive value of the pitching moment at sp = 0 and a negative pitching moment slope are two
necessary conditions for longitudinal static stability of aircraft under control-fixed conditions. This is a
classical criterion of static stability developed for conventional aircraft and based on this criterion the
ornithopter is stable.
IV. FLIGHT TEST RESULTS AND DISCUSSIONS
A. Flights with Fixed Controls
In the course of flight experiments, telemetry data on the roll and pitch angles of the aircraft, its in-plane
location (longitude and latitude), and altitude were acquired by the autopilot sensors and transmitted to
the ground station computer. In order to maximize the downlink bandwidth, flights were conducted at
the distance of 1020 m from the ground station. The rate of the data transmission between the computer
and modem was set at 38,400 bps and the length of the transmitted data sample was 96 bit. Due to the
specifics of the radio transmission protocol and deteriorating effects of distance on the quality of the
signal, some samples were missing, resulting in a sampling rate in the range of 510 Hz.
In order to study the open-loop dynamics of the ornithopter, several fixed controls flights were
performed at a constant throttle setting of 67% and with elevons fixed in the trimmed positions as
described in the previous section. With the ornithopter airborne, the pilot brought the ornithopter to a
specified altitude by using radio controls. Then, the pilot let the ornithopter fly for 1020 sec in level
dashes, with controls held at the trimmed positions. The data acquired during Flights 1 and 2 are
presented and discussed here. The telemetry data were acquired at sampling frequencies of 10 Hz and
6 Hz, respectively. It was observed that both flights were sustained and stable in the course of
experiments.
The Flight 1 altitude variation and in-plane trajectory are shown in Figs. 8 and 9, respectively. The
length of the dash was about 100 m. During the flight, both the altitude and flight direction changed in
a slow oscillatory manner within a period of about 10 sec.
In-flight pitch and roll angle data are presented in Figs. 10 and 12, respectively. The maximum
amplitude of pitch oscillations is about 30 and the average value of higher frequency oscillations is
about 6. The data were analyzed using the fast Fourier transform (FFT) function from MATLABTM.
Pitch spectrum and roll spectra are shown in Fig. 11 and 13, respectively. They are very similar in
appearance, with a distinct peak frequency at about 1.5 Hz.
Volume 1 Number 1 2009

44

Flight Dynamics of a Flapping-Wing Air Vehicle

25

20

Y, m

Altitude, m

20

15

10

10
Time, s

15

20

40

Figure 9. Flight trajectory in Flight 1.

Figure 8. Altitude history in Flight 1.

Pitch, deg

0
X, m

20

30

10

20

10

4
2

10
Time, s

15

20

Figure 10. Pitch angle history in Flight 1.

0
0

2
3
Frequency, Hz

Figure 11. Power spectrum of pitch in Flight 1.

50

10

Roll, deg

8
0

6
4
2

10

15

Time, s

Figure 12. Roll angle history in Flight 1.

20

Frequency, Hz

Figure 13. Power spectrum of the roll in Flight 1.

The flight altitude variation and in-plane trajectory for Flight 2 are shown in Figs. 14 and 15,
respectively. This flight was shorter in duration than the first one. The altitude oscillations are around
an average flight altitude of about 22.5 m, with a period of oscillations of 5 s. The flight trajectory seen
in Fig. 15 is a semi-circular one with a radius of 15 m. For this flight, the elevons were fixed slightly
off the trim. This explains the semi-circular flight path.
Pitch and roll angle data shown in Fig. 16 and 18 were also analyzed using the fast Fourier
transform. Pitch and roll spectra shown in Figs. 17 and 19, respectively, feature a dominant frequency
at about 1.5 Hz, which is similar to the peak observed in Flight 1. Both spectra are less dense in
comparison to corresponding spectra in Flight 1. It can be explained by the smaller sampling frequency
for Flight 2.
The ornithopter flights performed with the controls fixed in a predefined position demonstrated that
a sustained and stable flight of the flapping-wing aircraft is possible. These experiments showed a similar
type of low-frequency oscillatory dynamics in both roll and pitch motions. Peaks at about 1.5 Hz were
exhibited in both pitch and roll spectra.
International Journal of Micro Air Vehicles

Roman Y. Krashanitsa, Dmitro Silin, Sergey V. Shkarayev, and Gregg Abate

45

24

Altitude, m

Y, m

23

22

21

4
6
8
Time, s
Figure 14. Altitude history in Flight 2.

10

0
X, m

Figure 15. Flight trajectory in Flight 2.

