Professional Documents
Culture Documents
InTech-Quad Rotor Unmanned Aerial Vehicle Helicopter Modelling Control
InTech-Quad Rotor Unmanned Aerial Vehicle Helicopter Modelling Control
InTech-Quad Rotor Unmanned Aerial Vehicle Helicopter Modelling Control
1.Introduction
1.1Background
1.2QuadrotorHelicopterStructure
Figure1.Proposedquadrotorhelicopterstructure
1.3SystemIntegration
Theelectronicsystemintegrationispresentedinfigure2.
The key components are a microcontroller, electronic
speed controllers (ESCs), an attitude and heading
reference system (AHRS), a communication interface
between the UAV and the user, a vision system and
suitablepowerdistribution.
TheAHRSmeasurestheinertialmovementsoftheUAV.
Itcomprisesofaninertialmeasurementunit(IMU)anda
globalpositioningsystem(GPS).TheIMUcontains3axis
accelerometers and gyroscopes to measure translational
and rotational body motion respectively. It also consists
of a magnetometer, which acts as a digital compass and
determines the heading of the UAV. The GPS is used to
determine the location of the rotorcraft. Another
component which could be added to the AHRS is a
pressure sensor to measure altitude. However, these
sensors are only suitable for high altitude applications
and therefore, it would be more appropriate to use
rangefinder to determine the altitude of the UAV. These
rangefinders could be either sonar or laser and form as
part of the vision system. The vision system is used to
detect possible obstacles in the path of the rotorcraft, as
well as transfer visual data to the user via the
communicationinterface.
1.4FlightDynamics
SONAR
RANGEFINDERS
CAMERA
MODULE
VISION
SYSTEM
XBEE WIRELESS
RADIO
CONTROLLER
COMMUNICATION
SYSTEM
MICROCONTROLLER
GPS
AHRS
MAGNETOMETER
3 AXIS
GYROSCOPE
3 AXIS
ACCELEROMETER
IMU
ESC 1 ESC 2 ESC 3 ESC 4
POWER
DISTRIBUTION
LIPO
BATTERIES
PDB
MOTOR
1
MOTOR
1
MOTOR
1
MOTOR
1
Figure2.Electronicsystemintegration
Figure3.Freebodydiagramofaquadrotorhelicopter
A quadrotor setup is controlled by manipulating thrust
forces from individual rotors as well as balancing drag
torque. For hovering, all rotors apply a constant thrust
forceasillustratedinfigure4(c),thuskeepingtheaircraft
balanced.Tocontrolverticalmovement,themotorspeed
is simultaneously increased or decreased, thus having a
lowerorhighertotalthrustbutstillmaintainingbalance.
Forattitudecontrol,theyawangle()maybecontrolled
bymanipulatingthetorquebalance,dependingonwhich
direction the aircraft should rotate. The total thrust force
still remains balanced, and therefore no altitude change
occurs. This can be shown in figures 4(a) and 4(b). In a
similarway,therollangle( )andpitchangle()canbe
manipulated applying differential thrust forces on
opposite rotors, as illustrated in figure 4(d) (Stepaniak,
2008;JKim,2010).
1.5RotorAerodynamics
searchandrescuerobot,therotorcraftwillbeexposedto
harsh environments where it should be able to produce
enough thrust force to counter any bursts of external
forces applied to it in order for it to stabilise itself.
Further, it should be able to carry the payload of
equipmentsuchascameras,sensors,etc.
2.RotorAerodynamics:Momentum
TheoryandBladeElementTheory
2
2
i
Av T
(1)
WhereTisthethrust,isthedensityofair,Aisthearea
oftherotor,andvi istheinducedairvelocityattherotor.
Blade element theory (PPounds, 2004; JFDouglas, 2005)
is particularly useful for airfoil and rotor performance.
Theforcesandmomentdevelopedonauniformwingare
modelledby,
c v C L
L
2
2
1
(2)
c v C D
D
2
2
1
(3)
2 2
2
1
c v C M
M
(4)
R v (5)
Byintegratingliftanddragalongthelengthoftheblade,
equivalent rules may be produced for the entire rotor
(PPounds,2004).
