International Journal of Control and Automation

Vol. 10, No. 10 (2017), pp.37-48

http//dx.doi.org/10.14257/ijca.2017.10.10.04

Variable Structure-Model Reference Adaptive Control for

Stabilization the Attitude of UAV

Zain Anwar Ali1 and Sujeet Pandey2.

1
Collège of Automation Engineering, Nanjing University of Aeronautics &
Astronautics, Nanjing, Jiangsu, China;
2
Gyancity Research Lab, Motihari, India.
*Correspondence: [email protected]; Tel.: +86-1301-693-1051,
[email protected]

Abstract
In this article, Variable Structure – Model Reference Adaptive Control (VS-MRAC)
algorithm with integrator in the feedback loop is proposed called as (VS-MRACI), for
stabilization the attitude of the tri-rotor Unmanned Aerial Vehicle (UAV). The proposed
controller uses control input as a reference, such that translation and rotational velocities
which are used to derive the model of the UAV. The stability of the proposed scheme is
presented using Lyapunov candidate function. The validity of the proposed algorithm for
stabilizing the attitude of UAV is simulated on Simulink/Matlab. The effectiveness of the
proposed algorithm is compared with MRAC. It shows that the proposed algorithm have
fast error convergence, robustness and a zero steady state error in the presence of model
uncertainties.

Keywords: Variable structure, model reference adaptive control, the dual controller
and UAV.

1. Introduction
Autonomous control of flying robots commonly called as Unmanned Aerial Vehicle
(UAV) is a challenging task. Since, the last decade it attracts many researchers of the
world to work on the designing of rotor crafts as well as its control algorithms [1]. It has
an obvious advantage over helicopters due to its rapid direction changing ability, Vertical
Takeoff and Landing (VTOL) capability and simple construction. It is widely used in
different arenas of our daily life like surveillance, rescue, monitoring of traffic,
environmental pollution and for the military secret tasks [2].
UAV’s are classified in two types with respect to its body frame, rotary wings and
fixed wing [3]. As compared to fixed wing UAV’s, rotor craft UAV has two main
advantages, VTOL and rapid change in the direction of flight. However, the controlling of
rotor crafts is quite complex due to Multiple Input Multi Output (MIMO), highly
nonlinear and an under-actuated system along with parameter uncertainties in the dynamic
model of the aircraft [4]. The rotor crafts have a wide range of categories including dual-
rotor, tri-rotor, quad-rotor and hex-rotor. In this paper, our main concern is with the tri-
rotor aerial vehicle that has three rotor fixed at the end of each arm. The tri-rotor aerial
robot is an under-actuated system along with four control inputs ( ) to control
the altitude, roll, pitch, yaw and nine outputs: the Euler angles ( ),the translational
velocity ( ) and rotational velocity ( ) with the six degree of freedom (6
DOF). Real time control of the aerial vehicle is required to ensure the stability of the
UAV [5]. The tri-rotor aerial vehicle has three motors (actuators) that can be used under
the effect of torque to achieve roll and pitch motion [6].

ISSN: 2005-4297 IJCA

Copyright © 2017 SERSC Australia
International Journal of Control and Automation
Vol. 10, No. 10 (2017)

