Anais do XIX Congresso Brasileiro de Automtica, CBA 2012.

CONTROL ARCHITECTURE FROM PEGASUS AUTOPILOT: DESIGN AND SOFTWARE-IN-THE-LOOP

SIMULATION
ADRIANO BITTAR, NEUSA M.F. OLIVEIRA, DIEGO A. CARVALHO
Instituto Tecnolgico de Aeronutica
Pa Mal. Eduardo Gomes, 12.228-900, So Jos dos Campos, SP.
E-mails: [email protected], [email protected], [email protected].
Abstract This paper describes a detailed autopilots control block design of a small fixed-wing unmanned air vehicle. The
block is composed of feed-backs and feed-forwards controls. The feedback control guarantees attitude stabilization, including
pitch and roll, and maintains the aircraft in a desired heading, altitude, and airspeed. The feed-forward control compensates two
coupling effects, the first one between the lateral-directional and longitudinal motion, and the second one between airspeed and
altitude hold. The proposed approach follows the standard practice of Software-In-the-Loop implementation, where the control
is developed on MATLAB/SIMULINK and the airplane response is simulated in the flight simulator X-Plane. In order to make
the system seem more real, the target aircraft Piper J3 Cub 1/6 scale was modeled in Plane-Maker and was used to verify,
validate, and optimize the controllers implemented on MATLAB/SIMULINK.
Keywords UAVs, Control, SIL, X-Plane, Matlab.
Resumo Este artigo descreve detalhadamente o projeto do bloco de controle do piloto automtico de um pequeno veculo
areo no tripulado de asa fixa. O bloco composto por malhas de feedback e feed-forward. O controle de feedback garante o
controle de atitude, incluindo o angulo de arfagem e rolagem, e mantm a aeronave em uma posio desejada, velocidade,
altitude. O controle feed-forward compensa dois efeitos de acoplamento, entre o movimento lateral-direcional e longitudinal, e
entre velocidade e controle de altitude. A abordagem proposta segue a prtica padro de implementao em Software-In-theLoop , onde o controle desenvolvido em MATLAB / SIMULINK e a resposta avio simulada no simulador de voo X-Plane.
A fim de tornar o sistema similar ao real, a aeronave alvo, um Piper J3 Cub escala 1/6, foi modelada no Plane-Maker e foi
utilizado para verificar, validar e otimizar os controladores implementados em MATLAB / SIMULINK
Palavras Chaves UAVs, Controle, SIL, X-Plane, Matlab.

1 Introduction
An autopilot is an electrical, mechanic, or
hydraulic system used to guide a vehicle without
human assistance. It can be coupled to any
autonomous, land, water, or air vehicle.
A Small Unmanned Aerial Vehicle (SUAV), which
is the focus of this work, must use an autopilot to
perform its missions. Due to great responsibilities
involved, a common practice is to test and evaluate
the autopilot by using a flight simulator, reducing the
risk of project failure (Chao, 2010)
The Pegasus AutoPilot has been developed to create
an efficient, complete, and low cost autopilot to be
used for all kinds of missions, especially in fixedwing unmanned aircrafts.
This paper will describe the architecture and design
of the control block used in the implementation of
Pegasus. The PID (proportional, integral, and
derivative) controllers were used, although not all
three gains are always present. The variables
controlled by feed-back closed-loop control are pitch,
roll, heading angles, climb rate, forward speed, and
altitude. In addition, the feed-forward control
establishes a connection between the forward speed
and altitude, the heading and altitude, and the
heading using both aileron and rudder.
The designed control is validated thought a SoftwareIn-the-Loop (SIL) simulation, using Matlab/Simulink
and the flight simulator X-Plane, where a target
aircraft was modeled. The navigation system, the
aircraft response, and visual are presented by the

ISBN: 978-85-8001-069-5

flight simulator, whereas the control runs on Matlab.

