See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/339045499

Robust Integral Sliding Mode Control Design for Stability Enhancement of

Under-actuated Quadcopter

Article in International Journal of Control Automation and Systems · February 2020

DOI: 10.1007/s12555-019-0302-3

CITATIONS READS
82 2,987

5 authors, including:

Safeer Ullah Adeel Mehmood

Quaid-e-Azam College of Engineering and Technology Sahiwal University of Hull
36 PUBLICATIONS 357 CITATIONS 46 PUBLICATIONS 876 CITATIONS

SEE PROFILE SEE PROFILE

Qudrat Khan Sakhi Rehman

COMSATS University Islamabad University of Science & Technology Bannu
108 PUBLICATIONS 1,135 CITATIONS 1 PUBLICATION 81 CITATIONS

SEE PROFILE SEE PROFILE

All content following this page was uploaded by Safeer Ullah on 08 June 2020.

The user has requested enhancement of the downloaded file.

International Journal of Control, Automation and Systems 18(7) (2020) 1671-1678 ISSN:1598-6446 eISSN:2005-4092
http://dx.doi.org/10.1007/s12555-019-0302-3 http://www.springer.com/12555

Robust Integral Sliding Mode Control Design for Stability Enhancement

of Under-actuated Quadcopter
Safeer Ullah, Adeel Mehmood, Qudrat Khan, Sakhi Rehman, and Jamshed Iqbal* 

Abstract: In this paper, a robust backstepping integral sliding mode control (RBISMC) technique is designed for
the flight control of a quadcopter, which is an under-actuated nonlinear system. First, the mathematical model of
this highly coupled and under-actuated system is described in the presence of dissipative drag forces. Second, a
robust control algorithm is designed for the derived model to accurately track the desired outputs while ensuring
the stability of attitude, altitude and position of the quadcopter. A step by step mathematical analysis, based on the
Lyapunov stability theory, is performed that endorses the stability of both the fully-actuated and under-actuated sub-
systems of the aforementioned model. The comparison of proposed RBISMC control algorithm, with fraction order
integral sliding mode control (FOISMC), affirms the enhanced performance in terms of faster states convergence,
improved chattering free tracking and more robustness against uncertainties in the system.

Keywords: Backstepping, integral sliding mode control, quadcopter, under-actuated, unmanned aerial vehicle.

1. INTRODUCTION and rotational position of a quadcopter. The response time

observed minor overshoot whereas the steady-state error
The quadcopter, a special case of Unmanned Ariel Ve- was nearly equal to zero. Cowling et al. [4] applied linear
hicles (UAV), has distinctive characteristics of flying in quadratic regulator (LQR) controller to achieve an exact
nearly all types of weather condition. These vehicles have reference tracking in simulation even in the presence of
attenuating ability against the undesirable torques pro- disturbances. However, PID and LQR based control tech-
duced by the reverse rotation of propellers installed on the niques have some performance limitations when applied
opposite sides. High degrees of maneuverability enable to the quadcopter with unmodeled system’s dynamics [5].
them to hover with enhanced payload capacity and faster In [6], a nonlinear technique i.e., integral backstepping
mobility [1]. Their longitude, altitude and attitude are con- was used to control the angular rotations and linear trans-
trolled by four contra-rotating propellers. The quadcopter lations of quadcopter in the presence of aerodynamic ef-
has a wide range of applications in defense, military, me- fects. Shi et al. [7] presented the design of fractional-order
dia, agriculture and other industries including but not lim- backstepping SMC for the control of attitude and altitude
ited to; mining, oil and gas, logistics and transportation. of the OS4 quadcopter. Their project provided the capabil-
The dynamic model of the aforesaid aerial vehicle is in- ity to hover and collision avoidance. It is evident that the
herently known as an under-actuated for the fact that they order reduction in SMC algorithm provides the insensitiv-
have lesser number of control inputs than the degrees of ity to matched disturbance (i.e., matched disturbance re-
freedom. Therefore, the control design of such a system jection) [8], nevertheless, the high-frequency oscillations
is a very challenging task [2]. In this regards, a lot of issue may degrade the performance of system [9]. This
work has been conducted on the control design and imple- problem is overcame via different advanced higher-order
mentation of the quadcopter. Some relevant work is men- SMC techniques (see for instance [10–12]). However, the
tioned here. Proportional Integral Derivative (PID) con- key performance factor i.e. robustness was unfortunately
trol is the most widely used linear control scheme in the lost while alleviating the problem of high-frequency os-
industry. For example, in [3], a gain-scheduled PID was cillations. In the existing literature, some differentiators
applied for the reference tracking of both transnational were used to overcome the unwanted effects of uncer-

Manuscript received April 23, 2019; revised August 1, 2019 and October 22, 2019; accepted November 15, 2019. Recommended by
Associate Editor Guangdeng Zong under the direction of Editor Hamid Reza Karimi.

Safeer Ullah and Adeel Mehmood are with the Department of Electrical & Computer Engineering, COMSATS University, Islamabad, Pak-
istan (e-mails: [email protected], [email protected]). Qudrat Khan is with Center for Advanced Studies in Telecommu-
nication, COMSATS University, Islamabad, Pakistan (e-mail: [email protected]). Sakhi Rehman is with Electrical Engineering
Department, University of Science and Technology, Bannu, Pakistan (e-mail: [email protected]). Jamshed Iqbal is with the Depart-
ment of Electrical and Electronic Engineering, University of Jeddah, Saudi Arabia and Department of Electrical Engineering, FAST National
University of Computer and Emerging Sciences, Islamabad, Pakistan (e-mail: [email protected]).
* Corresponding author.