In reference [7-10] previously many control algorithms were used to control the
dynamics of UAV, based on Fuzzy Logic Control (FLC), Proportional Integral Derivative
(PID) control, the Linear Quadratic Regulator (LQR) control. In [11] a nonlinear model of
the quadrotor aircraft is controlled by using the adaptive controller which has an ability to
give better performance to control the dynamics of the aircraft in the presence of
saturation and uncertainties in the model of UAV. In [12-14] the MRAC control
algorithm is used to control the model of the quadrotor, in which saturation in the actuator
was considered to amplify the linear controller and the stability of the controller was
guaranteed by Lyapunov stability criteria.
In [15] a dual controller approach for controlling the dynamics of UAV in which
feedback/ feedforward and neural network classifier is designed. A novel scheme is
proposed for searching landmark and detection for the autonomous navigation of UAV,
such that the key contribution in that article is to combine the entropy of an image with a
dual feed-forward / feedback controller for the searching and detection of an object [16].
In reference [17] a nonlinear base dual controller approach is used to control the
formation of flight for multiple UAV’s under rigid environment. However, the dual
controller approach consists of particle swarm optimization (PSO) with state feedback
control for the feasibility and stability of multiple UAV’s formation flight.
Formerly, many hybrid and dual control algorithms were used to stabilize the dynamics
of UAV. In reference [18] gain scheduling PID (GS-PID) with MRAC is used to control
the altitude and attitude of UAV. In [19] a novel fuzzy base hybrid algorithm was
designed to control the dynamics for the stabilization of tri-rotor UAV. It consists of
Regulation, Pole-placement and Tracking (RST) plus Model Reference Adaptive Control
(MRAC) such that the gains of the RST controller are being fine-tuned by the fuzzy logic
controller. Another hybrid approach was designed for controlling the attitude and altitude
of UAV [20]. In which the hybrid controller consists of fuzzy regulation, pole-placement
and tracking (F-RST). To stabilize the nonlinear characteristics of UAV, the adaptive
hybrid controller is utilized, such that the tuning of the RST controller is achieved by the
gains of the adaptive fuzzy logic controller.
In this article, the robustness of VS-MRAC with feedback integrator is compared with
classical MRAC of [21] without feedback integrator. To stabilize the attitude of UAV the
main contributions are as follows.
i. VS-MRAC is designed by adding the feedback integrator in the controller in which
VS-MRAC handles the uncertainties in the system and integrator removes the
external disturbance.
ii. The proposed VS-MRACI controller uses translational and rotational velocities as
an input constraints, which shows more closeness with real world application.
iii. The stability of overall design of the controller is proved by using Lyapunov
stability criteria.
The rest of the article is structured as follows. Section II, defines the system modeling
& its preliminaries are presented. The complete controller designing is described in
section III. Section IV defines the simulation, results and discussion. Lastly, section V,
concludes the whole article.

2. System Modeling & Its Preliminaries

The tri-rotor aerial robot has three arms and at the end of each arm the rotor is placed.
The forces and aerodynamic components of tri-rotor UAV is defined in Figure 1, such
that (X, Y, Z) axis of UAV defines the straight direction from the view, the direction at
the right side and direction downward from the center of gravity.

38 Copyright © 2017 SERSC Australia

 International Journal of Control and Automation
Vol. 10, No. 10 (2017)

Figure 1. The Structure of the Tri-rotor Aerial Vehicle

Figure 2. The Relationship between Earth’s & the Tri-Rotor Coordinate

System

The tri-rotor aerial robot coordinate system depends on the Earths coordinate system
which defined in Figure 2. The relation between the Earths and the tri-rotor coordinate
system are defined by the rotational matrix “R” of the UAV. The roll, pitch and yaw (the
attitude of the UAV) or Euler angle which is indicated with ( ) respectively [22].
The rotation of the UAV is applied with respect to the Earth frame and the complete
rotation is done with the roll angle ( ) with X axis, pitch angle ( ) with Y axis, and the
yaw angle ( ) with the Z axis. Now the rotation matrix is defined as,

[ ][ ][ ]

[ ]
{
(01)
The actuators, which is fixed in the triangular frame of the aerial robot which produces
the thrust force, which helps to hover the UAV. Therefore, the blades of the UAV have
fixed rotation angles, such that the direction of the airflow be governed by the rotation of
the blades [23]. The overall force and torque of the aerial vehicle is defined as,
{ (02)
Therefore, for the rotors of the aerial vehicle.
The rotational or angular velocities and the tilt angle of the aerial robot will offer the
UAV to control ( ) its four movement: such that three are the rotor
velocities and the last one is the tilt angle of one rotor out of three. Where as to control