10

20

10

Pitch, deg

20

6
0
4
2
0

4
6
Time, s

10

Figure 16. Pitch angle history in Flight 2.

2
3
Frequency, Hz

Figure 17. Power spectrum of pitch in Flight 2.

20

Signal magnitude 10

10

Roll, deg

4
6
Time, s

10

Figure 18. Roll angle history in Flight 2.

8
6
4
2
0
0

2
3
Frequency, Hz

Figure 19. Power spectrum of roll in Flight 2.

B. Autonomous Flights
Six fully autonomous flights were completed. The data acquired from one of the flights with a total
duration of 350 sec are shown in Figs. 2028. The flight plan was to navigate the aircraft between two
waypoints. The only constraint put on the navigation was to fly within 5 m from a waypoint. The
autonomous flight trajectory is shown in Fig. 20, where waypoints are labeled as waypoint 1 and
waypoint 2. The ornithopter performed elliptical courses or a Figure-8, depending on the wind direction.
150
12

Ground speed, m/s

waypoint 2

Y, m

100
50
0

50

100

X, m

Figure 20. Autonomous flight trajectory.

Volume 1 Number 1 2009

8
6
4

waypoint 1
0

10

100

200
Time, s

Figure 21. Ground speed variation.

300

46

Flight Dynamics of a Flapping-Wing Air Vehicle

The ground speeds were measured using an on-board GPS unit and are presented in Fig. 21. Since
a wind was present during the flight, the ground speed was changing for the flight segments in the wind
and against the wind, and was varying within a range of 510 m/sec.
The throttle was controlled by the autopilot for the duration of the flight. As can be seen in Fig. 22,
the throttle was varying in the range of 40%-80% for the major portion of the flight.
A plot of the on-board 3-cell battery voltage versus time is shown in Fig. 23. The flight started at a
voltage value of 11.2V and, at the end of the flight, battery voltage was at 10 V.
100

11.5

Battery voltage, V

Throttle, %

80
60
40
20
0

100

200
Time, s

11
10.5
10
9.5

300

Figure 22. Throttle time history.

100

Figure
23.
during flight.

200
Time, s

Battery

voltage

300

change

The throttle setting and elevator were controlled by the inner loop of the altitude control algorithm.
Results of the altitude control are shown in Fig. 24 as the actual altitude versus commanded altitude.
Maximum altitude error for the duration of the flight was about 7 m and the average error was about 1
m. After 280 s of flight time, the desired altitude was increased from 60 m to 70 m (Fig. 24) by a
command from the ground station. The ornithopter reacted by gaining 10 m in about 40 s. The desired
altitude was reached rather slowly, but without any overshoot. The overall quality of the flight plan
tracking can be regarded as satisfactory, and it is believed that it can be improved by further
optimization of the flight control system.

actual

Altitude, m

70

commanded

65

60

55
0

50

100

150

200

250

300

Time, s

Figure 24. Actual and commanded altitude variations.

Time histories of actual versus commanded pitch and roll angles are shown in Figs. 25 and 26,
respectively. The data for the actual values of roll and pitch were taken at a sampling frequency of about
4 Hz. Average amplitudes of oscillations in roll were more than 2 times higher than in pitch and can be
explained by lower inertia of the ornithopter in the roll axis. Overall, proportional control laws have
shown satisfactory performance during flight tests.
Elevator and aileron command histories are shown in Figs. 27 and 28, respectively. The average
values of aileron and elevator deflections were less than 10% of the total range for the servo for the
duration of the flight.
International Journal of Micro Air Vehicles

Roman Y. Krashanitsa, Dmitro Silin, Sergey V. Shkarayev, and Gregg Abate

40

actual

20

actual
commanded

Roll, deg

Pitch, deg

commanded
10
0

100

200

20

300

100

200

300

Time, s

Time, s

Figure 25. Measured and commanded pitch.

Figure 26. Measured and commanded roll.

50

50

Aileron, %

Elevator, %

47

100

200

300

Time, s

Figure 27. Elevator command history.

100

200
Time, s

300

Figure 28. Aileron command history.

