You are on page 1of 8

Nonlinear trajectory control of a apping-wing

micro aerial vehicle


Esmaeel Khanmirza, Aghil Youse-Koma and Bahram Tarvirdizadeh
School of Mechanical Engineering, College of Engineering, University of Tehran, Tehran, Iran
Abstract
Purpose The purpose of this paper is to develop a nonlinear control system for ight trajectory control of apping Micro Aerial Vehicles (MAVs),
subjected to wind.
Design/methodology/approach In the dynamic study and fabrication of the MAV, biomimetic principles are considered as the best inspiration for
the MAVs ight as well as design constraints. The blade element theory, which is a two-dimensional quasi-steady state method, is modied to consider
the effect of MAVs translational and rotational velocity. A quaternion-based dynamic wrench method is then developed for the dynamic system.
Findings The apping ight dynamics is highly nonlinear and the system is under-actuated, so any linear control strategy fails to meet any desired
maneuver for trajectory tracking. In this study, a controller with quaternion-based feedback linearization method is designed for the dynamical
averaged system. It is shown that the original system is bonded to a stable limit cycle with desired amplitude and the controller inputs are bounded.
Practical implications The effectiveness of a synthesized controller is proved for the cruse and the Cuban-8 maneuver.
Originality/value The authors major contribution is developing feedback linearization quaternion-based controller and deriving some essential
mathematics for implementing quaternion model in the synthesis of controller. A piezoelectric-actuated wing model is developed for the control system.
Results of cursing and turning modes of the ight indicate the stability of the ight. Finally, an appropriate controller is designed for the Cuban-8
maneuver so that the MAV would follow the trajectory with a bounded uctuation.
Keywords Control systems, Flight control, Flapping ight, Feedback linearization, Trajectory control, Piezoelectric actuator
Paper type Research paper
Introduction
The recent interest in micro aerial vehicles (MAVs), largely
motivated by the need for aerial reconnaissance robots inside
buildings and conned spaces, has galvanized the development
of inch-size apping wing MAVs that could mimic the insect
ight (Kummari et al., 2010; Hamamoto et al., 2010;
Rakotomamonjy et al., 2010; Go and Hao, 2010). This is a
challenging endeavour for several reasons. For instance,
aerodynamics for inch-size apping robots differs
substantially from manmade xed or rotary-wing vehicles.
Moreover, sensor types and size constraints add complexity to
the design of MAVs.
The goal of this paper is to develop a control for MAVs to have
a stable trajectory tracking. Also, we propose a piezoelectric
actuator and its model for actuating the wing. This study
includes a comprehensive modelling of the MAVaerodynamics,
body dynamics, and an electromechanical actuation system.
The time-varying (TV) dynamics of the insect ight caused by
the apping wings is approximated by a time-invariant (TI)
system. This approximation is based on two approximations
that can be formalized within the framework of high-frequency
control theory. First, the frequency of the aerodynamic forces
acting on the insect is much higher than the bandwidth of the
body dynamics, therefore only the mean aerodynamic forces
and torques over one wingbeat affect the insect dynamics.
Second, the wing trajectory is parameterized by biologically
inspired wing kinematic parameters which affect the
distribution of aerodynamic forces within one wingbeat, thus
modulate the total forces and torques acting on the insect
(Figure 1).
These parameters appear as virtual inputs in the TI
approximation of the ight dynamics. Finally, we show how
the parameters of the TI approximation can be identied
directly from sensors measurements and actuators input
voltages obtained from experiments of the original TV system.
This approach is particularly suitable for the apping ight
since it does not require the knowledge of the exact and
complex aerodynamics models.
Our major contribution is developing feedback linearization
quaternion-based controller and derive some essential
mathematics for implementing quaternion model in the
synthesis of controller. This controller allowed us to track
nonlinear trajectory without hazardous of any singularity in
model which can be found in Eulers equation of motion. Also,