X-Plane has the built-in capability of transmitting
flight parameters and receiving command for aircraft
flight control surfaces over Ethernet using the User
Datagram Protocol (UDP). This capability is also
available on Matlab/Simulink, which facilitated the
implementation.
2 Autopilot
An autopilot consists of four main blocks:
guidance, navigation, control, and ground station
(Kendoul, 2009). Fig. (1) shows the block diagram of
the systems embedded in the aircraft and the
interaction between these systems.
The navigation block is responsible for providing the
current state (position, forward speed, and attitude)
of the aircraft and making them available for use by
guidance and control blocks (Jang, 2006). The data is
obtained through appropriate sensors. In SUAVs, the
most commonly sensors used are the Instrument
Measurement
Unit
(IMU),
composed
of
accelerometers, gyros, and magnetometers on the
three axes; a Global Positioning System (GPS); a
static pressure sensor; and a dynamic pressure sensor
(Yu,2007). The Pegasus AutoPilot uses all these
sensors.
The guidance block is responsible for generating
reference commands to the control block. It
interprets the data path to be followed, and generates
the desired values of heading, altitude, and airspeed
for the aircraft (Santos, 2011).

4146

Anais do XIX Congresso Brasileiro de Automtica, CBA 2012.

Fig. 2a. Real Aircraft

Figure 1. Blocks of an autopilot

The control block is responsible for reading the

reference values given by the block of guidance.
Using data from the navigation block, it operates in
the servos of the aircraft in order to reach the
reference values (Chen, 2011).
The control block is also responsible for maintaining
a constant attitude. It is called "Hold Autopilot"
when it also controls and keeps constant the heading,
altitude, and airspeed (Nelson, 1998).
3 The Aircraft
The model airplane chosen was a Piper J-3 CUB
48S, manufactured by "The World Models." The
main features of the aircraft are found in the
manufacturer's manual model, as follow:
1. Wingspan (Standard): 1.800mm [71.0 in]
2. Wing Area (Standard): 45.0 dm2 [698 sq in]
3. Weight: 2500g [5.5lbs]
4. Length: 1.200 mm [47 in]
5. Brushless Engine: 540W
6. Propeller: 12x7
The main factors for choosing this aircraft were the
easy acquisition, the stability of aircraft flight and for
having an experienced human pilot used to fly the
real aircraft.
The flight simulator model was performed using
Plane-Maker (v 9.22), which is a program bundled
with X-Plane. This tool allows the user to design any
imaginable aircraft. Plane-Maker provides an
interface to design a vehicle based on the physical
specifications of the airplane (weight, wing span,
control deflections, engine power, airfoil sections,
etc.), while the X-Plane simulator predicts how that
built plane flies. Fig. (2a) illustrates the real model,
whereas Fig. (2b) shows the model built in PlaneMaker.
The position of the surfaces on X-plane varies
linearly between -1 and +1, and the engine between 0
and +1. The real deflection of the surface on the
aircraft model is configured in Plane-Maker. For this
model X-Planes deflection of -1 corresponds to a
deflection of -20 degrees on the airplane models
surface and the +1 to a deflection of +20 degrees. For
the engine, the deflection of +1 means an output of
540 watts of power to the airplane, and 0 turns off
the engine.

ISBN: 978-85-8001-069-5

Fig. 2b. X-Plane Model Aircraft

4 Feed-Back Control
The feed-back closed-loop is constituted from the
longitudinal, the lateral-directional, and the airspeed
control, which together performs six hold controls.
All six closed-loops were implemented on
Matlab/Simulink, which communicates with X-Plane
through the UDP, the native communication interface
of this simulator (Ernest, 2007). The system operates
at a sample rate of 20Hz configurable on the flight
simulator.
4.1 The Lateral-Directional Control
The main problem with yaw control is that the
heading adjusted directly by aileron produces a very
unstable airplane behavior. This occurs because only
the aileron deflection cannot control the roll angle.
Using only the aileron deflection and not being
careful in control the roll angle, there is a risk that
the roll angle becomes excessive and it can cause
loss of air lift on the wings. Therefore we have to
consider the roll limits of the airplane in the heading
loop system to avoid this problem. Then it was
implemented inside the yaw outer loop control an
inner loop that controls the roll angle.
The outer control loop provides a roll reference angle
as the input signal to the inner loop. This roll
reference signal (
) is the input signal to the inner
loop.
The inner loop, receives the roll reference signal
(
), as it inputs. It generates the appropriate
command signal to the ailerons in order to the
aircrafts behaves as desired, respecting the roll limits
of the airplane and avoiding loss of air lift.
However, this nested loop structure brings more
difficulties to tune the controllers parameters. The
outer loop, the heading hold loop, uses a PID
controller, which makes necessary to set three
parameters. The inner loop, the roll hold loop, uses a
PD controller, with two parameters to be set. The
appropriate tuning of all parameters was done by

