You are on page 1of 15

Available online at www.sciencedirect.

com

Journal of the Franklin Institute 349 (2012) 685–699


www.elsevier.com/locate/jfranklin

Super twisting control algorithm for the attitude


tracking of a four rotors UAV
L. Derafaa, A. Benallegueb,n, L. Fridmanc,1
a
Automatic Control Laboratory EMP, Bordj El Bahri 16000, Algeria
b
Engineering Systems Laboratory of Versailles, 10-12 avenue de l’Europe, 78140 Velizy, France
c
Departamento de Control Automatico, CINVESTAV-IPN Mexico D.F. AP-14-740, Mexico
Received 25 January 2011; received in revised form 19 August 2011; accepted 17 October 2011
Available online 25 October 2011

Abstract

This paper deals with the design and implementation of a nonlinear control algorithm for the
attitude tracking of a four-rotor helicopter known as quadrotor. This algorithm is based on the
second order sliding mode technique known as Super-Twisting Algorithm (STA) which is able to
ensure robustness with respect to bounded external disturbances. In order to show the effectiveness
of the proposed controller, experimental tests were carried out on a real quadrotor. The obtained
results show the good performance of the proposed controller in terms of stabilization, tracking and
robustness with respect to external disturbances.
& 2011 The Franklin Institute. Published by Elsevier Ltd. All rights reserved.

1. Introduction

Unmanned Aerial Vehicles (UAV) are attracting the interest of many researchers all over the
world [1–5]. This popularity may be attributed to their potential use in many applications such
as, search and rescue missions, surveillance, law enforcement, inspection, mapping, and aerial
cinematography. The quadrotor is a four-rotor VTOL (Vertical Take-Off and Landing) aircraft

n
Corresponding author.
E-mail addresses: derafa@lisv.uvsq.fr (L. Derafa), benalleg@lisv.uvsq.fr (A. Benallegue),
lfridman@servidor.unam.mx (L. Fridman).
1
On leave from Departamento de Ingenierı́a de Control y Robótica, División de Ingenierı́a Electrica, Facultad
de Ingenierı́a UNAM, Mexico.

0016-0032/$32.00 & 2011 The Franklin Institute. Published by Elsevier Ltd. All rights reserved.
doi:10.1016/j.jfranklin.2011.10.011
686 L. Derafa et al. / Journal of the Franklin Institute 349 (2012) 685–699

which has several advantages over the traditional helicopters in terms of manoeuvrability,
motion control and cost [1].
In order to accomplish high level human-planned missions, robust flight control systems are
required to track desired trajectories in the presence of wind or other disturbances. Most of the
existing flight control systems have been designed by applying classical synthesis techniques
(such as single-loop PID systems, root locus, Bode plots, etc.) to an approximate linear model
of the vehicle dynamics. But the trend of escalating performance, the increasing manoeuvr-
ability, the unpredictable changes in the environment, the stronger dynamic coupling and
nonlinearities necessitate more sophisticated control systems [3].
In practical applications, the position in space of the UAVs is generally controlled by an
operator through a remote-control system, while the attitude can be automatically stabilized
via an onboard controller. The attitude controller is an important feature since it allows the
vehicle to maintain a desired orientation and, hence, prevents the vehicle from flipping over
and crashing when the pilot performs the desired maneuvers [4].
A wide class of controllers have been proposed for the attitude control problem (see for
instance [4–8], and the list is not exhaustive). Most of them presented only simulation results,
and generally, the control strategies are based on simplified models without compensation
of modeling errors and external disturbances. One attempt to address this problem is given in
[9]. In this work, only simulation results are given and the controller is based on the estimation
of the disturbance which is obtained by filtering the output of a first order sliding mode
observer.
The main objective of the proposed controller is to overcome the disturbance by
using Super-Twisting Algorithm [10–12], without any a priori estimation. In fact, it has
been shown in [11,13] that this algorithm, which is based on the second order sliding
mode technique, ensures robustness with respect to modeling errors and external
disturbances while reducing the chattering phenomenon caused by all first order sliding
mode based controllers (for the general problem of chattering see [14] and for chattering
problems in quadrotor see [5]). The stability and finite time convergence characteristics
of the used algorithm have been recently proved by means of Lyapunov functions [15–18],
so the stability analysis of the proposed controller has been conducted in the same way.
In order to show the effectiveness of the controller, experimental tests were carried
out on a quadrotor. The obtained results show the good performance of the pro-
posed controller in terms of stabilization, tracking and robustness with respect to wind
perturbations.
The paper is organized as follows: in Section 2, a dynamic model for a miniature four-rotor
helicopter is developed. In Section 3, based on a simplified version of the obtained model, we
design a dynamic feedback controller which ensures robustness with respect to modeling errors
and external disturbances. In Section 4, some experimental tests are carried out to show the
effectiveness of the proposed controller in the presence of disturbances and parametric
uncertainties. Both attitude stabilization and trajectory tracking have been presented. Finally,
Section 5 concludes this work.

