TEKNOFEST

AVIATION, SPACE AND TECHNOLOGY FESTIVAL

FIGHTER UAV COMPETITION 2022

CRITICAL DESIGN REPORT

TEAM NAME: AL-BURAQ

AUTHORS: All Team Members
Contents
1 BASE SYSTEM SUMMARY 1
1.1 System Description . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1
1.2 System Final Performance Specifications . . . . . . . . . . . . . . . . . . . . . 1

2 ORGANIZATION SUMMARY 3
2.1 Team Organization . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3
2.2 Timeline and Estimated Budget . . . . . . . . . . . . . . . . . . . . . . . . . . 4
2.2.1 Schedule timeline . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4
2.2.2 Budget Planning . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4

3 DETAILED DESIGN SUMMARY 6

3.1 Final System Architecture . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6
3.1.1 System Architecture . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6
3.1.2 Components selection . . . . . . . . . . . . . . . . . . . . . . . . . . . 7
3.2 Subsystems Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 12
3.3 Aircraft Performance Summary . . . . . . . . . . . . . . . . . . . . . . . . . . 12
3.4 3D Design of Aircraft . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 17
3.5 Aircraft Weight Distribution . . . . . . . . . . . . . . . . . . . . . . . . . . . . 19

4 AUTONOMOUS MISSIONS 20
4.1 Autonomous Lockdown . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 21
4.2 Kamikaze Mission . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 26

5 GROUND STATION AND COMMUNICATION 29

6 USER INTERFACE DESIGN 35

7 AIRCRAFT INTEGRATION 38
7.1 Structural Integration . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 38
7.2 Mechanical Integration . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 40
7.3 Electronic Integration . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 41

8 TEST AND SIMULATION 43

8.1 Sub-System Tests . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 43
8.2 Flight Test and Flight Checklist . . . . . . . . . . . . . . . . . . . . . . . . . . 49

9 SAFETY 50

References 51
FIGHTER UAV COMPETITION 2022

1 BASE SYSTEM SUMMARY

1.1 System Description
Within the scope of the Fighting UAV competition, autonomous take-off, flight, landing,
locking and real-time video and telemetry data stream tasks were examined and a gen-
eral system architecture was constructed. In this setup, a system that can fulfill the tasks
has been established, this system has been designed to meet the requirements, literature
research has been done, it has been conceptually designed, produced and sub-tested.
With high power GPU based ground station and processor for high frequency decision
making and communication between multi-nodes of competition server, on-board system.
High speed multi-core on-board processing unit to control vehicle attitude and tracking in
dog-fight mode and lock in kamikaze mission. Server API node is designed to communi-
cate with the server competition. beside a message broker to communicate between inter-
nal nodes and on-board companion computers, a multi-channel high power RF telemetry
communication module to communicate with both flight controller and companion com-
puter in real-time and with antennas covering a high range of angles to ensure commu-
nication in any vehicle’s orientation during maneuvers. Also independent FPV real-time
video stream with two good fitting antennas to ensure high rate high resolution video
stream all flight time.
System software and control algorithms conceptually designed, simulated and sub-tested
to fit with our UAV and communication chart. Communication diagram, system archi-
tecture, mechanical design, aerodynamic system, and control algorithms and software is
discussed in detail in following sections.
Al-Buraq UAV with simple yet effective aerodynamics to achieve the required task. Such
as, low aspect ratio for higher maneuverability, back swept wing to allow high speeds,
raked wingtip to avoid the wingtip vortices, semi symmetric airfoil to get advantage of
lift and maneuverability. Moreover, relying on a battery powered motor with a high pitch
pusher propeller to ensure high thrust to weight ratio.

1.2 System Final Performance Specifications

Cruise speed 35 m/s

Rate of climb 8 m/s
Cd 0.064
Cl /Cd 40
Thrust / weight 1

AL-BURAQ 1
FIGHTER UAV COMPETITION 2022

Table 1 continued from previous page

Cruise speed 35 m/s
Stall speed 12 m/s
Max speed at 0 AoA 45 m/s
Max AoA 12°
Max bank angle 60°
Max rate of decent 8 m/s
Flight time 27 minute
Take off weight 5270 gr

Table 1: System Final Performance Specifications

These results proved that the vehicle is well suited for the required task:

• A low drag value resulting in low power consumption and long endurance.

• A wide range of speed offering good handling for both low and high speed scenar-
ios.

• A high maneuverability due to sufficient climbing rate and wide angle of movement
on three axis ( pitch, yaw and roll )

With the outstanding aerodynamic and propulsion performance that optimized for the
task of dog and kamikaze fight missions, Flight controller system of Pixhawk cube and
companion computer to ensure high rate attitude-velocity control of the vehicle during
autonomous flights and missions along side good ground station relays on powerful gpu
for video processing and decision making; software system of filters, estimation algo-
rithms, detection AI models, decision making algorithms, between nodes message bro-
ker, competition server API, and position-attitude controllers well designed to have high
speed high resolution smooth behavior of the vehicle during the required missions. Outer
control and prediction loop of frequency of 25 HZ with internal attitude control loop of
400HZ, detection and video stream of 30 FPS, with high rate and bandwidth communi-
cation channels; Our flight system can achieve good tracking performance with tuned
controller and limit coefficients.

AL-BURAQ 2
FIGHTER UAV COMPETITION 2022

2 ORGANIZATION SUMMARY
2.1 Team Organization

Salah Gallo Yasin Arslan

(Team Leader) (Advisor)
Albaraa Mohamed
Mechanical Team Omar Abdelmoaty Abdelbasset Sedjai
(Team Leader)
Ahmed ElGreetly
Software Team Youssef Darahem Anas Badawi
(Team Leader)
Halid Biltaci
Control Team Moustafa Hassanein Ahmed Elsheikh
(Team Leader)
Table 2: Team Structure

Sallah Gallo – Team Leader, B.Sc. Electrical and Electronics Engineering, Istanbul Uni-
versity, MBA
Yasin Arslan – Advisor, Istanbul Technical University, PhD. Mechanical Engineering
Mechanical Team:
Description: The Designing and the manufacturing of the UAV body, wings and mecha-
nisms.
Team Members:

• Baraa Tawakol, İstanbul Technical University, M.Sc. Aeronautics and Astronautics

Engineering

• Omar Abdelmoaty, Bilgi university, B.Sc. Mechatronics Engineering

• Abdelbasset Sedjai,

Software Team:
Description: Implementation of the detection algorithms, communication between the
UAV and ground station, and between ground station and competition server.
Team Members:

• Ahmed Elgreetly, Near East University, B.Sc. Electrical and Electronics Engineering

• Youssef Darahem, Medipol University, B.Sc. Computer Engineering

• Anas Badawi, Bilgi university, B.Sc. Computer Engineering

AL-BURAQ 3
FIGHTER UAV COMPETITION 2022

Control Team:
Description: Flight controller configuration, components and power system integration,
and mission and tracking motion control.
Team Members:

• Halid Biltacı, İstanbul Technical University, M.Sc. Aeronautics and Astronautics En-
gineering

• Moustafa Hassanein, İstanbul Technical University, B.Sc. Control and Automation

Engineering

• Ahmed Elsheikh, Medipol University, B.Sc. Electrical and Electronics Engineering