4147

Anais do XIX Congresso Brasileiro de Automtica, CBA 2012.

dividing the task in two steps, which allowed dealing

with single loop system in each step. First the roll
controller was tuned and then the heading controller.
1) Roll Hold Controller: The roll hold, illustrated in
Fig. 3, is responsible for changing and keeping the
aircrafts roll stable according to the input value, the
reference roll value (
). It presents three feedback
lines: the current roll (
) and the angular
velocity (p), feed the P-D controller (
and
gains), and the current aileron position (
),
calculates the exact amount of deflection that must
be applied on the aircrafts ailerons. The P-D control
also guarantees the stable angular velocity.
The feedback from the current position of the servo
comes from the fact that this closed-loop generates a
increment on the deflection (
), as in (1), to be
imposed on the aileron. However, on X-Plane, it is
necessary to set the position of the servo (as in a real
servo).
Hence, the current position is added to the deflection
calculated by the control generating the next servos
position ( ), accordingly to (2).
The feedback from the current position of the servo
comes from the fact that this closed-loop generates a
increment on the deflection (
), as in (1), to be
imposed on the aileron, however, on X-Plane, it is
necessary to set the position of the servo (as in a real
servo). Hence, the current position is added to the
deflection calculated by the control generating the
next servos position ( ), accordingly to (2).
(

(1)
(2)

Two saturators are used; the first is used to restrict

the maximum amount of variation that the servo of
the aileron may have, preventing the plane from
making rude movements when a new reference roll is
imposed. The second is to limit the maximum servo
position, which limits (by software) how many
degrees the aileron can deflect. This prevents the
case of a very high value being imposed to the
aileron, causing a loss of lift on the wings. The
references and the values of X-Plane are given in
degrees, but the Matlab has been configured to work
in radians, so it is necessarily to convert these
quantities properly.

ref

e
+

Aircraft

a
Kpa

current

SAT

SAT

acurrent
Kd a

acurrent

Figure 3. Roll Hold Control System Inner Loop

ISBN: 978-85-8001-069-5

2) Heading Hold Controller: This controller

changes the aircrafts heading and holds it in the
assigned reference value. It generates the roll
reference (
) to the roll hold system. Fig. (4)
shows the heading hold control system. This system
has two control loops, an inner and an outer loop.
The inner loop is the roll hold control system
presented previously.
The outer loop controls the heading indirectly by
ailerons through roll angle, controlled by the inner
loop.
The choice of this kind of scheme is motivated
by the necessity to obtain an efficient yaw axis
control by aileron actuation.
The error (
) between the desired heading
(
) and the current heading (
) is
calculated as in (3).
(

(3)

The roll reference is given by (4).

(4)

4.2 The Longitudinal Control

The longitudinal control consists on hold the
altitude. The control implemented in this paper
works as a chain where the only input is the altitude
reference. The altitude hold generates the reference
for the climb rate, which generates the reference for
the pitch hold. The pitch hold then generates the
elevators servo position. This architecture makes
sure that all these variables (altitude, climb rate, and
pitch) are controlled, improving the UAV stability.
This controller has two PIDs (altitude hold and climb
rate hold) and one P-D (pitch hold), which are
necessary to configure individually.
The terms climb rate and vertical velocity will be
used indistinctly along the text. Also the terms
horizontal speed, airspeed and forward speed.
1) Pitch Hold Controller: The pitch hold control
loop was implemented as illustrated in Fig. (5) and
is responsible for maintaining the pitch of the aircraft
according to the desired reference pitch (
).
Similar to the roll hold system, it uses a P-D
controller and three feedback lines.
The airplanes current pitch (current) and its angular
velocity (q) on the y axis, the pitch rate, are used to
feed the P-D controller, generating the elevator
deflection (e) as in (5).

ref

e
+

PID

Inner Loop

Kp
Ki

SAT

ref

Aircraft
a

Kd

current

Roll Control
HOLD

Figure 4. Heading Hold Control System Outer Loop

4148

Anais do XIX Congresso Brasileiro de Automtica, CBA 2012.