the use of feedback linearization and tuning the gain based on
designer knowledge about the characteristics of the system,
result in the bounded actuator activity. In addition, we drive the
model for piezoelectric actuated wing and implement it in
design. The states which feed back to controller are the
orientation and position of MAV measured by gyros and
accelerometer. The characteristic of MAV fabricated at the
Advanced Dynamic and Control Systems Laboratory
The current issue and full text archive of this journal is available at
www.emeraldinsight.com/1748-8842.htm
Aircraft Engineering and Aerospace Technology: An International Journal
84/1 (2012) 5865
q Emerald Group Publishing Limited [ISSN 1748-8842]
[DOI 10.1108/00022661211194988]
The authors would like to thank Iranian National Science Foundation
(INSF) for their nancial support.
58
(ADCSL) are employed in the modelling and control
development.
Modelling
Insect ight aerodynamics, which belongs to the regime of
Reynolds number between 30 and 1,000, has been a very
active area of research in the past decades after the seminal
work of Ellington (1984) (Ramamurti and Sandberg, 2002;
Sun and Tang, 2002). Results obtained from these works
identied three main aerodynamics mechanisms peculiar to
the unsteady state nature of apping ight: delayed stall
(Anderson et al., 1998; Triantafyllou et al., 2004; Leishman,
2003), rotational lift (Sane and Dickinson, 2002; Walker,
2002) and wake capture (Sane and Dickinson, 2002). By
substituting the normal, tangential and rotational coefcients
(equations (3)-(6)), published by Sane (2003)) into the
aerodynamic forces which is presented by Kuethe and Chow
(1986), The net drag and lift force of the wings (equations
(1)-(2)) can be obtained as:
F
D
t 2
1
2
rA
w
U
cp
t
C
N
atU
cp
t C
rot
C
max
_ at
_ _ _ _
sinat
_
C
T
atU
2
cp
t
_ _
cosat
1
F
L
t
1
2
rA
w
U
cp
t
C
N
atU
cp
t C
rot
C
max
_ at
_ _ _ _
cosat
_
2 C
T
atU
2
cp
t
_ _
sinat
2
C
N
a 3:4 sina 3
C
T
a
0:4 cos
2
2a 0 # a # 458
0 otherwise
_
4
U
cp
t ^ rL
_
ft v
b
t 5
C
rot
2p
3
4
2 ^ x
_ _
; ^ r
_
L
0
crr
2
dr
R
2
A
w
6
where c is the cord width of the aerofoil, r is the density of air,
a is the angle of attack dened as the angle between the wing
prole and the wing velocity, U
cp
at the centre of pressure
relative to the uid, C
N
and C
T
are the dimensionless force
coefcients which were derived using experimental results
given by Sane (2003), C
rot
is the rotational force coefcient,
^ x
0
is the dimensionless distance of the longitudinal rotation
axis from the leading edge (In most ying insects ^ x
0
is about
1/4), f is the stroke angle, A
w
is the wing area, L is the wing
length, ^ r
2
is the normalized centre of pressure, C
max
is
maximum wing chord width, and v
b
(t) is the velocity vector of
the insect body relative to the inertial frame represented in the
wing frame coordinate system.
The mathematical aerodynamic modelling presented above
is a combination of an analytical model, based on quasi-steady
state equations for the delayed stall and rotational lift, and an
empirically matched model with the estimation of the
aerodynamic coefcients based on experimental data
(Kuethe et al., 2004) (Figure 2).
The wing forces can be assumed to be applied at a distance,
r
cp
^ rL, from the wing base. According to thin aerofoil theory,
the centre of pressure r
cp
lies about 1/4 of chord length fromthe
leading edge (Ramamurti and Sandberg, 2002). Because of
trajectory tracking, the body velocity is entered in MAV model
in our work (equation (5)) which adds more complexity in
dynamic. In addition, the wind model could be considered by
adding it to body velocity. Complexity in dynamic grows from
inertial coordinate system representation of wind model.
Body dynamics
The body dynamic equations compute the evolution of the
position of the insect centre of mass and the orientation of the
insect body, with respect to an inertial frame. This evolution is
the result of the wings inertial forces, and the external forces,
specically aerodynamic forces, body damping forces, and the
force of gravity. Since the mass of the wings is only a small
percentage of the insect body mass and as they move almost
symmetrically, their effect on insect body dynamics is likely
Figure 1 MAV fabricated at ADCSL
Figure 2 Denition of wing kinematic parameters
4 3
4 1
V
W
f
W
LEADING EDGE
TRAILING EDGE
SIDE VIEW
Y
Z
B
X
W
n
t
Note: Side view of wing perpendicular to wing axis
of rotation r

