Lo Quadrotor With MPC Model Implementation

Preprints (www.preprints.org) | NOT PEER-REVIEWED | Posted: 10 November 2022 doi:10.20944/preprints202211.0181.
v1
Disclaimer/Publisher’s Note: The statements, opinions, and data contained in all publications are solely those of the individual author(s) and
contributor(s) and not of MDPI and/or the editor(s). MDPI and/or the editor(s) disclaim responsibility for any injury to people or property resulting
from any ideas, methods, instructions, or products referred to in the content.
Article
Towards Optimization of Energy Consumption of Tel-

lo Quadrotor with MPC Model Implementation
Rabab Benotsmane 1, József Vásárhelyi 2
1 Institute of Automation and Info-Communication; [email protected]
2 Institute of Automation and Info-Communication; [email protected]
Abstract: For a decade, the studies of dynamic control for unmanned aerial ve-
hicles took a large interest, where drones as a useful technology in different are-
as were always suffering from several issues like instability-high energy con-
sumption of batteries - inaccuracy of tracking targets. Different approaches are
proposed for dealing with the nonlinearity issues which present the most im-
portant features of this system. This paper describes our focus on the most
common control strategies, known as model predictive control MPC, by devel-
oping a model based on the sensors embedded in our Tello quadrotor used for
indoor purposes. The original controller of Tello quadrotor is supposed to be a
slave, where the designed model predictive controller is created in MATLAB
and imported to another embedded system, considered as a master; the objec-
tive of this model is to track the reference trajectory, almost keeping the stability
of the system and ensure the low energy consumption. In the first part, a pro-
found description of the modeling process of a dynamic model for drones is
presented, explaining the design of MPC controller with both linear and nonlin-
ear strategies built in MATLAB. In the final part, simulation and results are dis-
cussed regarding its behavior and performance, highlighting the MPC model's
important role on drones' energy consumption profile.
Keywords: dynamic control; UAV; model predictive control; nonlinear MPC;

trajectory tracking; energy consumption
1. Introduction
Drones, abbreviated as UAVs (Unmanned Aerial Vehicles), are
unmanned aircraft capable of carrying out a mission autonomously by
being self-piloted or semi-autonomous using a remote control [1], [2].
The primary function of these vehicles is to extend human vision be-
yond the natural horizon to accomplish works at risk or in hostile envi-
ronments. The military has implemented the first applications of drones
for surveillance or reconnaissance missions without the risk of human
losses. More recently, civil applications have appeared, such as the pre-
vention of forest fires, the inspection of engineering structures, the ener-
gy flow monitoring of motorway traffic, or the collection of meteorolog-
ical data. However, the use of civilian drones remains limited because
non-military drones still need to be fully integrated into civil airspace
[2], [3].
The classification of the existing UAVs differs according to the
countries and their army [4]. The UAVs can be classified according to
© 2022 by the author(s). Distributed under a Creative Commons CC BY license.

Preprints (www.preprints.org) | NOT PEER-REVIEWED | Posted: 10 November 2022 doi:10.20944/preprints202211.0181.v1
2 of 28
size, endurance, flight altitude, function, mass, payload, etc. The most
accepted classification divides UAVs according to their size and endur-
ance, where mostly five categories of UAVs exist, from aircraft UAVs
with a wingspan of about 30 meters to flapping wings a few millimeters
long.
In this paper, our interest mainly focuses on quadrotor UAVs,
known as drones with multiple rotors. Multi-rotor UAVs are certainly
the best-known and most widespread aeromechanics configuration
among autonomous aerial vehicles [5]. These aircraft are usually
equipped with four rotors, but there are also some with six or even eight
rotors. The mechanical simplicity of this type of vehicle makes it a wide-
ly used configuration for the realization of experimental platforms at
reduced costs [6]. The operation of multi-rotor drones is quite particular,
where the configuration ca be plus or cross configuration [7]. First, to
compensate for the torque reaction, the rotation of the rotors' direction is
reversed two by two. In addition, the translation and the rotation
movements according to the three axes are done carefully by manipulat-
ing the rotational speeds of the different rotors. In fact, it is the differ-
ence in the lift that determines the inclination of the aircraft around the
roll and pitch angles, allowing translational flight, Figure 1 shows the
movement states of a quadrotor with cross configuration regarding the
X, Y, and Z axes [8].
Figure 1. The possible motions of quadrotor with cross configuration
Due to the very good distribution of lift in the horizontal plane,

multi-rotor UAVs are particularly suitable for hovering and low-speed
flights. On the other hand, this configuration is not recommended for
high-speed translational flights nor for driving in windy conditions.
Moreover, this rotor configuration is restricted to small-size aircraft be-
cause the size of the rotors increases with the increase of the mass of the
drone.
3 of 28
The main goal of drones is to achieve a high degree of autonomy to

make decisions and be in a stable state, this can be done based on the
embedded systems in the body part, which are the measurement system
and the controller. The measurement system in the drone is responsible
for the collection of information on the state of the drone and its envi-
ronment, where in the drone, we can find three types of sensors [9], [10]:
• Sensors for measuring attitude, such as Gyroscopes to measure an-
gular position and angular velocity, Accelerometers to measure ac-
celeration, Magnetometers to measure the direction of the magnetic
field and scan the area and detect metals in space,
• Sensors for measuring the velocity and position of the drone as an
Altimeter, GPS, and Camera,
• Sensors for detecting obstacles around the drone as Ultrasonic sen-
sors and Telemeter laser.
The obtained information from the measurement system is trans-
ferred to the control part as digital signals, where the data is processed
by the control system. The control system based on a mathematical al-
gorithm generates the control signals allowing the drone to move in an
appropriate way. This is based on a mathematical representation of the
drone's mechanical body and on the measurement signals from the
onboard sensors. Its role is, above all, to guarantee the stability of the
device during the autonomous flight phase. Despite a general principle
that is relatively simple to understand, the design of such a system is ra-
ther complex and requires special attention. Indeed, UAVs, and more
particularly rotary wing UAVs such as quadrotors, are under-actuated
systems, sensitive to aerodynamic disturbances, and whose dynamics
are highly non-linear [11], [12]. Moreover, it presents a significant cou-
pling between the system's state variables and its control inputs. The
coupling characterizes the fact that any change in a control input affects
not only the variable of interest, but also the others [13].
Based on Equations of Newton Euler or Energetic Lagrange deriva-
tive, many scientists suggested how to design the dynamic modeling of
drones, where from literature review, we can find the mathematical
model for the kinematic and dynamic models, as it appears in [13]–[15];
taking into consideration the behavior of the quadrotor in different
flight conditions (stationary, in translation, or in rotation). However, the
main problem with these dynamic models is the difficulty to design a
"simple" control algorithm due to the complexity of the model which is
nonlinear. The first mission of the quadrotor is how to track a position
from one point to another point in the 3D environment respectively to
the given orientation. Under all the previous cited challenging criteria,
this task makes the object deal with the study of control theory, by de-
veloping controllers that can success to track the trajectory and avoiding
obstacles, all by optimizing the battery life of the drone.
The control strategy for the quadrotor scope represents a big issue
in this research field, where many scientists suggested different imple-
mentations, some of them were focusing on linear control strategies as
PID – LQ – H infinity – Linear MPC, where others preferred to deal with
nonlinear control approaches as Backstepping control – Sliding mode
control – Nonlinear MPC [16]–[21].
4 of 28
At the energy profile, one of the biggest drawbacks of the quad-

rotor is presented in the battery life, where usually the fact that time av-
erage mission does not exceed 15 min. Since existing controllers are typ-
ically designed to maximize the performance of tracking while provid-
ing a sufficient margin of stability. However, extracting the maximum
performance generates greater consumption and, therefore, a reduction
in the duration of battery life. For this reason, several researchers have
focused on regulation using the linear quadratic controller since it min-
imizes, for its quadratic nature, energy consumption. In addition, sever-
al strategies can improve fuel efficiency. We can cite as an example:
• Minimization of errors: indeed, this is the classic approach; the less
we commit errors in the pursuit of the desired trajectory, the less
energy is spent [22];
• Fuzzy logic applied to the altitude control: Indeed, the fact of
climbing or descending consumes double and even more because
this movement causes the rotation of the four engines simultane-
ously, so one could think of establishing fuzzy logic between flight
height and battery state of charge [23];
• The search for an optimal trajectory: energy consumption can be
reduced using or avoiding air currents and thus finding the flight
path least costly in terms of energy compared to a direct flight to
the destination [23].
This paper presents our focus on developing a customized, power-
ful embedded control system for Tello EDU quadrotor model [24] quad-
rotor already has a basic embedded system that fulfills simple com-
mands for take-off and landing motions with low accuracy and stability.
The paper includes the following parts: first, the process to identify
the dynamic mathematical model of Tello drone with the actual parame-
ters is presented, this model is purely nonlinear due to the strong cou-
pling between the state variables and the control inputs, but in addition,
its dynamics are under-actuated, i.e., the number of information which
is 4 is less than the number of outputs which is six that presents the de-
grees of freedom. In the second part, after getting the dynamic model,
we will show the methodology to build a control strategy. Two ap-
proaches are taken into consideration in our research: the first assumes
that the system is linear, so we apply model predictive control. This ap-
proach requires linearizing the dynamic model around an equilibrium
point hover position. The second approach assumes that we are dealing
with reality and that the system is nonlinear. So, we applied nonlinear
model predictive control to track the reference trajectory. A detailed
comparison will be presented regarding the motion tracking, stability,
and energy profile. We will show the reader how the inputs will lead to
the effect of predictive model control on energy consumption. The input
variables for the drone present the angular velocities of the four rotors.
Then in the last section, the simulation results are discussed, and a com-
parison is made to understand the potential of using a nonlinear con-
troller instead of a linear one.
2. Materials and Methods