(5)

The current position of the elevator servo

is used to calculate the new elevators position (
as in (6).

)
),

ref e
+

(7)

Then, pitch deflection is added to the current pitch of

the airplane to generate the reference pitch,
accordingly to (8).

Kpe

SAT

SAT

ecurrent

current

(6)
Just like the roll hold controller, this loop requires
two saturators. The first one restricts the maximum
amount of variation that the servo of the elevator
may have, and the second one is used to limit the
maximum and minimum values of the elevator
position.
2) Climb Rate Hold Controller: The climb rate
loop, illustrated in Fig. (6), has two feedback lines.
The current climb rate (VVcurrent) feedback feeds the
PID controller and the other one, the current pitch
(
), is used to calculate the new reference
pitch value. This control loop calculates the climb
rate error ( ), which is the difference between the
climb rate reference (VVref) and the current climb rate
(VVcurrent). Furthermore, this error is used to generate
a pitch deflection, as in (7).

Aircraft

Kd e

ecurrent

Figure 5. Pitch Hold Control System Inner Loop

VVref eVV
+

PID
Kp
Ki

Kd

VVcurrent

Inner Loop

SAT

SAT

current

ref

Aircraft

VV

VV

Pitch Control
HOLD

Figure 6. Climb Rate Hold Control

This closed-loop incorporates the climb rate hold

system, and consequently the pitch hold system, as
illustrated by Fig. (7). The error altitude ( ), which
is the difference between the reference altitude
(
) and current altitude (
), as in (9), is
used to feed the PID controller. The reference climb
rate is given by (10).
(

(9)

(8)
Then the reference pitch value, is used by the inner
loop, the pitch hold control explained before, to
generate the appropriate elevator servo position
which makes the aircraft reaches the desired climb
rate.
The addition of the current pitch smoothies the
movement of the airplane. Without this feedback the
airplane would shake, making it impossible to
control the climb rate within acceptable response.
As in the pitch hold loop, two saturators are needed.
The first one limits the maximum pitch variation that
the controller could generate and the second one is
used to limit the minimum and maximum pitch
reference value. These two saturators are essential to
avoid sudden changes in the airplanes attitude. A
high pitch reference, if reached by the airplane, could
make the airplane to become unstable.
3) Altitude Hold Controller: This controller is on the
top of the longitudinal control chain. It is based on
the same concept of the heading controller and its
work is to follow a desired altitude (
) given by
a guidance block. It generates the climb rate
reference (
) for the climb rate hold system,
controlling indirectly the aircrafts altitude.

ISBN: 978-85-8001-069-5

(10)

In this controller, the feedback of the climb rate is

not needed due to the fact that when the altitude
reaches the desired altitude, the error generated is
zero, which means the climb rate should also be zero.
The altitude hold requires only a saturator to limit
maximum and minimum vertical velocity that the
aircraft can reach, avoiding great changes in the
attitude.
The reference climb rate is then used by the pitch
hold control to properly actuate on the elevator, so
the airplane can reach the desired altitude.
This loop also provides the error between the current
and desired altitude, to be used to calculate
corrections caused by coupling error between the
longitudinal movement and the airspeed control.
4.3 The Forward Speed Control
The forward speed control is composed of a
single loop which controls the power applied to the
engine, as showed in Fig. 8. Two feedback lines are
used to feed the system.