Nonlinear trajectory control of a apping-wing MAV


Esmaeel Khanmirza, Aghil Youse-Koma and Bahram Tarvirdizadeh
Aircraft Engineering and Aerospace Technology: An International Journal
Volume 84 Number 1 2012 5865
59
to cancel out within a single wingbeat (Murray et al., 1994;
Example 7.2). The equations of motion for a rigid body subject
to an external wrench F
b
[ f
b
,t
b
]
T
applied at the centre of mass
and specied with respect to the body coordinate frame,
are given as:
dq
dt

1
2
Vq
1
2
0 2V
1
2V
2
2V
3
V
1
0 V
3
2V
2
V
2
2V
3
0 V
1
V
3
V
2
2V
1
0
_
_
_
_
_
_
_
_
_
_
_
_
_
_
q
0
q
1
q
2
q
3
_
_
_
_
_
_
_
_
_
_
_
_
_
_
7
d
dt
V
1
V
2
V
3
_
_
_
_
_
_
_
_

I22I3
I1
V
2
V
3

t
b
1
I1
I32I1
I2
V
3
V
1

t
b
2
I2
I12I2
I3
V
1
V
2

t
b
3
I3
_
_
_
_
_
_
_
_
_
_
_
_
_
_
8
Qf
b
mI
id

P 9
Q 2
q
2
0
q
2
1
2q
2
2
2q
2
3
=2 q
1
q
2
2q
0
q
3
q
1
q
3
q
0
q
2
q
1
q
2
q
0
q
3
q
2
0
2q
2
1
q
2
2
2q
2
3
=2 q
2
q
3
2q
0
q
1
q
1
q
3
2q
0
q
2
q
2
q
3
q
0
q
1
q
2
0
2q
2
1
2q
2
2
q
2
3
=2
_

_
_

_
10
where m is the mass of the insect, I is the insect body inertia
matrix relative to the centre of mass, I
id
is the 3 3 identity
matrix, V is the angular velocity matrix in body frame
coordinates, q is the quaternion vector and P is the absolute
position in inertial frame. The values for the body and wing
morphological parameters, such as lengths and masses, used in
our simulations are those of a fabricated MAVat ADCSL.
The total forces and torques in the body frame are given by
the sum of the three external forces: the aerodynamic forces,
f
b
a
, the body damping forces, f
b
d
, and the gravity force, f
b
g
.
Since the lift and drag forces given by equations (1) and (2)
are calculated relative to the stroke plane frame, a coordinate
transformation is necessary before obtaining the forces and
torques acting on the body frame. The insect body frame is
dened as the coordinate system attached to the body centre
of gravity and with x-axis oriented from tail to head, the y-axis
from right wing hinge to left wing hinge, and the z-axis from
ventral to dorsal side of the abdomen. Since these are the axes
of symmetry of the insect, the matrix of inertia is almost
diagonal in the body frame. The stroke plane frame is the
coordinate system attached to the centre of the thorax at
the centre of the wings base, whose x-y plane is dened as the
plane to which the wing motion is approximately conned
during apping ight. The viscous damping exerted by the air
on the insect body is approximately given by:
f
b
d
t
b
d
_
_
_
_

2bv
b
0
_ _
11
where b is the viscous damping coefcient. The reason for the
linearity in the velocity of the drag force is that the velocity of
the insect is small relative to insect size; therefore viscous
damping prevails over quadratic inertial drag. Empirical
evidence for linear damping has been recently observed by
Sane (2003), by analyzing the free ight dynamics of true fruit
ies. Moreover, these data indicate that rotational damping
of the insect body is negligible relative to aerodynamic
forces even during rapid body rotation and can therefore be
neglected.
In this paper, each wing is moved by the mechanism,
a complex trapezoidal structure actuated by two piezoelectric
actuators at its base. A piezoelectric actuator employed in
Wing structure is modelled in Ansys and the transfer functions
obtained (Khanmirza, 2006). For instance, the transfer
function between the piezoelectric voltage and the stroke
angle is:
G
ACT
s
2:471
2:527s
2
400s 7:01 10
5
12
Insect ight control
The wrench, i.e. the forces and torques applied to the centre of
mass, is a nonlinear function of the instantaneous position and
velocity of the wing stroke (apping) angle f and the angle of
attack aof both wings, but it does not depend explicitly on time.
The aerodynamic forces and torques can be written as:
f
b
a
t f
b
a
f
r
;
_
f
r
; a
r
; _ a
r
; f
l
;
_
f
l
; a
l
; _ a
l
f
b
a
u; _ u
t
b
a
t t
b
a
f
r
;
_
f
r
; a
r
; _ a
r
; f
l
;
_
f
l
; a
l
; _ a
l
t
b
a
u; _ u
13
where u (w
r
, w
l
, a
r
, a
l
) and the lower scripts r and l stand for
right and left wing, respectively. The stroke angle f is the angle
between the wing radial axis and the y-axis of the stroke plane.
The rotation angle w is dened as the angle between the vertical
plane and the wing prole, which corresponds to the
complement of the angle of attack a, i.e. a 908 2 jwj.
The wingbeat period is much smaller than the responsiveness of
the insect body; therefore, intuitively speaking, only mean
forces and torques are relevant. In fact, this approximation has
been formalized by averaging theory (Avadhanula et al., 2002)
and has been widely used in different applications including
helicopter aerodynamics (Leishman, 2003). In particular, these
tools model the systemdynamics as an afne systemof the form
_ x f
o
x