2.2 Timeline and Estimated Budget

2.2.1 Schedule timeline

Figure 1: Schedule timeline

2.2.2 Budget Planning

Component Name Price ($) QTY Total Price Department

5mm*5mm basla 4,5 10 45 Mechnical
4mm yuvarlak balsa 7,5 20 150 Mechnical

AL-BURAQ 4
FIGHTER UAV COMPETITION 2022

Table 3 continued from previous page

Component Name Price ($) QTY Total Price Department
push rod steel 6,25 4 25 Mechnical
landing gear anti vibrations 11 2 22 Mechnical
shrink film kaplama 13 10 130 Mechnical
1mm basla 4,75 40 190 Mechnical
Servo Connectors 1,6 10 16 Mechnical
8*6 mm carbon fiber 2,25 20 45 Mechnical
12mm alumınuım 2,5 6 15 Mechnical
menteşe 0,5 64 32 Mechnical
motor 720kv 82,5 2 165 Mechnical
props 7,5 10 75 Mechnical
retract landing gear 62,5 2 125 Mechnical
landing gear 27,5 2 55 Mechnical
servo motors 11,5 8 92 Mechnical
ısı Shrink Film iron 60 1 60 Mechnical
epoksi 52,5 2 105 Mechnical
carbon fiber kumas 300 1 300 Mechnical
ESCs 43,33 6 259,98 Mechnical
battery 135 2 270 Mechnical
carbon fiber 10*8 7,5 4 30 Mechnical
glass fiber 370 1 370 Mechnical
3mm balsa 180 1 180 Mechnical
carbon fiber 6*4 10 10 100 Mechnical
cam fiber 4mm 3,6 10 36 Mechnical
standard pix tele 75 1 75 Electrical & Control
Remote control 460 1 460 Electrical & Control
pix cube 400 1 400 Electrical & Control
HERE GPS 190 1 190 Electrical & Control
RC transmitter module 50 1 50 Electrical & Control
RC reciever 50 1 50 Electrical & Control
Herelink 1000 1 1000 Electrical & Control
servo for AT 5 1 5 Electrical & Control
Battery charger 150 1 150 Electrical & Control
air pressure sensor 80 1 80 Electrical & Control
rocket dish antenna 160 1 160 Electrical & Control
UBEC (5V 5A) 4 5 20 Electrical & Control
HDMI action camera 80 1 80 Electrical & Control

AL-BURAQ 5
FIGHTER UAV COMPETITION 2022

Table 3 continued from previous page

Component Name Price ($) QTY Total Price Department
Telemetry 285 1 285 Electrical & Control
CSI HQ camera 100 1 100 Electrical & Control
Total $5.997,98

Table 3: Budget Planning

3 DETAILED DESIGN SUMMARY

3.1 Final System Architecture
3.1.1 System Architecture

433 MHz Telemetry

Air Speed Sns. Here_GPS

Servos
Retfact
I2C

USB.
915 Mhz Telemetry
Steering
Aileron R
Aileron L
Raspberry Pi 4 Pixhawk Cube
Elevator
750 KV
ESC Brushless Motor
6V, 5A
Battery Sensor UBEC
PPM
Telemetry

5V, 3A 5V, 2A RC Receiver

UBEC UBEC

Herelink s.Bus
Air Unit
Power
Power

Action HDMI

Camera
Antennas
Power Power input Emergency 65, 14000 mAh
Battery Sensor
Module Switch Lipo

Figure 2: Architecture of the on plane system

• Companion computer is responsible for the task of controlling the flight controller
attitude in dog-fight mode and kamikaze fight mode, location estimation and track-
ing of enemies in both fight-dog and search mode. Companion computer is not
responsible for detection; it’s only for position and tracking prediction and attitude
control. Companion computer used instead of direct ground station control for high
frequency estimation and control.

• HereLink is the video transmission channel with the ground station; it’s used for its
good full-deblieux two antennas to ensure coverage in all orientations of the vehicle
with high frequency rate of 30fps.

AL-BURAQ 6
FIGHTER UAV COMPETITION 2022

• GoPro Hero 4 camera with hdmi live output for damped focused high frequency
frames with the high speed motion and accelerations of the vehicle.

• Power module and emergency switch for safety and to monitor the current and volt-
age of the battery.

• UBEC and BEC used to regulate voltage with enough current for all components.

4.2 GHz Antenna

Wifi

915 Mhz
Telemetry
Here
Link
USB/UART

Competition Ethernet Switch Ethernet Ground Station Wifi

Server Router Master PC Adapter

Mission
AC Power Ethernet 433 MHz
Planner
Telemetry
Laptop

Figure 3: Architecture of the ground station system

• The PC computer is the master unit in the system; it’s responsible for communicating
with the competition server ,flight controller and on-board companion computer.
It’s designed to maintain communication, decision making and mode change, video
processing and object detection AI models.

• Ground station consists of the main unit (PowerFul PC) and laptop for the mission
planner interface as second channel for monitor and control. Communicate with
flight controller and camera stream through Here-Link, and with on-board computer
with the 915Mhz RF module.

3.1.2 Components selection

• Motor:

– Lightweight X3520-520kv Sunnysky motor that supports enough thrust for ac-
robatic, 3D flying, maneuvering, diving, and recovering.

AL-BURAQ 7
FIGHTER UAV COMPETITION 2022

• Servo Motors:

– High resolution and high torque RC Servo Motors that have digital metal gear
to ensure smooth movement of the control surface (ailerons, elevator, and rud-
der), which provides stable and safe flight.

• Electronic Speed Control(ESC):

– 80A ESC that has a Battery Eliminator Circuit (BEC). This feature ensures that
a second set of RC batteries is not required to power the RC receiver and RC
servos. This feature obviously saves weight and makes for a safer installation.

• Battery :

– lightweight 6s 14000mAh LiPo battery serves as the main power source for the
on-plane system and actuators. With a high discharge rate of 25C to be able to
provide the required current.

• 90A Power module:

– As a sensor for having feedback measurements from the battery it simply takes
the current and voltage and connects to the controller to make monitoring the
energy dissipated in the system power module an advantage for keeping the
UAV safe and protecting the system from short circuits.

• Pixhawk flight controller:

– Pixhawk Cube Flight controller unit with multi high resolution IMUs and CPU;
provides smooth flying. An open-source flight controller gives us the privilege
of editing and adding new features to the code with a lot of available libraries
to be used, with support for multiple sensors, actuators, and communication
modules.

Cube Orange Features:

• Faster H7 SOC with 1MB ram

• Upgraded triple redundant IMU sensors for extra redundancy

• 2 sets of IMU are vibration-isolated mechanically, reducing the effect of frame vibra-
tion to state estimation

• IMUs are temperature-controlled by onboard heating resistors, allowing optimum

working temperature of IMUs

AL-BURAQ 8
FIGHTER UAV COMPETITION 2022

• The entire flight management unit(FMU) and inertial management unit(IMU) are
housed in a relatively small form factor (a cube).

• Fully Cubepilot carrierboard compatible, all inputs and outputs go through a 80-
pin DF17 connector, allowing a plug-in solution for manufacturers of commercial
systems. Manufacturers can design their own carrier boards to suit their specific
needs now and in the future.