4149

Anais do XIX Congresso Brasileiro de Automtica, CBA 2012.

ealt

Altref

Inner Loop

Kp

Aircraft

Inner Loop

PID
Ki

VVref ref ref

SAT

Kd

Alt

Alt

general, a turn down of aircraft muzzle (Wu, 2003)

The tendency to tilt the plane's nose down losing
altitude must be compensated by increasing the
elevator from a value proportionally to the reference
roll, as in (13).
|

Pitch Control
HOLD

Climb rate Control

HOLD

Altcurrent

(13)

Where:
Elevator is the value applied on the elevator
surface;

Fig. 7. Altitude Hold Control

Vref

PID
+

thr

Kp
Ki

Aircraft

SAT

Kd

SAT

thr current

Vcurrent

thr

Airspeed

thr current

Fig. 8. Forward Speed Hold Control

The PID controller uses the error (ev) between the

desired forward speed (Vref ) and the current forward
speed (Vcurrent), which is the first feedback line, to
calculate the increase or decrease in amount of power
(
) that is needed to maintain the airspeed in the
desired value, as in (11).

(11)

The second feedback line is the current power

(
) applied on the engine. This power is
added to the calculated change of power (
),
resulting in the new power value to be applied in the
engine, (
), as in (12).

is the value calculated in (6);

Kp_(e-r) is the coupling gain between the
lateral and longitudinal control;
|_ref | is the absolute value of the roll
reference;
The term feed-forward stands for the fact that the
coupling errors are corrected before they happen,
such as in the example above, the pitch and altitude
would remain constant in a turn movement.
In addition to the coupled movement already
mentioned, there are two others. The forward speed
decreases, if the airplane increases its altitude, hence
it is necessary to increase the power applied to the
engine proportionally to the error between the
derived altitude and the current altitude, as in (14).
(14)
Where:
is the value applied on airplane.
is the value calculated in (12);
is the coupling gain between the
forward speed and longitudinal control;
The last effect to compensate is the airplanes side
slip as the roll reference command is applied. To
compensate this effect a feed-forward control is used
to command the rudder surface, as in (15).
(15)

(12)
Where:
As the airspeed control is a very sensitive system, a
regulator is necessary to divide the airspeed error, in
this case by 10^5, to minimize the power variation,
reaching a stable airspeed output.
Despite the plane having an electric motor, the term
throttle is used to designate the power applied to
the motor, because the equations and closed-loop
established can be also used in an airplane that uses a
combustion engine (Glow).

Rudder is the value applied on airplane.

is the coupling gain between the
lateral and directional control;
The ailerons surface is not used to correct any
coupling gains, the value calculated in (2) on the
ailerons is applied, as in (16)

(16)
Where:

5 Feed-Forward Control
The feed-forward closed loops are used to
correct errors caused by coupling movements
(Andrievsky, 2002), such as undesirable pitch effects
when aircraft turns left or right producing a loss of
lift that involves a diminution of pitch angle, or in

ISBN: 978-85-8001-069-5

Ailerons is the value applied on the ailerons

surface.
is the value calculated by (2)
Fig. 9 shows the three feed-forward controllers, using
only proportional controllers.

4150

Anais do XIX Congresso Brasileiro de Automtica, CBA 2012.

Feed-Back
Control

Feed-Forward
Control

Airspeed
Thr
Hold
Control
Altitude
Hold
Control

ealt

Aircraft

Throttle

Kpthr

Elev

Elevator

Fig. 10. Graphical User Interface

Heading
Hold
Control

|ref|

Kpe-r

ref

Kpr-r

Ailrn

Rudder

Ailerons

Fig. 9. Feed-Forward Controllers

6 Implementation
The control law used in the proposed framework
was developed in the Matlab/Simulink environment.
This experimental framework employs the
Matlab/Simulink running the autopilot controller to
be tested, and another PC running the X-Plane flight
simulator contained the unmanned aircraft to be
commanded (Ribeiro, 2011).
In order to facilitate the operation and change of
controller characteristics and desired references, a
reliable ground station software was developed using
the Graphics User Interface (GUI) tool available in
Matlab. This ground station software is used to
upload the reference values of altitude, heading, and
also define the desired UAV airspeed. Fig. 10 shows
the User Interface developed.
The parameters and control loop gains can be tuned
or changed through of User Interface, and they are
instantly modified in the control algorithm. The user
interfaces also allows the user to turn on or off any of
the hold controls individually. This means that it is
possible to evaluate the altitude hold control with the
airspeed hold control turned off or contrariwise.
To guarantee the airplane stability more than one
control law presented before are functioning in the
system. However, the tests done to validate the
designed control laws were applied individually. This
mean that although more than one control law was
working at the same time, only the loop in analysis is
submitted to a change in the reference value. It is
necessary to operate more than one law at the same
time to guarantee the airplanes stability.