ICROS,
c KIEE and Springer 2020
1672 Safeer Ullah, Adeel Mehmood, Qudrat Khan, Sakhi Rehman, and Jamshed Iqbal

tainties and provide a precise estimation of the outputs’

derivatives of the system [13, 14]. Apart from this, ma-
jority of the nonlinear systems may result in undesirable
behavior of the reaching phase of SMC (and even in the
instability of the system’s dynamics) [15]. Therefore, a
reaching-phase free ISMC scheme was presented in [16]
that provided both the chattering reduction and robustness
improvement. It is also worthy to mention that [16] only
controlled the altitude subsystem that is not much suffi-
cient for the full flight maneuver of a quadcopter. Conse-
quently, the above literature discussions motivate to pro-
pose a nonlinear control scheme that must provide the full
flight control of a quadcopter with reduced chattering and
improved robustness simultaneously. Fig. 1. The inertial and body frames of a quadcopter.
In this paper, a synthesized approach of backstepping
and integral sliding mode strategies is presented for a lational states (x, y, z), three rotational states (ϕ, ϑ , ψ) and
highly nonlinear and complex under-actuated quadcopter their derivatives (ẋ, ẏ, ż, ϕ̇, ϑ̇ , ψ̇) as shown in Fig. 1.
system. The main control task is tracking the system on The rotational motion represents the mathematical ex-
the desired trajectories. In other words, the system’s out- pression of roll ϕ, pitch ϑ and yaw ψ while the transla-
put (x, y, z, ϕ, ϑ , ψ) are needed to track the desired tra- tional motion is the combination of position in the hori-
jectories for both the fully-actuated and under-actuated zontal plane (x, y) and altitude z. The quadcopter’s motion
control loops of the quadcopter system. The main contri- is considered as a rigid body motion which is described by
butions in this article are three fold. At first, the overall inertial earth frame [X,Y, Z] and a fixed body frame [x, y, z].
model of the quadcopter with under-actuated and fully- The explicit expression of the rotational motion dynamics
actuated loops is considered. Secondly, we have utilized [18], in terms of angular positions and velocities, are ex-
the good features of the two nonlinear control techniques pressed as follows
i.e., (in case of backstepping) easy nonlinear control de- ⎫
IR ϑ̇ ω̄ (Iy − Iz ) bl K1 ⎪
sign for each state and (in case of sliding modes) the ϕ̈ = + ϑ̇ ψ̇ + U2 − ϕ̇ , ⎪
2
⎪
⎪
invariance property. Moreover, instead of conventional Ix Ix Ix Ix ⎪
⎪
⎪
⎬
SMC, an ISMC approach is used to enhance the robust- −IR ϕ̇ ω̄ (Iz − Ix ) bl K2
ϑ̈ = + ϕ̇ ψ̇ + U3 − ϑ̇  ,
2
ness from the very start of the process i.e., invariance Iy Iy Iy Iy ⎪
⎪
⎪
⎪
property is ensured from the initial time instant which (Ix − Iy ) d K3 ⎪
⎪
ψ̈ = ϕ̇ ϑ̇ + U4 − ψ̇ ,
2 ⎪
⎭
never happens in case of conventional SMC. In addition,
Iz Iz Iz
the unwanted chattering phenomena is also alleviated by (1)
the use of a strong reachability law. Finally, the proposed
RBISMC is compared with the standard control strategy where b is the thrust coefficient, d is the drag factor, IR is
i.e., FOISMC presented in [17]. The result comparison the rotor inertia, l is the arm’s length, ω̄ = ω1 + ω3 − ω2 −
demonstrates that the proposed control scheme is more ap- ω4 is an aerodynamic disturbance, ωi (i = 1, 2, 3, 4) repre-
pealing in term of regulation and trajectory tracking. The sents the angular speed of each brush-less DC motor, K1 ,
rest of the paper is organized as follows: In Section 2, the K2 , K3 are the coefficients of aero-dynamic frictions, U2 =
mathematical model of a quadcopter system is presented. −ω22 + ω42 , U3 = ω12 − ω32 and U4 = −ω12 + ω22 − ω32 + ω42
The RBISMC nonlinear control strategy is derived in Sec- are the thrust produced by the respective motors which are
tion 3. This section also includes the simulations and dis- required for the motion along ϕ, ϑ and ψ, respectively.
cussions. Finally, Section 4 concludes the paper. Now, by following the Euler-Newtonian equations of
motion [18], one can represent the translational motion by
2. MATHEMATICAL MODEL OF QUADCOPTER the following equations
b K̄d
Mathematical model of the aforementioned system is ẍ = (sin ψ sin ϕ + cos ψ sin ϑ cos ϕ) U1 − x ẋ2 ,
m m
essential for the design of control law which must be ca-
b K̄ d
pable enough to ensures the closed loop stability. Vari- ÿ = (sin ψ sin ϑ cos ϕ − cos ψ sin ϕ) U1 − y ẏ2 ,
ous modeling based control strategies for this system have m m
been discussed recently by many researchers, as described b K̄d
z̈ = (cos ϑ cos ϕ) U1 − g − z ż2 , (2)
previously. This section describes the rigid body dynam- m m
ics, kinematics of fixed and body reference frames and where g is the gravitational acceleration, m is mass of the
forces applied to it. The quadcopter system has three trans- quadcopter, b is the thrust factor of the propeller and U1
 Robust Integral Sliding Mode Control Design for Stability Enhancement of Under-actuated Quadcopter 1673