2
i i
b F
(6)
2
i i
d (7)
Wherebanddarethethrustanddragmomentconstants
respectively.
2
AR C b
T
(8)
3
AR C d
Q
(9)
Fi isthethrustproducedbyrotori,i isthedragmoment
and R is the rotor radius. CT and CQ are dimensionless
thrust and drag moment coefficients. It is also evident
from equation (6) and equation (7) that the thrust force
and rotor torque is directly proportional to the angular
velocity of the rotor. This is a useful relationship, as the
rotor angular velocity can be controlled by the motor,
thus, the thrust force and drag moment can also be
controlled by the motor. Smaller rotors require higher
speeds and more power than larger rotors for the same
thrust (PPounds, 2004). The respective total thrust force
androtortorqueofthesystemis,
4
1 i
i
F f
(10)
Besides the thrust force and the drag moment, which are
the predominant aerodynamic forces and moments
created by a rotor, there exist three other external
aerodynamicinfluences(RSiegwart,2007)whichactona
propeller.Thesemaybeillustratedinfigure5.Thefirstis
called ground effect. This refers to the variation of the
thrust coefficient when the rotor is in close proximity to
the ground. The ground effect thrust force may be
describedas,
2
R A C F
i
IGE
T IGE
(11)
Where, CT
IGE
is the thrust coefficient near ground level.
Another influential aerodynamic force occurs as a result
of horizontal forces acting on all the blade elements,
knownasthehubforce,
2
) ( R A C H
i H
(12)
CHisthehubforcecoefficient.
R R A C R
i R m
m
2
(13)
Where,
m
R
C istherollingmomentcoefficient.
Figure5.Aerodynamicforcesandmomentsonarotor
Int J Adv Robotic Sy, 2011, Vol. 8, No. 4, 139-149 142
www.intechweb.org
www.intechopen.com
3.CoordinateReferenceFramesandtheRotationMatrix
Sincethebodymotionsensorswillbeattacheddirectlyto
the reference frame, the readings will be with relation to
the rotated frame. Therefore, it is important to obtain
these coordinates in order to successfully model the
system.Thisisdonethrougharotationmatrix(ATewari,
2007;PCastilloR.a.,2005).Ifunitvectors,i;jandk,inthe
directionofthex;yandzaxesrespectively,arerotatedto
theorientationofx;yandz,withunitvectorsi,jandk
respectively, then these new unit vectors can be
represented in terms of the original orientation with the
useofarotationmatrix(ATewari,2007),
(
(
(
=
(
(
(
'
'
'
k
j
i
R
k
j
i
(14)
Where,
(
(
(
Figure6.Euleraxisandprincipalangleonarotatingcoordinateframe
The rotation matrix R is orthogonal, which implies that
the matrix transposed would be equivalent its inverse
form
.
Foreveryrotation,arotationmatrixexists.Iftheco
ordinate frame had to go through a second rotation, R,
then the resultant rotation from its original orientation
wouldbeR,where,
Thisisanextremelyusefulcharacteristic,asafullrotation
ofasystemcanbedescribedasaproductoftherotations
aboutitsx,yandzaxes,
z y x
R R R R =
(17)
(
(
(
=
| |
| |
cos sin 0
sin cos 0
0 0 1
x
R
(18)
(
(
(
=
u u
u u
cos 0 sin
0 1 0
sin 0 cos
y
R
(19)
(
(
(
=
1 0 0
0 cos sin
0 sin cos
z
R
(20)
ResultinginarotationmatrixR,wherecrepresentscos
andsrepresentssin,
(
(
(
+
+
=
u | | u | | u |
u | | u | | u |
u u u
c c c s s s c s s c s c
c s c c s s s s c c s s
s s c c c
R
(21)
4.QuadRotorDynamicsModelling
| |
T
z y x q u | =
(22)
| |
T
z y x = and | |
T
u | q = (23)
Therefore,
| |
T
q q = (24)
ALagrangianisobtainedbymodellingtheenergyofthe
system, where the difference between kinetic and
potential energy is taken. Kinetic energy of the system is
modelled for both translational and rotational motion.