5 of 28
The main aim of this work is to build a customized embedded sys-

tem for Tello EDU quadrotor. The customized embedded system is
based on a System on Chip from Xilinx-AMD and implements a control
strategy based on MPC. Figure 2 presents the plan for our main target.
To start the process, the research began by developing a suitable model
predictive control for Tello EDU quadrotor. This type of drone is helpful
for education and research purposes. The control model is based on the
dynamic mathematical model of the drone and its fundamental parame-
ters; therefore, this paper presents a methodology for building a suitable
control strategy based on the following points:
• the conditions are taken into consideration to identify the dynamic
mathematical model.
• parameter identification of Tello EDU quadrotor.
• use of linear and nonlinear model predictive control
• deep comparison to show the effectiveness of a nonlinear controller
- how the optimization of energy consumption can be achieved re-
garding the control part.
The simulation is done using MATLAB Simulink, offering different
functions that could help us achieve the research target efficiently, espe-
cially since setting parameters and analyzing the results in Simulink is
more friendly than using MATLAB Script.
In the following, the paper will highlight the methods used to build
the optimal controller, starting by modeling the dynamics of Tello quad-
rotor, then dealing with the control strategy part by getting familiar
with a model predictive controller in general and how we succeed in
tuning both control types (linear and nonlinear).
Figure 2. The proposed plan for building the customized embedded system for
Tello Edu
2.1. Tello Quad-rotor Model

Tello EDU quadrotor model [24] is an impressive and programma-
ble valuable drone for educational purposes. The system is dedicated to
the students to learn programming languages such as Scratch, Python,
and Swift and develop AI functions like a fly-in swarm, upgraded with
6 of 28
SDK 2.0, embedded with DJI’s flight control technology, and supports
electronic image stabilization.
Figure 3 presents the Tello quadrotor with its specifications. The
drone features a vision positioning system and an onboard camera
based on an advanced flight controller to hover in a precise place, suita-
ble for flying indoors, in which its maximum time flight is around 13
minutes, and its maximum flight distance is 100 meters. These specifica-
tions are critical to see the advantages and the drawbacks that they
have. For example, when we say the drone is for indoor activities, this
characteristic will neglect the wind and gyroscopic effects as a disturb-
ance in the system. Also, it is a motivation for us to optimize the energy
profile of this drone where the time flight can be much longer in case of
using optimal control strategies.
From Tello drone, our interest, as we mentioned earlier, is to devel-
op a new predictive control model using its basic sensors identified as a
complete system: the vision positioning system helps the aircraft to
maintain its current position, where the main components are the cam-
era and the 3D infrared module located on the underside of the plane.
Figure 3. Tello drones basic components
2.2. Quadroto Mathematical Model

Designing a suitable controller for an appropriate drone requires
first modeling the mathematical dynamics of the plant. This helps to
understand how the drone moves and what forces and torques are ap-
plied. How many motions do we have? Which velocities are needed to
execute different flight modes? Therefore, one can identify the inputs,
outputs, the state’s variables, and the disturbances of the central system.
Figure 4 describes the dynamic structure of Tello drone in the body
and inertial frames, where the corresponding angular velocities, tor-
ques, and forces created by the four rotors (numbered from 1 to 4) are
presented. The drone's movement is generated by differential thrust
forces between F2 and F4 that generate the rotation around the X axis,
known as roll motion. On the other hand, differential thrust forces be-
tween F1 and F3 will generate the rotation around the Y axis, known as
pitch motion. In contrast, the rotation around the Z axis, the yaw mo-
tion, is caused by differential torques between clockwise and anticlock-
wise rotors, i.e., (𝜏𝜏1 − 𝜏𝜏2 + 𝜏𝜏3 − 𝜏𝜏4 ).
The modeling of UAVs is a delicate task since the system's dynam-
ics are strongly nonlinear and fully coupled. Many literature reviews
7 of 28
explained the methodology for modeling such systems (references?).

This paper presents the exact strategy to get both linear and nonlinear
dynamic models based on the Tello drone's main parameters. The calcu-
lation of the mathematical model is based on some simplifications; here
are the different working hypotheses [25]:
• The structure of the quadrotor is assumed rigid and symmetrical,
which induces that the matrix of inertia will be assumed diagonal.
• The propellers are assumed rigid in order to be able to neglect the
effect of their deformation during rotation.
• The center of mass and the origin of the frame linked to the struc-
ture coincide.
• The lift and drag forces are proportional to the squares of the ro-
tors' rotation speed, which is a very close approximation of the aer-
odynamic behavior.
• To evaluate the quadrotor's mathematical model, we use two
frames, a fixed inertial frame to the earth R f and the second mobile
frame fixed in the quadrotor R m . The transformation matrix gives
the passage between the body frame and the inertial frame T,
which contains the orientation and the mobile position reference
relative to the fixed reference.
Figure 4 shows the axis convention taken for Tello drone to model
the mathematical dynamic, where the forces, inertial moments, and an-
gular velocities are described.
Figure 4. Modeling frame assignments for Tello quadrotor [25]
The quadrotor system is a light structural flight vehicle. Therefore,

the gyroscopic effects resulting from the rotation of the rigid body and
the four propellers should be included in the dynamic model. However,
the dynamic model of the system is obtained under the assumption that
the vehicle is a rigid body in the spatial domain with different forces
and torques. The mathematical model presented here is based on New-
ton and Euler equations for the rigid body 3D motion [25].
8 of 28
Let us describe [x y z ϕ θ ψ] T the vector containing the linear and

angular position of the quadrotor in the inertial frame Fi , and
[u v w p q r] T the vector containing the linear and angular velocities in
the body frame Fb . From 3D body dynamics, it follows that the two ref-
erence frames are linked by the following relations:
𝑣𝑣𝑥𝑥𝑖𝑖 𝑥𝑥̇ 𝑢𝑢
𝑣𝑣 = �𝑣𝑣𝑦𝑦𝑖𝑖 � = �𝑦𝑦̇ � = 𝑅𝑅 ∙ 𝑣𝑣𝑏𝑏 = 𝑅𝑅 ∙ � 𝑣𝑣 � (1)
𝑣𝑣𝑧𝑧𝑖𝑖 𝑧𝑧̇ 𝑤𝑤
Ω𝑏𝑏𝑥𝑥 𝑝𝑝 𝜙𝜙̇
𝑏𝑏
Ω = �Ω𝑦𝑦 � = �𝑞𝑞 � = 𝑇𝑇 ∙ � 𝜃𝜃̇ � (2)
Ω𝑏𝑏 𝑟𝑟 𝜓𝜓̇
𝑧𝑧
where v is the linear velocity and Ω the rotation velocity, R is the rota-
tion matrix describes the rotation from the body reference system to the
inertial reference, T is the matrix for angular transformation, where to
obtain [ϕ̇ θ̇ ψ̇] T , we need the inverse of matrix T.
𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶 𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶 − 𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶 𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶 + 𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆
𝑅𝑅 = 𝑅𝑅𝑏𝑏𝑖𝑖 = � 𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶 𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆 + 𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶 𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶 − 𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆� (3)
−𝑆𝑆𝑆𝑆 𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆 𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶
1 0 −𝑆𝑆𝑆𝑆 1 𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆 𝐶𝐶𝜙𝜙𝜙𝜙𝜙𝜙

T = �0 𝐶𝐶𝐶𝐶 𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶 �, 𝑇𝑇 −1 = �0 𝐶𝐶𝐶𝐶 −𝑆𝑆𝑆𝑆 � (4)
𝑆𝑆𝑆𝑆 𝐶𝐶𝐶𝐶
0 −𝑆𝑆𝑆𝑆 𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶 0 𝐶𝐶𝐶𝐶 𝐶𝐶𝐶𝐶
where Cθ and Sϕ and Tϕ represent cosinus and sinus and tangant.
From the two relations, we can get the kinematic model for the quad-
rotor as follow:
𝑥𝑥̇ = 𝑢𝑢[𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶] + 𝑣𝑣[𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶 − 𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶] + 𝑤𝑤[𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶 + 𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆]

⎧
𝑦𝑦̇ = 𝑢𝑢[𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶] + 𝑣𝑣[𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆 + 𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶] + 𝑤𝑤[𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶 − 𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆]
⎪
⎪ 𝑧𝑧̇ = 𝑢𝑢[−𝑆𝑆𝑆𝑆] + 𝑣𝑣[𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆] + 𝑤𝑤[𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶]
𝜙𝜙̇ = 𝑝𝑝 + 𝑞𝑞[𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆] + 𝑟𝑟[𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶] (5)
⎨
⎪ 𝜃𝜃̇ = 𝑞𝑞[𝐶𝐶𝐶𝐶] + 𝑟𝑟[−𝑆𝑆𝑆𝑆]
⎪ 𝑆𝑆𝑆𝑆 𝐶𝐶𝐶𝐶
⎩ 𝜓𝜓̇ = 𝑞𝑞 � � + 𝑟𝑟 � �
𝐶𝐶𝐶𝐶 𝐶𝐶𝐶𝐶
The vector [u v w p q r] T is obtained using IMU system, with the
data obtained from the accelerometer and gyroscop sensors. In order to
get the dynamic model, let us call all the physical effects, forces and the
inertia moments applied on the system:
• The weight force of the quadrotor given by:
• P= m∙g (6)
• where m is the total mass of the rigid body and g is the gravity.
• The thrust forces created by the four rotors.
• These forces are perpendicular to the plane of the propellers, and
are proportional to the square of the rotational speed of the motors,
the thrust forces are given by:
• Fi = b ∙ ω2i (7)
9 of 28
• where i = 1 … 4; b is the coefficient of lift it depends on the shape

and the number of blades and the air density; ω is the angular ve-
locity of the engine.
• The drag force is the coupling between the pressure force and the
viscous frictional force.
In this case, two drag forces are acting on the system, the drag in the
propellers and the drag along the axes (x, y, z).
The drag in the propellers acts on the blades; it is proportional to
the density of the air, to the shape of the edges, and to the square of the
rotation velocity of the propeller, and given by the following relation:
𝐹𝐹𝑑𝑑𝑑𝑑 = 𝑑𝑑 ∙ 𝜔𝜔2 (8)
where: d is the drag coefficient; it depends on the manufacture of the

propeller.
The drag along the axes (x, y, z) is due to the movement of the body
of the quadrotor:
𝐹𝐹𝑑𝑑𝑑𝑑 = 𝐾𝐾𝑓𝑓𝑓𝑓 ∙ v (9)
with: K ft is the translational drag coefficient, and v is the linear velocity.