is the control input for the translational motion genera- ϑ ). The main objective of this work is to follow a full
tion. The drag force constant is defined by K̄d = ρAKd , flight desired trajectory by the quadcopter in the presence
where ρ is mass density of the airflow that depends on the of matched uncertainties. For this purpose, an RBISMC
height above the sea level, A is the specific area of the control strategy is presented in Section 3.
quadcopter, ẋ, ẏ and ż are the speeds of translational axis
in space relative to the air flow and Kd is the coefficient of 3. CONTROL DESIGN
air drag force and it depends on the structure of body.
Now, the overall mathematical model of a quadcopter Since the quadcopter is an under-actuated system with
system, in state space, can be described as follows: strong coupling effect among the rotors’ inputs, therefore,
 a nonlinear RBISMC scheme is proposed to achieve the
ż1 = z2 , ż2 = (cos ϑ1 cos ϕ1 )b0 U1 +Δz −g−Kz z22 , stability of both fully-actuated and under-actuated subsys-
(3a)
ψ̇1 = ψ2 , ψ̇2 = a5 ϑ2 ϕ2 +b3 U4 +Δψ −Kψ ψ22 , tems.
⎫
ẋ1 = x2 , ẋ2 = ux b0 U1 − Kx x22 , ⎪
⎪ 3.1. Fully-actuated subsystem
⎪
⎪
ẏ1 = y2 , ẏ2 = uy b0 U1 − Ky y2 ,
2 ⎬ The error between the reference and actual trajectory
ϕ̇1 = ϕ2 , ϕ̇2 = a1 ψ2 ϑ2 + a2 ω̄ ϑ2 + b1 U2 − Kϕ ϕ22 , ⎪⎪
⎪
serve as control variable in the controller, which in turn
⎪
2 ⎭
generates the control inputs. The control inputs, when ap-
ϑ̇1 = ϑ2 , ϑ̇2 = a3 ψ2 ϕ2 + a4 ω̄ ϕ2 + b2 U3 − Kϑ ϑ2 , plied to the actuator, vary the speed of the relative mo-
(3b) tors. Consequently, the desired motion of a quadcopter is
achieved. For this purpose, the reference tracking errors
where ux = (sin ψ sin ϕ + cos ψ sin ϑ cos ϕ), uy =
are defined as follows:
(sin ψ sin ϑ cos ϕ − cos ψ sin ϕ), b0 = mb , b1 = Ilx , b2 = Ily , 
(Iy −Iz ) εz1 = z1 − zd ⇒ ε̇z1 = z2 − żd ,
b3 = Iz , a1 =
1
Ix , a2 = IIRx , a3 = (IzI−I x)
, a4 = −I
Iy ,
R
(4)
εψ1 = ψ1 − ψd ⇒ ε̇ψ1 = ψ2 − ψ̇d ,
y
(Ix −Iy ) K(1,2,3)
a5 = Iz , K(x,y,z) = ρAK m , K(ϕ,ϑ ,ψ) = I(x,y,z) . In addi-
d

tion, Δz and Δψ are the matched uncertainties. where zd and ψd are the reference outputs. By taking the
Assumption 1: The uncertainty terms Δz and Δψ are derivatives of Lyapunov candidate functions Vz1 = 12 εz21
assumed to be matched and norm bounded by their re- and Vψ1 = 12 εψ2 1 along (3b) and (4), one has
positive constants λz and λψ i.e., |Δz | ≤ λz and 
spective
 V̇z1 = εz1 ε̇z1 ⇒ V̇z1 = εz1 (z2 − żd ),
Δψ  ≤ λψ . (5)
V̇ψ1 = εψ1 ε̇ψ1 ⇒ V̇ψ1 = εψ1 (ψ2 − ψ̇d ).
The system parameters, used in the simulation, are
given in Table 1. Now, the variables z2 and ψ2 are treated as virtual control
It is worthy to mention that for the ease of control law inputs in such a way that they will ensure the derivative of
design, the proposed control law is derived by dividing the Lyapunov function negative definite. Thus, the following
system model into two subsystems i.e., the fully-actuated choices lead us
subsystem and the under-actuated subsystem. In the fully- z2 = żd − κz1 εz1 , ψ2 = ψ̇d − κψ1 εψ1 . (6)
actuated subsystem, the number of outputs (z, ψ) are equal
to the number of control inputs (U1 , U4 ), whereas in the By substituting (7) in (5), one has
under-actuated subsystem, the number of control inputs
V̇z1 = −κz1 εz21 , V̇ψ1 = −κψ1 εψ2 1 ,
U2 and U3 are less than the number of outputs (x, y, ϕ,
where κz1 and κψ1 are the positive design constants.
Table 1. Parameters of quadcopter system. Now, proceeding toward the next step, we will consider
the aforementioned z∗2 and ψ2 (in (6)) as new modified
Parameter Symbol Value Unit references for the state z2 and ψ2 in the second step
Quadcopter mass m 0.650 Kg
z2 = εz2 + żd − κz1 εz1 , ψ2 = εψ2 + ψ̇d − κψ1 εψ1 , (7)
Ix =Iy 7.5 × 10−3
Inertia constants kg.m2
Iz 1.3 × 10−2 where εz2 = z2 − z2 and εψ2 = ψ2 − ψ2 .
Thrust coeff. b 3.13 × 10−5 N.s2 At the second step, the interest is that z2 should track z∗2
Drag factor d 7.5 × 10−7 N.m.s2 and, similarly, ψ2 should track ψ2 . In order to meet these
Rotor inertia Ir 6 × 10−5 kg.m2 requirements, integral manifolds of the following form are
Arm length l 0.23 m designed
Kx =Ky 5.56 × 10−4
Drag coeff. N/m/s βz2 = z2 − z2 + zz , βψ2 = ψ2 − ψ2 + zψ , (8)
Kz 6.35 × 10−4
 