V. SUMMARY
Advancements of flapping-wing MAV technology depend on obtaining reliable laboratory and flight
data. The present study contributes to the understanding of flapping flight dynamics. More precisely,
essential stability parameters and basic dynamic modes were investigated through extensive laboratory
and flight measurements on an autonomous ornithopter.
An experimental ornithopter was built and an autopilot was integrated into the vehicle. The autopilot
provides stabilization and navigation of the vehicle and in-flight data acquisition. The flight control
system utilizes GPS for waypoint navigation and proportional longitudinal and lateral-directional
controls. The autonomous ornithopter is a unique test bench for the investigations of aerodynamic
forces, kinematics, and automatic controls in flapping flight.
Wind tunnel tests were conducted on the ornithopter. Values of aerodynamic forces and pitching
moment were measured. Aerodynamic coefficients and their derivatives were determined in order to
analyze the performance and stability of the ornithopter. Aerodynamic forces obtained in wind tunnel
tests match very closely the forces corresponding to those observed in balanced level flight. The
longitudinal static stability under control-fixed conditions was analyzed with the help of a classical
theory developed for conventional aircraft. Based on this analysis, the ornithopter is statically stable.
Several control-fixed flights were performed in order to examine passive aerodynamic stability
mechanisms of the flapping-wing apparatus. These flights demonstrated that a sustained, stable flight
of the flapping-wing aircraft with only passive aerodynamic stability is possible. Telemetry data were
acquired, including altitude, in-plane trajectory, and pitch and roll angles, and were analyzed using fast
Fourier transform techniques. These experiments showed a similar type of low-frequency oscillatory
dynamics in both roll and pitch motions. Peaks at a frequency of about 1.5 Hz were found in both pitch
and roll spectra. This frequency is about 6 times lower than the exciting frequency due to flapping.
Further studies are needed in order to identify mode shapes (phugoid, short period, etc.) corresponding
to this frequency.
During the autonomous flights, the ornithopter performed waypoint and altitude navigation,
demonstrating stable performance. Average amplitudes of oscillations in roll were more than 2 times
higher than in pitch and can be explained by lower inertia in the roll axis. The overall quality of the
Volume 1 Number 1 2009

48

Flight Dynamics of a Flapping-Wing Air Vehicle

flight plan tracking can be regarded as satisfactory, and it is believed that it can be improved by further
optimization of the flight control system.
VI. ACKNOWLEDGMENTS
This work has been sponsored under a research grant from the Air Force Research Laboratory,
Munitions Directorate, Eglin AFB, FL. The authors also would like to thank David Addai and Gavin
Kumar for their contributions to the flight experiments.
REFERENCES
1.
Mueller, T. J., and DeLaurier, J. D., An Overview of Micro Air Vehicle Aerodynamics, In: Fixed
and Flapping Wing Aerodynamics for Micro Air Vehicle Applications, edited by T. J. Mueller,
Vol. 195, AIAA, Reston, VA, 2001, pp. 112.
2.
Ellington, C. P., The Novel Aerodynamics of Insect Flight: Applications to Micro Air Vehicles,
The Journal of Experimental Biology, Vol. 202, 1999, pp. 34393448.
3.
Taylor, G. K., Mechanics and Aerodynamics of Insect Flight Control, Biological Reviews. 76,
2001, pp. 449471.
4.
Taylor, G.K., and Thomas, A.L.R., Dynamic Flight Stability in the Desert Locust Schistocerca
gregaria, The Journal of Experimental Biology, Vol. 206, 2003, pp. 28032829.
5.
Etkin, B., and Reid, L. D., Dynamics of Flight Stability and Control, Third Edition, John Wiley
and Sons, Inc., New York, NY, 1996.
6.
Sun, M. and Xiong, Y., Dynamic Flight Stability of a Hovering Bumblebee, The Journal of
Experimental Biology, Vol. 208, 2005, pp. 447459.
7.
Taylor, G.K., and Zbikowski, R., Nonlinear Time-Periodic Models of the Longitudinal Flight
Dynamics of Desert Locusts Schistocerca Gregaria, J. R. Soc. Interface, No. 2, 2005, pp.
197221.
8.
Taylor, G. K., Bomphrey, R. and J.t Hoen, Insect Flight Dynamics and Control, 44th AIAA
Aerospace Sciences Meeting and Exhibit, Reno, Nevada, January 912, 2006, 12 p.
9.
Parga, J. R., Reeder, M. F., and Leveron, T., Experimental Study of a Micro Air Vehicle with a
Rotatable Tail, Journal of Aircraft, Vol. 44, No. 6, NovemberDecember, 2007, pp. 1761768.
10. Rayner, J. M. V., Form and function in avian flight. Current Ornithology, Vol. 5 (ed. R. F.
Johnston), New York, London: Plenum Press, 1988, pp. 166.
11. Aki, M., Waszak, M., and Shkarayev, S. Development of Micro Air Vehicles with In-Flight
Adaptive Wing, Chapter 6, Introduction to the Design of Fixed-Wing Micro Aerial Vehicles,
edited by T. J. Mueller, J. C. Kellogg, P. G. Ifju, and S. V. Shkarayev, AIAA, Reston, VA, 2006.
12. Krashanitsa, R., Platanitis, G., Silin, D., and Shkarayev, S., Autopilot Integration into Micro Air
Vehicles, Introduction to the Design of Fixed-Wing Micro Aerial Vehicles, edited by T. J.
Mueller, J. C. Kellogg, P. G. Ifju, and S. V. Shkarayev, AIAA, Reston, VA, 2006.
13. Dietl, J. M., and Garcia, E., Stability in Ornithopter Longitudinal Flight Dynamics, Journal of
Guidance, Control, and Dynamics, Vol. 31, No. 4, 2008, pp. 11571163.
14. Wang, Z. J., Birch, J., and Dickinson, M., Unsteady Forces and Vorticity Field in Hovering
Flight: Two Dimensional Computations vs. Robotic Wing Experiments, The Journal of
Experimental Biology, Vol. 207, No. 3, 2004, pp. 449460.
15. Malladi, B., Krashanitsa, R., Silin, B., and Shkarayev, S., Dynamic Model and System
Identification Procedure for Autonomous Ornithopter, The 3rd US-European Competition and
Workshop on Micro Air Vehicles, September 2007, Toulouse, France.
16. Buler, W., Loroch, L., Sibilski, K., and Zyluk, A., Modeling and Simulation of the Nonlinear
Dynamic Behavior of a Flapping Wings Micro-Aerial-Vehicle, 42nd AIAA Aerospace Sciences
Meeting and Exhibit, Reno, Nevada, Jan. 58, 2004, AIAA-2004-541.
17. Keennon, M., and Grasmeyer, J., Development of Two MAVs and a Brief Vision for the Future
of MAVs Design, AIAA-2003-2901, AIAA International Air and Space Symposium and
Exposition: The Next 100 Years, Dayton, Ohio, July 1417, 2003.
18. DeLaurier, J., Ornithopter Research, 2003 Bioflight Workshop, NASA Langley Research
Center, August 78, 2003.
International Journal of Micro Air Vehicles