n
i1
f
i
xu
i
, where u
i
are the control inputs.
Moreover, these systems are underactuated, i.e. the number
of available inputs u
i
is smaller than the degrees of freedom
(Sussmann and Liu., 2001). The application of geometric
control theory to underactuated system with nonzero f
0
is
limited (Vela, 2003; Martinez, et al., 2003; Colgate and Lynch,
2004).
We propose to parameterize the wing motion based on
biomimetic principles to design our periodic inputs based on
Dickinson and his group (Balint and Dickinson, 2004)
research, i.e. we propose:
un; t gt Gtn; n n
r
1
; n
l
1
; n
r
2
; n
l
2
; n
r
3
; n
l
3
14
g
g
f
g
f
g
w
g
w
_

_
_

_
; G
g
1
0 0 0
0 g
1
0 0
0 0 g
b
0
0 0 0 g
b
g
2
0
0 g
2
0 0
0 0
_

_
_

_
15
g
f
t
p
3
cos
2p
T
t
_ _
; g
w
t
p
4
sin
2p
T
t
_ _
g
1
t g
b
t
p
15
sin
3
p
T
t
_ _
; g
2
t
p
15
0 # t # T
16
Nonlinear trajectory control of a apping-wing MAV
Esmaeel Khanmirza, Aghil Youse-Koma and Bahram Tarvirdizadeh
Aircraft Engineering and Aerospace Technology: An International Journal
Volume 84 Number 1 2012 5865
60
It has been proved that the kinematic parameters: timing of
rotation, mean angle of attack, stroke angle amplitude, stroke
angle offset, downstroke deviation may sufce to generate any
ight maneuver and directly obtained by the appreciated value
of v. By applying averaging theory to approximate the
complex TV dynamics with the average TI dynamics, we
show that there is a direct map between the proposed
kinematic parameters and the mean forces and torques. The
kinematic parameters appear as virtual inputs in the averaged
dynamics. In fact, by averaging the body dynamic equations
in one wingbeat period, all terms directly change to same
average term except wrench. We obtain a static map w :
R
6
!R
6
from the wings parameters v [ R
6
to the mean
wrench

f
b
a
; t
b
a
_ _
[ R
6
. This is a nonlinear map and cannot be
computed analytically since the aerodynamic force and torque
are complex functions of wing position and velocity. However,
one could look for an afne approximation around the origin
of the wings parameters:
wv w
o
Wv dv 17
where w
o
[ R
6
, W[ R
66
, and d(v) is the approximation error.
Although, it is not possible to linearize analytically equations (1)
and (2) to obtain w
o
and W directly, it is possible to randomly
select different values for the parameter vector v, substitute it
into the parameterization given by equation (14), and nally
compute the true mean wrench

f
b
a
; t
b
a
_ _
via simulations using
the exact wing aerodynamics. The approximating w
o
and Wcan
then be found by rewriting equation (17) as a least square (LS)
problem where (w
o
,W) are the unknowns. The approximating
afne map is found to be as follows:
wO f O
0:0049
0
1:00297
0
0:0000261
0
_

_
_

_
;
W
Wf
Wt
_
_
_
_

20:00737f O 20:00737f O 0:0789f O 0:0789f O 0:00191f O 0:00191f O


20:00708f O 0:00708f O 0:018198f O 20:0181f O 20:00822f O 0:00822f O
20:1064f O 20:106f O 0:0467f O 0:0467f O 0 0
0:0417tO 20:0417tO 20:02042tO 0:0204tO 20:00451tO 0:00451tO
0:0133tO 0:0133tO 0:0141tO 0:0141tO 0:04526tO 0:04526tO
20:004173tO 0:004173tO 0:0413tO 20:0413tO 0:0010tO 20:0010tO
_