Kϕ = Kϑ 5.56 × 10−4 where zz = γz εz2 dt, zψ = γψ εψ2 dt and γz and γψ are the
Aero coeff. N/rad/s
Kψ 6.35 × 10−4 positive constants.
1674 Safeer Ullah, Adeel Mehmood, Qudrat Khan, Sakhi Rehman, and Jamshed Iqbal

Remark 1: At this stage, the objective of the proposed ≤ εz1 εz2 − κz1 εz21 − κz2 βz22 − (κz3 − |Δz |)|βz2 |,
control scheme is to drive the outputs (z2 , ψ2 ) of the fully- V̇z2 ≤ εz1 εz2 − κz1 εz21 − κz2 βz22 − ηz |βz2 |,
actuated subsystem on the desired trajectory, which is pos-
sible by steering their sliding manifolds to zero in finite V̇ψ2 = εψ1 εψ2 − κψ1 εψ2 1 − κψ2 βψ22
time with the help of control inputs U1 = U10 + U11 and + (Δψ − κψ3 sign(βψ2 ))βψ2
U4 = U40 + U41 . Note that U10 and U40 , are the continuous ≤ εψ1 εψ2 − κψ1 εψ2 1 − κψ2 βψ22 − (κψ3 − |Δψ |)|βψ2 |,
control components which are designed via pole place-
ment method. These components governs the system dy- V̇ψ2 ≤ εψ1 εψ2 − κψ1 εψ2 1 − κψ2 βψ22 − ηψ |βψ2 |. (13)
namics in sliding mode whereas the other components U11
These inequalities (13) remain true only if κz3 − |Δz | ≥ ηz
and U41 are designed by integral sliding mode approach
and κψ3 − |Δψ | ≥ ηψ , where ηz and ηψ are small positive
such that these components establish sliding mode from
numbers. The control components U10 and U40 are used
the very beginning. The design of these control compo-
to steer their respective mismatches εz2 and εψ2 to zero,
nents, for their respective dynamics, is presented in the
which result in vanishing of the terms εz1 εz2 and εψ1 εψ2
forthcoming theorem.
(see [19] for more detail). Consequently, it confirms the
Theorem 1: Consider that the fully-actuated subsys- negative definiteness of the augmented Lyapunov func-
tem (3b) remains true subject to the Assumption 1. If the tions V̇z2 and V̇ψ2 . Now, moving a step back, the backstep-
backstepping procedure based virtual control laws z2 and ping based virtual controllers z2 and ψ2 ensured the en-
ψ2 and the integral sliding manifolds are chosen accord- forcement of z1 and ψ1 to zd and ψd , respectively. Hence,
ing to (7) and (8), respectively, then the following control this results in the asymptotic convergence of the states of
laws will ensure finite time sliding mode enforcement and the fully-actuated subsystem of the quadcopter to their de-
consequently asymptotic convergence of the errors states. sired references. Similarly, convergence holds for the in-
⎫ equalities (13).
1
U11 = g + Kz z22 + ż∗2 − κz2 βz2 ⎪
⎪
⎪
b0 (cos ϑ1 cos ϕ1 ) ⎪
⎪
⎪
⎪ 3.2. Under-actuated subsystem
⎪
⎪
− κz3 sign(βz2 ) , ⎬ Now, for the ease of control design, the model of an
1 ⎪
⎪ under-actuated subsystem (3b) can be re-written in the
U41 = − a5 ϑ2 ϕ2 + Kψ ψ22 + ψ̇2∗ − κψ2 βψ2 ⎪
⎪
b3 ⎪
⎪ form
⎪
⎪ 
⎪
⎭
− κψ3 sign(βψ2 ) , χ̇1 = χ2 ; χ̇2 = 1 + 1 Ω1 ,
(14)
(9) θ̇1 = θ2 ; θ̇2 = 2 + 2 Ω2 ,
where κz2 , κz3 , κψ2 and κψ3 are the positive design con- x1 x −Kx x22 u
where χ1 = , χ2 = 2 , 1 = , Ω1 = x ,
stants. y1 y2 −Ky y22 uy
Proof: To prove this theorem, consider the time deriva- 1 0 ϕ1 U2 ϕ2
 1 = b0 U 1 , θ1 = , Ω2 = , θ2 = ,
tive of sliding manifolds βz2 and βψ2 (represented in (9)) 0 1 ϑ1 U3 ϑ2
along (3b), one obtains a ψ ϑ + a2 ω̄ϑ2 − Kϕ ϕ22 b 0
 2 = 1 2 2 and 2 = 1 .
a3 ψ2 ϕ2 + a4 ω̄ϕ2 − Kϑ ϑ22 0 b2
β̇z2 = cos ϑ1 cos ϕ1 b0 U1 + Δz − g − Kz z22 − ż∗2 + żz ,
The reference tracking errors between the desired refer-
β̇ψ2 = a5 ϑ2 ϕ2 + b3 U4 + Δψ − Kψ ψ22 − ψ̇2∗ + żψ .
ence and actual outputs are defined as
(10)