Specifications:

• Processor:

– 32bit ARM@ STM32H753 Cortex(R)-M7 (with DP-FPU)

– 400 Mhz/1 MB RAM/2 MB Flash
– 32 bit STM32F103 failsafe co-processor

• Sensors:

– Three redundant IMUs (Accelerometers/Gyroscopes), Two Barometers, One

Magnetometer
– ICM 20649 integrated accelerometer / gyro, MS5611 barometer on baseboard
– InvenSense ICM20602 IMU,ICM20948 IMU/MAG, MS5611 barometer on tem-
perature controlled, vibration isolated board
– All sensors connected via SPI

• Power:

– Redundant power supply with automatic failover

– Servo rail high-power (7 V) and high-current ready
– All peripheral outputs over-current protected, all inputs ESD protected

• Interfaces:

– 14x PWM servo outputs (8 from IO, 6 from FMU)

– S.Bus servo output
– R/C inputs for CPPM, Spektrum / DSM and S.Bus
– Analogue / PWM RSSI input
– 5x general purpose serial ports, 2 with full flow control
– 2x I2C ports

AL-BURAQ 9
FIGHTER UAV COMPETITION 2022

– SPI port (un-buffered, for short cables only not recommended for use)
– 2x CAN Bus interface
– 3x Analogue inputs (3.3V and 6.6V)
– High-powered piezo buzzer driver (on expansion board)
– High-power RGB LED (I2C driver compatible connected externally only)
– Safety switch / LED
– Optional carrier board for Intel Edison (now obsolete)

• Radio controller:

– Frsky Taranis X9 9-channel radio controller joystick for controlling the UAV in
manual mode and switching between different modes, with a high range R9M
module.

• Ground control station:

– As the Ground Control Station is based on Mission Planner software, a full fea-
tured and open source platform it allows multi mode command, safe commu-
nication link and high aspect integration. Beside the powerful master ground
control station PC which helps in detection and communication with competi-
tion servers. PC with high GPU power and CUDA cores will be used to preform
this task.

• HERE+ GPS:

– Here+ GPS kit is for having an accurate position estimation compared to the
normal one because it uses centimeter-accurate ranging with millisecond laten-
cies system which enables high precision flight and safe take-off and landing.

• Airspeed pressure sensor:

– The airspeed pressure sensor is for measuring the true airspeed, which enables
the UAV of more precise control.

• Ground station:

– Powerful PC with high GPU power to communicate with all nodes and detec-
tion on live feed.

• Raspberry pi 4:

AL-BURAQ 10
FIGHTER UAV COMPETITION 2022

– Raspberry pi is low power consumption and provides good processing speed

for our task, and has multiple IØpins and ports that could be used to control
multi-sensors, with the USB3 port enabling high-speed.

• GoPro Hero 4:

– Forward visual camera to track and detect the targets; action camera for high
resolution and high FPS with auto-focusing and damping in high speed motion
and acceleration on air due the maneuvers. Also support HDMI live output to
be passed to Here-link to ground station.

• HereLink:

– A video feed and telemetry channel between the flight controller and forward
camera, and the ground station computer high speed and high range. With
high speed video processing and API for RTSP video stream and telemetry
data, HereLink provides a high rate stream of 30 FPS and high range of 20 Km,
high transmitted power and receiving sensitivity with circular polarized an-
tenna and two patch antennas that ensure coverage in any orientation during
flight.

• 433Mhz telemetry:

– 433MHz is used as a redundant telemetry channel to ground stations with 1W

transmission power for high range. It is used for safety to monitor and control
the vehicle from a laptop as a second channel in case of system failure.

• 915MHz telemetry:

– 915MHz is used with a circular polarized antenna to ensure coverage during ve-
hicle’s maneuvers, with 500mW transmission power for high range. This chan-
nel used to communicate between the ground station and companion computer
to send modes, enemy’s locations and detection resulting in dog-fight mode.

• 90A emergency switch:

– A box for high current 80A fuse to protect the short circuit and fit with the
power consumption of the system. Safety switch with XT60 Cable to be easy to
pull and cut power in case of emergency.

AL-BURAQ 11
FIGHTER UAV COMPETITION 2022

3.2 Subsystems Summary

Communication architecture in Figure 17 , had changes to ensure coverage data stream
during all orientations and maneuvers of the vehicle and high data rate. This architecture
is based on the software architecture where the video processing and detection is on the
ground station PC, and attitude controller on board.
The architecture is considering the minimization of the body size and weight of the on-
board components for minimization of the stall speed and drag.
processing is divided into two units;

• on-board processing prediction and estimation of the position of the enemy in global
coordinate and image frame, vehicle’s location and attitudes. Beside the attitude
PID controller with the position attitude and enemy’s position in frame errors.This
processing needs to be done on-board to ensure high frequency rate of prediction
and control with callback attributes for the EKF flight controller data.

• PC ground station for decision making ,AI detection models and communication
with all nodes.

Video stream with a high rate of 30 fps and delay of 150 ms is fitting the rate of the object
detector model on the pc.
For the small delay of video stream and possible rate of detection less than 20 fps, an on
board system contains predictions using measurements of both enemy’s location and de-
tection result to compensate for that delay to ensure real-time.
On-board computer with high frequency and multi-core processors is used for the high
rate attitude control of the vehicle during the fight mission, and position tracking during
search and track mode.

3.3 Aircraft Performance Summary

To ensure a long endurance for the vehicle the body was built with lightweight material,
the body structure weight is only 800gr. The total take-off weight with payload installed
is 5270gr. The task requested from the vehicle requires a high capability of maneuver and
good climbing rate, thence the motor chosen to provide thrust is capable of delivering a
thrust to weight ratio equal to 1. the climb rate can reach a max of : 8 m/s The slender and
clean body design generates a small force of the drag, the wing tips are also equipped with
a steam line device that helps to reduce wingtip vortices and decrease their overall effect
on the drag. based on simulations and numeric calculation to lift to drag ratio Cl /Cd = 40

AL-BURAQ 12
FIGHTER UAV COMPETITION 2022

The stall speed is calculated according to the following equation:

s
2×W
v= = 12.5m/s (1)
ρ · Cd · S

Following real life testing the stall speed was measured at 12 m/s
Other aspects of flight performance were calculated to evaluate the maneuver capability
of the aircraft.
The max speed at zero angle of attack:
s
2 × Tmax
v= = 45m/s (2)
ρ · Cd · S

The max tested bank angle: 60◦ degree Radius of turn:

speed2
r= (3)
g · tan(bankangle )

minimum radius of turn is 13 m, to make one turn in 6 seconds.

Max rate of descent:

W · sin(angle − descent) · V − D · V
= 8m/s (4)
W
The following figure demonstrates a plot of thrust vs efficiency in correlation with voltage
levels.

Figure 4: Thrust VS efficiency For X4120/ KV550 motor 14” / 7” propeller

AL-BURAQ 13
FIGHTER UAV COMPETITION 2022

The cruise speed is calculated based on the highest speed acquired at the lowest power
consumption while considering drag force.
The optimal efficiency point for the designated motor is at 7411 rpm.
cruise speed of 35 m/s:
s
2 × TmaxPer f ormance
v= = 35m/s (5)
ρ · Cd · S

Voltage(V) Amps (A) Thrust(g) Watts (W) Efficiency(g/W) RPM

22.51 2.2 432 49.61 8.71 3248
22.49 4.21 776 94.7 8.19 3906
22.47 5.84 1010 131.2 7.69 4560
22.43 7.86 1260 176.32 7.14 5040
22.35 10.06 1514 224.77 6.73 5537
22.31 11.95 1728 266.69 6.48 5955
22.24 15.18 2073 337.53 6.14 6313
22.18 17.13 2261 379.89 5.95 6758
22.16 19.27 2455 427.08 5.74 7050
22.09 22.87 2742 505.1 5.42 7411
22.02 26.05 3023 573.64 5.27 7640
21.86 33.07 3498 722.82 4.83 8260
21.69 41.19 4041 893.32 4.52 8748
21.19 66.31 5277 1405.08 3.75 9780

Table 4: Motor test report for 6s battery and 14x7 propeller

Flight time:
Used battery is 14000mAh 6s battery. Power of 320 Wh power capacity. 505 watt motor
power consumption for cruise and maximum used during take-off and high rate of climb
times of 1400 watt averaging 650 watt consumption of motor. 5 V: 2.5A average consump-
tion of on-board processing units + 3A average servos consumption + 1 A for herelink
unit + 1 A for other telemetries; totaling around 40 watt system consumption.