Regarding the inertial moments applied to the system, we have the fol-
lowing:
• The inertial moments due to the thrust force: The rotation around
the x and y axes is due to the inertial moment created by the differ-
ence between the thrust forces of rotors 2 and 4 and 1 and 3, respec-
tively:
𝑀𝑀𝑥𝑥 = 𝑙𝑙(𝐹𝐹4 − 𝐹𝐹2 ) = 𝑙𝑙𝑙𝑙(𝜔𝜔42 − 𝜔𝜔22 )
� (10)
𝑀𝑀𝑦𝑦 = 𝑙𝑙(𝐹𝐹3 − 𝐹𝐹1 ) = 𝑙𝑙𝑙𝑙(𝜔𝜔32 − 𝜔𝜔12 )
where: l is the length of the arm between the rotor and the
center of gravity of the quadrotor.
• The inertial moment due to the drag forces: The rotation around
the z axis is due to a reactive torque by the drag torques in each
propeller:
𝑀𝑀𝑧𝑧 = 𝑑𝑑(𝜔𝜔12 − 𝜔𝜔22 + 𝜔𝜔32 − 𝜔𝜔42 ), (11)
• The inertial moment resulting from aerodynamic friction is given

by:
𝑀𝑀𝑎𝑎 = 𝐾𝐾𝑓𝑓𝑓𝑓 ∙ Ω2 (12)
where: 𝐾𝐾𝑓𝑓𝑓𝑓 the coefficient of aerodynamic friction angular

velocity.
• Gyroscopic effect This effect is defined as the difficulty of modify-
ing the position or the orientation of the plane of rotation of a rotat-
ing mass. In our case there are two gyroscopic moments. The gyro-
scopic moment of the helices is given by the expression:
𝑀𝑀𝑔𝑔ℎ = ∑4𝑖𝑖=1 Ω ∧ 𝐽𝐽𝑟𝑟 [0 0 (−1)𝑖𝑖+1 ⋅ 𝜔𝜔𝑖𝑖 ]𝑇𝑇 (13)
The gyroscopic moment due to the movements of the quadrotor

is written:
10 of 28
𝑀𝑀𝑔𝑔𝑔𝑔 = Ω ∧ 𝐽𝐽Ω (14)
where: Jr is the moment of inertia of the rotor, J is the moment of inertia

of the system.
Other aerodynamic effects, which can disturb the movement of the
quadrotor, are not taking into consideration in this article, as their ef-
fects at low speed can be neglected as: blade flapping effect- air friction-
ground effect.
By using Newton – Euler equations, the dynamics of a rigid body
under external forces and moments in the fb can be formulated as fol-
lows:
• Dynamic equation of translation motion
𝒎𝒎𝐯𝐯̇ 𝒊𝒊𝒊𝒊𝒊𝒊𝒊𝒊𝒊𝒊𝒊𝒊𝒊𝒊𝒊𝒊 = ∑ 𝑭𝑭𝒆𝒆𝒆𝒆𝒆𝒆𝒆𝒆𝒆𝒆𝒆𝒆𝒆𝒆𝒆𝒆 , (15)
where m is the masse of the rigid body and v𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖 and 𝐹𝐹𝑒𝑒𝑒𝑒𝑒𝑒𝑒𝑒𝑒𝑒𝑒𝑒𝑒𝑒𝑒𝑒 are the
linear velocity and the forces projected in the inertial frame respectively.
By using the rotation matrix to transform forces from the rigid body
frame to the inertial body frame, we get:
𝒎𝒎𝝃𝝃̈ = 𝑭𝑭𝒕𝒕 + 𝑭𝑭𝒅𝒅𝒅𝒅 + 𝑭𝑭𝒈𝒈, (16)
𝐹𝐹𝑡𝑡 = 𝑅𝑅[0 0 ∑4𝑖𝑖=1 𝐹𝐹𝑖𝑖 ]𝑇𝑇 ; 𝐹𝐹𝑔𝑔 = [0 0 −𝑚𝑚𝑚𝑚]𝑇𝑇 (17)
−𝐾𝐾𝑓𝑓𝑓𝑓𝑓𝑓 0 0 𝑥𝑥
𝐹𝐹𝑑𝑑𝑑𝑑 = � 0 −𝐾𝐾𝑓𝑓𝑓𝑓𝑓𝑓 0 � 𝜉𝜉̇ ; where 𝜉𝜉 = �𝑦𝑦� (18)
0 0 −𝐾𝐾𝑓𝑓𝑓𝑓𝑓𝑓 𝑧𝑧
By replacing all the terms that we have presented, we get:
𝑥𝑥̈ 𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶 + 𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆 𝐾𝐾𝑓𝑓𝑓𝑓𝑓𝑓 𝑥𝑥̇ 0

𝑚𝑚 �𝑦𝑦̈ � = �𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶 − 𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆� ∑𝑖𝑖=1 𝐹𝐹𝑖𝑖 − �𝐾𝐾𝑓𝑓𝑓𝑓𝑓𝑓 𝑦𝑦̇ � − � 0 �,
4
(19)
𝑧𝑧̈ 𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶 𝐾𝐾𝑓𝑓𝑓𝑓𝑓𝑓 𝑧𝑧̇ 𝑚𝑚𝑚𝑚
4
⎧𝑥𝑥̈ = 1 (𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶 + 𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆) �� 𝐹𝐹 � − 𝐾𝐾𝑓𝑓𝑓𝑓𝑓𝑓 𝑥𝑥̇
𝑖𝑖
⎪ 𝑚𝑚 𝑚𝑚
⎪ 𝑖𝑖=1
⎪ 4
1 𝐾𝐾𝑓𝑓𝑓𝑓𝑓𝑓
𝑦𝑦̈ = (𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶 − 𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆) �� 𝐹𝐹𝑖𝑖 � − 𝑦𝑦̇ (20)
⎨ 𝑚𝑚 𝑚𝑚
𝑖𝑖=1
⎪ 4
⎪ 1 𝐾𝐾𝑓𝑓𝑓𝑓𝑓𝑓
⎪ 𝑧𝑧̈ = (𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶) �� 𝐹𝐹𝑖𝑖 � − 𝑧𝑧̇ − 𝑔𝑔
⎩ 𝑚𝑚 𝑚𝑚
𝑖𝑖=1
• Dynamic equation of rotation motion

By application of Euler’s rotation equation, we have:
𝐉𝐉𝛀𝛀̇ + 𝛀𝛀 × 𝑱𝑱𝛀𝛀 = � 𝑴𝑴𝒆𝒆𝒆𝒆𝒆𝒆𝒆𝒆𝒆𝒆𝒆𝒆𝒆𝒆𝒆𝒆 (21)
where Ω in this equation represent the rotation velocity presented in the

inertial axes, J is the inertial tensor of the symmetric rigid body around
11 of 28
its center of mass. Mexternal represent the sum of moments acting on the
vehicle,
𝐽𝐽𝑥𝑥𝑥𝑥 0 0
𝐽𝐽 = � 0 𝐽𝐽𝑦𝑦𝑦𝑦 0� (22)
0 0 𝐽𝐽𝑧𝑧𝑧𝑧
where Jxx , Jyy , Jzz are calculated according to the mass distribution, by
replacing the previous inertial moments, we get:
𝐉𝐉𝛀𝛀̇ = −𝛀𝛀 × 𝑱𝑱𝛀𝛀 − 𝑴𝑴𝒂𝒂 − 𝑴𝑴𝒈𝒈𝒈𝒈 + 𝑴𝑴𝒇𝒇 , (23)
𝐾𝐾𝑓𝑓𝑓𝑓𝑓𝑓 𝜙𝜙̇ 2
𝑀𝑀𝑥𝑥 = 𝑙𝑙𝑙𝑙(𝜔𝜔42 − 𝜔𝜔22 )
where: 𝑀𝑀𝑓𝑓 = � �; 𝑀𝑀𝑎𝑎 = � 𝐾𝐾𝑓𝑓𝑓𝑓𝑓𝑓 𝜃𝜃̇ 2 �
𝑀𝑀𝑦𝑦 = 𝑙𝑙𝑙𝑙(𝜔𝜔32 − 𝜔𝜔12 ) (24)
2 2 2 2
𝑀𝑀𝑧𝑧 = 𝑑𝑑(𝜔𝜔1 − 𝜔𝜔2 + 𝜔𝜔3 − 𝜔𝜔4 ) 𝐾𝐾𝑓𝑓𝑓𝑓𝑓𝑓 𝜓𝜓̇ 2
By replacing all the terms in equation, we get:

𝐽𝐽𝑥𝑥𝑥𝑥 0 0 𝜙𝜙̈ 𝜙𝜙̇ 𝐽𝐽𝑥𝑥𝑥𝑥 0 0 𝜙𝜙̇ � 𝑟𝑟 𝜃𝜃̇
−𝐽𝐽𝑟𝑟 Ω 𝐾𝐾𝑓𝑓𝑓𝑓𝑓𝑓 𝜙𝜙̇ 2 𝑙𝑙𝑙𝑙(𝜔𝜔42 − 𝜔𝜔22 )
�0 𝐽𝐽𝑦𝑦𝑦𝑦 0 � � 𝜃𝜃̈ � = − � 𝜃𝜃̇ � ∧ �� 0 𝐽𝐽𝑦𝑦𝑦𝑦 0 � � 𝜃𝜃̇ �� − � 𝐽𝐽 Ω� ̇
𝑟𝑟 𝑟𝑟 𝜙𝜙
� − � 𝐾𝐾𝑓𝑓𝑓𝑓𝑓𝑓 𝜃𝜃 ̇ 2
� + � 𝑙𝑙𝑙𝑙(𝜔𝜔32 − 𝜔𝜔12 ) � (25)
0 0 𝐽𝐽𝑧𝑧𝑧𝑧 𝜓𝜓̈ 𝜓𝜓̇ 0 0 𝐽𝐽𝑧𝑧𝑧𝑧 𝜓𝜓̇ 0 𝐾𝐾𝑓𝑓𝑓𝑓𝑓𝑓 𝜓𝜓̇ 2 𝑑𝑑(𝜔𝜔12 − 𝜔𝜔22 + 𝜔𝜔32 − 𝜔𝜔42 )
Such that:
� 𝑟𝑟 𝜃𝜃̇ − 𝐾𝐾𝑓𝑓𝑓𝑓𝑓𝑓 𝜙𝜙̇ 2 + 𝑙𝑙𝑙𝑙(𝜔𝜔42 − 𝜔𝜔22 )