The following choice of the integral dynamics żz and żψ , εχ1 = χ1 − χd ⇒ ε̇χ1 = χ2 − χ̇d ,
(15)
with positive constants ρz , ρψ εθ1 = θ1 − θd ⇒ ε̇θ1 = θ2 − θ̇d ,
żz = −cos ϑ1 cos ϕ1 b0 U10 with U10 = −ρz (z2 − z2 ), T T
where χd = xd yd and θd = ϕd ϑd are the ref-
żψ = −b3 U40 with U40 = −ρψ (ψ2 − ψ2 ) (11) erence outputs. The derivatives of Lyapunov functions
(Vχ1 = 12 εχ21 ) and Vθ1 = 12 εθ21 along (14) is defined as
reduces (10) to the following forms
 
β̇z2 = cos ϑ1 cos ϕ1 b0 U11 + Δz − g − Kz z22 − ż∗2 , V̇χ1 = εχ1 ε̇χ1 ⇒ V̇χ1 = εχ1 (χ2 − χ̇d ),
(12) (16)
β̇ψ2 = a5 ϑ2 ϕ2 + b3 U41 + Δψ − Kψ ψ22 − ψ̇2∗ . V̇θ1 = εθ1 ε̇θ1 ⇒ V̇θ1 = εθ1 (θ2 − θ̇d ),

Now, consider the time derivatives of extended Lyapunov where the variables χ2 and θ2 are chosen as virtual control
functions Vz2 = Vz1 + 12 βz22 and Vψ2 = Vψ1 + 12 βψ22 along (12), inputs that will ensure the stability of χ1 and θ1 . Hence,
one may get the virtual control inputs will be

V̇z2 = εz1 εz2 − κz1 εz21 − κz2 βz22 + βz2 (Δz − κz3 sign(βz2 )) χ2 = χ̇d − κχ1 εχ1 , θ2 = θ̇d − κθ1 εθ1 , (17)
 Robust Integral Sliding Mode Control Design for Stability Enhancement of Under-actuated Quadcopter 1675

where κχ1 and κθ1 are the positive design constants. By Vχ2 = Vχ1 + 12 βχ22 and Vθ2 = Vθ1 + 12 βθ22 along (18) and (24)
substituting (19) in (16), one gets are as follows:
⎫
V̇χ1 = −κχ1 εχ21 , V̇θ1 = −κθ1 εθ21 . (18) V̇χ2 =εχ1 εχ2 − κχ1 εχ21 − κχ2 βχ22 − κχ3 βχ2 sign(βχ2 )⎪⎪
⎪
⎪
≤εχ1 εχ2 − κχ1 εχ21 − κχ2 βχ22 − κχ3 |βχ2 |, ⎬
Now, proceeding towards the last step, we will consider
⎪
V̇θ2 =εθ1 εθ2 − κθ1 εθ21 − κθ2 βθ22 − κθ3 βθ2 sign(βθ2 ) ⎪
the following χ2∗ and θ2 (in (17)) as new modified desired ⎪
⎪
⎭
references for the state χ2 and θ2 in the form ≤εθ1 εθ2 − κθ1 εθ1 − κθ2 βθ2 − κθ3 |βθ2 |.
2 2

(25)
χ2 = εχ2 + χ̇d − κχ1 εχ1 , θ2 = εθ2 + θ̇d − κθ1 εθ1 ,
(19) Once again, by a proper choice of these constants, one
may get the negative definite V̇χ2 and V̇θ2 . Consequently,
where εχ2 = χ2 − χ2 and εθ2 = θ2 − θ2 . Now, to ensure the the asymptotic convergence of εχ2 and εθ2 to zero is en-
convergence of εχ2 and εθ2 , the desired sliding surfaces βχ2 sured. In similar fashion, the convergence of χ1 and θ1
and βθ2 will be to the desired χd and θd is also ensured via backstepping
based virtual control laws θ2 and χ2 .
βχ2 = χ2 − χ2 + zχ , βθ2 = θ2 − θ2 + zθ , (20) Thus, taking into account equations (14) and (21), the
  applied control inputs for under-actuated subsystem can
where zχ = γχ εχ2 dt and zθ = γθ εθ2 dt and γχ and γθ be expressed by the following forms
are the positive constants. ⎫
1  ⎪
Theorem 2: Consider that the dynamics of under- ux1 = Kx x2 + ẋ2 − κx2 βx2 − κx3 sign(βx2 ) ,⎪
2 ∗
⎪
⎪
actuated subsystem (in (3b)) are transformed into (14). If b0 U1 ⎪
⎪
⎪
⎪
the backstepping procedure based virtual control laws χ2 1   ⎪
uy 1 = Ky y2 + ẏ2 − κy2 βy2 − κy3 sign(βy2 ) ,⎪
2 ∗ ⎪
⎪
and θ2 and the integral sliding manifolds are chosen ac- b0 U1 ⎪
⎪
⎪
⎪
cording to (19) and (20), respectively, then the following 1 ⎪
⎪
U21 = Kϕ ϕ2 − a1 ψ2 ϑ2 − a2 ω̄ ϑ2 + ϕ̇2
2 ∗ ⎬
control laws will ensure finite-time sliding mode enforce- b1
⎪
⎪
ment and consequently asymptotic convergence of the er- − κϕ2 βϕ2 − κϕ3 sign(βϕ2 ) , ⎪
⎪
⎪
⎪
rors states. ⎪
⎪
⎫ 1 ⎪
⎪
U31 = Kϑ ϑ2 − a3 ψ2 ϕ2 − a4 ω̄ ϕ2 + ϑ̇2
2 ∗ ⎪
⎪
1  ∗  ⎪
Ω11 = χ̇2 − 1 − κχ2 βχ2 − κχ3 sign(βχ2 ) ,⎪⎪
⎬ b2 ⎪
⎪
⎪
1 ⎪
⎪
− κϑ2 βϑ2 − κϑ3 sign(βϑ2 ) . ⎭
1  ∗  ⎪⎪
Ω21 = θ̇2 − 2 − κθ2 βθ2 − κθ3 sign(βθ2 ) . ⎭ (26)
2
(21)
The desired Euler angles ϕd , ϑd and ψd will be obtained
Proof: To proceed to the proof, consider the time from the basic trigonometry of quadcopter (as shown in
derivative of integral manifolds (βχ2 , βθ2 in (20)) along Fig. 1). By following the defined expression of [20], the
(14), one has desired ϕd , ϑd and ψd can be obtained with the following
expressions