2. Quadrotor attitude model

The attitude dynamical model of the considered mini quadrotor, shown in Fig. 1, is
described in details by [4,19–22].
L. Derafa et al. / Journal of the Franklin Institute 349 (2012) 685–699 687

F2
F3

F1
F4 Eb
Body-fixed b
O
frame eb eb
eb

(Pitch) p

q g (Roll)
v r u
(Yaw)
a
E a e a
O w
Earth-fixed a
e x
frame
ea 


y
z
Fig. 1. Quadrotor helicopter [22].

The attitude is represented by Euler angles Y ¼ ½f,y,cT , corresponding to an aeronautical


convention. The attitude angles are respectively called Yaw angle (c rotation around z-axis),
Pitch angle (y rotation around y-axis) and Roll angle (f rotation around x-axis). The angular
_ y,
_ ¼ ½f,
velocities and accelerations are given respectively by Y _ c
_ and Y € y,
€ ¼ ½f, € c.

The equation of the attitude dynamics of the quadrotor will be

Y _
€ ¼ ðJMðYÞÞ1 ½Tprop JNðY, YÞA _ _
T ðUÞðMðYÞYÞ  ðJMðYÞYÞ ð1Þ

_ are given by
where MðYÞ and NðY, YÞ
2 3
1 0 sinðyÞ
6 7
MðYÞ ¼ 4 0 cosðfÞ sinðfÞ cosðyÞ 5
0 sinðfÞ cosðfÞ cosðyÞ

2 3
cosðyÞy_ c_
_ ¼6
NðY, YÞ 4 sinðfÞf_ y_ þ cosðfÞ cosðyÞf_ csinðfÞ
_ 7
sinðyÞy_ c_ y_ c_ 5
cosðfÞf_ ysinðfÞ
_ cosðyÞf_ ccosðfÞ
_ sinðyÞ

The matrix J is the inertia matrix of the quadrotor. Tprop is the torque vector of the
propeller system. The aerodynamic functions AT(U) are highly nonlinear and dependent
on numerous physical variables such as the angle between airspeed and the body-fixed
frame and geometric form of the helicopter. Generally, they are approximated using the
non-dimensional coefficients Ci as ATi ðUÞ ¼ 12 rCi U 2 where r is the air density [23,24]. U is
the velocity of the helicopter with respect to the air.
688 L. Derafa et al. / Journal of the Franklin Institute 349 (2012) 685–699

3. Attitude controller design

In this section, we design a controller for the attitude of the aircraft using Super Twisting
Algorithm. The objective is to ensure the convergence of the attitude positions defined by
Euler angles ffðtÞ,yðtÞ,cðtÞg to the bounded desired trajectories ffd ðtÞ,yd ðtÞ,cd ðtÞg. In this
case, let us use Eq. (1) and define the following attitude state variables:
x1 ¼ Y; _
x2 ¼ Y
Then the state-space form of this model is given by
(
x_ 1 ¼ x2
ð2Þ
x_ 2 ¼ f ðxÞ þ gðxÞu þ wðtÞ

where wðtÞ ¼ ½w1 w2 w3 T is considered as external disturbance vector, the control input is
u ¼ Tprop ¼ ½uf ,uy ,uc T , the vector f(x) is given by
_
f ðxÞ ¼ ðJMðYÞÞ1 ½JNðY, YÞðMðYÞ _  ðJMðYÞYÞ
YÞ _ ð3Þ
and the matrix g(x) is given by
gðxÞ ¼ ðJMðYÞÞ1
One can synthesize the control law forcing the state x1 to follow the desired trajectory
xd1 ðtÞ ¼ ½fd ðtÞ,yd ðtÞ,cd ðtÞT . To this end, the following assumptions are needed:
_ can be measured or estimated by on-board sensors.
Assumption 1. The signals Y, and Y
Assumption 2. The desired trajectories and their first and second time derivatives are bounded.
Assumption 3. The velocity U and the acceleration U_ of the helicopter with respect to the
air are bounded.
Assumption 4. The roll, pitch and yaw angles are limited to ðp=2ofop=2Þ, ðp=2oy
op=2Þ and ðpocopÞ. It means that the acrobatic behavior is not allowed.

