Programmable Force-Feedback Side-Stick for Flight

Simulation
Mohamed Guiatni, Abdelmadjid Ournid, Mohamed Amine Boulahlib and Abdelkrim Abane
Control and Command Laboratory
Ecole Militaire Polytechnique
Algiers, 16111 Algeria
Email: mohamed.guiatni@gmail.com

AbstractIn this paper, we develop and integrate a forcereflection side stick controller for a research flight simulator,
including a 2 degrees of freedom (DOF) joystick and the related
software. The developed force-reflection joystick provides twoway communication in both position and force, and is very
helpful for the user to interact with the simulation system.
Thus, the design must allow the restitution of aerodynamic
forces onto the hand of the pilot. This is an important feature,
which gives the pilot the natural feel of traditional mechanical
aircraft control. An automatic controller is implemented based on
fuzzy logic systems which provide a control strategy from expert
knowledge. For demonstration, we integrate the developed side
stick controller system to pilot a virtual aircraft using Microsoft
Flight Simulator software (MSFS).

I. I NTRODUCTION
Simulation is a useful tool for aviation research. It has
evolved and matured over the last forty years in equal pace
with developments in the aerospace industry. When flight simulation and research are combined, the objective is to measure
the human performance in the simulated environment [1].
Research will pose certain requirements on the simulation
hardware and software used; it requires generic tools that can
be adjusted to the evolving insight in topics. This implies that
flight simulators (hardware) and simulation models (software)
used for research will often be a compromise between realism
and flexibility [2]. A flight simulator must include an aircraft
model, a display capability and control hardware for the
pilot [3]. Three examples of commonly used flight simulator
packages are Microsofts Flight Simulator (MSFS), FlightGear
and XPlane. The display is a three dimensional environment
shown on a monitor or projected onto a screen. The flight
controls are hardware providing input to the aircraft model.
In most cases they are aircraft-like control input such as a
joystick, yoke and rudder pedals.
The interaction between the pilot and the flight control
system is one of the subjects of concern. In manual control of
the aircraft, the stick is the primary input device to the flight
control system. This stick usually provides the pilot only with
an impression of the input to the flight control system, the
stick force or stick displacement. This is in contradiction to
the conventional control systems were the stick force and stick
displacement have a direct relation to the control surface hinge
moments and deflections.
The aim of this work is to develop a new control interface to

978-1-4577-1772-7/12/$26.00 2012 IEEE

be used in a research flight simulator. The advantage of such

a device is that the stick force and stick displacement can
be made independent. This yields the possibility to influence
subjects output generation by changing the dynamic relation
between stick force, stick deflection, airspeed and aircrafts
motion. This control interface will be used to control the pitch
and the bank motions of the aircraft and at the same time it
must be able to reproduce the force feedback from the aerodynamic forces. We propose to use a Fuzzy Logic Controller to
generate desired torques/forces based on information feedback
from the simulated aircraft using an expert knowledge. Also,
we integrate required sensors with a Proportional-IntegralDerivative (PID) controller to achieve desired torques. We
used a designed two degree-of-freedom (2-DOF) device and
Microsoft flight simulator 2004 from/to which we have to
extract and send the concerned data. .
II. BACKGROUND
The air flowing over the elevator of any aircraft creates
a pressure distribution on its surface that causes a moment
about the elevator hinge line called the elevator hinge moment.
To keep the elevator at a particular angle, a control moment
opposite in sign to the hinge moment must be provided by
the pilot in form of a force on the stick. The magnitude of
the applied force depends on the size of the airplane and the
speed for which it is designed. For small, slow airplanes, the
pilot is connected directly to the elevator (reversible control
system), and he feels the full effect of the moment acting
on the elevator. For large, slow airplanes, the elevator may
be so heavy that the pilot cannot move it by himself. In this
case the airplane is designed with a hydraulic or electrical
system which provides part or even all the necessary control
moment. This is also true for small and large, fast airplanes
where the aerodynamic hinge moment becomes very large due
to high dynamic pressure. With the development of fly-bywire technology, the traditional mechanical linkages between
the cockpit controls and the aircrafts control surfaces have
been eliminated. They were replaced by control computers
and actuators [4]. However, without the traditional mechanical linkages, the pilot also lost his natural "feel" for the
aerodynamic forces and the flight conditions of his aircraft.
Instead, to give the pilot some feel about how much force
is being created, an artificial feel system is provided. This

GS
Control
Column

FS
Gearing
H

Control moment

Control
Position
Sensor

Trim Actuator

Fig. 2.