β̇χ2 = 1 + 1 Ω11 − χ2∗ + żχ , ⎫
(22) ϕd = sin−1 (ux sin ψ1 − uy cos ψ1 ), ⎪
β̇θ2 = 2 + 2 Ω21 − θ2∗ + żθ .  ⎪⎪
⎪
−1 u x cos ψd − uy sin ψd ⎪
⎬
ϑd = sin ,
The derivative of the integral dynamics żχ and żθ , with cos ϕd (27)
  ⎪
⎪
positive constants ρχ and ρθ , can be chosen as ⎪
−1 yd − y1 ⎪
⎪
⎭
ψd = tan ,
 xd − x1
żχ = −1 Ω10 with Ω10 = −ρχ (χ2 − χ2 ),
(23) where ux and uy are obtained using the proposed control
żθ = −2 Ω20 with Ω20 = −ρθ (θ2 − θ2 )
scheme (26).
to reduce (22) to the following forms In the next subsection, the simulation results of the
quadcopter based on RBISMC scheme are presented.
β̇χ2 = 1 + 1 Ω11 − χ̇2∗ , β̇θ2 = 2 + 2 Ω21 − θ̇2∗ .
(24) 3.3. Simulation results
In this section, the proposed RBISMC control scheme
Now, to prove the stability of the aforesaid subsystem, has been simulated to control the dynamic model of quad-
the time derivatives of the augmented Lyapunov functions copter in MATLAB/Simulink environment in the presence
1676 Safeer Ullah, Adeel Mehmood, Qudrat Khan, Sakhi Rehman, and Jamshed Iqbal

3
Error along x axis via Efe
2
Error along x axis via RISMC

x position (m)
1

-1

-2

-3
0 25 50 75 100 125
Time (s)
Fig. 2. Generic control system. 4
Error along y axis via Efe
2 Error along y axis via RISMC
of matched uncertainties. The behavior of the aforemen-

Position (m)
0
tioned control scheme (which is shown in Fig. 2) is com-
pared with the FOISMC technique available in the litera- -2
ture [17].
-4
In Fig. 2, the (U − to − ω) conversion block shows
the conversion of applied forces/torques (U1 , U2 , U3 , U4 ) -6
0 25 50 75 100 125
into the motor speeds (ω1 , ω2 , ω3 , ω4 ). When the angular Time (s)
speeds of both the right and left propellers are increased 4
(or decreased) and that of back and front propellers are Error along z axis via Efe
Error along z axis via RISMC
decreased (or increased), then the motion along yaw angle
Position (m)

2
“ψ" is produced. Similarly, if the thrust (produced by the
four propellers rotating at same angular speed ωi ) is equal
to the quadcopter’s weight then it will counterbalance the 0

acceleration due to gravity in a hover condition. In addi-

tion, the forces/torques (required for the stable flight of -2
0 25 50 75 100 125
the quadcopter in all feasible conditions) are delivered by
Time (s)
the high thrust propellers. Note that the high torque servos
and gears are used to achieve the desired angular speed of Fig. 3. Comparative errors regulation of the proposed
the brush-less DC motors. RBISMC with standard literature [17].
In Fig. 3, a quite effective comparison is made in term
of reference tracking errors (between the desired and real sequently, degrades the dynamics’ performance. The ini-
trajectories) along x, y and z axis. In addition, the com- tial values of control inputs and angular velocities of the
parative performances in term of regulation for the Euler system are Uı0 = 0 (ı = 1, 2, 3, 4) and ϕ̇ = ϑ̇ = ψ̇ = 0.
angles i.e., roll, pitch and yaw, are shown in Fig. 4. The The controller parameters κχ1 , κχ2 , κθ1 , κθ2 , κψ1 , κψ2 , κz1
proposed RBISMC control law significantly decreases the and κz2 , are selected as to force the system’s states to the
overshoot and settling time with zero steady-state error. sliding surface. The parameters κχ3 , κθ3 , κψ3 and κz3 are
Note that the convergence for these Euler angles to the chosen to ensure robustness in the system against matched
equilibrium is much better as compared to the standard re- uncertainties.
sults [17]. It is also clear that the proposed control scheme
steers the corresponding tracking errors (along x, y and 4. CONCLUSION
z) and Euler angles (ϕ, ϑ , ψ) to their equilibrium val-
ues without having chattering problem, which makes the The mathematical model, based on the Euler-
proposed control scheme more appealing as compared to Newtonian and Euler-Lagrangian equations, is compre-
FOISMC presented in [17]. Besides these advantages, the hensively presented in this work. For the control design,
performance of the RBISMC controller is robust against the overall system is subdivided into fully-actuated and
the unwanted matched uncertainties Δz = 0.3 sin(z1 )+z1 z2 under-actuated subsystems. The design of control law
and Δψ = 0.1 sin(ψ1 ) + ψ1 ψ2 from the very start of the has been simplified with the effective combination of
process. backstepping and ISMC based strategies, mainly by cap-
In Fig. 5, the control inputs of the proposed control italizing the positive characteristics of both the nonlinear
scheme (required for fulfilling the main objective) are al- control techniques during the full flight maneuver of the
most chatter and oscillations free whereas the counterpart quadcopter The overall closed loop stability is presented
presented oscillations with a considerable magnitude. The step by step for all the subsystems. The proposed con-
oscillations are dangerous for the system’s health and con- trol scheme claims improved robustness by eliminating
 Robust Integral Sliding Mode Control Design for Stability Enhancement of Under-actuated Quadcopter 1677