According to these assumptions and to the expression of the function given by Eq. (1), it
should be noted that following inequalities are satisfied ði ¼ 1; 2,3Þ:
jw_ i ðtÞjrdi ð4Þ
According to Assumption 4, the matrix g is non-singular and its inverse is given by
g1 ¼ JMðYÞ ð5Þ
The controller is designed in order to obtain the error dynamics in the form of Super-
Twisting given in [15,18]. Its goal is to enforce the sliding mode on the manifold
s ¼ e_ þ le ð6Þ
33
where e ¼ xd1 x1 , e_ ¼ xd2 x2 ¼ x_ d1 x_ 1 and l 2 R is a diagonal positive definite matrix.
The proposed controller is given by the following:
 Z t 
u ¼ g1 x€ d1 l_e K1 jsj1=2  sgnðsÞK2 sgnðsðtÞÞ dtf ðxÞ ð7Þ
0
L. Derafa et al. / Journal of the Franklin Institute 349 (2012) 685–699 689

where the function sgnð:Þ denotes the usual sign function, the gain matrices K1 ¼ diagðk11 ,
k12 ,k13 Þ and K2 ¼ diagðk21 ,k22 ,k23 Þ, with k1i and k2i are positive gains chosen as follows:
k2i 4di
2 ði ¼ 1; 2,3Þ ð8Þ
k1i 44k2i

The closed loop error dynamics is given by


Z t
s_ ¼ K1 jsj1=2  sgnðsÞK2 sgnðsðtÞÞ dt þ wðtÞ ð9Þ
0

Let us take
z1i ¼ si
Z t
z2i ¼ k2i sgnðsi ðtÞÞ dt þ wi ðtÞ
0

w_ i ðtÞ ¼ ri ðtÞ ð10Þ


then Eq. (9) can be written in scalar form ði ¼ 1; 2,3Þ as
z_ 1i ¼ k1i jz1 j1=2  sgnðz1i Þ þ z2i
ð11Þ
z_ 2i ¼ k2i sgnðz1i Þ þ ri ðtÞ

The proof of finite time convergence to zero of the variables z1i and z2i is explicitly given
in appendix and this results are taken from [16]. It can be concluded that if the conditions
on the gains given by Eq. (8) are satisfied, we obtain si ¼ 0 in finite time. Therefore, it can
be concluded, according to Eq. (6), that limt-1 e ¼ 0 and limt-1 e_ ¼ 0.
Remark 1. According to Eq. (10) and the finite time convergence to zero of the signals z2i
ði ¼ 1; 2,3Þ, it can be concluded that the perturbations wi(t) are estimated in finite time as
follows:
Z tZT0
wi ðtÞ ¼ k2i sgnðsi ðtÞÞ dt
0

and is, therefore, compensated by the controller.


Remark 2. In real applications, the control inputs in Eq. (2) are the four rotor forces
(F1, F2, F
P3 and F4) of the aircraft (see Fig. 1). It is, thus, necessary that the total thrust
Fprop ¼ i ¼ 14 Fi is available in order to use it for the calculation of these forces. One
manner to design it, is to control the altitude (z-position) of the quadrotor. However, for
the sake of simplicity of implementation in this work, Fprop has been fixed to compensate
the gravity force.

4. Experimental results

In order to validate the proposed controller, we implemented the control law given by
Eq. (7) on a PC Pentium II at 200 MHz, equipped with a dSpace DS1103 PPC real-time
controller board using Matlab and Simulink software as shown in Fig. 2. The sampling
frequency has been fixed to 1 kHz. The mechanical structure of the quadrotor is that of the
690 L. Derafa et al. / Journal of the Franklin Institute 349 (2012) 685–699

Fig. 2. Experimental setup.

Table 1
Quadrotor physical parameters.