Screen saving of the MSFS environment.

Stability
Augmentation
Actuator

Begin
Open the RS232 link
Open the FSUIPC link

Fig. 1.

Ts time out

Traditional mechanical linkages

artificial feel can be created by a spring of controlled stiffness

and damping opposing the rotation of the control stick. These
springs and dampers now provide the pilot with only a rough
approximation of the forces. However, this force is completely
independent from the aerodynamic forces nor flight conditions.
The flight control system protects the aircraft from stalling,
over-speeding, excessive attitudes and even wind-shear. The
stiffness and damping effects are usually produced using
hydraulic or electrical actuators [5].
To illustrate the analysis of stick force, hinge moment,
etc., consider the reversible control system (no power assist)
shown in Fig.1 where the region marked gearing represents
the mechanical linkage of the control system, and assume that
the airplane is flying at a low subsonic speed (Mach number
effects negligible in the aerodynamics).
A positive stick force is a pull, and a positive stick deflection
is aft. At the elevator, a positive hinge moment is clockwise,
same as the pitching moment [6]. The stick force can be
expressed as:
Fs = G.H
(1)
where G > 0 is the gearing and H is the aerodynamic hinge
moment. The gearing includes the moment arm which converts
the stick force to a moment.
III. M ICROSOFT F LIGHT S IMULATOR (MSFS)
A. MSFS Features
MSFS is a flight simulator program, marketed and often
seen as a video game. However, it is less a game than an
immersive virtual environment (since it is very realistic, see
Fig. 2).
For our application we have chosen to use MSFS 2004
for many reasons: First, this flight simulator is commercially
available with relatively low cost. Adding to that, MSFS is
used by the U.S. Navy and the Flight Safety International
Academy (USA) to teach theirs trainee pilots, which is a proof
of its reliability. Despite of its good realism provided by a
very good three-dimensional displaying and dynamic aircraft
models, MSFS is not very greedy of computation power.
Adding to that, the long history, the consistent popularity and

YES

NO

Data reception via

RS232

Read data from

FS2004

Data processing

Data processing

Write data in
FS2004

Send data via

RS232

NO
Terminate
YES
Disconnect the RS232 link
Close the FSUIPC link
End

Fig. 3.

Data flow from flight simulator 2004.

the open nature of flight simulator structure have encouraged

a very large body of freeware and payware add-on packages
to be developed. These add-ons, widely available over the
internet, are very helpful because they, not only, permit to
change internal aspects of the simulator (airplanes, scenery...),
but also to interface it with external software and hardware
such as homebuilt cockpits.
B. Communication with MSFS
We extract/send data from/to MSFS 2004 thanks to a
dynamic link library add-on made by Pete Dowson [7],
called Flight Simulator Universal Inter-Process Communication (FSUIPC). FSUIPC is developed to succeed to FS6IPC,
which is developed by Adam Szofran to interface to FS95.
Both modules are designed to allow external (i.e. separate)
programs to communicate with and control MSFS in real time.
In other words, they permit to read from and write in MSFS
while its running and place it in a 64 Kb buffer.
In our manipulation, we need to extract the deflection
(of the aileron and the elevator), the pitch, the bank (roll)
angles and the airspeed of the simulated aircraft in the MSFS
environment, in real-time using a software based on the SDK
software and FSUIPC. For example, table I indicates in which
offsets we can read the pitch, the bank angles, in which format
and what are the unit conversion operations necessary to obtain
exploitable data.
The data are communicated via RS232 connection in raw

TABLE I
D ATA EXTRACTION AND CONVERSION EXAMPLE .
Offset
0x0578
0x057C

Format/Size
Integer
4 Bytes
Integer
4 Bytes

Use
Pitch,360/(65536 65536) f or degrees
0 = level,ve = pitch up,+ve = pitch down
Bank,360/(65536 65536) f or degrees
0 = level,ve = bank right,+ve = bank le f t

form (without unit conversion) in order to accelerate computation, (see Fig.3). Reading-sending process is cadenced using
a basic software timer that permit to the user to set the sample
time. Data are sampled at a frequency of 1KHz.
IV. PID I MPLEMENTATION FOR T ORQUE C ONTROL
Torque is deduced from the actuators current measured
using Hall-Effect sensor based on calibration experiments. We
selected experimentally the PID parameters those keep the
system stable (without exceeding value in the transient phase),
eliminate the steady state error and reduce the response time,
(see Fig. 4).