REFERENCES
30 Roll angle ( ) presented by Efe
Roll angle ( ) via RISMC [1] Z. Ma, H. Li, Y. Gu, Z. Li, and Q. Li, “Flight and hover
Roll angle (deg)

20
control system design for a mini-quadrotor based on multi-
10
sensors,” International Journal of Control, Automation and
0 Systems, vol. 17, no. 2, pp. 486-499, February 2019.
-10 [2] Y. Díaz-Méndez, M. S. de Sousa, G. Gomes, S. Cunha, and
-20 A. Ramos, “Analytical design and stability analysis of the
0 25 50 75 100 125 universal integral regulator applied in flight control,” In-
Time (s)
ternational Journal of Control, Automation and Systems,
10 vol. 17, no. 2, pp. 391-404, February 2019.
[3] M. Wasim, M. Ullah, and J. Iqbal, “Gain-scheduled propor-
Pitch angle (deg)

0 tional integral derivative control of taxi model of unmanned

aerial vehicles,” Revue Roumaine des Sciences Techniques-
Serie Electrotechnique et Energetique, vol. 64, no. 1, pp.
-10
75-80, January 2019.
Pitch angle ( ) presented by Efe
Pitch angle ( ) via RISMC
[4] I. D. Cowling, O. A. Yakimenko, J. F. Whidborne, and
-20
0 25 50 75 100 125 A. K. Cooke, “A prototype of an autonomous controller for
Time (s) a quadrotor UAV,” Proc. of European Control Conference
40 (ECC), IEEE, pp. 4001-4008, 2007.
Yaw angle ( ) presented by Efe
30 Yaw angle ( ) via RISMC [5] S. A. Ajwad, J. Iqbal, R. U. Islam, A. Alsheikhy, A. Alme-
Yaw angle (deg)

shal, and A. Mehmood, “Optimal and robust control of

20
multi DOF robotic manipulator: Design and hardware real-
10 ization,” Cybernetics and Systems, pp. 1-17, January 2018.

0 [6] S. Bouabdallah and R. Y. Siegwart, “Full control of a

quadrotor,” Proc. of International Conference on Intelli-
-10
0 25 50 75 100 125 gent Robots and Systems, IROS 2007, San Diego, CA,
Time (s) IEEE, pp. 153-158, Oct. 29-Nov. 2, 2007.
Fig. 4. The comparison of Euler angles trajectories of [7] X. Shi, Y. Cheng, C. Yin, S. Dadras, and X. Huang, “De-
RBISMC with the standard literature [17]. sign of fractional-order backstepping sliding mode control
for quadrotor uav,” Asian Journal of Control, vol. 21, no. 1,
Angular speed [rad/s]

Angular speed [rad/s]

pp. 156-171, January 2019.

240 240
[8] W. Qi, G. Zong, and H. Karimi, “Sliding mode control for
220 220 nonlinear stochastic singular semi-markov jump systems,”
IEEE Transactions on Automatic Control, vol. 65, no. 1,
200 200
Speed ( ) history Speed ( ) history pp. 361-368, January 2020.
1 2
[9] W. Qi, G. Zong, and H. R. Karim, “Observer-based adap-
0 25 50 75 100 125 0 25 50 75 100 125
tive smc for nonlinear uncertain singular semi-markov
Time (s) Time (s)
jump systems with applications to dc motor,” IEEE Trans-
Angular speed [rad/s]

Angular speed [rad/s]

240 240 actions on Circuits and Systems I: Regular Papers, vol. 65,
no. 9, pp. 2951-2960, February 2018.
220 220
[10] F. Muñoz, E. S. Espinoza, I. González-Hernández,
200 200 S. Salazar, and R. Lozano, “Robust trajectory tracking for
Speed ( ) history Speed ( ) history
3 4 unmanned aircraft systems using a nonsingular terminal
modified super-twisting sliding mode controller,” Journal
0 25 50 75 100 125 0 25 50 75 100 125
of Intelligent & Robotic Systems, vol. 93, no. 1-2, pp. 55-
Time (s) Time (s)
72, February 2019.
Fig. 5. History of control inputs in term of speeds.
[11] S.-L. Shi, J.-X. Li, and Y.-M. Fang, “Fractional-
disturbance-observer-based sliding mode control for frac-
the sensitive reaching phase of the conventional sliding tional order system with matched and mismatched distur-
modes. In addition, the chattering phenomena is allevi- bances,” International Journal of Control, Automation and
ated by the proposed control law. A detailed simulation Systems, vol. 17, no. 5, pp. 1184-1190, May 2019.
study is carried out and the obtained results are compared [12] D. Shi, Z. Wu, and W. Chou, “Super-twisting extended
with the results from standard literature. It is, therefore, state observer and sliding mode controller for quadrotor
concluded that the proposed design is very effective for uav attitude system in presence of wind gust and actuator
such kind of complex electro-mechanical systems. faults,” Electronics, vol. 7, no. 8, p. 128, August 2018.
 1678 Safeer Ullah, Adeel Mehmood, Qudrat Khan, Sakhi Rehman, and Jamshed Iqbal