−(𝐽𝐽𝑧𝑧𝑧𝑧 − 𝐽𝐽𝑦𝑦𝑦𝑦 )𝜃𝜃̇ 𝜓𝜓̇ − 𝐽𝐽𝑟𝑟 Ω
⎧ 𝜙𝜙̈ =
⎪ 𝐽𝐽𝑥𝑥𝑥𝑥
⎪
−(𝐽𝐽𝑧𝑧𝑧𝑧 − 𝐽𝐽𝑥𝑥𝑥𝑥 )𝜙𝜙𝜓𝜓 − 𝐽𝐽𝑟𝑟 Ω𝑟𝑟 𝜙𝜙̇ − 𝐾𝐾𝑓𝑓𝑓𝑓𝑓𝑓 𝜃𝜃̇ 2 + 𝑙𝑙𝑙𝑙(𝜔𝜔32 − 𝜔𝜔12 )
̇ ̇ �
𝜃𝜃̈ = (26)
⎨ 𝐽𝐽𝑦𝑦𝑦𝑦
⎪ ̇ ̇ ̇2 2 2 2 2
⎪𝜓𝜓̈ = −(𝐽𝐽𝑦𝑦𝑦𝑦 − 𝐽𝐽𝑥𝑥𝑥𝑥 )𝜙𝜙𝜃𝜃 − 𝐾𝐾𝑓𝑓𝑓𝑓𝑓𝑓 𝜓𝜓 + 𝑑𝑑(𝜔𝜔1 − 𝜔𝜔2 + 𝜔𝜔3 − 𝜔𝜔4 )
⎩ 𝐽𝐽𝑧𝑧𝑧𝑧
where: �Ω𝑟𝑟 = 𝜔𝜔1 − 𝜔𝜔2 + 𝜔𝜔3 − 𝜔𝜔4
• The matrix that relies between forces/ inertial moments and angu-
lar velocities of four motors:
2
𝑢𝑢1 𝐹𝐹𝑡𝑡 𝑏𝑏 𝑏𝑏 𝑏𝑏 𝑏𝑏 ⎡𝜔𝜔1 ⎤
𝑀𝑀𝑥𝑥 2
𝑢𝑢2
�𝑢𝑢 � = �𝑀𝑀 � = � 0 −𝑙𝑙𝑙𝑙 0 𝑙𝑙𝑙𝑙 � ⎢𝜔𝜔2 ⎥, (27)
3 𝑦𝑦 −𝑙𝑙𝑙𝑙 0 𝑙𝑙𝑙𝑙 0 ⎢𝜔𝜔32 ⎥
𝑢𝑢4 ⎢ ⎥
𝑀𝑀𝑧𝑧 𝑑𝑑 −𝑑𝑑 𝑑𝑑 −𝑑𝑑 ⎣𝜔𝜔2 ⎦
4
The relation presented in this matrix is very useful for building

controllers, were u1 , u2 , u3 , u4 are the four inputs parameters that con-
trol the system. In our case, Tello EDU quadrotor is dedicated to the in-
door application; therefore, in the following, the linear drag force Fdl
and the inertial moment of the gyroscopic effect Mgh , are considered as
external disturbances, which are totally neglected in order to simplify
the model.
To summarize the previous work, the model plant is based on 4 in-
puts U and 6 outputs Y expressed as follows:
𝑢𝑢1 𝐹𝐹𝑡𝑡
𝑢𝑢2 𝑀𝑀𝑥𝑥
Inputs: �𝑢𝑢3 � = �𝑀𝑀 � Outputs: [𝑥𝑥 𝑦𝑦 𝑧𝑧 𝜙𝜙 𝜃𝜃 𝜓𝜓]𝑇𝑇 , (28)
𝑦𝑦
𝑢𝑢4 𝑀𝑀𝑧𝑧
The model is simplified as follow:
12 of 28
1 4
⎧𝑥𝑥̈ = 𝑚𝑚 (𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶 + 𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆)�∑𝑖𝑖=1 𝐹𝐹𝑖𝑖 �
⎪ 1
𝑦𝑦̈ = (𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶 − 𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆)�∑4𝑖𝑖=1 𝐹𝐹𝑖𝑖 �, (29)
𝑚𝑚
⎨ 1
⎪ 𝑧𝑧̈ = (𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶)�∑4𝑖𝑖=1 𝐹𝐹𝑖𝑖 � − 𝑔𝑔
⎩ 𝑚𝑚
−(𝐽𝐽𝑧𝑧𝑧𝑧 − 𝐽𝐽𝑦𝑦𝑦𝑦 )𝜃𝜃̇𝜓𝜓̇ + 𝑙𝑙𝑙𝑙(𝜔𝜔42 − 𝜔𝜔22 )

⎧ 𝜙𝜙̈ =
⎪ 𝐽𝐽𝑥𝑥𝑥𝑥
⎪
−(𝐽𝐽𝑧𝑧𝑧𝑧 − 𝐽𝐽𝑥𝑥𝑥𝑥 )𝜙𝜙̇𝜓𝜓̇ + 𝑙𝑙𝑙𝑙(𝜔𝜔32 − 𝜔𝜔12 )
𝜃𝜃̈ = (30)
⎨ 𝐽𝐽𝑦𝑦𝑦𝑦
⎪ −(𝐽𝐽 − 𝐽𝐽 )𝜙𝜙̇ 𝜃𝜃̇ + 𝑑𝑑(𝜔𝜔12 − 𝜔𝜔22 + 𝜔𝜔32 − 𝜔𝜔42 )
⎪𝜓𝜓̈ = 𝑦𝑦𝑦𝑦 𝑥𝑥𝑥𝑥
⎩ 𝐽𝐽𝑧𝑧𝑧𝑧
2.2.1. Identification parameters of Tello EDU drone

The mathematical dynamic model for the quadrotor taking into
consideration the six outputs [x y z ϕ θ ψ] T is standard, the main
change is reflected regarding the neglected disturbances and the param-
eter of every model, in our case, the main parameters that we had to
identify for the dynamic model are presented in Table 1:
Table 1. Parameters of Tello quadrotor
Parameters Values
Mass 0.8 [Kg]
Jxx 0.0097
Jyy 0.0097
Jzz 0.017
b 1
d 0.08
l 0.06 [m]
g 9.81
2.3. Control Strategy of Quadrotor System

The controller system is a mathematical algorithm that processes
how the quadrotor should act to maintain the measurement equal to the
set point. Despite its general principle being relatively simple, the de-
sign of such a system requires particular attention, given the complex
and under-actuated dynamics of the quadrotor [5]. The literature on this
subject is vast and very varied and can lead to a better understanding of
the performance of the proposed controllers [26]. This part is devoted to
some methods of linear and non-linear controls applied to the quad-
rotor.
• Linear Control Strategy: Linear control techniques are the first used
for UAV flight control. Linear controllers are applied to a linear-
ized model of the quadrotor; the linearization is made around an
equilibrium point using Taylor series expansion and the Jacobean
method, which will be presented later; among the command's laws
known in this field [27]–[30]
• Nonlinear Control Strategy: Nonlinear controllers perform better
because they are applied to the accurate quadrotor model, which
presents all the critical nonlinearities of the dynamics. The follow-
13 of 28
ing presents the most used method in UAV flight control. Usually,
the application of a nonlinear controller can achieve higher perfor-
mance with a nonlinear model of the UAV plant, but the main dis-
advantage is that it needs higher power computation [20], [31]–[33].
Table 2 summarizes the most linear and nonlinear control strategies
used in UAV flight control in the literature review. In our case, we are
using MPC strategy, where two scenarios can be taken:
• 1- the first scenario assumes that the dynamic model is linear, so
MPC is linear also; this is very familiar in science, where the dy-
namic model should be linearized around a defined point called
the equilibrium point.
• 2- the second scenario: suggests taking the nonlinear model as it is
and, according to the constraints, then building an efficient nonlin-
ear MPC. The second method presents different complications and
requires much more computation power.
Table 2. Parameters of Tello quadrotor
Controller Stability and Computation

Linearity type
type time response time
Low stability in higher disturb-
PID Controller Linear Low
ances, acceptable response time.
Linear Quadratic Good stability, with slow re-
Linear Very low
Regulator sponse time.
H-infinity Optimal High stability, medium response
Linear Medium
Control time.
Not robust when the uncertain-
Backstepping
Nonlinear ties are present in the model of Medium
Control
the plant
Sliding mode Robust even with the presence of
Nonlinear High
Control uncertainties and disturbances.
Good stability with the presence
Adaptive Control Nonlinear High
of uncertainties.
Stability is not guaranteed, high
Model Predictive
It can be both optimization for different con- High
Control (MPC)
straints.
In this paper both extensions are presented in detail, to show the

reader the main differences and results obtained from the simulation,
especially at the energy profile level.
The obtained dynamic model shown above is purely nonlinear, to
go forward with the control field, first, we identify the states of the sys-
tem as follow, where we can have plenty suggestions. In This paper we
will go forward with this notation:
𝑋𝑋 = [𝑥𝑥 𝑦𝑦 𝑧𝑧 𝜙𝜙 𝜃𝜃 𝜓𝜓 𝑥𝑥̇ 𝑦𝑦̇ 𝑧𝑧̇ 𝜙𝜙̇ 𝜃𝜃̇ 𝜓𝜓̇]𝑇𝑇 , (31)
𝑋𝑋⃗̇ = 𝑓𝑓̂(𝑋𝑋⃗, 𝑢𝑢
�⃗) (32)
14 of 28
where f̂�X
�⃗� is the state variables function, and f̂(u
�⃗) is the input control
variables function.
2.3.1. Model Predictive Control Concept
Model predictive control MPC is an advanced automatic control
technique useful mostly for industrial applications. MPC concept is
based on using the linear or nonlinear dynamic model of the plant in-
side the controller in real-time, this is to anticipate the future behavior of
the system. The main particularity of MPC is that the output calculation
must be solved online. The optimization process is based on the in-
puts/outputs of the system that minimize a cost function to predict the
future. The prediction is executed from the internal model of the con-
troller over a finite time window called the prediction horizon. The solu-
tion of the optimization problem is presented as a control vector. Using
updated system data, the controller can solve the problem again on the
next interval period, therefore MPC is called also moving horizon or re-
ceding horizon optimal control.
An MPC solves an optimization problem, specifically a quadratic
program (QP), it consists of the following three components: a state dy-
namics model; a cost function, which is the objective that we would like
to achieve; and constraints, which are presented as boundaries to limit
the future behavior for the inputs and outputs, where can set soft or
hard constraints.
MPC operates as follows: using the open-loop plant model and a
chosen prediction horizon N, calculate the set of future states
xk+1 , … … xk+N and control inputs uk ,⋯, uk+N−1 that minimizes a cost
function subject to state and input constraints. Then, the first element uk
in the set of calculated inputs is used as the input to the plant, and the
sequence of calculations is repeated at the next sampling instant.
MPC controller has shown its efficiency in the application with
short sampling periods where in the field of numerical controls has giv-
en good results in terms of speed and accuracy [34]. Figure 5 shows the
internal closed loop of MPC with the plant, where Figure 6 shows the
internal behavior concept of MPC.
Figure 5. MPC closed control loop