320W battery power

× 60 = 27.5minutes (6)
Average total power consumption of 700W

of flight at most power consumption scenarios. which is 35% more than needed for round
time as a factor of safety.

AL-BURAQ 14
FIGHTER UAV COMPETITION 2022

Figure 5: Reynold number calculator

The following plot demonstrates the performance of the airfoil chosen for the main wing.
Each figure shows the data for two different Reynolds numbers: 250 000 and 500 000
respectively. The plot shows the good performance in a wide range of AoA while giving
a good lift performance in correlation to the drag force.

AL-BURAQ 15
FIGHTER UAV COMPETITION 2022

Figure 6: airfoil NASA’s data

A simulated calculation was performed to define the drag amount of each major compo-
nent of the vehicle on OpenVSP alongside Ansys Fluent CFD simulations.
The simulation was run based on the following conditions:

• Air speed: 20 m/s

• Temperature: 25◦ C

• Altitude: 100 m AGL

Component Reynold
Swet (cm2 ) Lre f (cm) t/c or d/l f (cm2 ) Cd % Total
Name number
Wing 9417.27 26.79 0.11 314328.6 65.84 0.0142 22.31
Camera 162 3 0.41 35197.39 160.14 0.0347 54.27
Fuselage 4815.26 146 11.76 1712964 20.31 0.0044 6.88
Tail 3096.95 22.99 0.09 269820.3 21.58 0.0046 7.31
Rudder 755.09 13.61 0.09 159742 5.87 0.0012 1.99
Motor support 118.09 5 0.88 58662.31 12.27 0.0026 4.16
Motor 122.74 8.85 1.79 103863.8 2.47 0.0005 0.83
Propeller 114.20 2.11 0.5 24760.72 6.55 0.0014 2.22
Totals: 295.07 0.0640 100

Table 5: Coefficients of drag

Conclusions made of this simulation and output that surface like camera result in high Cd
value, so all surface is flitted and minimize all perpendicular areas.

AL-BURAQ 16
FIGHTER UAV COMPETITION 2022

3.4 3D Design of Aircraft

Using the CAD program we were able to create the 3D model of our UAV. And designing
all the parts with the real dimensions and weights which helped us to choose the suitable
fuselage area and to distribute the components based on its weight.

Figure 7: 3D Design

(a) Top (b) Back

Figure 8: 2D Drawings

AL-BURAQ 17
FIGHTER UAV COMPETITION 2022

Figure 9: Fuselage internals

Figure 10: Prototype 1

AL-BURAQ 18
FIGHTER UAV COMPETITION 2022

3.5 Aircraft Weight Distribution

c
Figure 11: Fuselage Component Distribution

Component weight (gram) distance from nose (mm)

Airspeed sensor 20 -29
Camera & housing 150 18
Nose landing gear 200 86
Antennas x2 50 300
Battery 2000 340
Fuse 20 458
Power module 28 458
BEC 10 537
HearLink airUnit 70 537

AL-BURAQ 19
FIGHTER UAV COMPETITION 2022

Table 6 continued from previous page

Component weight (gram) distance from nose (mm)
Wing 800 642
Telemetry 12 655
BEC 10 655
Pixhawk Cube 35 655
telemetry 12 744
RX 6 744
GPS 40 822
Landing Gear 100 822
Raspberry pi 46 973
BEC 10 973
ESC 100 1181
Tail 200 1328
Rudder 120 1300
Motor 400 1480
fuselage 900 1527

Table 6: Weight Distribution

CG at 150mm from the wingtip.

4 AUTONOMOUS MISSIONS
Autonomous Take-Off is planned in auto mission from mission planner, vehicle armed to
trigger and start take-off mission.
Loiter Mode:
After finishing successful take-off the vehicle is going to start loitering at specific altitude
with specific radius waiting for another vehicle to be launched. Vehicle won’t go out of
this mode unless other vehicle positions are received and proved their air-seed they are
on air.
search-track mode:
Decision making of which enemy flight is near and with less (bearing-heading) angle to
follow it by estimating its position and direction of motion. Ground station should decide
which vehicle to follow, then send its location in real-time to an on-board system that will
start to track it. On-board computer will initialize the kalman filter to predict the position
of the vehicle in real-time.

AL-BURAQ 20
FIGHTER UAV COMPETITION 2022

In a loop of prediction with high

rate and correcting with incom-
ing data which has lower rate; the
vehicle is going to track the pre-
dicted position of the other vehi-
cle and track it by L1 tracker.

4.1 Autonomous Lockdown

During the search-track mode we
are trying to detect the other ve-
hicle using the image processing
YOLOR algorithm and starting
fight mode when first confirma-
tion of existence is done. In fight
mode, we continue using YOLOR
that works to bring and keep the target into the field of view by following the target.
Confirmation of target existence is done by detecting it using image processing YoloR al-
gorithm then start locking.
YOLOR algorithm:
Because of the differences in authorship, architecture, and model infrastructure, YOLOR is
a state-of-the-art machine learning algorithm for object detection. It differs from YOLOv1-
YOLOv5. YOLOR stands for ”You Only Learn One Representation,” not ”You Only Look
Once,” as in YOLO versions 1–4.
Object detection is the sole purpose of YOLOR, as opposed to other machine learning ap-
plications such as object identification or analysis. This is because object detection is based
on general identifiers that classify an object into a certain group or class.
Humans may learn and understand the physical world by vision, hearing, and touch (ex-
plicit knowledge), as well as previous experience (implicit knowledge). As a result, hu-

AL-BURAQ 21
FIGHTER UAV COMPETITION 2022

mans can efficiently interpret wholly new facts by drawing on a wealth of prior learning
experience that is earned through normal learning and stored in the brain.
The YOLOR research article is founded on this concept, and it offers a method for com-
bining explicit knowledge, which is defined as learning based on supplied facts and in-
put, with implicit knowledge, which is learnt subconsciously. As a result, the YOLOR
concept is based on simultaneously recording implicit and explicit knowledge, similar to
how mammalian brains process implicit and explicit knowledge. The suggested unified
network in YOLOR provides a unified representation that may serve a range of jobs at the
same time.

Kernel space alignment, prediction refinement, and a convolutional neural network (CNN)
with multi-task learning are three notable procedures that make this design functional.
According to the findings, adding implicit knowledge to a neural network that has al-
ready been trained with explicit knowledge improves the network’s performance on a
variety of tasks.
Multitask learning contains tasks such as object detection, multi-label picture classifica-
tion, and feature embedding. That is why we selected YOLOR
Convolutional Neural Networks (CNN) are typically used to achieve a single goal, but
they may be trained to tackle numerous issues at once, which is exactly what YOLOR is
aiming for. CNNs are frequently built with a specific purpose in mind. YOLOR aims to
have CNNs learn both (1) how to get outputs and (2) what all the potential outputs could
be, while CNNs learn how to evaluate inputs to produce outputs. It can have multiple
outputs rather than simply one.
YOLOR accuracy and performance:
The novel YOLOR algorithm seeks to complete jobs with a fraction of the extra expenses
that comparative algorithms are expected to incur. As a result, YOLOR is a unified net-
work capable of processing implicit and explicit knowledge simultaneously and produc-
ing a refined general representation as a result of the technique.