_
_

_
18
where f
o
mg, t
o
mgL, m is the mass of the insect, L is the
length of the wing, and the zero entries correspond to estimated
values negligible relative to the largest entries inthe matrix. This
approximation is quite accurate for kinematic parameters
smaller than unity, kvk
1
# 1. Figure 3 shows that the estimated
mean wrench, ^ wv : w
o
Wv predicts quite accurately the
true mean wrench obtained from simulations, thus validating
our approach.
Quaternion-based feedback linearization
controller
Following the guidelines described in the previous section, we
can now develop a controller for trajectory tracking by
designing a feedback law v h(x) such that the origin of the
averaged system is exponentially stable.
At rst, we derive Attitude equations which are depend on
quaternion and its derivatives. From classical dynamics, we
have:
dIV
dt
IV Vt
b
V 2q
c
dq
dt
; V 0; V; t
b
0; t
b

where I
33
is inertial matrix in body frame, V and V are
angular velocity vector and quaternion in body frame. From
quaternion manipulation it is obvious that:
IV V ; I q
c
dq
dt
q
c
dq
dt
2q
c
dq
dt
Iq
c
dq
dt
_ _
I
44

1 0
0 I
_ _
This grows from the following lemma:
Lemma. The cross-product can be dened as the pure
quaternion:
u v ; 0; u v
uv 2vu
2
u 0; u; v 0; v
where uv is the multiplication of two quaternions. By
rewriting the Euler equations in quaternion form, we have:
d
dt
2q
c
dq
dt
_ _
q
c
dq
dt
q
c
dq
dt
2I
21
q
c
dq
dt
Iq
c
dq
dt
_ _
I
21
t
b
And by little algebraic manipulation:
d
2
q
dt
2

1
2
dq
dt
q
c
dq
dt
2qI
21
q
c
dq
dt
Iq
c
dq
dt
2q
dq
c
dt
dq
dt
_ _

1
2
qI
21
t
b
Now, by considering the translation equation and averaging,
we have:

q g q;
_
q
1
2
qI
21
t
b

P 2
c
m
_

P 2g
1
m
Qf
b
By considering the following denition:
t
b
0
0; t
b
0
; v 0; v; W
t

1 0
0 W
t
_ _
And substitute averaged wrench (equation 18), we have:

q g q;
_
q
1
2
qI
21
t
b
0

1
2
qI
21
W
t
v

P 2
c
m
_

P 2g
1
m
Qf
b
0

1
m
QW
f
v
19
Assume that x q;

P
T
is state vector, therefore:
x
g q;
_
q
1
2
qI
21
t
b
0
2
c
m
_

P 2g
1
m
Qf
b
0
_
_
_
_

1
2
qI
21
W
t
1
m
QW
f
_
_
_
_
v 20
We can represent this equation in following form:
x f x; _ x bxv 21
Nonlinear trajectory control of a apping-wing MAV
Esmaeel Khanmirza, Aghil Youse-Koma and Bahram Tarvirdizadeh
Aircraft Engineering and Aerospace Technology: An International Journal
Volume 84 Number 1 2012 5865
61
where:
f x; _ x
g q;
_
q
1
2
qI
21
t
b
0
2
c
m
_

P 2g
1
m
Qf
b
0
_
_
_
_
bx
1
2
qI
21
W
t
1
m
QW
f
_
_
_
_
22
Now, we use feedback linearization method for controller
design. On the other hand, by appropriate choice of v, we
simplify system dynamic into linear form. Thus, we choose v
in the following form:
v bx
21
v
*
2f x; _ x 23
By substitute in equation (21) we have:
x v
*
24
where ~ x x 2x
d
is considered as error equation, which x
d
is
desired state, v
*
obtained from:
v
*
2C
_
~ x 2K~ x 2M
_
~ x 25
where coefcients C, K, and M are chosen so that, following
system would have appropriate response:

~ x C
_
~ x K~ x M
_
~ x 0 26
It must be paid attention that feedback linearization have
undesirable effect on actuators so we must bonded the control
signal. In this paper we use normal velocity Lyapunov
function to satisfy stability condition and make near f x; _ x.
For cruising mode of ight x
d
assumed with:

P;

Q;
_

P;
_

Q
_ _


P; 0;
_

P
0
; 0
_ _
27
Respectively (
_

P
0
is desired cursing velocity, Q is a
vector including Euler Angles f, u, and c around X, Y, and Z
axis). The conversion from Euler angle to quaternion is
performed by:
q
0
cosfcosucosw 2sinfsinusinw
q
1
cosfcosusinw sinfsinucosw
q
2
sinfcosucosw cosfsinusinw
q
3
cosfsinucosw 2sinfcosusinw
Results
For trajectory tracking mode of ight, we have organized the
steering mode rst and an 8-shape trajectory tracking second.
Designing feedback laws that generate steering behaviors
can be done along the same lines as the design of stabilizing
Figure 3 Simulations of exact mean wrench w (y-axes) versus the predicted mean wrench w (x-axes)
0.3
0.2
0.1
0
0.1
0.1
0.05
0
0.05
0.1
0.05 0.05 0.1 0.15 0.2 0
0
0.04
0 0.05 0.05
0.05
0
0.05 1.4
1.2
1
0.8
0.7 0.8 0.9
Note: The green and red lines indicate the upper and lower bounds, respectively
1 1.1
0.05
0.05
0 0.02 0.04 0.02
0.1 0.1 0.05 0 0.05
F
x
from linearization
F
x

f
r
o
m

s
i
m
u
l
a
t
i
o
n
F
y

f
r
o
m

s
i
m
u
l
a
t
i
o
n
F
z

f
r
o
m

s
i
m
u
l
a
t
i
o
n
F
y
from linearization
F
z
from linearization

x

f
r
o
m

s
i
m
u
l
a
t
i
o
n

Y

f
r
o
m

s
i
m
u
l
a
t
i
o
n

z

f
r
o
m

s
i
m
u
l
a
t
i
o
n

Y
from linearization

z
from linearization

x
from linearization
0.3
0.2
0.1
0
0.05 0.1 0.15 0.2 0
0.1
Nonlinear trajectory control of a apping-wing MAV
Esmaeel Khanmirza, Aghil Youse-Koma and Bahram Tarvirdizadeh
Aircraft Engineering and Aerospace Technology: An International Journal
Volume 84 Number 1 2012 5865
62
feedback laws for cruising (equation (27)). In steering mode,
we command to rotate around roll axis in sinusoidal manner
with 0.5 rad degree amplitude (Figure 4). It must be noted
that from averaging theory, the dynamics is stabilized in
asymptotic manner to limit cycle. The uctuating in Insect
body angles are very small and cannot be understood by
human sensory organs. The voltage of one actuator on the left
wing is shown in Figure 5. Other actuators voltage is the same
as this one. This gure shows that the actuator voltage
remains bounded.
As predicted by the averaging theory, the dynamic for
trajectory tracking stabilization, exponentially converges to a
periodic orbit. This can be clearly seen in the oscillation of x,y
positions shown in Figure 6. In this maneuver, the insect is
commanded to follow the Cuban-8 maneuver trajectory. As
shown in Figure 6, a small offset in MAVs Z position occurs
in this maneuver which must be hold at zero. In fact the MAV
controller must track the trajectory as well as dominate gravity
force and hold the MAV in the same altitude. So, this offset is
the direct effect of applying averaging theory on a highly
nonlinear system. In Figures 7 and 8 the Euler angles for
Cuban-8 maneuver and the model outputs are shown.
From classical nonlinear theory we know that these gures
prove the existence of limit cycle in the system. Our nonlinear
controller drives the average system to be stable, and thus the
system is driven to the stable limit cycle as discussed in the
previous section.
Conclusion
In this paper a mathematical model for apping ight inch
size MAVs has been presented. The aerodynamics, the
electromechanical architecture, and the sensory system for
these vehicles differ considerably from larger rotary and xed-
winged aircrafts, and require specic modelling.
Mathematical modelling and simulations have been
presented for the aerodynamics, the insect body dynamics,
and the electromechanical wing-thorax dynamics.
The apping ight dynamics is highly nonlinear and the
system is underactuated (number of inputs are less than the
number of outputs). Using averaging theory in high frequency
nonlinear systems, the original system was approximated by
an averaged system and some added virtual inputs (virtual
inputs are obtained by parameterization of control inputs)
Figure 4 The steering mode of ight, ADCSL lab
0.6
0.5
0.4
0.3
0.2
0.1
0
0 1 2 3 4 5
0.1
Time (sec)
R
o
l
l

