Aerospace and Electronic Systems Magazine January 2018 - 51

with sonar, a laser scanner, and a camera, was modeled by using a
six degrees of freedom (6DoF) model from the Simulink Aerospace
toolbox. A virtual environment was simulated in [10] to implement
the simultaneous localization and mapping (SLAM) algorithm on
an airplane. A 3D graphical engine integrated with a 6DoF dynamic
simulator generated real-time image sequences from the virtual
flight environment, and the sequences were processed offline over
Matlab to implement the SLAM algorithm in an SITL framework.
Prado et al. [11] proposed a strategy for safely controlling the position tracking of a quadcopter UAV; the strategy features the use of
a linear state-space model predictive controller. Low-cost software
was used in the loop simulation scheme to evaluate the performance
of the controller. The proposed scheme involves the simulation of
vehicle dynamics and flight environments in an X-Plane simulator installed in a computer; the implementation of the controller in
Matlab/Simulink, which is installed in another computer; and the
interconnection of two computers via the User Datagram Protocol.
In [12], an SITL-based simulation of a UAV was implemented in
Matlab/Simulink to verify and develop a navigation and control
system. Meyer et al. [13] presented an SITL scheme to simulate
a quadcopter flight test and used the Gazebo open source simulator to simulate onboard sensors, such as inertial measurement unit
(IMU) and a camera, as well as a virtual environment. The authors
applied the Robot Operating System as the control software for
simulating the flight of a flying robot to implement Light Detection and Ranging-based and visual SLAM approaches based on the
SITL framework. Gaujens et al. [14] put forward a hybrid method
that combines SITL and model in the loop strategies for cooperative tasks performed by aerial and ground autonomous vehicles.
In the simulation, the authors interfaced a 3D physical simulator
with Matlab/Simulink. A navigation algorithm for both aerial and
ground robots was designed, and cooperative missions of the robots
were tested. The proposed approach was successfully evaluated and
validated as a feasible method for implementation on actual targets. When the performance of designed guidance, navigation, and
control algorithms are not perfected before actual testing, damage
to expensive equipment may occur because of probable failures in
actual tests. This problem prompted researchers to develop a method [15] that implements a nonlinear path-following algorithm on a
UAV in an SITL simulation that uses the Microsoft Flight SimulaJANUARY 2018

tor. Figueiredo et al. [16] also conducted an SITL simulation of an
aircraft and a virtual flight environment, with the authors using the
X-Plane flight simulator and Matlab/Simulink as the platforms in
which control laws were implemented.

SOFTWARE IN THE LOOP SIMULATION
In the present study, an SITL simulation was conducted to evaluate
the performance of the guidance, navigation, and control system of
an autonomous vehicle. The high probability of failures and crashes necessitates the simulation of UAV flight before actual tests are
carried out. Thus, we compiled the APM open source code written
in C to simulate a quadcopter equipped with GPS, IMU, a barometer, and compass sensors. The APM firmware served as the autopilot system for simulating the motors, electronic speed controllers, and sensors for controlling vehicle movement. The autopilot
system also sends data to the ground station for further analysis
and receives commands from the ground station to accomplish a
predetermined mission. It operates as a software program in the
simulation loop and communicates with the ground station software, MAVProxy, through the Mavlink protocol (Figure 2).

DYNAMIC MODEL OF QUADCOPTER
A full dynamic model of a flying robot considers issues such as
aero elastic effects, propeller deformation, dynamic inertia in motors, and some other variable factors that make it too hard to control
the vehicle. A favorable approach is to use a simplified model that
consists of the least number of inputs and states but maintains the
main characteristics of a system. The quadcopter simulated in this
work was controlled by changing the angular velocities of rotors
that apply thrust and torque on the vehicle for roll, pitch, and yaw
movements. The angular equations of motion can be written as

 = θφ(
  I yy − I zz ) − J θΩ
 + l U
Ø
2
I xx
I xx
I xx
I −I
J 
l
θ = Ø
 φ(
 zz xx ) −
θΩ +
U3
I yy
I yy
I yy

 θ  I xx − I yy  + l U
 = Ø
φ
4
 I zz  I zz

IEEE A&E SYSTEMS MAGAZINE

(1)
(2)
(3)
51

Table of Contents for the Digital Edition of Aerospace and Electronic Systems Magazine January 2018