ISBN: 978-85-8001-069-5

Therefore, when tuning the pitch hold control gains,

the roll hold control must be turned on with zero
degrees of reference to avoid changes in the lateral
movement that could affect the study of the
longitudinal control. In this way when selecting the
altitude hold control, automatically the pitch hold
control and the climb rate control are turned on. The
same occurs selecting the heading hold control, when
the roll hold control starts to work instantly.
The communication between Matlab/Simulink and
X-Plane, performing a SIL simulation, is very
common in a situation where it is necessary to save
time and efforts to evaluate the control architecture
(Garcia, 2009). It allows a pre-test of the system
before a flight.
7 Results
The results reached in this works show the behavior of the airplanes model created on PlaneMaker when running the control laws described before.
In the test in which the airspeed hold control was not
the focus, the throttle was fixed at 70% of the
maximum value (540W) that can be applied on the
airplanes engine.
The roll hold control must be the first control loop
validated to guarantee a minimum level of attitude
stabilization. This initial stabilization is essential to
validate the other controls; otherwise the airplane can
reach an undesirable roll that interferes during the
other control test. To evaluate this loop, the
airplanes roll angle was fixed at zero degrees and
then a step input of 12 degree was applied. All the
other controls were turned off and the gains were
tuned by using the user interface, until a acceptable
response was reached, as shown in Fig. 11.
The pitch control was evaluated applying a reference
of 10 degrees and had the behavior shown in Fig. 12.
Even though the roll and pitch control response
present overshoot, approximately 50% and 30%
respectively, it is worth to be noticed that those
values are not a problem, because the target vehicle

4151

Anais do XIX Congresso Brasileiro de Automtica, CBA 2012.

is an unmanned aircraft, so comfort issues are not

considered.
With the attitude of the airplane controlled, it was
possible to start testing the other control loops. The
heading was tested by changing the reference from 0
to 81 degrees, and was reached without any
overshoot in the response shown in Fig. 13.
The climb rate control law was validated by applying
a step of 50 ft/min, which made the airplane change
its vertical velocity from 102.5 ft/min to 152.5
ft/min. Fig. 14 shows the vertical velocity response.
The altitude hold control could only be tested after
the pitch and climb rate validation, because it
depends on them. Fig. 15 shows a step input of 400
feet on the altitude hold control.
Similar to the other test, evaluation of the horizontal
speed was done by applying a reference of 36 knots
on the airspeed hold control input. Unlike in other
tests where the throttle was fixed, in this test the
throttle was commanded by the control law presented
in (12) and (14). Fig. 16 shows the response of the
airplane airspeed.
The performance parameters obtained for each loop
is shown in Table 1.

Fig. 14. Climb Rate Response

Fig. 15. Altitude Response

Fig. 16. Forward Speed Response

Fig. 11. Roll Response

TABLE 1. PERFORMANCE PARAMETERS

Fig. 12. Pitch Response

Roll
Pitch
Heading
Climb
Rate
Altitude
Airspeed

Over
shoot
(%)
48.26
28.71
---

0.29
0.36
0.14

Rise
Time
(s)
0.046
0.08
0.068

49.3

0.308

0.02

0.0

--5.9

0.4375
7.7

0.185
2.36

0.00
0.9

Settling
Time (s)

Error
(%)
0.0
0.0
0.44

In order to illustrate the airplane in a

controlled situation, Fig. 17 (a) shows the Piper J3
Cub modeled on Plane-Maker during a climb, Fig. 17
(b) and (c) show straight level flight after the aircraft
has reached the desired altitude, heading, and
forward speed, which are the main values given by a
guide block (in this case by the user interface).
Fig. 13. Heading Response

ISBN: 978-85-8001-069-5

4152

Anais do XIX Congresso Brasileiro de Automtica, CBA 2012.

7 CONCLUSION
The architecture of the control block from
Pegasus Autopilot was deeply detailed, explaining
each feedback line and detailing completely the feedforward control.