AL-BURAQ 22
FIGHTER UAV COMPETITION 2022

The YOLOR obtained equivalent object detection accuracy as the Scaled YOLOv4 when
combined with state-of-the-art approaches, but inference speed was raised by 88 percent.
YOLOR is one of the quickest object detection algorithms in modern computer vision as a
result of this. On the MS COCO dataset, YOLOR has a 3.8 percent greater mean average
precision than PP-YOLOv2 at the same inference speed.

YOLOR explicit knowledge:

These are fact-based data that could be obtained with the use of shallow neural networks.
Explicit knowledge is the result of conscious learning.

YOLOR Implicit knowledge:

Implicit knowledge is the extraction of detailed features from provided data, whereas ex-
plicit knowledge is the extraction of rough features. Deep neural networks can be used to
extract implicit knowledge. Implicit learning is the result of subconscious learning.

When implicit knowledge is combined with explicit knowledge, performance in tasks

such as kernel space alignment, prediction refinement, and multitask learning improves.

AL-BURAQ 23
FIGHTER UAV COMPETITION 2022

In general, a neural network serves only one purpose, hence different neural networks
are required to do diverse tasks. By combining implicit and explicit knowledge, YOLOR
solves this limitation and serves the aim of multitasking and achieving numerous objec-
tives at the same time.
y = f θ (x) + ϵ −→ minimize ϵ

In the previous formula, y represents the task’s target, x represents the observation, (theta)
represents a collection of neural network parameters, and f represents the function that
performs the provided operation. The error rate, on the other hand, is represented as
(Epsilon). The error rate must be reduced constantly. To meet the provided goal, the error
rate (ε) must be kept to a minimum. The relaxing of ε is done to achieve the generalization
of the model being able to solve all problems in domain T, and finally the modeling of
error rate is done to obtain ti .

Implicit and explicit knowledge general formula:

y = f θ ( x ) ⋆ gϕ ( z ) (7)

Using the combining operator, the implicit and explicit knowledge formulations are com-
bined. The addition or concatenation operator is used as a combining operator.
Modeling Implicit knowledge:
Implicit knowledge can be represented in three different ways:

1. Vector: used as an implicit representation

2. Neural network: uses vector and adds weight to it then uses linear combination or
non linearization for representing implicit knowledge.

3. Matrix factorization: Multiple vectors are combined as an implicit representation of

knowledge using a prior base of knowledge and a coefficient.

AL-BURAQ 24
FIGHTER UAV COMPETITION 2022

Vector is used as direct implicit representation and Detection results are as in the following
figure, and errors from the center of the frame are calculated.

Figure 12: Image error calculation method

After first detection, fight mode started and a kalman filter with two measurements (En-
emy’s location based on frame pixel position estimator, and detection result) initialized
to predict then track the position of the enemy in the frame (in pixels) with high rate and
smoothed output. Error of prediction output is calculated as the above image too. High
rate control loop of 20 HZ should take place.

Figure 13: kalman filter algorithm

After error calculated, vehicle attitude control diagram is as following:

AL-BURAQ 25
FIGHTER UAV COMPETITION 2022

Figure 14: vehicle attitude control diagram in dog fight mode

All errors are normalized with the frame size and multiplied by maximum acting angle.
In the simulation environment PID coefficient tuned for smooth rapid attitude tracking.
multiplied angle is changed with the change of the linear displacement of the enemy’s
vehicle from ours to have these shown maximum angle at the most near displacement of
10 m. and divided by the (displacement/10).
prevent from stall functions with stall speed angles data is used in throttle control and
pitch control to prevent stall either from high angle of attack or stall speed.

4.2 Kamikaze Mission

descentangle = atan2(attitudedifference, (XY)diagonaldifference) (8)

Where,

θ = bearing angle between UAV location and QR location - UAV’s heading angle

With these calculated angles errors PID attitude controller loop takes action to compensate
for these errors. Beside these errors, the thrust control will try to keep up a reference
velocity.
In the simulation environment PID coefficient tuned for smooth rapid attitude tracking.
Filter for attitude prediction may be used for a high rate control loop of 20 HZ.

AL-BURAQ 26
FIGHTER UAV COMPETITION 2022

Figure 15: descent attitude error calculation

Figure 16: vehicle attitude control diagram in dog fight mode

AL-BURAQ 27
FIGHTER UAV COMPETITION 2022

With the Zbar QR read, detect and en-

code library with OpenCV, QR code pose
in frame will be detected and read then en-
coded and sent to the competition server.
when QR detected encoded data with lock
information sent to the server.

To achieve the mission we need

to dive at speed 8 m/s and 25
degree angle.Then we need to re-
cover before touching the ground.
While recovering the CG region
will reach a high stress level. So
the wing and the fuselage are sus-
ceptible to being broken. As a re-
sult, this critical region is reinforced in the fuselage and the wing is by carbon fiber cloth
which has a yield strength equal to 2500 MPa.

AL-BURAQ 28
FIGHTER UAV COMPETITION 2022

5 GROUND STATION AND COMMUNICATION

RabbitMQ:
RabbitMQ is a message broker that will be used to establish a communication between
different interfaces and modules of the aircraft which should result in a much cleaner
structure compared to making the modules communicate directly to one another. The
desired result is isolating each module of the aircraft and routing all the communications
through the RabbitMQ.
General structure:
Modules are divided into two main groups (Ground station and plane). On the Ground
Station there are the image detection module, message broker, maps module, flight control
module, finally the database module.
While on the plane there is just the live video module which is connected directly to the on
board cameras also there is the vehicle connection module which handles communication
between plane and ground station.

• Communication with the jury’s server:

– The jury’s server publishes data to the RabbitMQ module on the other side
both the ground station and the maps modules are subscribed to the jury’s data
through the broker module. This results in the locations of the rival aircrafts
displayed using the maps module while the necessary calculations are being
computed on the ground station module. The broker can also publish to the
jury’s server the current location of the aircraft and the live stream. So both
the jury’s server and the ground station are publishers and subscribers to each
other through the message broker.

• Communication between aircraft and ground station:

– On the plane there is a vehicle connection module that sends and receives the
data between the plane and the rabbitMQ broker on the ground station. The
broker is required to handle which module on the ground station receives the
coming data. For example as mentioned above when the aircraft sends its loca-
tion the broker should direct it to a couple of channels one of which is the jury’s
server.

• Communication between different modules on the ground station.

– On the ground station channels are created by the message broker and modules
would be ethir subscriber or publisher or both to these channels.
Examples would be:

AL-BURAQ 29
FIGHTER UAV COMPETITION 2022

Between the image detention module and the vehicle connection module, ini-
tially the vehicle connection module will send the live video to the message
broker on the ground station which will direct it to the image detection mod-
ule. After the rival aircraft is detected the result should be passed back to the
broker which will be directed to the flight controller module which will make
a decision and send the command dback to the broker which will finally pass
it back to the vehicle connection module in order for the aircraft to execute the
command.
Another example is the connection between all the modules and the database
module. All modules would send all the messages - such errors, faults and
successes - to the broker which will pass it to the database module to log it.