Parameter Description Value Unit

m Mass 0.420 kg
d Distance 20.5 cm
c Proportionality factor 1.08 cm
Ix Roll inertia ðfÞ 3.83  103 kg m2
Iy Pitch inertia ðyÞ 3.83  103 kg m2
Iz Yaw inertia ðcÞ 7.13  103 kg m2

four-rotor mini-helicopter manufactured by Draganfly Innovations, Inc. (http://www.


rctoys.com). The physical parameters of the used quadrotor are given in Table 1 [25].
The objective of this experiment is to safely test the proposed attitude controller, so we
decided to use a stationary ball joint base, as shown in Fig. 2. This base gives the aircraft
unrestricted yaw movement and around 7301 of pitch and roll, while restricting the aircraft to
a fixed point in the three-dimensional space [4]. To measure the aircraft angles and the
angular velocity we used an Inertial Measurement Unit (IMU) XSENS MTI-28A53G35
(http://www.xsens.com). The IMU is connected to the dspace DS1103 serial communica-
tion port EIA-RS-232.
The four DC permanent-magnet mini motors are current amplified with intelligent
microcomputer speed controllers of type IMCS 25 and driven by PWM signals.
To explore the effectiveness and the robustness of the proposed STA controller, four
experiments have been performed on the quadrotor. Experiment 1 involves the aircraft
attitude stabilization from some given initial angles. Initial conditions are set for the
aircraft angles as f ¼ 231,y ¼ 131 and the yaw c ¼ 201. In experiment 2, attitude tracking
L. Derafa et al. / Journal of the Franklin Institute 349 (2012) 685–699 691

test was carried out. The reference signals are of sinusoidal form with magnitude equal to
0.1 rad for the roll, 0.2 rad for the pitch and 0.3 rad for the yaw. In experiment 3, the attitude
is externally disturbed to explore the disturbance rejection performance in stabilization around
zero and in tracking case. In experiment 4, the quadrotor is in free flight situation where the
attitude is stabilized around zero with the altitude controlled at around 1 m over the floor by
an operator through a remote-control system.
For this application the chosen gains are given as
pffiffiffi
li ¼ 3 ði ¼ 0; 1,2Þ, K1 ¼ 4:5 d, K2 ¼ 1:1d
with d ¼ diag½12; 12,8.

30 Roll
Pitch
Yaw
20
Quadrotor Angles (Deg)

10

−10

−20

−30

10 15 20 25 30 35 40 45 50
Time (sec)

Fig. 3. Euler angles f, y, c (stabilization test).

1
Uφ (N.m)

−1
10 15 20 25 30 35 40 45 50
Time (sec)
1
Uθ (N.m)

−1
10 15 20 25 30 35 40 45 50
Time (sec)
1
Uψ (N.m)

−1
10 15 20 25 30 35 40 45 50
Time (sec)

Fig. 4. Controller outputs (stabilization test).


692 L. Derafa et al. / Journal of the Franklin Institute 349 (2012) 685–699

4.1. Stabilization experiment

Figs. 3 and 4 show the attitude response, the corresponding controller outputs and the
applied forces. One can notice the effectiveness of the proposed controller. The pronounced
frequency oscillations in the control signals are due to sensor noise.

4.2. Tracking experiment

Attitude tracking experiment has been carried out. Figs. 5 and 6 show the attitude response,
the corresponding controller outputs and the applied forces signals. The obtained results
clearly show the effectiveness of the controller.

30

Pitch
20
Quadrotor Angles (Deg)

10 Roll

−10

−20

−30 Yaw

0 10 20 30 40 50
Time (sec)

Fig. 5. Euler angles f, y, c (tracking test).

0.5
Uφ (N.m)

0
−0.5
0 10 20 30 40 50
Time (sec)

0.5
Uθ (N.m)

0
−0.5
0 10 20 30 40 50
Time (sec)
0.5
Uψ (N.m)

0
−0.5
0 10 20 30 40 50
Time (sec)

Fig. 6. Controller outputs (tracking test).


L. Derafa et al. / Journal of the Franklin Institute 349 (2012) 685–699 693

4.3. Disturbance rejection experiment

Figs. 7 and 8 illustrate the Euler angles response and the corresponding controller outputs.
As shown in Fig. 7 the disturbances are externally applied at time instants 10 s, 16 s for the
pitch and 22 s for the roll and the yaw angle.
Figs. 9 and 10 show respectively the Euler angles response where the attitude is externally
disturbed at time 24 s and 35 s for the roll and 33 s for the yaw and the corresponding controller
outputs.

Pitch
30 Roll
Yaw
20
Quadrotor Angles (Deg)

10

−10

−20

−30

0 10 20 30 40 50
Time (sec)

Fig. 7. Euler angles f, y, c (disturbance test).