15 of 28
Figure 6. Concept of Model Predictive Control
The standard cost function denotes Θ usually is the sum of four

terms, each focusing on a particular aspect of controller performance, as
follows:
Every term in Θ represents an objective sub-function:
Θ(𝓏𝓏𝑘𝑘 ) = Θ𝑦𝑦 (𝓏𝓏𝑘𝑘 ) + Θ𝑢𝑢 (𝓏𝓏𝑘𝑘 ) + Θ∆𝑢𝑢 (𝓏𝓏𝑘𝑘 ) + Θ𝜀𝜀 (𝓏𝓏𝑘𝑘 ), (33)
where:
• Θy (𝓏𝓏k ): Cost function for Output Reference Tracking
• Θu (𝓏𝓏k ): Cost function for Manipulated variable tracking
• Θ∆u (𝓏𝓏k ): Cost function for manipulated variable move suppression
• Θε (𝓏𝓏k ): Cost function for constraint violation
• 𝓏𝓏k : The quadratic program
In our case we are interested in minimizing the following two
terms Θy (𝓏𝓏k ), Θu (𝓏𝓏k ), where we would like to track the reference output,
so the error between the reference and the output signal should be min-
imized, where the second term is minimized in order to the controller
keep selected manipulated variables (MVs) at or near specified target
values, in general the problem statement is written as follow:
min {∑𝑁𝑁−1 2 2
𝑘𝑘=0 ‖𝑟𝑟(𝑡𝑡) − 𝑦𝑦𝑡𝑡+𝑘𝑘 ‖ + 𝜌𝜌‖𝑢𝑢𝑡𝑡+𝑘𝑘 − 𝑢𝑢𝑟𝑟 (𝑡𝑡)‖ } (34)
𝑢𝑢𝑡𝑡 ,....,𝑢𝑢𝑡𝑡+𝑁𝑁−1
Subject to:
• xt+k+1 = f(xt+k , ut+k ) presents the dynamic of the system.
• yt+k = g(xt+k , ut+k ) presents the outputs of the system
• umin ≤ ut+k ≤ umax
• ymin ≤ yt+k ≤ ymax
• xt = x(t), k = 0, … . , N − 1 presents state variables
In the following, we will present the design of linear and Nonlinear
MPC using MATLAB. The software provides some facilities regarding
16 of 28
MPC controller, where the optimization problem is solved using a

quadratic program (QP) – at each control interval. The solution deter-
mines the manipulated variables (MVs) to be used in the plant until the
next control interval. The definition of our cost function in the software
is written as follow:
Θ(𝓏𝓏𝑘𝑘 ) = Θ𝑦𝑦 (𝓏𝓏𝑘𝑘 ) + Θ𝑢𝑢 (𝓏𝓏𝑘𝑘 ), (35)
𝑛𝑛𝑦𝑦 𝑝𝑝 𝑦𝑦 2
𝑤𝑤𝑖𝑖,𝑗𝑗
Θ(𝓏𝓏𝑘𝑘 ) = � � �� 𝑦𝑦 �𝑟𝑟𝑗𝑗 (𝑘𝑘 + 𝑖𝑖|𝑘𝑘) − 𝑦𝑦𝑗𝑗 (𝑘𝑘 + 𝑖𝑖|𝑘𝑘)��
𝑠𝑠𝑗𝑗
𝑗𝑗=1 𝑖𝑖=1
𝑦𝑦 2 (36)
𝑤𝑤𝑖𝑖,𝑗𝑗
+ � 𝑦𝑦 �𝑢𝑢𝑗𝑗 (𝑘𝑘 + 𝑖𝑖|𝑘𝑘) − 𝑢𝑢𝑗𝑗,𝑡𝑡𝑡𝑡𝑡𝑡𝑡𝑡𝑡𝑡𝑡𝑡 (𝑘𝑘 + 𝑖𝑖|𝑘𝑘)��
𝑠𝑠𝑗𝑗
where: k is the current control interval; p is the Prediction horizon

(number of intervals); ny : Number of plant output variables. 𝓏𝓏k : Quad-
ratic program decision given by:
𝓏𝓏𝑘𝑘 𝑇𝑇 = [𝑢𝑢(𝑘𝑘|𝑘𝑘)𝑇𝑇 … . . 𝑢𝑢(𝑘𝑘 + 𝑝𝑝 − 1|𝑘𝑘)𝑇𝑇 𝜀𝜀𝑘𝑘 ] (37)
yj (k + i|k): Predicted value of jth plant output at ith prediction horizon

step, in engineering units. rj (k + i|k): Reference value for jth plant out-
y
put at ith prediction horizon step, in engineering units; sj : Scale factor
y
for jth plant output, in engineering units; wi,j : Tuning weight for jth
plant output at ith prediction horizon step (dimensionless).
2.3.2. Linear Model Predictive Control
Linear MPC is the traditional model which the basic of its cost
function is a quadratic function, the model traits linear dynamic sys-
tems, and does not require higher computation to solve the optimization
problem. The basic step to continue with linear MPC is to linearize the
dynamic model of the quadrotor, so the methodology is presented as
follow:
• Linearization of the quadrotor system

Our aim from this part is to generate a linear state space represen-
tation from a nonlinear model written in eq (29) and eq (30) (The non-
linear model is linearized around an equilibrium point Xe “ hover posi-
tion “, the linearization is performed on a simplified model, where small
oscillations are considered as: sin(angle) = angle and cos(angle) = 1,
and using Taylor series expansion the new state space representation af-
ter the linearization is written in the following form:
⃗̇ ⃗′ �⃗′ where 𝑋𝑋⃗ ′ = 𝑋𝑋⃗ − 𝑋𝑋⃗𝑒𝑒 ,

�𝑋𝑋 = 𝐴𝐴𝑋𝑋′ + 𝐵𝐵𝑢𝑢 (38)
𝑦𝑦⃗ = 𝐶𝐶𝑋𝑋⃗ + 𝐷𝐷𝑢𝑢
�⃗′ �⃗′ = 𝑢𝑢
𝑢𝑢 �⃗ − 𝑢𝑢
�⃗𝑒𝑒
where �X⃗ is the ‘State vector’ of (n × 1), �u⃗ is the ‘Input (or control) vec-
tor’of (m × 1), y �⃗ is the ‘Output vector’ of (p × 1), A is the ‘System Ma-
trix’ of (n × n), B is the ‘Input Matrix’ of (n × p), C is the ‘Output Matrix’
of (q × n), D is the ‘Feed forward Matrix’ of (q × p).
17 of 28
In hover position, we have the following state vector and input vec-
tor:
𝑋𝑋𝑒𝑒 = [𝑥𝑥𝑒𝑒 𝑦𝑦𝑒𝑒 𝑧𝑧𝑒𝑒 0 0 0 0 0 0 0 0 0]𝑇𝑇 , (39)
𝑢𝑢𝑒𝑒 = [𝑚𝑚𝑚𝑚 0 0 0]𝑇𝑇 (40)
After the linearization calculation we get the following four matri-

ces for the state space representation:
0 0 0 0 0 0 1 0 0 0 0 0
⎡ 0 0 0 0 0 0 0 1 0 0 0 0⎤
⎢ 0 0 0 0 0 0 0 0 1 0 0 0⎥
⎢0 0 0 0 0 0 0 0 0 1 0 0⎥
⎢ ⎥
⎢0 0 0 0 0 0 0 0 0 0 1 0⎥
𝜕𝜕𝜕𝜕(𝑋𝑋,𝑢𝑢)
𝐴𝐴 = �𝑥𝑥=𝑥𝑥𝑒𝑒 = ⎢0 0 0 0
0
0
𝑔𝑔
0
0
0 0 0 0 0 1⎥,
(41)
𝜕𝜕𝜕𝜕 ⎢0 0 0 0 0 0 0 0 0⎥
𝑢𝑢=𝑢𝑢𝑒𝑒 ⎢0 0 0 −𝑔𝑔 0 0 0 0 0 0 0 0⎥
⎢0 0 0 0 0 0 0 0 0 0 0 0⎥
⎢0 0 0 0 0 0 0 0 0 0 0 0⎥
⎢0 0 0 0 0 0 0 0 0 0 0 0⎥
⎣0 0 0 0 0 0 0 0 0 0 0 0⎦
0 0 0 0
⎡ 0 0 0 0 ⎤
⎢ 0 0 0 0 ⎥
⎢ 0 0 0 0 ⎥
𝜕𝜕𝜕𝜕(𝑋𝑋,𝑢𝑢) ⎢ ⎥
⎢ 0 0 0 0 ⎥
𝐵𝐵 = � 𝑥𝑥=𝑥𝑥𝑒𝑒 = 0 0 ⎥ (42)
𝜕𝜕𝜕𝜕 ⎢ 0 0
𝑢𝑢=𝑢𝑢𝑒𝑒 ⎢ 0 0 0 0 ⎥
⎢1/𝑚𝑚 1/𝐽𝐽𝑥𝑥𝑥𝑥 0 0 ⎥
⎢ 0 0 1/𝐽𝐽𝑦𝑦𝑦𝑦 0 ⎥
⎣ 0 0 0 1/𝐽𝐽𝑧𝑧𝑧𝑧 ⎦
The outputs variable that we would like to track, and control are
the following: [x y z ϕ θ ψ], therefore the matrix C and D are written as
follow:
1 0 0 0 0 0 0 0 0 0 0 0
⎡0 1 0 0 0 0 0 0 0 ⎤
0 0 0
⎢ 1 ⎥
0 0 0 0 0 0
𝐶𝐶 = ⎢0 0 0 0 0⎥,
(43)
⎢0 0 0 1 0 0 0 0 0 0 0 0⎥
⎢0 0 0 0 1 0 0 0 0 0 0 0⎥
⎣0 0 0 0 0 1 0 0 0 0 0 0⎦
𝐷𝐷 = 𝑧𝑧𝑧𝑧𝑧𝑧𝑧𝑧(6,4) (44)
If the disturbance is taking into consideration, so the matrix dist is

added to the equation of the state space as follow:
𝑋𝑋⃗̇ = 𝐴𝐴𝑋𝑋⃗ ′ + 𝐵𝐵𝑢𝑢

�⃗′ + 𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑⃗, (45)
𝜕𝜕𝜕𝜕(𝑋𝑋, 𝑢𝑢, 𝑑𝑑)

𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑 = � 𝑥𝑥=𝑥𝑥𝑒𝑒 (46)
𝜕𝜕𝜕𝜕
𝑢𝑢=𝑢𝑢𝑒𝑒
where d�⃗ is the vector of the disturbances applied on the quadrotor. In

the case presented in this paper, we neglect all the disturbances.
• Design Linear MPC for Tello Quadrotor System
Using MATLAB Simulink, we designed the dynamic system and
the linear MPC to track the following outputs variables: [x y z ϕ θ ψ], we
started by reconfigure the main inputs and outputs of the controller.
Figure 7 shows the necessary inputs and outputs for MPC controller, as
18 of 28
inputs we have: - mo: 6 measured outputs that present the state vector
[x y z ϕ θ ψ], - md: 0 measured disturbances where we neglected all the
disturbances applied on the quadrotor, as outputs: - mv: 4 manipulated
variables [u1 , u2 , u3 , u4 ], which present the input control variables.
Inside MPC plant model, different recommendations can be taking
into consideration, depending on the characteristics of the dynamic sys-
tem that we have, in our case the constraints are the weights, and the
control horizon were setting as presented in the Table 3. The weight re-
fers to the priority of the variables, where the high weight indicates the
higher priority to track.
Figure 7. MPC controller inputs - outputs
Table 3. The required parameters of MPC for Tello quadrotor
MPC Parameters
Sample time (𝑇𝑇𝑠𝑠 ) 0.1 s
Prediction Horizon (P) 20 s
Control Horizon (M) 3s
MPC Constraints
Boundaries of 𝑢𝑢1 [-10 – 12] N.m
MPC Weights
Translation motion along 𝑥𝑥 1
Translation motion along 𝑦𝑦 1
Translation motion along 𝑧𝑧 1
Pitch motion along 𝜙𝜙 0.1
Roll motion along 𝜃𝜃 0.1
Yaw motion along 𝜓𝜓 0.1
2.3.3. Nonlinear Model Predictive Control

The principle of nonlinear model predictive control is the same as
the linear one, the key differences are:
• the prediction model can be nonlinear and include time-varying
parameters.
19 of 28
• the equality and inequality constraints can be nonlinear.

• the scalar cost function to be minimized can be a nonquadratic (lin-
ear or nonlinear) function of the decision variables.
When the dynamic model is linear, the global analytical solution of
the optimization problem exists and is relatively simple to determine
[35]. But when this model is nonlinear, it is not certain that the optimiza-
tion algorithm converges to the global analytical solution and if a solu-
tion is found, there is not sure that it is fast enough. Indeed, the use of a
nonlinear model involves solving a nonlinear optimization problem
whose complexity is related to the degree of accuracy of the model.
Several numerical techniques were developed to guarantee the
convergence of the optimization algorithm towards an approximation of
the optimum overall. While some of these methods, such as Eleanor's
method, have proven their efficiency, they are however not appropriate
for control of drones due to their running time. Among the techniques
that can be carried on board vehicles we can cite the gradient method,
the quadratic programming method sequential or the particle swarm
method. Some techniques are based also on the approximation of certain
signals by their Taylor series expansion [36]–[40].
While traditional linear MPC controllers optimize control actions to
minimize a quadratic cost function, nonlinear MPC controllers support
generic custom cost functions. For example, the cost function can be
specified as a combination of linear or nonlinear functions of the system
states and inputs. To improve computational efficiency, we can also
specify an analytical Jacobian for the custom cost function [41].
Using a custom cost function, the algorithm of NMPC will be able
to maximize profitability and minimize energy consumption.
• Design Nonlinear MPC for Tello quadrotor system
As we did in the previous part with linear MPC, the design of
NMPC was done in MATLAB Simulink environment, MATLAB Sim-
ulink already has a basic architecture block function of such controller.
We had to reconfigure the internal structure according to our system
and tune the main parameters to achieve high performance. According
to the nonlinear dynamic system of Tello quadrotor, the controller
should receive different input signals: the reference trajectory – 12 state
variables – last 4 manipulated variables that present the four control in-
puts variables[u1 ′, u2 ′, u3 ′, u4 ′] as presented in Figure 8, in our case we
neglected the gyroscopic effect and the other disturbances applied on
the system, whereas outputs we receive the new updated 4 manipulated
variables.
The design process for NMPC was as follows: - an object function
nlobj has been created in the internal block with setting the necessary
parameters and using nlmpc function a nonlinear MPC controller is de-
signed. As drawback of nonlinear controllers is the required huge time
simulation, to avoid this issue and guarantee a high performance for the
system an algorithm to calculate Jacobian function was inserted to the
controller block function, Jacobian function guarantee to have fast and
efficient simulation.
20 of 28
Figure 8. Nonlinear MPC controller inputs – outputs
Table 4 identify the necessary parameters for the nonlinear MPC that fit
Tello quadrotor. The weight parameter in the controller refers to the
priority of the variables to track; usually in the quadrotor, we set higher
priority for the position variables than the orientation variables, in our
case we are looking for the best performance, so we set higher priority
for all six variables.
Table 4. The required parameters of Nonlinear MPC for Tello quadrotor
Nonlinear MPC Parameters

Sample time (𝑇𝑇𝑠𝑠 ) 0.1 s
Prediction Horizon (P) 18 s
Control Horizon (M) 3s
Nonlinear MPC Constraints
Boundaries of 𝑢𝑢1 [0 – 12] N.m
Nonlinear MPC Weights
Translation motion along 𝑥𝑥 1
Translation motion along 𝑦𝑦 1
Translation motion along 𝑧𝑧 1
Pitch motion along 𝜙𝜙 1
Roll motion along 𝜃𝜃 1
Yaw motion along 𝜓𝜓 1
3. Simulation Results
In this section, several simulations for three different trajectories
are presented to validate the comparison between linear and nonlinear
MPC based on the performance analyses of the following inputs and
outputs:
• error position tracking
• velocity tracking
• control inputs tracking.
21 of 28
The design of the linear and nonlinear dynamic model of the drone
beside the linear and nonlinear MPC models were designed in
MATLAB Simulink, where different functions were written in MATLAB
Script as: trajectory function, which provides to the control loop the ref-
erence desired trajectory; Jacobian function, which guarantees solving
the optimization problem and avoiding the singularities.
Three generated trajectories are defined: the first represents a
straight curvature motion of the drone, the second trajectory is present-
ed as a circle motion, while the third trajectory is a helical motion, the
initial states for both controllers were set as follow: 𝑋𝑋𝑖𝑖 =
[0 2 2 0 0 0 0 0 0 0 0 0]𝑇𝑇 .
For Linear MPC, the control input bounds were set as presented in
Table 3 between minimum value of -10 and maximum value of 12. This
choice leads to give the controller more freedom to track the trajectory,
where if the minimum value is set to 0. The controller will behave in an
aggressive scenario. For Nonlinear MPC the control input boundaries
were set between minimum value of 0 and maximum value of 12. An-
other setting was made regarding the control inputs tracking, where for
Nonlinear MPC, the tracking control inputs can be done successfully so
the track values where set to [4.9 4.9 4.9 4.9], while for Linear MPC, this
option was neglected because it will definitely disturb the controller to
achieve the most priority task, which is tracking the position and orien-
tation outputs. From the first attempt and without looking to the simu-
lation results, the recognition of these limitations could identify that
Nonlinear MPC is more flexible than Linear MPC in term of setting
conditions.
3.1. Straight Curvature Reference Trajectory

Figure 9 presents the simulation results of the straight curvature ref-
erence trajectory, the error position tracking, the velocity tracking, and
the control inputs tracking for linear and nonlinear MPC. From Figure 9,
we observe that regarding error position tracking LMPC performs better
regarding the three axis X, Y, and Z, where for orientations angles, the
signal is disturbed comparing to NLMPC where the oscillations are
minimized. These oscillations describe the instability of the drone. Also,
at level of tracking velocity, where both of controllers are behaving
nearly the same but always NMPC is able to minimize the peak ampli-
tude of the oscillations. From the energy profile perspective, the com-
parison of the control inputs tracking leads to identify that both control-
lers reach the maximum and the minimum boundaries set earlier, to
know which controller is using more torque, the mean value of ∆𝑢𝑢 for
every execution is calculated, this will be presented in the following ta-
ble (Table 5).
3.2. Circle Reference Trajectory

Figure 10 presents the simulation results of circle motion as refer-
ence trajectory. From this Figure, the differences start to rise clearly,
where it is obvious that NLMP is tracking the positions and the veloci-
ties in stable behavior, where the oscillations are more presented with
LMPC, otherwise NLMP has succeed to minimize them. Also at control
inputs tracking level, LMPC is using the maximum and minimum
22 of 28
Figure 9. Straight curvature trajectory there is problem in the picture
boundaries so it can reach the minimum value – 10 N.m where NMPC is

acting well with a limited boundaries with minimum value 0 N.m. Also
NMPC is able to track the set target control inputs [4.9 4.9 4.9 4.9].
3.3. Helical Reference Trajectory

Figure 11 presents the simulation results of helical motion as refer-
ence trajectory. This motion describes typically the real life, where the
motion is depending on the three axes simultaneously, therefore we ob-
serve that NMPC succeed to track the three positions smoothly and with
minimum peak amplitude, while in the orientation tracking an over-
shoot is appeared in the first two seconds, regarding the velocity track-
ing. NMPC also provide good results comparing to LMPC, where the
tracking control inputs is well achieved by NMPC. LMPC is always us-
ing all the boundaries to execute the suitable control behavior.
23 of 28
Figure 10. Circle reference trajectory
3.4. Performance measurments ∆𝑢𝑢 and ExTime