The protocol which all the above communication would go through is MQTT.
MQTT is a light weight broker protocol that follows the publish/subscribe model to pro-
vide a message protocol for minimal network bandwidth application ex. (iot applica-
tions).

Air vehicle

Jury server

Image processing
Massage broker
System Log

Decision Maker
Targets tracker

Ground Station UI

Communication diagram between ground station and on-board system is relays on multi-
spectrum redundant and independent chart;

AL-BURAQ 30
FIGHTER UAV COMPETITION 2022

Figure 17: Communication Diagram

• Remote controller: We decided to use the Frsky x8r RC receiver to connect the UAV
with the pilot to have manual control of the UAV. It has 16 channels with 8 channel
PWM and all 16 channel dedicated S.BUS.

• HereLink is a device we chose to provide telemetry data to ground stations and

to ensure access to the camera of UAV. in real time at a high rate. It has 2 HDMI
ports for cameras with 1080p for 30fps/60fps .It transmits data over 20KM with
minimum latency 110ms. It has a processor with eight core performance named
S1. In addition Solex and QGS which are two types of GCS we can choose among
them. So HERELINKE could be concluded as professional UAV monitoring and
datastream link.

• Telemetry data which connects UAV to pc. It has around 3 km range, 433mhz and
1000mw. With the telemetry module we will configure the drone in the mission

AL-BURAQ 31
FIGHTER UAV COMPETITION 2022

planner and assign the parameter. Also will be used to monitor the vehicle on air
and command it in case of system failure.

• RF 915 Mhz channel to send mode data, enemies location data, detection results,
and main control commands to the companion computer is used to ensure real-time
reliable in motion data feed to on-board system.

• GPS receiver: Here+ is used for 50cm accuracy of position with high frequency of
5-10 Hz.

Figure 18: Antennas configuration

All antennas are configured and placed on the bottom and oriented to ensure coverage in
all cases, also the receiver antenna is placed to be free holded from bottom to freely move
and ensure good orientation facing the remote control transmitter antenna.
Frsky R9 receiver is 16 channel with high gain receiving antenna ensuring coverage all
time and in all orientations.
Circular polarized antenna:
No matter what direction or angle your mini quad is flying at in relation to your receiv-
ing antenna, circular polarized signals always overlap (no signal loss regardless of the
antennas alignment is). Circular polarized antennas have therefore become the industry
standard in FPV flying. It also has the benefit of being able to reject multipath interference.
The most prevalent cause of poor video quality is multipath interference, which manifests
itself as unpredictable color changes, static, confused images, and drop-outs. When a sig-
nal is reflected from an object, it becomes distorted and phase delayed, interfering with
the main signal. Left-hand circular polarized antennas (LHCP) and right-hand circular
polarized antennas (RHCP) are two types of circular polarized antennas (RHCP). The an-
tennas on the transmitter and receiver must be compatible; otherwise, severe signal loss
may occur. It’s good for Low altitude flying -Acrobatic flying when the aircraft’s attitude
and angle are continually changing (proximity flying)

AL-BURAQ 32
FIGHTER UAV COMPETITION 2022

Figure 19: circular polarized antenna gain with angles in two planes

Figure 20: range calculator of the device with worst cases and sensitivity

Herelink antenna:

• Frequency Range: 700-800MHz / 824-960MHz / 1575.42MHz / 1710-2170MHz /

2400-2800MHz

• Input Impedance: 50 Ω

• VSWR: ¡2.5:1

• Radiation: Omnidirectional

• Maximum Power: 50Watts

AL-BURAQ 33
FIGHTER UAV COMPETITION 2022

Figure 21: antenna gain with angles in two planes

Figure 22: antenna gain with angles 3D

Figure 23: range calculator of the device with worst cases and sensitivity

AL-BURAQ 34
FIGHTER UAV COMPETITION 2022

6 USER INTERFACE DESIGN

Like shown in the communication-interface diagram in figure 17 there is three interfacing
units on the ground with on-board system/UAV;
PC interfacing:
• Camera Interface:

– The camera interface is considered to be one of the simplest interfaces as it only

contains the live video from the aircraft. Its two main responsibilities are to
monitor the live video or to display the detection and lock of the target aircraft
while in the fight mode. In addition to that the interface will show the time of
the flight and the state of both recording and broadcasting.
• Monitoring:

– For monitoring the live video should be constantly showing on the screen in
real time as it could be used to manually control the aircraft if needed at any
given moment. The live stream also should be broadcasted and sent to the
jury server, such action will be achieved by sending the stream to the message
broker (RabbitMQ) which will handle sending it to the jury’s server separately.
• Detection and locking:

– While the aircraft is in searching or fighting modes the live video should show
the rival aircraft surrounded by a rectangle which should be drawn as soon as
a rival aircraft is detected in our FOV and it should be drawn by the ground
station. While the aircraft is trying to maintain a successful lock on the rival
aircraft a recording of this process should be initiated and saved on the ground
station to be sent to the jury’s server. The camera interface should alert the user
that a recording is taking place at the moment.
Example of the interface:

AL-BURAQ 35
FIGHTER UAV COMPETITION 2022

• Map tracker UI:

– This is a real time map that gives information about enemy aircrafts relative to
our aircraft position, every aircraft is represented by a spot on the map, our air-
craft is represented by green color, target aircraft is represented by red color and
other aircrafts are represented in blue. This gives us the ability to understand
in real time how our system works and if it’s functioning in the right way.

Remote Control Interface:

• 4-axis Joysticks to control thrust, yaw, roll, pitch in manual control modes.

• Two three position switches to change between different flight modes and to switch
to manual, return to launch, or failsafe fall in case of emergency.

• Analog channels used to tune the control coefficients on air.

Mission Planner laptop: The following image shows the mission planning interface of
the mission planner where takeoff points, pitch angle, loiter and autonomous navigation
waypoints, and landing approaching points with altitudes and land points are located.
Also, the radius of the waypoint that it’s considered reached is specified. Different maps
and hybrid maps could be selected. Also, the polygon fence, which is the border of the
flight area, is drawn and uploaded from this page.

AL-BURAQ 36
FIGHTER UAV COMPETITION 2022

The main page where real time attitude, velocity position of the vehicle could be moni-
tored.

Actions menu where basic actions of arm/disarm, level sensors attitude estimator initial-
ization, and mode change is made.

The basic tuning of coefficient, attitude limitations, and cruise is defined.

AL-BURAQ 37
FIGHTER UAV COMPETITION 2022

7 AIRCRAFT INTEGRATION
7.1 Structural Integration
Wing:

The wings of our UAV are made out of

balsa and carbon fiber. Using a CAD de-
sign program to model a 3D wing and ex-
port 2D files ready for laser cutting. Then,
assemble the balsa ribs with the carbon
fiber rods, using fast adhesive and epoxy
to fix them together. Using balsa sheets to
cover the surface of the wing and sanding the sheets very well to be ready for shrinkable
skin covering .Furthermore, covering the wing with heat shrinkable skin provides high
strength, a smooth surface, and good shape.

Based on the simulation that we have made (shown in simulation and test section), the
AFT-SWEPT wingtip is chosen. So, The PLA 3D printed AFT-SWEPT wingtip is glued
with the wing using epoxy and covered by heat shrinkable skin.