Fig. 17(a). The aircraft climbing

Fig. 17(b). Straight level (Side View)

Fig. 17(c). Straight level (Front View)

The feed-back control was used to individually

control the movements of the airplane, but it proved
to be incomplete, and it was necessary to create
corrections due to the coupling errors between these
movements. The feed-forward control was, then,
established in order to guarantee that the aircraft
reaches or maintains a determined action in a
situation that a desired maneuver implies in a
coupled undesired airplane behaviour, such as
climbing movement results also in velocity decrease.
This allowed, in advance, correction of the coupling
error, attaining a high level of safety for the airplane
and robust control for SUAV.
The system, tested and implemented through a SIL
Simulation, was controlled using PID controllers
only. The great advantage of using PID, compared
with other controllers, is its easy implementation in
an embedded system. The early test of controllers in
a SIL allows evaluation of the system before a flight,
saving time and money, because tuning the
controllers from zero on a flight increases the chance
of a possible plane crash. With the controllers tested
and evaluated in this type of simulation, only a fine
adjustment is necessary when flying. The creation of
a model in Plane-Maker has increased the credibility

ISBN: 978-85-8001-069-5

of the controllers. The same program will be used in

future work where the autopilot will be implemented
in the hardware to be embedded in the airplane.
REFERENCES
Andrievsky, B., Fradkov, A. (2002), .Combined
Adaptive Autopilot for an UAV Flight Control,
In: 2002 IEEE International Conference on
Control Applications, pp. 290 291.
Chao,H., Cao Y., and Chen, Y., (2010). Autopilots
for Small Unmanned Aerial Vehicles: A
Survey., Int. J. Control, Automation and
Systems, vol. 8, pp. 36-44.
Chen, J., Wang. Y.,(2011).The Guidance and Control
of Small Net-recovery UAV. Seventh
International Conference on Computational
Intelligence and Security, Sanya, Hainan, China.
Ernst, D., Valavanis, K., Garcia, R., and Craighead,
J., (2007).Unmanned Vehicle Controller Design,
Evaluationand Implementation: From MATLAB
to Printed Circuit Board, IEEE J. Intelligent &
Robotic Systems, vol. 49, Number 1, pp. 85108.
Garcia R., and Barnes L., (2009) Multi-UAV
Simulator Utilizing X-Plane, IEEE J. Intelligent
& Robotic Systems, vol. 57, pp. 393-406.
Jang, J. S.; Liccardo, D.(2006). Automation of
Small UAVs using a Low Cost Mems Sensor and
Embedded Computing Platform. IEEE/AIAA 25th
Digital Avionics Systems Conference, San Jose,
CA.
Kendoul, F.; Zhenyu, Y., Nonami, K., (2009).
Embedded autopilot for accurate waypoint
navigation and trajectory tracking: Application to
miniature rotorcraft UAVs. In: IEEE International
Conference on Robotics and Automation, Kobe,
Japan.
Nelson, R. C. (1998). Flight Stability and Automatic
Control. McGraw-Hill, Second Edition.
Ribeiro L. R., and Oliveira, N. M. F., (2010). UAV
(Unmanned
Aerial
Vehicle)
Autopilot
Controllers Test Platform using Matlab/Simulink
and X-Plane. Proccedings of ASEE/IEEE
Frontiers in Education Conference, Washington
DC.
Santos S. R. B., Nascimento Jr. C. L., Givigi Jr. S.
N., Bittar A., and Oliveira. N. M. F., (2011).
Experimental Framework For Evaluation Of
Guidance And Control Algorithms For UAVs.
Proceedings of 21th International Congress of
Mechanical Engineering, Natal, Brazil.
Yu, Z.,
Choe, R.,(2007). Development &
Verification of a Low-Cost IMU for Vehicle
Attitude Determination. Aerospace Technology
Seminar Singapore, Marc.
Wu, H. Y., Zhou , Z. Y., Sun, D. (2003),
Autonomous Hovering Control and Test for
Micro Air Vehicle. In: Proceedings of the
International Conference on Robotics and
Automation, pp.528-533, Taiwan, 2003.

4153