Copyright © 2017 SERSC Australia 39

International Journal of Control and Automation
Vol. 10, No. 10 (2017)

the roll movement, to control the pitch movement, to control the yaw movement and
to control the altitude or yaw movement of UAV. To estimate the tri-rotor as a non-
holonomic, rigid body, and multivariable along with six Degree Of Freedom (6 DOF),
such that the states of the aerial robot is defined as: ( ), ( ) and ( ) are
the linear or translational velocity component, angular or rotational velocity
subcomponent and the Euler angles of the body.
The external forces and mass that can be exerted in a body of aircraft which is,

( ) ( ( ) )
( )
(√ ) ( )
( ) ( ( ) ( ) ( ))
{ ( ) ( )
(03)
Where is the distance from the center of the body to the arm, ( ) are
control inputs of the tri-rotor aerial vehicle. The translational and rotational velocity of the
aircraft is described below,
̇ ( )( )
{ ̇ ( )( ( ) ( ) ( ))
̇ ( )( ( ) ( ) ( ))
(04)
̇ ( )( ) ( )(√ )( )
{ ̇ ( )( ) ( )( )( ) (05)
̇ ( )( ) ( )( )( )
Where is the mass, ( ) the external forces which is exerted on the body
( ) and ( )the forces and torques that is produced by the rotors of the UAV
which is already shown in Figure 1.

3. Controller Designing
The VS-MRAC was proposed by (Hsu) [24], the main advantages of this algorithm is
deal with uncertainties, disturbances and well transient behavior with good robustness.
But in this article, VS-MRACI is proposed to control the altitude and attitude of the tri-
rotor aerial robot. The overall working of the control system is shown in Figure 3.

Figure 3. Block Diagram of Control System

Now, the actual model of the system is described as,

40 Copyright © 2017 SERSC Australia

 International Journal of Control and Automation
Vol. 10, No. 10 (2017)

̇
{ (06)
The reference model of the system is,
̇
{ (07)

The integral base output error is,

̅ (08)
The objective of the controller is to force the actual system output to track the reference
output signal, to regulate the ̅ to be zero, although observe the close-loop control signals
which are also bounded. Now, the control law which is written as,
(09)
That is same as the classical model reference adaptive control. Furthermore, switching
Laws are designed to enhance the system robustness and transient performance of the
system. If and are known, the ideal parameters to control the roll of the UAV are
(̅ ̅ ) that can be written as using the following condition.
⁄ ⁄ (10)
Equation (10) shows that if our controller fulfills the requirements and the desired path
is achieved. Consequently, the controller can be written as,
̅ ̅ (11)
To analyze the equation (06) & (07) the output of the system can be written as.

̇
{ (12)
̇
Now using addition and subtraction to make the ideal control parameter of (09).

̅ ̅ ̅ ̅ (13)
Grouping the terms of the equation (13),
( ̅ ) ( ̅ ) ̅ ̅ (14)
Now,
̃ ̃ ̅ ̅ (15)
However, ̃ and ̃are the ideal control parameters derivative of and . By
substituting (15) in equation (12),
̇ (̃ ̃ ̅ ̅ )
{ (16)
̇ ̃ ̃ (̅ ̅ )
which gives,
̇ ( ̅ ) ̅ (̃ ̃ ) (17)
Now, from (17) the reference model input can be written as,
( ̅ ̅ ) (18)
By using equation (16) and (18) in the equation (12), we get.

̇ ( ) ( ̅ )( ̅ ̅ ) (19)

Moreover, by comparing (17) & (16) using equation (10) the reference controller must
be,

̅ ( )
{ (20)
̅
Above the constraints of the system shows that the output converges at the desired
state, therefore and are known, the criteria for designing the controller is called
“Matching conditions”. Instead of known conditions our main goal are concerned with