Roman Y. Krashanitsa, Dmitro Silin, Sergey V. Shkarayev, and Gregg Abate

19.

20.

21.

22.
23.

24.
25.

49

Thompson, P., Ward, G., Kelman, B., and Null, W., Design Report: Development of
Surveillance, Endurance, and Ornithoptic Micro Air Vehicles, 2004 International Micro Air
Vehicle Competition, Tucson, Arizona, 10 p., April 2004.
Olson, D. H., Silin, D., Aki, M., Murrieta, C., Tyler, J., Kochevar, A., Jehle, A., and Shkarayev,
S., Wind Tunnel Testing and Design of Fixed and Flapping Wing Micro Air Vehicles at the
University of Arizona, Micro Air Vehicle Design Papers, Konkuk University, South Korea, 9 p.,
2005.
Yanagita, T., Lind, T., Malladi, B., Silin, D., Coopamah, D., Berka, T., Han, B., Rodriguez, S., and
Shkarayev, S., Simulation Wind Tunnel Testing, and Design of Fixed and Flapping Wing Micro
Air Vehicles, Micro Air Vehicle Design Papers, 10th International Micro Air Vehicle
Competition, May 2006, Provo, Utah.
Drouin, A., and Brisset, P., PaparaDzIY: Do-It-Yourself UAV, 4th European Micro-UAV
Meeting, Toulouse, France, Sept. 1517, 2004.
Krashanitsa, R., Platanitis, G., Silin, B., and Shkarayev, S., Aerodynamics and Controls Design
for Autonomous Micro Air Vehicles, AIAA Atmospheric Flight Mechanics Conference and
Exhibit, August 2124, 2006, Keystone, Colorado, AIAA 2006-6639.
Barlow, J. B., Rae, W. H., Jr., and Pope, A., Low-Speed Wind Tunnel Testing, 3rd Ed., Wiley,
New York, 1999.
Krashanitsa, R., Silin, D., Shkarayev, S., and Abate, G., Flight Dynamics of Flapping-Wing Air
Vehicle, AIAA AFM Conference and Exhibit, August 1821, 2008, Honolulu, Hawaii, AIAA2008-6698.

Volume 1 Number 1 2009