Carbon fiber rods are

used to integrate the
wings to the fuselage.So,
it’s easy and fast to be as-
sembled and disassem-
bled. And this white 3D
printed part is used to
prevent the wings from
moving out during the
flight. So, it can’t move
up and down or forward and backward because of carbon fiber rods and it can’t go out
because of the 3D printed part.

AL-BURAQ 38
FIGHTER UAV COMPETITION 2022

Tail and Rudder:

Tail and Rudder are made out

of balsa and carbon fiber. And
we have followed the same pro-
cedure that we used for making
wings. Creating 3D design and
cutting the tail and rudder into
sections. Exporting 2D drawings
for laser cutting. Then assembling
the balsa sections with the carbon fiber rods using fast adhesive and epoxy to fix them to-
gether. Then covering it with balsa sheets and sanding to be smooth and ready for heat
shrink skin. Also we used carbon fiber rods to integrate it into the fuselage.So, it’s easy
and fast to be assembled and disassembled.The control surfaces are made out of 5mm
balsa and mounted by hinges to the tail.

Fuselage :

The fuselage is made out of balsa

and has 6 carbon fiber rods 2 for
wings 2 for tail 2 for rudder. The 2
sides are symmetric, the top has 3
covers. 2D drawings are exported
from the 3D design and cut by cnc
laser. All the sides are interlaced
together using wood glue. Also
the critical places such as under the motor base and the main landing gear are covered
by carbon fiber. Then, the whole body is covered by heat shrinkable film.

Landing gears:

The main landing gear is commer-

cially available and it is made out of
carbon fiber and fixed with the body
by 2 M4 bolts. And the steering is
made out of aluminum and has a
shock absorber and retracting motor,
so after take off it will go inside the fuselage to reduce the drag.

AL-BURAQ 39
FIGHTER UAV COMPETITION 2022

7.2 Mechanical Integration

Motor:

The motor is firmly mounted on

3D printed support fixed in the
fuselage rear end while taking
in consideration the moment arm
that could be exerted by the motor
thrust.

Antenna:

The antenna of the herelink is

fixed on 3D printed support at 45 degree by double face tape.

The ESC, BECs, RX, the telemetry, and the herelink air unit are fixed in the fuselage using
double-faced tape. The GPS is fixed on a 5mm carbon rod, and the rod is fixed by 2 screws
to the fuselage. Also, the pixhawk cube is fixed on a double plate shock absorber for the
pixhawk.
The Raspberry Pi is fixed by a screw. The battery and the camera are fixed by both zip ties
and velcro tape.

AL-BURAQ 40
FIGHTER UAV COMPETITION 2022

7.3 Electronic Integration

In terms of the dependability and operability of these systems, regular placement of elec-
tronic equipment in aircraft is critical. The weight of the devices, their connections, and
their operation needs were all taken into account while creating the arrangement. The
proper positioning of power and signal connections, in addition to the equipment em-
ployed, is critical for data security and flight safety. The electronic components of the
aircraft are introduced in this part, as well as their location in the aircraft and their con-
nections.
The pre-stated final system architecture components discussed in section 3.1, have been
installed into the fuselage considering the proper positioning of power and signal connec-
tions.

The camera is installed into the nose of the UAV due to the competition specification. The
battery is installed into the nose to tune the center of gravity point. Emergency switch has
been installed between the battery and the power module. Herelink, Pixhawk, GBS, and
the Raspberry pi have been distributed along the fuselage. Each UBEC has been installed
beside the component it supplies and the ESC has been installed in the tail to power up
the DC brushless motor. The volume of the UAV’s fuselage is well balanced to contain all
the components and power/signal connections.

AL-BURAQ 41
FIGHTER UAV COMPETITION 2022

The herelink antennas are installed in the bottom nose of the UAV to cover all the angles
during the maneuvers, and the circular antenna has been installed in the bottom center
for the same reason.

For power distribution, the Pixhawk 90A power module is used. The emergency switch
is capable of cutting the power in less than 0.7 ms in case of an emergency and is installed
outside the fuselage. It is easily reachable and yellow in color. The wiring has been orga-
nized inside the fuselage and 20 awg cables have been used for the servo motors, 18 awg
cables have been used for the UBECs, 14 awg cables have been used for the motor ESC,
and a 10 awg cable is used for the main power line to withstand the current. All the wires
are silicon for flexibility.

AL-BURAQ 42
FIGHTER UAV COMPETITION 2022

8 TEST AND SIMULATION

8.1 Sub-System Tests
Wing load test:

Structural Test:

The purpose of this test is to sim-

ulate the aerodynamic loads that
occur on the wings during flight
and to prevent any failure that
may occur. So, we have made a
structural simulation using Solid-
works to check the stresses on the
wing. So, by adding the weight of
the UAV (42 N) to half of the wing. As a result, at the wingroot the minimum factor of
safety 2.1 and the maximum deflection is 40mm at the wingtip which provides 3 degrees
of a dihedral angle which increases the lateral stability and the flight performance. Also
after building the UAV we have made a real wing load test to compare between the com-
puted results and the real ones. So we have put the wing tips on two chairs and increased
the weight of the UAV by 4kg so the total weight is 8 kg. As a result we have a successful
wing load test mathematically and by experiment.
The following image shows a wing load test to check the structure and the strength of the
wing to hold the vehicle weight in the air. a small dihedral angle like the simulated one
has shown.

AL-BURAQ 43
FIGHTER UAV COMPETITION 2022

CFD simulation:

According to Bernoulli’s princi-

ple: higher pressure regions have
lower fluid speed and lower pres-
sure regions have higher fluid
speed. Because of the difference
in pressure between the upper
and lower surfaces of a wing vor-
tices will be created which in-
creases the induced drag. Usually
winglets are used to reduce the vortices but that means extra weight will be added which
is undesirable.So,using Ansys (fluent) we were able to simulate the performance of the
vortices at the wingtip. The purpose of the simulation is to find proper wingtip shape to
reduce the vortices as much as we can. And the difference is clearly noticeable between
the CUT-OFF wingtip and the AFT-SWEPT wingtip.

UAV YOLOR detection:

YOLOR detection tested to achieve fps of 10 at low power GPU (low power GTX 1050) at
laptop achieve 12 FPS with 80 class complicated model. Despite this fact we are training
our model of one class tiny mode that will achieve more than max stream fps of 30 on
our ground station PC with (GTX 1070 8GB or RTX 3060 12 GB) high power graphic card.
YOLOR tested in a lot of pictures to prove that it’s the best choice for small size objects in
the frame with high confidence and fps rate fit our need.

Dogfight tracking:

Simulation of dog-fight tracking is made by simulating two independent vehicles on SITL

with visualization on mission planner and plotted graph. Used camera focal length is

AL-BURAQ 44
FIGHTER UAV COMPETITION 2022

multiplied by the displacement of the enemy from our vehicle in the body coordinate
frame to give the enemy’s place in the camera frame in pixels.

(a) screenshot of two simulation plane in dog-(b) simulated competitor detection using posi-
fight mode tions and camera focal length

When the object size in the frame is more than 2%, it is tested to be detected in detec-
tion model dog-fight mode initialized with the simulated detection and kalman filter for
prediction and tracking with a high rate.

(a) competitor position and speed

(b) vehicle position and speed in tracking

The PID controller is primarily tuned to control the enemy’s vehicle in high speed tracking
and nearly smooth with no hard turning to avoid instability and high G-force. Tracking
for 10 seconds made successfully multi-times in the simulation and the following image
shows the response (attitude of the simulation of our vehicle).