Copyright © 2017 SERSC Australia 41

International Journal of Control and Automation
Vol. 10, No. 10 (2017)

parameters of the system that are unknown, or with parameters of the system along with
uncertainties, which need the adaptive laws for regulating the parameters of the actual
system.
Now derivate the integral base output error signal of equation (08)
̇̅ ̇ ̇ (21)
Now, equation (10) with equations (12), (20) and (21).
̇̅ ( ̅ )( ̅ ̅ ) ( ) (22)
Rearranging the equation (22),
̇̅ ( ) ( ̅ )( ̃ ̃ ) (23)
Thus,
̇̅ ̅ ( ̅ )( ̃ ̃ ) (24)
By, considering the lyapunov candidate function,
̇( ̅) ( ) ( ̅) (25)
Its derivative is,
̇( ̅) ̅ ̇̅ (26)
Now substitute (24) in to (26)
̇( ̅) ( ̅ )( ̅ )( ̃ ̃ ) ̅ (27)
Now, the switching laws for the roll control strategy is written as,
̂ ( ̅ )
{ (28)
̂ ( ̅ )
The (sgn) function can be written as,
( ̅ )
( ̅ ) { ( ̅ )
( ̅ )
( ̅ )
( ̅ ) { ( ̅ )
( ̅ )
To obtain the desired altitude and attitudes of UAV can be written as,
̇( ̅) ̅ ( ̅ ) (̂ ̅ ̅ ̅ ) (̂ ̅ ̅ ̅ )
̇( ̅) ̅ ( ̅ ) (̂ ̅ ̅ ̅ ) (̂ ̅ ̅ ̅ )
̇( ̅) ̅ ( ̅ )( ̂ ̅ ̅ ̅ ) ( ̂ ̅ ̅ ̅ ) (29)
̇
{ ( ̅) ̅ ( ̅ ) [( ̂ ̅ ̅ ̅ ) (̂ ̅ ̅ ̅ )]
If the conditions,
̂ ̅ ̂ ̅ are satisfied for the roll control can be written as,
̇( ̅) ̅ (30)
The same conditions is designed for pitch, yaw and altitude control.
Which guarantees that “ ̅ = 0” (GAS) globally asymptotically stable. Equilibrium
point become (30) is a negative definite function.

4. Simulation Results & Discussion

This section of the manuscript, defines the robustness and efficiency of the proposed
VS-MRACI controller. Moreover, the validity of the designed VS-MRACI controller is
compared with the MRAC controller of [21]. According to the model of UAV, the
parameters for the simulation conditions are shown in table.

42 Copyright © 2017 SERSC Australia

 International Journal of Control and Automation
Vol. 10, No. 10 (2017)

Table 1. Constraints of UAS

Parameters Values Si Units
(Mass)“m” 0.785 kg
(G.F) “g” 9.81 m/s2
l 0.3050 m
Ix 0.3105 Kgm2
Iy 0.2112 Kgm2
Iz 0.2215 Kgm2
0.007 Nms2/rad2
0.0156 Nms2/rad2

The desired state of the UAV uses the input control commands as a reference signals,
which are constant and the location of UAV is stable, such that the translational and
rotational velocity components are also used to reach the UAV at desired location. The
overall performance of the system is governed by VS-MRACI and the stability is ensured
by Lyapunov candidate function. The initial position of the UAV is zero and it will reach
at the desired height about 1m which is shown in Figure 4.

Figure 4. The Desired Altitude of UAV

Copyright © 2017 SERSC Australia 43

International Journal of Control and Automation
Vol. 10, No. 10 (2017)

Figure 5. Control Input Commands of UAV

Figure 5, shows the control input (u1, u2, u3, u4) responses of the UAV, initially it has
some oscillations and after that it will converge to the desired (2, 0, 0, and 0) input path
and behaves linearly. Furthermore, which confirms that is time invariant in a finite
period of time. The initial altitude of the UAV is (z = 0) and (-1, -0.01, -0.01) rad, after
that the desired position and attitude are: (z = 1)m and ( )rad, which is
shown in Figure 6. The proposed algorithm stabilize the UAV at predefined altitude and
attitude in a fair amount of time. The altitude z and yaw rate are prominently shows that
it converge to the desired states. Moreover, the altitude z, reaches to the predefined value
at u1. The roll and yaw rates have little oscillations to converge at the referred state,
which is shown in Figure 6.