t
r
a
j
e
c
t
o
r
y
MAV GROUP
Figure 5 The actuator voltage of left wing in hovering, ADCSL lab
0 0.5 1 1.5 2
4
3
2
1
0
1
2
3
4
v
o
l
t
a
g
e
Actuator voltage(mv)
Nonlinear trajectory control of a apping-wing MAV
Esmaeel Khanmirza, Aghil Youse-Koma and Bahram Tarvirdizadeh
Aircraft Engineering and Aerospace Technology: An International Journal
Volume 84 Number 1 2012 5865
63
that facilitated the controller design. In the dynamic model a
quaternion was used; but most importantly, the necessary
mathematical equations for the direct use of quaternion
in the feedback linearization algorithm are extracted for the
rst time. An appropriate controller was then designed for the
averaged system and the stability of the original system
(with this controller) was proved. Also we modelled the
piezoelectric actuator for the wing thorax structure. Finally a
Figure 6 x, y position for the trajectory tracking (left), z position (right)
0.06
0.02 0.15 0.1 0.05 0 0.05 0.1 0.15
0.04
0.02
0
Y

p
o
s
i
t
i
o
n
Z

p
o
s
i
t
i
o
n

(
m
)
X position time (sec)
0.02
0.04
0.06
Target
Model output
0
1
2
3
4
5
6
7
0 1 2 3 4 5
10
3
Target
Figure 7 Angle of the trajectory tracking
4
3.5
3
2.5
2
1.5
1
0.5
0
0.5
1
0 0.5 1 1.5 2 2.5 3 3.5 4 4.5 5


A
n
g
l
e


A
n
g
l
e
Target
Model output
time
3.5
3
2.5
2
1.5
3 3.1 3.2 3.3 3.4 3.6 3.7 3.8 3.9 4 3.5
time
Figure 8 Angles of the trajectory tracking
10
5
0
5
0.1 0 0.2 0.3 0.4
time
0.5
Model output
Model output