Uφ (N.m)

0.5
0
−0.5
0 10 20 30 40 50
Time (sec)

0.5
Uθ (N.m)

0
−0.5
0 10 20 30 40 50
Time (sec)
Uψ (N.m)

0.5
0
−0.5
0 10 20 30 40 50
Time (sec)

Fig. 8. Controller outputs (disturbance test).


694 L. Derafa et al. / Journal of the Franklin Institute 349 (2012) 685–699

30
Pitch
20

Quadrotor Angles (Deg) 10 Roll

−10

−20

Yaw
−30

0 10 20 30 40 50
Time (sec)

Fig. 9. Euler angles f, y, c (tracking disturbance test).

0.5
Uφ (N.m)

0
−0.5
0 10 20 30 40 50
Time (sec)
Uθ (N.m)

0.5
0
−0.5
0 10 20 30 40 50
Time (sec)
Uψ (N.m)

0.5
0
−0.5
0 10 20 30 40 50
Time (sec)

Fig. 10. Controller outputs (tracking disturbance test).

4.4. Free flight stabilization test


In order to show the effectiveness of the controller in real situation a free flight
stabilization test has been carried out (see Fig. 11). A disturbance is applied by hand to the
yaw angle c at time t ¼ 8 s. Figs. 12 and 13 show respectively the Euler angles response and
the corresponding controller outputs. It can be noticed that the controller is reacting
against the disturbance in opposite direction in order to compensate it.
L. Derafa et al. / Journal of the Franklin Institute 349 (2012) 685–699 695

Fig. 11. Free flight quadrotor.

Attitude Stabilization test (Super Twisting Algorithm)


30
φ
θ
25
ψ
Quadrotor Angles (Deg)

20

15

10

−5

−10
0 10 20 30 40 50
Time (sec)

Fig. 12. Euler angles f, y, c (free flight test).

5. Conclusion

In this paper, a super twisting controller algorithm has been proposed and successfully
implemented on a small scale quadrotor aircraft. The controller has been designed using
second order sliding mode approach in order to avoid chattering phenomenon and
to ensure robustness with respect to model uncertainties and external disturbances.
696 L. Derafa et al. / Journal of the Franklin Institute 349 (2012) 685–699

0.1

Uφ (N.m)
0

−0.1
0 10 20 30 40 50
Time (sec)
0.1
Uθ (N.m)

−0.1
0 10 20 30 40 50
Time (sec)
0.5
Uψ (N.m)

−0.5
0 10 20 30 40 50
Time (sec)

Fig. 13. Controller outputs (free flight test).

The experimental results obtained on a quadrotor system clearly show the effectiveness of
the proposed controller in the stabilization, tracking and disturbance rejection cases.

Appendix A

A.1. Lyapunov stability analysis of the STA [16]

Consider the standard Super-Twisting Algorithm (STA) with a perturbation term


(
z_ 1 ¼ k1 jz1 j1=2 sgnðz1 Þ þ z2
ðA:1Þ
z_ 2 ¼ k2 sgnðz1 Þ þ rðt,zÞ

where z1 2 R and z2 2 R and the perturbation term r is uniformly bounded ðjrjodÞ.


Let us prove the stability of the equilibrium point ðz1 ,z2 Þ ¼ ð0; 0Þ. The proof is stated as
in [15–18]. Consider the Lyapunov function presented in [15]
V ¼ zT Pz ðA:2Þ
where z ¼ ½jz1 j1=2 sgnðz1 Þ,z2 T and P is a positive definite matrix.
Notice that V ðz,tÞ is continuous and differentiable except when z1 ¼ 0. In fact, when
z1 a 0, V_ exists and is negative definite. However, before arriving at the equilibrium point
ðz1 ,z2 Þ ¼ ð0; 0Þ, the solution of system (A.1) crosses the line z1 ¼ 0 when z2 a 0. This means
that the derivative of the Lyapunov function exists almost everywhere while V(t) decreases
until the system reaches the equilibrium. As presented in [15], V(t) is a strong Lyapunov
function for (A.1) in the form of Eq. (A.2).
Moreover, this Lyapunov function is positive definite but radially unbounded
lmin ðPÞJzJ2 rV rlmax ðPÞJzJ2 ðA:3Þ
where JzJ22 ¼ jz1 j þ z22 represents the Euclidian norm of z.
L. Derafa et al. / Journal of the Franklin Institute 349 (2012) 685–699 697

The construction of suitable positive definite matrices P ¼ PT , provided in [15], is based


on the following algebraic LMI equation:
" #
AT P þ PA þ EP þ d2 C T C PB
o0 ðA:4Þ
BT P 1
where
2 3
1 1  
 k1 0
A¼4 2 2 5; B¼ ; C ¼ ½1 0
k2 0 1

with k1 and k2 are positive gains.