A. Fuzzification
For the implementation of this project, two controllers (one
for each axe) were designed. Tree variables were used as
input for fuzzy controller those are, the airspeed, the angular
velocities and the deflection angles of the control surfaces.
The output of the controller is the desired force to be provided
by the joystick. There were 3 linguistic terms for each input
variable. These terms are:

Airspeed: SA (Small airspeed), MA (Medium Airspeed),

BA (Big airspeed) for the,
Deflection angles (for the elevator and the aileron):
NBD (Negative Big), NSD (Negative Small), ZZD
(around Zero), PSD (Positive Small), PBD (Positive Big).
Angular velocities (for the pitch and the roll axis):
NBV (Negative Big), NSV (Negative Small), ZZV
(around Zero), PSV (Positive Small), PBV (Positive Big).

Triangular functions are used for each variable.

----- Desired
----- Obtained

Torque (N.m)

Torque (N.m)

B. Rule-based system

Time (s)

Time (s)

Fig. 4.

Torque control example.

V. F UZZY L OGIC C ONTROLLER D ESIGN

Fuzzy Controllers represent an ever-growing area in controller design because they allow for "intelligent" control of
a system. Fuzzy logic, as defined by Yen and Langari [8],
refers to all of the theories and technologies that employ fuzzy
sets, classes with unsharp boundaries. This makes a fuzzy
controller ideal for dynamically reproducing the aerodynamic
forces for the pilot using an expert knowledge. As part of the
design of the experiment, the underlying factor for the entire
project was the sensed force that results from the aerodynamic
effects. The fuzzy controller was designed to account for the
different factors those affect the perceived force and reproduce
a realistic input signal for the torque control loop which
provides force-feedback.

Once the proper inputs were created, the next step is to

create the output characteristic behavior of the signal being
received by the force/torque control loop. The output variable
(Force) was classified as : MNF (Maximal Negative Force),
VBNF (Very Big Negative Force), BNF (Big Negative Force),
MNF (Medium Negative Force), SNF (Small Negative Force),
VSNF (Very Small Negative Force), ZF (Zero Force), VSPF
(Very Small Positive Force), SPF (Small Positive Force), MPF
(Medium Positive Force), BPF (Big Positive Force), VBPF
(Very Big Positive Force), MPF (Maximal Positive Force).
The next step is to create a rule base that would govern
the operation of the fuzzy controller. The proper conditions
must be created to implement a system that will allow for
perceptible force reproduction without causing damage to the
joysticks user. The fuzzy rules base are derived from expert
as shown in figure II.
C. Defuzzification
In order to produce an acceptable value, a method of defuzzification was chosen that would produce a accurate results.
Upon researching 3 commonly used defuzzification methods,
center of gravity (COG), mean of maximum (MOM), and max
criterion (MC) [9], it was determined that of consequent rules
was created. The COG defuzzification method would be used
for the purpose of this project since it has been proven to give
accurate results.
Once the result was defuzzified, it is then output to the
torque control loop.

TABLE II
RULE BASE SYSTEM .
Airspeed/Deflection
Angular velocity

SA/ZZD

SA/PSD

SA/PBD

MA/ZZD

MA/PSD

MA/PBD

BA/ZZD

BA/PSD

BA/PBD
MPF

NBV

VSNF

ZF

VSPF

VSNF

VSPF

SPF

VSNF

SPF

NSV

VSNF

VSPF

SPF

VSNF

SPF

MPF

VSNF

MPF

BPF

ZZV

ZF

SPF

MPF

ZF

MPF

BPF

ZF

BPF

VBPF

PSV

VSPF

MPF

BPF

SPF

BPF

VBPF

MPF

VBPF

MPF

PBV

SPF

BPF

VBPF

MPF

VBPF

MPF

BPF

MPF

MPF

PID Controller

Graphic station

Control station

P
Fuzzy
Controller

Flight Simulator Interface

Amp

2 DOF Side
Stick

Flight
simulator
2004

+
D
Control Signals

Current
Sensor

Q8 Control Board

Surface
Deflection
The developped
Side-Stick Controller

Airspeed
Aircraft Motion

Fig. 5.