The potential energy of the system is related only to the
altitude of the rotorcraft. The Lagrangian, L, can be
expressedas,
U T L = (25)
mgz I m q q L
T T
+ = q q
2
1
2
1
) , (
(26)
Thiscanbereducedto,
( )
( ) mgz I I I
z y x m q q L
zz yy xx
+ + +
+ + =
2 2 2
2 2 2
2
1
2
1
) , (
u |
(27)
F
q
L
q
L
dt
d
=
c
c
c
c
(28)
Where,theforceFrepresentsalltheexternalforcesacting
onthebodyoftherotorcraft.Again,thiscanbesplitinto
translational,
F
androtational,components.Where,
| |
T
z y x
F F F F =
(29)
And,
| |
T
u |
t t t t =
(30)
Therefore,
| |
T
F F t
=
(31)
ThePDEs
q
L
dt
d
c
c
and
q
L
c
c
canbeexpressedas,
(
(
(
(
(
(
(
(
+
+
+
=
c
c
u u
| |
zz zz
yy yy
xx xx
I I
I I
I I
z m
y m
x m
q
L
dt
d
(32)
And,
| |
T
mg
q
L
0 0 0 0 0 =
c
c
(33)
The only forces acting on the body are those from the
rotor. The translational forces are the thrust forces
resulting from each rotor, and the rotational forces are
duetothedragmomentaswellasthemomentcausedby
thrust forces from opposite rotors about the centre of
gravity. It must be noted that hub force, ground effect
and gyro effects were not taken into account, as the
model was developed for the purpose of designing a
control system around it, and thus should be kept as
simpleaspossible,withonlythemaineffectsbeingtaken
intoaccount.Thereforethetranslationalforceis,
| |
T
rotor
f F 0 0 =
(34)
rotor
F R F . =
(35)
Itcanbeseenthat,
(
(
(
=
u |
u |
u
cos cos
cos sin
sin
f
f
f
F
(36)
l F F
M M
) (
4 2
=
u
t 38)
Where,listhelineardistancefromthecentreoftherotor
to the centre of gravity. In the direction of yaw, the sum
ofallthedragmomentsproducedbytherotorsaretaken
intoaccount,
Mi
i
t t
4
1 =
=
(39)
Where,
Mi
t is the torque produced by rotor i. Therefore,
the total force function can be represented as the
followingvector,
( )
( )
(
(
(
(
(
(
(
(
=
=
Mi
i
M M
M M
l F F
l F F
f
f
f
F
t
u |
u |
u
4
1
4 2
1 3
cos cos
cos sin
sin
(40)
( )
( )
(
(
(
(
(
(
(
(
=
(
(
(
(
(
(
(
(
+
(
(
(
(
(
(
(
(
+
+
+
=
Mi
i
M M
M M
zz zz
yy yy
xx xx
l F F
l F F
f
f
f
mg
I I
I I
I I
z m
y m
x m
t
u |
u |
u
u u
| |
4
1
4 2
1 3
cos cos
cos sin
sin
0
0
0
0
0
(41)
Therefore,
( ) ( )
( ) ( )
(
(
(
(
(
(
(
(
(
(
(
(
(
(
|
.
|
\
|
=
(
(
(
(
(
(
(
(
=
=
t
u
|
u |
u |
u
u
|
zz Mi
i
zz
yy M M
yy
xx M M
xx
I
I
I l F F
I
I l F F
I
g
m
f
m
f
m
f
z
y
x
q
4
1
4 2
1 3
1
1
1
cos cos
cos sin
sin
(42)
5.QuadRotorControl
( )
2
4
2
3
2
2
2
1 1
e e e e + + + = b u
(43)
( )l b u
2
1
2
3 2
e e =
(44)
( )l b u
2
4
2
2 3
e e =
(45)
( )
2
4
2
3
2
2
2
1 4
e e e e + = d u (46)
Where,u1,u2,u3andu4arecontrolinputsforaltitude,roll,
pitch and yaw respectively. From this, the simplified
modelderivedinequation(42)becomes,
(
(
(
(
(
(
(
(
(
(
(
(
(
(
=
(
(
(
(
(
(
(
(
=
4
3
2
1
1
1
cos cos
cos sin
sin
u
I
u
I
u
I
g
m
f
m
f
m
f
z
y
x
q
zz
yy
xx
u |
u |
u
u
|
(47)
Itisevidentfromequation(47)thatthesimplifiedsystem
is linear and not coupled. Thus, a linear PD controller
designwaschosentobeused.