A
n
g
l
e


A
n
g
l
e
10
5
10
5
0.1 0 0.2 0.3 0.4
time
8
6
4
2
0
2
4
6
8
Nonlinear trajectory control of a apping-wing MAV
Esmaeel Khanmirza, Aghil Youse-Koma and Bahram Tarvirdizadeh
Aircraft Engineering and Aerospace Technology: An International Journal
Volume 84 Number 1 2012 5865
64
framework for the apping ight control and navigation in
biomimetic robotic insects was developed. A suitable
parameterization of the wing motion during the course of a
full wingbeat was employed and combined with the averaging
theory arguments. Consequently, for the trajectory tracking
mode of ight, an appropriate stable controller was designed
for Cuban-8 maneuver and the results showed the
effectiveness of the control system.
References
Anderson, J.M., Streitlien, K. and Triantafyllou, D.B.M.
(1998), Oscillating foils of high propulsive efciency,
J. Fluid Mech., Vol. 360, pp. 41-72.
Avadhanula, S., Wood, R., Campolo, D. and Fearing, R.
(2002), Dynamically tuned design of the MFI thorax,
Proc. of the IEEE International Conference on Robotics and
Automation, Washington, DC, USA, May.
Balint, C.N. and Dickinson, M.H. (2004), Neuromuscular
control of aerodynamic forces and moments in the blowy,
Calliphora vicina, J. Exp. Biol., Vol. 207 No. 22,
pp. 3813-38.
Colgate, J.E. and Lynch, K.M. (2004), Mechanics and
control of swimming: a review, IEEE J. Ocean. Eng.,
Vol. 29 No. 3, pp. 660-73.
Ellington, C.P. (1984), The aerodynamics of hovering insect
ight. I-VI, Philos. Trans. R. Soc. Lond. B. Biol. Sci.,
Vol. 305, pp. 1-181.
Go, T.H. and Hao, W. (2010), Investigation on propulsion
of apping wing with modied pitch motion, Aircr. Eng.
Aerosp. Technol., Vol. 82 No. 4, pp. 217-24.
Hamamoto, M., Ohta, Y., Hara, K. and Hisada, T. (2010),
A fundamental study of wing actuation for a 6-in-
wingspan apping microaerial vehicle, IEEE Trans.
Robot., Vol. 26 No. 2, pp. 244-55.
Khanmirza, E. (2006), Trajectory control of micro aerial
vehicles, Masters thesis, University of Tehran, Tehran.
Kuethe, A. and Chow, C. (1986), Foundations of
Aerodynamics, Wiley, New York, NY.
Kummari, B., Lal, K., Li, D., Guo, S. and Huang, Z. (2010),
Development of piezoelectric actuated mechanism for
apping wing micro-aerial vehicle applications, Adv. Appl.
Ceram., Vol. 109 No. 3, pp. 175-9.
Leishman, J. (2003), Principles of Helicopter Aerodynamics,
Cambridge Aerospace Series, University of Maryland,
College Park, MD.
Martinez, S., Cortes, J. and Bullo, F. (2003), On analysis
and design of oscillatory control systems, IEEE Trans.
Autom. Control, Vol. 48 No. 7, pp. 1164-77.
Murray, R.M., Li, Z. and Sastry, S. (1994), A Mathematical
Introduction to Robotic Manipulation, CRC Press,
Boca Raton, FL.
Rakotomamonjy, T., Ouladsine, M. and Moing, T.L. (2010),
Longitudinal modelling and control of a apping-wing
micro aerial vehicle, Control Eng. Practice, Vol. 18 No. 7,
pp. 679-90.
Ramamurti, R. and Sandberg, W. (2002),
A three-dimensional computational study of the
aerodynamic mechanisms of insect ight, J. Exp. Biol.,
Vol. 205 No. 10, pp. 1507-18.
Sane, S. (2003), The aerodynamics of insect ight, J. Exp.
Biol., Vol. 206 No. 203, pp. 4191-208.
Sane, S. and Dickinson, M. (2002), The aerodynamic effects
of wing rotation and a revised quasi-steady model of
apping ight, J. Exp. Biol., Vol. 205 No. 8, pp. 1087-96.
Sun, M. and Tang, J. (2002), Lift and power requirements of
hovering ight in Drosophila virilise, J. Exp. Biol., Vol. 205
No. 10, pp. 2413-27.
Sussmann, H.J. and Liu, W. (2001), Limits of highly
oscillatory controls and the approximation of general paths
by admissible trajectories, Proc. of the 30th IEEE Conference
on Decision and Control, New York, NY, USA.
Triantafyllou, M., Techet, A. and Hover, F. (2004), Review
of experimental work in biomimetic foils, IEEE J. Ocean.
Eng., Vol. 29 No. 3, pp. 585-95.
Vela, P.A. (2003), Averaging and control of nonlinear systems
(with application to biomimetic locomotion), PhD thesis,
California Institute of Technology, Pasadena, CA.
Walker, J. (2002), Rotational lift: something different or
more of the same?, J. Exp. Biol., Vol. 205 No. 24,
pp. 3783-92.
Further reading
Birch, J. and Dickinson, M. (2003), The inuence of wing
wake interactions on the production of aerodynamic forces
in apping ight, J. Exp. Biol., Vol. 206 No. 13,
pp. 2257-72.
Dickinson, W. and Dickinson, M. (2004), The effect of
advance ratio on the aerodynamics of revolving wings,
J. Exp. Biol., Vol. 207 No. 24, pp. 4269-81.
Hamamoto, M., Ohta, Y., Hara, K. and Hisada, T. (2010),
Basic design strategy for stiffness distribution on a
dragony-mimicking wing for a apping micro aerial
vehicle, Adv. Robot., Vol. 24 Nos 5/6, pp. 861-77.
Khalil, H.K. (2002), Nonlinear Systems, Prentice-Hall, Upper
Saddle River, NJ.
Sanders, J.A. and Verhulst, F. (1985), Averaging methods in
Nonlinear Dynamical Systems, Springer-Verlag, NewYork, NY.
Wie, B. (1998), Space Vehicle Dynamics and Control,
AIAA Educational Series, Reston, VA.
Corresponding author
Esmaeel Khanmirza can be contacted at: e.khanmirza@
gmail.com
Nonlinear trajectory control of a apping-wing MAV
Esmaeel Khanmirza, Aghil Youse-Koma and Bahram Tarvirdizadeh
Aircraft Engineering and Aerospace Technology: An International Journal
Volume 84 Number 1 2012 5865
65
To purchase reprints of this article please e-mail: reprints@emeraldinsight.com
Or visit our web site for further details: www.emeraldinsight.com/reprints

You might also like