In order to analyze the energy profile of both controllers, a calcula-
tion of the mean average value of the control effort ∆𝑢𝑢 and the execution
time of the simulation 𝐸𝐸𝐸𝐸𝑇𝑇𝑇𝑇𝑇𝑇𝑇𝑇 was made.
Table 5 shows the results of the two values when every trajectory is
executed. These two parameters identify the consumption energy for
both controllers, where regarding the execution time, we found that
LMPC is very slow comparing to NMPC, meanwhile the average of con-
trol effort for LMPC is less than the control effort required by NMPC
with average differences. This can be explained by the fact that NMPC
has different missions to achieve where it is tracked not only the trajec-
tory but also try to maintain a target value for the control inputs. A solu-
tion can be highlighted in the future by optimizing a cost function that
minimizes the energy consumption of the drone, by minimizing the ap-
plied control efforts directed to the four motors.
The calculation of ∆𝑢𝑢 was by mean function, where the execution
simulation time was calculated by timeit function.
24 of 28
Figure 11. Circle reference trajectory
Table 5. Calculation of the average control effort value and the execution time
Trajectory / Con- 1st Trajectory 1 2nd Trajectory 3rd Trajectory

troller
Δu[N.m]/ExTime[ms] Δu ExTime Δu ExTime Δu ExTime
LMPC 3.1678 64 2.7054 63 1.4229 62
NLMPC 5.9188 6.8 5.2379 6 4.8955 5.7
4. Summary and Conclusion

This paper presented a methodology to design MPC controller with
the purpose to track trajectory of the quadrotor system. The dynamic
mathematical modeling part for the quadrotor was deeply described in
the first section, where in the second section a theoretical comparison
between linear and nonlinear MPC was presented, which conducted us
to look after the efficiency of these both control laws especially regard-
ing the energy profile. The design of both controller models was built
and configured using MATLAB Script and Simulink, the controllers
were simulated under the execution of three different trajectories, to see
25 of 28
the resulted behavior. As a conclusion, we sum up that NMPC can pro-

vide a very promoting results comparing to LMPC, which is not flexible
and rise different limitations in the control drone field.
As future work, our main goal is to build a master controller hard-
ware embedded system for Tello quadrotor, the hardware will be em-
bedded by such a control law, as we identified how NMPC is fast to
process the commands, it would be great to increase the potential of it
regarding the optimization of energy consumption, which presents a
drawback of the quadrotor system, the target would be to add to NMPC
a new cost function that minimize the voltage from the control efforts
which is directed to the four rotors.
Author Contributions: “Conceptualization, R.B., V.J.; methodology, R.B.,
V.J.; formal analysis, R.B., V.J.; invited author by guest editor, V.J.; writing—
original draft preparation, R.B., V.J.; writing—review and editing; R.B., V.J.”
Funding: “This research received no external funding”
References
[1] R. K. Barnhart, S. B. Hottman, D. M. Marshall, and E. Shappee, Introduction to Unmanned Aircraft Systems,
1st ed. Boca Raton: CRC Press, 2016. doi: 10.1201/B11202.
[2] K. Nonami, F. Kendoul, S. Suzuki, W. Wang, and D. Nakazawa, Autonomous flying robots: Unmanned aerial
vehicles and micro aerial vehicles. Tokyo: Springer Japan, 2010. doi: 10.1007/978-4-431-53856-1/COVER.
[3] G. D. Padfield, C. A. Kluever, T. M. Young, A. J. Keane, A. Sóbester, and J. P. Scanlan, “Design of Un-
manned Aerial Systems,” Design of Unmanned Aerial Systems, Mar. 2020, doi: 10.1002/9781119508618.
[4] K. Dalamagkidis, Classification of uavs. Dordrecht: Springer Netherlands, 2014. doi: 10.1007/978-90-481-
9707-1_94/COVER.
[5] M. Idrissi, M. Salami, and F. Annaz, “A Review of Quadrotor Unmanned Aerial Vehicles: Applications,
Architectural Design and Control Algorithms,” Journal of Intelligent & Robotic Systems , vol. 104, no. 2, pp.
1–33, Jan. 2022, doi: 10.1007/S10846-021-01527-7.
[6] F. M. Agugliaro, M. Jacewicz, M. ˙ Zugaj, R. Gł˛, and P. Bibik, “Quadrotor Model for Energy Consump-
tion Analysis,” Energies 2022, Vol. 15, Page 7136, vol. 15, no. 19, p. 7136, Sep. 2022, doi:
10.3390/EN15197136.
[7] X. Zhang, X. Li, K. Wang, and Y. Lu, “A survey of modelling and identification of quadrotor robot,” Ab-
stract and Applied Analysis, vol. 2014, 2014, doi: 10.1155/2014/320526.
[8] X. Zhang, X. Li, K. Wang, and Y. Lu, “A survey of modelling and identification of quadrotor robot,” Ab-
stract and Applied Analysis, vol. 2014, 2014, doi: 10.1155/2014/320526.
[9] E. Balestrieri, P. Daponte, L. de Vito, and F. Lamonaca, “Sensors and Measurements for Unmanned Sys-
tems: An Overview,” Sensors 2021, Vol. 21, Page 1518, vol. 21, no. 4, p. 1518, Feb. 2021, doi:
10.3390/S21041518.
26 of 28
[10] S. Guan, Z. Zhu, and G. Wang, “A Review on UAV-Based Remote Sensing Technologies for Construction
and Civil Applications,” Drones 2022, Vol. 6, Page 117, vol. 6, no. 5, p. 117, May 2022, doi:
10.3390/DRONES6050117.
[11] G. Barbaraci, “Modeling and control of a quadrotor with variable geometry arms,” J Unmanned Veh Syst,
vol. 3, no. 2, pp. 35–57, 2015, doi: 10.1139/JUVS-2014-0012.
[12] Z. Tahir, M. Jamil, S. A. Liaqat, L. Mubarak, W. Tahir, and S. O. Gilani, “State Space System Modeling of
a Quad Copter UAV,” Indian J Sci Technol, vol. 9, no. 27, Jul. 2016, doi: 10.17485/IJST/2016/V9I27/96613.
[13] A. Chovancová, T. Fico, E. Chovanec, and P. Hubinský, “Mathematical Modelling and Parameter Identi-
fication of Quadrotor (a survey),” Procedia Eng, vol. 96, pp. 172–181, Jan. 2014, doi:
10.1016/J.PROENG.2014.12.139.
[14] M. R. Mokhtari and B. Cherki, “A new robust control for minirotorcraft unmanned aerial vehicles,” ISA
Trans, vol. 56, pp. 86–101, May 2015, doi: 10.1016/J.ISATRA.2014.12.003.
[15] Z. Bodó and B. Lantos, “Modeling and Control of Outdoor Quadrotor UAVs,” SISY 2018 - IEEE 16th In-
ternational Symposium on Intelligent Systems and Informatics, Proceedings, pp. 111–116, Nov. 2018, doi:
10.1109/SISY.2018.8524697.
[16] G. J. J. Ducard and M. Allenspach, “Review of designs and flight control techniques of hybrid and con-
vertible VTOL UAVs,” Aerosp Sci Technol, vol. 118, p. 107035, Nov. 2021, doi: 10.1016/J.AST.2021.107035.
[17] A. N. Chand, M. Kawanishi, and T. Narikiyo, “Adaptive pole placement pitch angle control of a flap-
ping-wing flying robot,” Advanced Robotics, vol. 30, no. 16, pp. 1039–1049, Aug. 2016, doi:
10.1080/01691864.2016.1196609.
[18] N. Chen, F. Song, G. Li, X. Sun, and C. Ai, “An adaptive sliding mode backstepping control for the mo-
bile manipulator with nonholonomic constraints,” Commun Nonlinear Sci Numer Simul, vol. 18, no. 10, pp.
2885–2899, Oct. 2013, doi: 10.1016/J.CNSNS.2013.02.002.
[19] A. Bogdanov et al., “State-dependent riccati equation control of a small unmanned helicopter,” AIAA
Guidance, Navigation, and Control Conference and Exhibit, 2003, doi: 10.2514/6.2003-5672.
[20] G. v. Raffo, M. G. Ortega, and F. R. Rubio, “Robust Nonlinear Control for Path Tracking of a Quad-Rotor
Helicopter,” Asian J Control, vol. 17, no. 1, pp. 142–156, Jan. 2015, doi: 10.1002/ASJC.823.
[21] J. P. How, E. Frazzoli, and G. V. Chowdhary, Linear Flight Control Techniques for Unmanned Aerial Vehicles.
Springer, Dordrecht, 2015. doi: 10.1007/978-90-481-9707-1_49.
[22] Z. Bodó and B. Lantos, “Integrating Backstepping Control of Outdoor Quadrotor UAVs,” Periodica Poly-
technica Electrical Engineering and Computer Science, vol. 63, no. 2, pp. 122–132, Feb. 2019, doi:
10.3311/PPEE.13321.
[23] F. Gao et al., “Optimal trajectory generation for quadrotor teach-and-repeat,” IEEE Robot Autom Lett, vol.
4, no. 2, pp. 1493–1500, Apr. 2019, doi: 10.1109/LRA.2019.2895110.
[24] Ryzerobotics User Manual, “User Manual 2018.09,” pp. 1–22, 2018.
27 of 28
[25] R. Benotsmane, A. Reda, and J. Vasarhelyi, “Model Predictive Control for Autonomous Quadrotor Tra-
jectory Tracking,” 2022 23rd International Carpathian Control Conference, ICCC 2022, pp. 215–220, 2022, doi:
10.1109/ICCC54292.2022.9805883.
[26] R. Roy, M. Islam, N. Sadman, M. A. P. Mahmud, K. D. Gupta, and M. M. Ahsan, “A Review on Compar-
ative Remarks, Performance Evaluation and Improvement Strategies of Quadrotor Controllers,” Technol-
ogies 2021, Vol. 9, Page 37, vol. 9, no. 2, p. 37, May 2021, doi: 10.3390/TECHNOLOGIES9020037.
[27] G. Farid, M. Hongwei, S. M. Ali, and Q. Liwei, “A REVIEW ON LINEAR AND NONLINEAR CON-
TROL TECHNIQUES FOR POSITION AND ATTITUDE CONTROL OF A QUADROTOR,” Mechatronic
Systems and Control 2017, vol. 45, no. 1, pp. 43–57, Jan. 2017, doi: 10.2316/JOURNAL.201.2017.1.201-2819.
[28] B. J. Emran and H. Najjaran, “A review of quadrotor: An underactuated mechanical system,” Annu Rev
Control, vol. 46, pp. 165–180, Jan. 2018, doi: 10.1016/J.ARCONTROL.2018.10.009.
[29] O. Araar and N. Aouf, “Full linear control of a quadrotor UAV, LQ vs H∞,” 2014 UKACC International
Conference on Control, CONTROL 2014 - Proceedings, pp. 133–138, Oct. 2014, doi:
10.1109/CONTROL.2014.6915128.
[30] A. Koszewnik, “The parrot UAV controlled by PID controllers,” Acta Mechanica et Automatica, vol. 8, no.
2, pp. 65–69, 2014, doi: 10.2478/ama-2014-0011.
[31] J. B. Rawlings, E. S. Meadows, and A. D. K. R. Muske, “Nonlinear Model Predictive Control: A Tutorial
and Survey,” IFAC Proceedings Volumes, vol. 27, no. 2, pp. 185–197, May 1994, doi: 10.1016/S1474-
6670(17)48151-1.
[32] A. Isidori, L. Marconi, and A. Serrani, “Robust nonlinear motion control of a helicopter,” IEEE Trans Au-
tomat Contr, vol. 48, no. 3, pp. 413–426, Mar. 2003, doi: 10.1109/TAC.2003.809147.
[33] S. J. Qin and T. A. Badgwell, “An Overview of Nonlinear Model Predictive Control Applications,” Non-
linear Model Predictive Control, pp. 369–392, 2000, doi: 10.1007/978-3-0348-8407-5_21.
[34] C. B. E. F. Camacho, Model Predictive Control, no. July. London: Springer London, 2000. doi: 10.1007/978-0-
85729-398-5.
[35] P. Ru and K. Subbarao, “Nonlinear Model Predictive Control for Unmanned Aerial Vehicles,” Aerospace
2017, Vol. 4, Page 31, vol. 4, no. 2, p. 31, Jun. 2017, doi: 10.3390/AEROSPACE4020031.
[36] M. Kamel, M. Burri, and R. Siegwart, “Linear vs Nonlinear MPC for Trajectory Tracking Applied to Ro-
tary Wing Micro Aerial Vehicles,” IFAC-PapersOnLine, vol. 50, no. 1, pp. 3463–3469, Jul. 2017, doi:
10.1016/J.IFACOL.2017.08.849.
[37] M. Ławryńczuk and R. Nebeluk, “Computationally Efficient Nonlinear Model Predictive Control Using
the L1 Cost-Function,” Sensors 2021, Vol. 21, Page 5835, vol. 21, no. 17, p. 5835, Aug. 2021, doi:
10.3390/S21175835.
[38] P. D. Domanski, “Performance Assessment of Predictive Control—A Survey,” Algorithms 2020, Vol. 13,
Page 97, vol. 13, no. 4, p. 97, Apr. 2020, doi: 10.3390/A13040097.
28 of 28
[39] A. Arce Rubio, A. Seuret, A. Mannisi, Y. Ariba, and A. Arce, “Optimal control strategies for load carrying
drones,” vol. 6, pp. 183–197, 2016, doi: 10.1007/978-3-319-32372-5.
[40] Y. al Younes and M. Barczyk, “Nonlinear Model Predictive Horizon for Optimal Trajectory Generation,”
Robotics 2021, Vol. 10, Page 90, vol. 10, no. 3, p. 90, Jul. 2021, doi: 10.3390/ROBOTICS10030090.
[41] R. Benotsmane and J. Vásárhelyi, “Nonlinear Model Predictive Control for Autonomous Quadrotor Tra-
jectory Tracking,” in Lecture Notes in Mechanical Engineering, 4th ed., K. , C. Á. Jármai, Ed. Springer,
Cham, 2023, pp. 24–34. doi: 10.1007/978-3-031-15211-5_3.