Thecontrolinputsdescribedinequation(43)toequation
(46)mayberepresentedinvectorform,
(
(
(
(
(
(
(
(
(
=
(
(
(
(
2
4
2
3
2
2
2
1
4
3
2
1
0 0
0 0
e
e
e
e
d d d d
bL bL
bL bL
b b b b
u
u
u
u
(47)
(
(
(
(
(
(
(
(
(
(
(
(
=
(
(
(
(
(
4
3
2
1
2
4
2
3
2
2
2
1
4
1
2
1
0
4
1
4
1
0
2
1
4
1
4
1
2
1
0
4
1
4
1
0
2
1
4
1
u
u
u
u
d bL b
d bL b
d bL b
d bL b
e
e
e
e
(48)
U=kp(proportionalerror)+kd(derivativeerror)
( ) ( ) y y k y y k U
ref d ref p
ALT ALT
+ =
1
(49)
( ) ( ) | | | |
+ =
ref d ref p
ROLL ROLL
k k U
2
(50)
( ) ( ) u u u u
+ =
ref d ref p
PITCH PITCH
k k U
3
(51)
( ) ( ) + =
ref d ref p
YAW YAW
k k U
4
(52)
Thesefeedbackcontrolloopswillbeimplementedusinga
microcontrollerasexpalinedinfigure2.
Figure7.Feedbackcontrolloop
6.ModelSimulation
Simulink
m kg 4
l m 0.3
R m 0.15
kg/m
3
1.204
CT 0.5
CQ 0.08
Table1.Constantmodelparameters
6.1DynamicModelWithoutControl
Simulation
Number
MotorSpeed(rad/s)
Motor1
(CW)
Motor2
(CCW)
Motor3
(CW)
Motor4
(CCW)
1 230 230 230 230
2 231 229 231 229
3 230 229 230 231
4 229 230 231 230
Table2.Simulationmotorspeedinputs
Figure8.GraphofAcceleration(verticalaxis)vsTime(horizontalaxis)
Figure9.GraphofSpeed(verticalaxis)vsTime(horizontalaxis)
Int J Adv Robotic Sy, 2011, Vol. 8, No. 4, 139-149 146
www.intechweb.org
www.intechopen.com
Figure10.GraphofYaw(verticalaxis)vsTime(horizontalaxis)
Figure11.GraphofYawRate(verticalaxis)vsTime(horizontalaxis)
Figure12.GraphofRoll(verticalaxis)vsTime(horizontalaxis)
Thesecondsimulationinvestigatedtheyawingmotionof
the rotorcraft. Here, both clockwise rotating rotors were
set to rotate faster than the counterclockwise rotating
rotors, resulting in a constant counterclockwise angular
acceleration of 0.1235rad/s
2
. This behaviour is described
infigure10andfigure11.
Asnotedinfigure12,therollangleincreasestoalmostan
entire revolution in the oneminute simulation time. This
behaviour is unstable, and control is therefore required.
The resultant translational motion is also unstable. An
imbalanceoftorquewasalsonoted.
6.2ModelSimulationWithControl
6.2.1Altitudecontrol
Figure13.GraphofAltitude(verticalaxis)vsTime(horizontalaxis)
Figure14.GraphofAltitude(verticalaxis)vsTime(horizontalaxis)
6.2.2.YawControl
Theyawcontrollerimplementedduringthissimulationis
basedonequation(52).Theaimofthissimulationwasto
obtain a desired yaw angle of zero. This was effectively
achievedwithminimalresponsetime,asshowninfigure
15 and figure 16. This is evident if one compares the
natural response, depicted in figure 10 and figure 11, to
thecontrolledresponse.
Figure15.GraphofYawvsTime
6.2.3Roll/PitchControl
Figure16.GraphofYawRate(verticalaxis)vsTime(horizontalaxis)
Figure17.GraphofRoll(verticalaxis)vsTime(horizontalaxis)
7.Discussion
8.Conclusion
9.References
Simulink
,2010