Figure 6. The Euler Angle Rotational Rates of UAV

44 Copyright © 2017 SERSC Australia

 International Journal of Control and Automation
Vol. 10, No. 10 (2017)

Figure 7. Translational Velocity Components

Figure 8. Rotational Velocity Components

Copyright © 2017 SERSC Australia 45

International Journal of Control and Automation
Vol. 10, No. 10 (2017)

Figure 7 shows the translation velocity components with the initial values of (u = 0.1
m/s, v = - 0.14 m/s, w = 0.15 m/s) and stabilizes about 5 seconds with little overshoot of
“w” velocity component. Figure 8 shows the rotational velocity components with the
initial conditions are (p = 4 rad/s, q = -0.12 rad/s, r = 1 rad/s) and will converge to zero in
a fair amount of time with any overshoot.

5. Conclusion
In this article, a variable structure-model reference adaptive control with integrator in
the feedback loop is proposed for the tri-rotor aerial vehicle. Moreover, the modified VS-
MRACI controller is used to stabilize the attitude of UAV. The proposed algorithm uses
translation and rotational velocities to derive the system model. The validity of the
controller is illustrated using Simulink/ Matlab and shows that the proposed scheme has
good transient behavior with very zero steady state errors to reach at the desired altitude.
Additionally, proposed algorithm shows fast convergence and better robustness in the
presence of model uncertainties.

Conflicts of Interest
The authors declare no conflict of interest.

References
[1] G. Cai, B. M. Chen and T. H. Lee, “An overview on development of miniature unmanned rotorcraft
systems”, Frontiers of Electrical and Electronic Engineering in China, vol. 5, no. 1, (2010), pp. 1-14.
[2] S. Azeem, “Autonomous Unmanned Aerial Vehicles”, Technical report, The George Washington
University, (2012).
[3] E. Çetinsoy, S. Dikyar, C. Hançer, K. T. Oner, E. Sirimoglu, M. Unel and M. F. Aksit, “Design and
construction of a novel quad tilt-wing UAV”, Mechatronics, vol. 22, no. 6, (2012), pp. 723-745.
[4] N. Amiri, “Control of an Unconventional VTOL UAV for Complex Maneuvers”, PhD diss., University
of Calgary, (2013).
[5] P. Castillo, A. Dzul and R. Lozano, “Real-time stabilization and tracking of a four-rotor mini rotorcraft”,
IEEE Transactions on control systems technology, vol. 12, no. 4, (2004), pp. 510-516.
[6] S. Bouabdallah, P. Murrieri and R. Siegwart, “Design and control of an indoor micro quadrotor”, In
Robotics and Automation, 2004. Proceedings, ICRA'04. 2004 IEEE International Conference on, IEEE,
vol. 5, (2004), pp. 4393-4398.
[7] L. Doitsidis, K. P. Valavanis, N. C. Tsourveloudis and M. Kontitsis, “A framework for fuzzy logic
based UAV navigation and control”, In Robotics and Automation, 2004. Proceedings, ICRA'04. 2004
IEEE International Conference on, IEEE, vol. 4, (2004), pp. 4041-4046.
[8] S. Kurnaz, O. Cetin and O. Kaynak, “Fuzzy logic based approach to design of flight control and
navigation tasks for autonomous unmanned aerial vehicles”, Journal of Intelligent and Robotic Systems,
vol. 54, no. 1-3, (2009), pp. 229-244.
[9] M. Santos, V. Lopez and F. Morata, “Intelligent fuzzy controller of a quadrotor”, In Intelligent Systems
and Knowledge Engineering (ISKE), 2010 International Conference on, IEEE, (2010), pp. 141-146.
[10] K. Bickraj, T. Pamphile, A. Yenilmez, M. Li and I. N. Tansel, “Fuzzy Logic Based Integrated Controller
for Unmanned Aerial Vehicles”, In Florida Conference on Recent Advances in Robotics, FCRAR,
(2006).
[11] G. M Hoffmann, H. Huang, S. L. Waslander and C. J. Tomlin, “Quadrotor helicopter flight dynamics
and control: Theory and experiment”, In Proc. of the AIAA Guidance, Navigation and Control
Conference, vol. 2, (2007), p. 4.
[12] I. Sadeghzadeh and Y. M. Zhang, “A review on fault-tolerant control for unmanned aerial vehicles
(UAVs)”, Infotech@ Aerospace, St. Louis, MO., (2011).
[13] Y. M. Zhang, A. Chamseddine, C. A. Rabbath, B. W. Gordon, C-Y. Su, S. Rakheja, C. Fulford, J.
Apkarian, and P. Gosselin, “Development of advanced FDD and FTC techniques with application to an
unmanned quadrotor helicopter testbed”, Journal of the Franklin Institute, vol. 350, no. 9, (2013), pp.
2396-2422.
[14] V. Kumar and N. Michael, “Opportunities and challenges with autonomous micro aerial vehicles”, The
International Journal of Robotics Research, vol. 31, no. 11, (2012), pp. 1279-1291.
[15] Zemalache, K. Meguenni and H. Maaref, “Controlling a drone: Comparison between a based model
method and a fuzzy inference system”, Applied Soft Computing, vol. 9, no. 2, (2009), pp. 553-562.