AL-BURAQ 45
FIGHTER UAV COMPETITION 2022

Figure 27: vehicle attitude angles in tracking

Kamikaze:

kamikaze mode simulated in SITL by predefined four points at the corner of the flight
area. During loiter flight displacements and angles to each point is calculated and weighted
by priority to climb and bearing angle then displacement and so on. when the total weight
exceeds the threshold nearest and easiest to go point is targeted to go. In Kamikaze fight-
mode ratio of climb and bearing angle to distance to the point of start fighting is calculated
with formula and if it exceeds threshold the vehicle loiter to ensure good heading perpen-
dicular heading to QR code point to be able to lock on. When some criteria are achieved
the vehicle enters the attitude control mode to descend toward the QR code. It starts the
descent from altitude of 100 m to ensure lock is done by altitude of 40 m. The following
image is a plot of two consecutive points of lock, when lock period is done, climb control
is made to ensure the vehicle gains velocity and altitude to not stall.

(a) screenshot of simulated vehicle during

(b) vehicle attitude angles and altitude in two
kamikaze mission
consecutive kamikaze missions

AL-BURAQ 46
FIGHTER UAV COMPETITION 2022

Figure 29: kamikaze descent attitude control input and output

Figure 30: pitch angle control data

(a) roll angle control data (b) yaw angle control data

Figure 32: Velocity control dat

AL-BURAQ 47
FIGHTER UAV COMPETITION 2022

Attitude Kalman filter prediction is made to be able to control the vehicle at a high rate:

The Kalman filter prediction and estimation data is simulated and plotted as follows:

(a) in frame image object position prediction (b) Global location prediction

Figure 34: zoomed global location prediction shows true smooth prediction

AL-BURAQ 48
FIGHTER UAV COMPETITION 2022

8.2 Flight Test and Flight Checklist

Center of gravity tests performed:

multi tests performed to ensure the center of gravity is stable and not heavy nose. Because
the criteria of stability needed to be not too high to give the vehicle the ability of easy
maneuver and to minimize the size hence the weight of the fuselage, the first design Cg is
made in the end of Cg region. After multiple tests of take-off and stall because over back
placed Cg, a good point for Cg is defined and tested for stable flight. Because it would be
too long a fuselage to maintain Cg at 10 cm from the tip of the wing or 55cm distance from
end of wing to tip of tail, tail area is enlarged to have the Cg at 14 cm from tip of wing.
Like it clarified in section 3.4. So changes were made on the fuselage and tail to reduce the
weight that would be increased and the area of the body that would result in drag. Also
some edits in integral coefficient of Yaw and Roll control to have rapid correction during
the first tests. That resulted in a drifted lagged response causing error in roll hence stall
after a few meters of take-off. After laser cutting new materials and assembling a new tail
and fuselage, a stable flight test successfully performed with a good seated PID roll/yaw
coefficient in three flight tests.

Communication:

During this flight tests RC communication channel and telemetry channel with both 433Mhz
and 915Mhz modules tested with good ranges and RSSI during the flight in a range of 320
meters.

High force movement:

• high climb angles of 25 and descent of -25 degree is tested in stabilize mode.

• high speed climb after descent is tested to ensure wing load strength, a quite more
dihedral angle has shown, but no structure failure happened. as a test for a kamikaze
mission.

Speeds:

Take off speed of around 13, like 12.53 calculated stall. max speed shown in long waypoint
shown 34, but it could be increased with a higher propeller of 15x8. Also the distance
between points is not long enough to reach the maximum velocity.
A lot of more tests of the UAV itself need to be done for stability and to stimulate the
dog-fight and kamikaze mission. Then a controller tests and system tests.

AL-BURAQ 49
FIGHTER UAV COMPETITION 2022

9 SAFETY
We prioritize safety When building and testing UAVs. Risks that compromise security
are evaluated based on their frequency of occurrence and consequences. We first identify
potential threats, then evaluate their likely consequences if they occur, and finally, we
estimate the likelihood that these threats will materialize. Risks are weighed against a
set of approved criteria. The after-measures threshold is one. The hazards are analyzed
and reassessed after the actions have been taken. Additional steps are performed if the
amount remains above the threshold.
Before Precaution After Precaution Risk Level Reduction Responsible
Department Risk Precaution
likelihood Severity likelihood Severity Before After Rate (%) Person
(0-10) (0-10) (0-10) (0-10) Precaution Precaution
Work Equipping a
3 8 1 4 2.4 0.4 83 Salah
injury first aid kit
General
Using simulation
wrong
3 8 software and real 2 4 2.4 0.8 67 all the team
calculation
life testing
Storms / Use airspeed meter
Rain / 2 8 to evaluate weather 2 4 2.4 2.8 67 Abdelbasit
Snow situation
Some parts
5 9 Use spare parts 2 5 4.5 1 78
Mechanical are broken Baran
wrong One way installation
4 7 1 2 2.8 0.2 93
installation way design applied
Autonomous
3 6 switch to rtl mode 1 5 1.8 0.5 72
system failure
GPS system
5 4 switch to manual control 1 2 2 0.8 60
failure
Control, Manual
Avionics, control 2 8 perform fail-safe maneuver 2 4 2.4 2.8 67 Halid
and wiring. link failure
Motor or
3 9 emergency landing 1 2 2.7 0.2 93
esc failure
battery
3 9 Well Packing 1 2 2.7 0.2 93
explosion
Wrong wiring 3 9 Use checklist before flying 1 9 2.7 0.9 67
drone collision 3 9 Use spare parts 1 2 2.7 0.2 93
Improper Use data Correction
5 4 2 4 2 0.8 60
mission-plan from control server
Battery out
6 7 Use battery level tester 3 3 4.2 0.9 79
of charge
Exit the flight
3 6 Use manual control 1 4 1.8 0.4 78
zone
Signal
interference 3 6 Change control channel 1 5 1.8 0.5 72
Communication & traffic Omar
Troubleshooting control
Telemetry lose 4 7 1 2 2.8 0.2 93
system
Connection
3 6 Use manual control 1 4 1.8 0.4 78
out of range
Critical
Software software 3 9 Troubleshooting software 1 9 2.7 0.9 67 Ahmed
error

Table 7: Risks assessment

AL-BURAQ 50
FIGHTER UAV COMPETITION 2022

References
[1] Ardupilot documentation. (n.d.). Retrieved from https://ardupilot.org/plane/index.html

[2] Beard, R. W., & McLain, T. W. (2012). Small unmanned aircraft: Theory and practice.
Princeton University Press.

[3] MAVLINK Common Message Set. MAVLink Developer Guide. (n.d.). Retrieved from
https://mavlink.io/en/messages/common.html

[4] Nvidia Jetson Developer Tools. NVIDIA Developer. (2019, July 30). Retrieved from
https://developer.nvidia.com/embedded/develop/

[5] OpenCV modules. OpenCV. (n.d.). Retrieved from https://docs.opencv.org/4.x/

[6] PX4 autopilot user guide (master). PX4 User Guide. (n.d.). Retrieved from
https://docs.px4.io/master/en/

[7] Python API reference. DroneKit. (n.d.). Retrieved from https://dronekit-

python.readthedocs.io/en/latest/automodule.html

[8] Raymer, D. P. (1992). Aircraft design:A conceptual approach.

[9] Ultralytics. (n.d.). Yolov5 . GitHub. Retrieved from

https://github.com/ultralytics/yolov5

AL-BURAQ 51