[13] L. Luque-Vega, B. Castillo-Toledo, and A. G. Loukianov, Qudrat Khan received his B.Sc. degree
“Robust block second order sliding mode control for a in mathematics from the University of Pe-
quadrotor,” Journal of the Franklin Institute, vol. 349, shawar in 2003, an M.Sc. and M.Phil. de-
no. 2, pp. 719-739, March 2012. grees in mathematics from Quaid-i-Azam
[14] H. Sun, L. Hou, G. Zong, and X. Yu, “Fixed-time attitude University, Islamabad, in 2006 and 2008,
tracking control for spacecraft with input quantization,” respectively, and a Ph.D. degree in non-
IEEE Transactions on Aerospace and Electronic Systems, linear control systems from Mohammad
vol. 55, no. 1, pp. 124-134, June 2018. Ali Jinnah University, Islamabad, in 2012.
He was a Post-doctoral Fellow at Inter-
[15] S. Irfan, A. Mehmood, M. T. Razzaq, and J. Iqbal, “Ad- national Islamic University, Malaysia, for one year. He is cur-
vanced sliding mode control techniques for inverted pendu- rently an Assistant Professor with the Center for Advanced
lum: Modelling and simulation,” Engineering Science and Studies in Telecommunications, COMSATS University, Islam-
Technology, An International Journal, vol. 21, no. 4, pp. abad. His research interests include robust nonlinear control, ob-
753-759, August 2018. servers/estimators design, and fault diagnosis of dynamic sys-
[16] K. Nonaka and H. Sugizaki, “Integral sliding mode alti- tems via sliding mode and its variants.
tude control for a small model helicopter with ground ef-
fect compensation,” Proceedings of the American Control
Sakhi Rehman received his B.Sc. Elec-
Conference, IEEE, pp. 202-207, 2011.
trical Engineering degree from the Uni-
[17] M. Ö. Efe, “Integral sliding mode control of a quadrotor versity of Engineering & Technology, Pe-
with fractional order reaching dynamics,” Transactions of shawar in 2010, and an M.S. degree in
the Institute of Measurement and Control, vol. 33, no. 8, Electrical Engineering from CECOS Uni-
pp. 985-1003, December 2011. versity of IT & ES, Peshawar, in 2015. He
[18] B. Samir, “Design and control of quadrotors with applica- is currently serving as a Lecturer at Uni-
tion to autonomous flying,” Ecole Polytechnique Federale versity of Science & Technology, Bannu,
de Lausanne, 2007. KP, Pakistan. His research interests in-
clude linear and nonlinear control system design.
[19] H.-P. Ren and X. Wang, “Experimental backstepping adap-
tive sliding mode control of hydraulic position servo sys-
tem,” Proc. of International Conference on Advanced Jamshed Iqbal holds Ph.D. in Robotics
Mechatronic Systems (ICAMechS), IEEE, pp. 349-354, De- from Italian Institute of Technology (IIT)
cember 2017. and three Master degrees in various fields
of Engineering from Finland, Sweden and
[20] E. Suiçmez, “Trajectory tracking of a quadrotor unmanned
Pakistan. He is currently working as a Re-
aerial vehicle (uav) via attitude and position control,” Edu-
search Associate Professor in University
cational & Industrial Television, July 2014.
of Jeddah, KSA. With more than 20 years
of multi-disciplinary experience in indus-
Safeer Ullah received his B.S. degree try & academia, his research interests in-
in Electronics Engineering from Interna- clude robot analysis and design. He has more than 60 journal
tional Islamic University, Islamabad in papers on his credit with H-index of 26. He is a senior member
2012 and an M.S. Electrical Engineering of IEEE USA.
from COMSATS University, Islamabad in
2016. He is currently pursuing a Ph.D. Publisher’s Note Springer Nature remains neutral with regard
degree in COMSATS University, Islam- to jurisdictional claims in published maps and institutional affil-
abad. His research interests are in analysis, iations.
observation and control of under-actuated
nonlinear systems using advanced nonlinear control approaches.

Adeel Mehmood holds a Ph.D. degree

in Nonlinear Control from the Technical
University of Belfort-Montbeliard, France.
He completed the M.S. in robotics and
embedded systems from the University
of Versailles Saint-Quentin en Yvelines,
France, in 2008, and B.S. in Mechatron-
ics engineering from the National Univer-
sity of Science and Technology, Pakistan,
in 2006. He also worked as a Post-doctoral Researcher with the
University of Haute-Alsace, France. He is currently working as
an assistant professor at COMSATS University Islamabad, Pak-
istan. His research interests include robotics, robust and nonlin-
ear control of servo systems.

View publication stats