46 Copyright © 2017 SERSC Australia

 International Journal of Control and Automation
Vol. 10, No. 10 (2017)

[16] O. Amidi, T. Kanade and R. Miller, “Vision-based autonomous helicopter research at Carnegie Mellon
Robotics Institute 1991-1997”, American Helicopter Society, (1998).
[17] Y. Sun, Y. Sun, N. Xian, N. Xian, H. Duan and H. Duan, “Linear-quadratic regulator controller design
for quadrotor based on pigeon-inspired optimization”, Aircraft Engineering and Aerospace Technology,
vol. 88, no. 6, (2016), pp. 761-770.
[18] I. Sadeghzadeh, A. Mehta, Y. Zhang and C.-A. Rabbath, “Fault-tolerant trajectory tracking control of a
quadrotor helicopter using gain-scheduled PID and model reference adaptive control”, In Annual
conference of the prognostics and health management society, vol. 2, (2011).
[19] Ali, Z. Anwar, D. Wang and M. Aamir, “Fuzzy-Based Hybrid Control Algorithm for the Stabilization of
a Tri-Rotor UAV”, Sensors, vol. 16, no. 5, (2016), p. 652.
[20] Ali, Z. ain Anwar, D. Wang, M. Aamir and S. Masroor, “Trajectory Tracking of a Tri-Rotor Aerial
Vehicle Using an MRAC-Based Robust Hybrid Control Algorithm”, Aerospace, vol. 4, no. 1, (2017), p.
3.
[21] A. Chamseddine, Y. Zhang, C.-A. Rabbath, C. Fulford and J. Apkarian, “Model reference adaptive fault
tolerant control of a quadrotor UAV”, AIAA Infotech@ Aerospace, St. Louis, Missouri, USA 2931,
(2011).
[22] S. K. Hong, “Fuzzy logic based closed-loop strapdown attitude system for unmanned aerial vehicle
(UAV)”, Sensors and Actuators A: Physical, vol. 107, no. 2, (2003), pp. 109-118.
[23] M. Lasek, J. Pietrucha, M. Zlocka and K. Sibilski, “Analogies between rotary and flapping wings from
control theory point of view”, In AIAA Atmospheric Flight Mechanics Conference and Exhibit, (2001),
pp. 6-9.
[24] L. Hsu, “Variable structure model-reference adaptive control (VS-MRAC) using only input and output
measurements: the general case”, IEEE Transactions on Automatic Control, vol. 35, no. 11, (1990), pp.
1238-1243.

Copyright © 2017 SERSC Australia 47

International Journal of Control and Automation
Vol. 10, No. 10 (2017)

48 Copyright © 2017 SERSC Australia