Using the vector z ¼ ½jz1 j1=2 sgnðz1 Þ,z2 T , the system (A.1) can be rewritten as
1
z_ 1 ¼ ~
ðAz þ BrðtÞÞ ðA:5Þ
jz1 j
where the transformed perturbation rðt,zÞ ~ ~
¼ jz1 j rðt,zÞ satisfies jrðt,zÞjrdjz 1 j. As a
~ ¼ r~ 2 ðt,zÞ þ d2 z21 Z0.
consequence, oðr,zÞ
As k1 and k2 are positive gains, the system (A,B,C) is observable and controllable, so we
can use the bounded-real lemma [26] to determine the condition on the gains k1 and k2
instead of using the circle criterion given in [16]. It is shown that the Linear Matrix
Inequality (A.4) is feasible if and only if the linear system defined by HðsÞ ¼ dCðsI AÞ1 B
is nonexpansive, i.e.
maxjHðjoÞjo1
o

This implies the following condition:


1
maxjGðjoÞjo
o d
where
1
GðsÞ ¼ CðsIAÞ1 B ¼ 2
s2 þ 12 k1 s þ 12 k2
The previous equality yields to two conditions on gains. By choosing one of them
1
maxjGðjoÞj ¼ if k12 44k2
o k2
we can then deduce conditions on gains k1 and k2 as follows:
k2 4d

k12 44k2 ðA:6Þ


Consider the Lyapunov function defined by Eq. (A.2). Its derivative writes
" #
1  T AT P þ PA PB  
V_ ðzÞ ¼ z r~ z r~
jz1 j BT P 0
( " # )
1  T AT P þ PA PB  
r z r~ z r~ þ oðr,zÞ
~
jz1 j BT P 0
698 L. Derafa et al. / Journal of the Franklin Institute 349 (2012) 685–699

" #
1  T AT P þ PA þ d2 C T C PB  
r z r~ z r~
jz1 j BT P 1
" #
1  T AT P þ PA þ EPEP þ d2 C T C PB  
r z r~ z r~
jz1 j BT P 1
E T
r z Pz
jz1 j

Finally
E T E
V_ r z Pz ¼  V ðzÞ ðA:7Þ
jz1 j jz1 j
From Eq. (A.3), we deduce the following inequality:
V 1=2 ðzÞ
jz1 jrJzJ2 r 1=2
lmin fPg
We can then conclude that V_ satisfies
V_ raV 1=2 ðzÞ
where
1=2
a ¼ Elmin fPg ðA:8Þ
The previous result guarantees the finite time convergence of vector z ¼ ½z1 ,z2 T to zero.
This time is bounded by
2V 1=2 ðzð0ÞÞ
T0 ¼ ðA:9Þ
a
where zð0Þ is the initial value of z and a is given by Eq. (A.8).

References

[1] V. Mistler, A. Benallegue, N.K. M’Sirdi, Linearisation exacte et decouplage entrees-sorties comparaison
entre l’helicoptere standard et l’helicoptere 4 rotors, in: Conference Internatinale Francophone d’Automatique
(CIFA’2002), Nantes, France, 8–10 juillet 2002.
[2] M. Chen, M. Huzmezan, A simulation model and h1 loop shaping control of a quad rotor unmanned
air vehicle, in: M.H. Hamza (Ed.), Modelling, Simulation, and Optimization, IASTED/ACTA Press, 2003,
pp. 320–325.
[3] S. Bouabdallah, P. Murrieri, R. Siegwart, Design and control of an indoor micro quadrotor, in: Proceedings
of IEEE International Conference on Robotics and Automation, New Orlean, USA, 2004.
[4] A. Tayebi, S. McGilvray, Attitude stabilization of a vtol quadrotor aircraft, IEEE Transactions on Control
Systems Technology 14 (3) (2006).
[5] T. Madani, A. Benallegue, Backstepping sliding mode control applied to a miniature quadrotor flying robot,
in: Proceedings of the 32nd Annual Conference of the IEEE Industrial Electronics Society (IECON’ 06),
Paris, France, November 7–10, 2006.
[6] O.-E. Fjellstad, T.I. Fossen, Comments on the attitude control problem, IEEE Transactions on Automatic
Control 39 (1994).
[7] S.M. Joshi, A.G.Kelkar, J.T.-Y.Wen, Robust attitude stabilization of spacecraft using nonlinear quaternion
feedback, IEEE Transactions on Automatic Control 40 (1995).
[8] F. Lizarraide, J.T. Wen, Attitude control without angular velocity measurement: a passivity approach, IEEE
Transactions on Automatic Control 41 (1996).
L. Derafa et al. / Journal of the Franklin Institute 349 (2012) 685–699 699