Lo Quadrotor With MPC Model Implementation

Uploaded by

Copyright:

Available Formats

Lo Quadrotor With MPC Model Implementation

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Lo Quadrotor With MPC Model Implementation

Uploaded by

Copyright:

Available Formats

Preprints (www.preprints.org) | NOT PEER-REVIEWED | Posted: 10 November 2022 doi:10.20944/preprints202211.0181.

Towards Optimization of Energy Consumption of Tel-

1 Institute of Automation and Info-Communication; [email protected]

2 Institute of Automation and Info-Communication; [email protected]

Keywords: dynamic control; UAV; model predictive control; nonlinear MPC;

© 2022 by the author(s). Distributed under a Creative Commons CC BY license.

Figure 1. The possible motions of quadrotor with cross configuration

Due to the very good distribution of lift in the horizontal plane,

The main goal of drones is to achieve a high degree of autonomy to

At the energy profile, one of the biggest drawbacks of the quad-

2. Materials and Methods

The main aim of this work is to build a customized embedded sys-

2.1. Tello Quad-rotor Model

Figure 3. Tello drones basic components

2.2. Quadroto Mathematical Model

explained the methodology for modeling such systems (references?).

Figure 4. Modeling frame assignments for Tello quadrotor [25]

The quadrotor system is a light structural flight vehicle. Therefore,

Let us describe [x y z ϕ θ ψ] T the vector containing the linear and

1 0 −𝑆𝑆𝑆𝑆 1 𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆 𝐶𝐶𝜙𝜙𝜙𝜙𝜙𝜙

𝑥𝑥̇ = 𝑢𝑢[𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶] + 𝑣𝑣[𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶 − 𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶] + 𝑤𝑤[𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶 + 𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆]

• where i = 1 … 4; b is the coefficient of lift it depends on the shape

where: d is the drag coefficient; it depends on the manufacture of the

with: K ft is the translational drag coefficient, and v is the linear velocity.

𝑀𝑀𝑧𝑧 = 𝑑𝑑(𝜔𝜔12 − 𝜔𝜔22 + 𝜔𝜔32 − 𝜔𝜔42 ), (11)

• The inertial moment resulting from aerodynamic friction is given

where: 𝐾𝐾𝑓𝑓𝑓𝑓 the coefficient of aerodynamic friction angular

The gyroscopic moment due to the movements of the quadrotor

𝑀𝑀𝑔𝑔𝑔𝑔 = Ω ∧ 𝐽𝐽Ω (14)

where: Jr is the moment of inertia of the rotor, J is the moment of inertia

𝒎𝒎𝐯𝐯̇ 𝒊𝒊𝒊𝒊𝒊𝒊𝒊𝒊𝒊𝒊𝒊𝒊𝒊𝒊𝒊𝒊 = ∑ 𝑭𝑭𝒆𝒆𝒆𝒆𝒆𝒆𝒆𝒆𝒆𝒆𝒆𝒆𝒆𝒆𝒆𝒆 , (15)

𝒎𝒎𝝃𝝃̈ = 𝑭𝑭𝒕𝒕 + 𝑭𝑭𝒅𝒅𝒅𝒅 + 𝑭𝑭𝒈𝒈, (16)

𝐹𝐹𝑡𝑡 = 𝑅𝑅[0 0 ∑4𝑖𝑖=1 𝐹𝐹𝑖𝑖 ]𝑇𝑇 ; 𝐹𝐹𝑔𝑔 = [0 0 −𝑚𝑚𝑚𝑚]𝑇𝑇 (17)

By replacing all the terms that we have presented, we get:

𝑥𝑥̈ 𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶 + 𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆 𝐾𝐾𝑓𝑓𝑓𝑓𝑓𝑓 𝑥𝑥̇ 0

• Dynamic equation of rotation motion

𝐉𝐉𝛀𝛀̇ + 𝛀𝛀 × 𝑱𝑱𝛀𝛀 = � 𝑴𝑴𝒆𝒆𝒆𝒆𝒆𝒆𝒆𝒆𝒆𝒆𝒆𝒆𝒆𝒆𝒆𝒆 (21)

where Ω in this equation represent the rotation velocity presented in the

𝐉𝐉𝛀𝛀̇ = −𝛀𝛀 × 𝑱𝑱𝛀𝛀 − 𝑴𝑴𝒂𝒂 − 𝑴𝑴𝒈𝒈𝒈𝒈 + 𝑴𝑴𝒇𝒇 , (23)

By replacing all the terms in equation, we get:

� 𝑟𝑟 𝜃𝜃̇ − 𝐾𝐾𝑓𝑓𝑓𝑓𝑓𝑓 𝜙𝜙̇ 2 + 𝑙𝑙𝑙𝑙(𝜔𝜔42 − 𝜔𝜔22 )

The relation presented in this matrix is very useful for building

−(𝐽𝐽𝑧𝑧𝑧𝑧 − 𝐽𝐽𝑦𝑦𝑦𝑦 )𝜃𝜃̇𝜓𝜓̇ + 𝑙𝑙𝑙𝑙(𝜔𝜔42 − 𝜔𝜔22 )

2.2.1. Identification parameters of Tello EDU drone

Table 1. Parameters of Tello quadrotor

2.3. Control Strategy of Quadrotor System

Table 2. Parameters of Tello quadrotor

Controller Stability and Computation

In this paper both extensions are presented in detail, to show the

𝑋𝑋 = [𝑥𝑥 𝑦𝑦 𝑧𝑧 𝜙𝜙 𝜃𝜃 𝜓𝜓 𝑥𝑥̇ 𝑦𝑦̇ 𝑧𝑧̇ 𝜙𝜙̇ 𝜃𝜃̇ 𝜓𝜓̇]𝑇𝑇 , (31)

Figure 5. MPC closed control loop

Figure 6. Concept of Model Predictive Control

The standard cost function denotes Θ usually is the sum of four

MPC controller, where the optimization problem is solved using a

Θ(𝓏𝓏𝑘𝑘 ) = Θ𝑦𝑦 (𝓏𝓏𝑘𝑘 ) + Θ𝑢𝑢 (𝓏𝓏𝑘𝑘 ), (35)

where: k is the current control interval; p is the Prediction horizon

yj (k + i|k): Predicted value of jth plant output at ith prediction horizon

• Linearization of the quadrotor system

⃗̇ ⃗′ �⃗′ where 𝑋𝑋⃗ ′ = 𝑋𝑋⃗ − 𝑋𝑋⃗𝑒𝑒 ,