Demonstration Experimental Setup.

Fig. 6.

PID-Fuzzy Logic Force control scheme.

VI. I NTEGRATION E XPERIMENTS

Velocity (deg/s)

A demonstration experiment consists in the use of the

MSFS software for scenery visualization of a Cessna C172SP
Skyhawk type aircraft and a our 2-DOF stick for its control
(see Fig. 5). The control setup consists in a Quansers Q8
Hardware-in-the-Loop (H.I.L.) Control Board which is a versatile and powerful real-time measurement and control board.
The power stage is designed based on two OPA548 operational
power amplifiers from Burr-Brown which are able to provide
a current about 3A. The angular positions of the active joints
are measured using incremental encoders type HEDS 5540.
Control loops are implemented successfully using PID and
Fuzzy Logic Controllers. The inner loops control the applied
torque where the outer loops generate desired torques from the
aircrafts acceleration, attitude, deflection and airspeed based
on the Fuzzy Logic Rules. The pilots control deflections are
sent to the virtual aircraft. Then, the virtual aircraft motions are
extracted from the aircraft space. The extracted data are used to
compute the force which is fed back to the pilot throw the stick
actuators. Figures 7 8 9 represents a sample graphics of the
aircraft motion, and the reproduced forces which are measured
using current sensors. The felt forces allow to the pilot to get
a realistic image of the aircraft motion and dynamics.

Velocity (deg/s)

A. Electronic Interface and Control Loop

Time (sec)

Time (sec)

Fig. 7. Evolution of the pitch velocity (left) and the roll velocity (right) of
the aircraft.

VII. C ONCLUSIONS
In this work, we contribute in the field of flight simulation
throw the proposition of a new interface for training pilots.
The proposed interface allows to the pilot to control a virtual
aircraft and to feel the force feedback from the aerodynamic
forces. We combined a commercial software that provides a
virtual environment modeling the aircraft dynamics and the
assembled side-stick that imitates the real aircraft controller.
Experiments were performed in order to evaluate the proposed
control loop based on a PID-Fuzzy Logic controller. The

[9] T. Jiang and Y. Li, Generalized defuzzification strategies and their

parameter learning procedures, in IEEE Transactions on Fuzzy Systems,
1996, pp. 6471.

Time (sec)

Time (sec)

Fig. 8. Evolution of the deflection angles of the elevator (left) and the aileron
(right).

---- Desired
---- Obtained

Time (sec)

Time (sec)

Fig. 9. Force feedback from the side-stick according to the pitch axis (left)
and the roll axis (right).

obtained results show that the force feedback gives a realistic

image of the sensations that a pilot feels with in a real aircraft.
This idea is a step to build our own flight simulator with visual,
force and inertial feedbacks.
In addition, our experimental setup and environment can
help in the validation of Hardware-In-the-Loop (HIL) applications such as controller implementation and evaluation and
for psychophysical studies.
R EFERENCES
[1] M. S. V. Clari, R. C. Ruigrok, B. W. Heesbeen, and J. Groeneweg,
Research flight simulation of future autonomous aircraft operations,
Proceedings of the Conference of European Aerospace Societies (CEAS),
pp. 11261134, 2002.
[2] J. Hoekstra, The smart software-simple hardware concept for maximum
flexibility in research flight simulation, Proceedings of the 2002 Winter
Simulation Conference, 1995.
[3] H. Teufel, H. Nusseck, K. Beykirch, J. Butler, M. Kerger, and H. Blthoff,
Mpi motion simulator: Development and analysis of a novel motion
simulator, in AIAA Modeling and Simulation Technologies Conference
and Exhibit, Hilton Head, South Carolina, 20-23 August 2007, pp. 111.
[4] G. Jayaraman and Z. Szulyk, Active sidestick design using impedance
control, in IEEE International Conference on Systems, Man and Cybernetics, 2009, pp. 44324437.
[5] C. Nam and Hong, Active stick control using frictional torque compensation, in Sensors and Actuators, A 117, 2005, pp. 194202.
[6] D. G. Hull, Fundamentals of airplane flight mechanics, Springer Berlin
Heidelberg New York, 2007.
[7] http://www.schiratti.com/dowson.
[8] J. Yen and R. Langari, In fuzzy logic: Intelligence, control and information, in Prentice-Hall, Inc., 1999.