[9] L. Besnard, Y.B. Shtessel, B. Landrum, Control of a quadrotor vehicle using sliding mode disturbance
observer, in: Proceedings of the 2007 American Control Conference, New York City, USA, July 11–13, 2007.
[10] A. Levant, Sliding order and sliding accuracy in sliding mode control, International Journal Control 58 (6)
(1993) 1247–1263.
[11] A. Levant, Robust exact differentiation via sliding mode technique, Automatica 34 (3) (1998) 379–384.
[12] A. Levant, High-order sliding modes: differentiation and output-feedback control, International Journal
Control 76 (9–10) (2003) 924–941.
[13] J. Davila, L.Fridman, A.Levant, Second order sliding mode observer for mechanical systems, IEEE Transaction
on Automatic Control 50 (11) (2005).
[14] V.I. Utkin, Sliding Mode and Their Application in Variable Structure Systems, Mir, Moscou, 1978.
[15] J.A. Moreno, M. Osorio, A lyapunov approach to second-order sliding mode controllers and observers, in:
Proceedings of 47th IEEE Conference on Decision and Control (CDC 2008), Cancun, Mexico, December 9–11,
2008.
[16] J.A. Moreno, A linear framework for the robust stability of a generalized super-twisting algorithm, in:
Proceedings of Sixth IEEE Conference on Electrical Engineering, Computing Science and Automatic Control
(CCE 2009), Toluca, Mexico, November 10–13, 2009.
[17] A. Davila, J.A. Moreno, L. Fridman, Optimal lyapunov function selection for reaching time estimation of
super twisting algorithm, in: Proceedings of 48th IEEE Conference on Decision and Control (CDC 2009),
Shanghai, PR China, December 16–18, 2009.
[18] A. Davila, J.A. Moreno, L. Fridman, Global non homogenous super-twisting controller for the quasi-linear
systems with unbounded uncertainties: a lyapunov design, in: Proceedings of the American Control
Conference (ACC 2010), Baltimore, MD, USA, June 30–July 2, 2010.
[19] T. Hamel, R. Mahony, R. Lozano, J. Ostrowski, Dynamic modeling and configuration stabilization for an
x4-flyer, in: Proceedings of IFAC World Congress, Barcelona, Spain, July 2002.
[20] G. Cai, B.M. Chen, K. Peng, M. Dong, T.H. Lee, Modeling and control of the yaw channel of a uav
helicopter, IEEE Transactions on Industrial Electronics 55 (9) (2008).
[21] P. McKerrow, Modelling the draganflyer four-rotor helicopter, in: IEEE International Conference on
Robotics and Automation (ICRA’04), New Orleans, LA, 2004.
[22] T. Madani, A. Benallegue, Sliding mode observer and backstepping control for a quadrotor unmanned aerial
vehicles, in: Proceedings of American Control Conference, New York City, USA, July 11–13, 2007.
[23] A. Gessow, G. Myers, Aerodynamics of the Helicopter, Frederick Ungar Publishing Co, New York, 1967.
[24] W. Johnson, Helicopter Theory, Dover, New York, 1994.
[25] L. Derafa, T. Madani, A. Benallegue, Dynamic modelling and experimental identification of four rotors
helicopter parameters, in: Proceedings of the IEEE International Conference on Industrial Technology
(ICIT’ 06), Mumbai, India, December 15–17, 2006.
[26] S. Boyd, L. El Ghaoui, E. Feron, V. Balakrishnan, Linear Matrix Inequalities in System and Control Theory,
Studies in Applied Mathematics, vol. 15, SIAM, Philadelphia, PA, 1994.

You might also like