International Journal of Advanced Robotic Systems

Composite Fuzzy Logic Control Approach

to a Flexible Joint Manipulator

Regular Paper



Mohd Ashraf Ahmad
1,*
, Mohd Zaidi Mohd Tumari
1
and Ahmad Nor Kasruddin Nasir
1

1 Control & Instrumentation (COINS) Research Group, Faculty of Electrical Engineering, University of Malaysia Pahang, Pekan, Malaysia
* Corresponding author E-mail: mashraf@ump.edu.my

Received 18 Jun 2012; Accepted 21 Aug 2012

DOI: 10.5772/52562

2013 Ahmad et al.; licensee InTech. This is an open access article distributed under the terms of the Creative
Commons Attribution License (http://creativecommons.org/licenses/by/3.0), which permits unrestricted use,
distribution, and reproduction in any medium, provided the original work is properly cited.


Abstract The raised complicatedness of the dynamics
of a robot manipulator considering joint elasticity
makes conventional modelbased control strategies
complex and hard to synthesize. This paper presents
investigations into the development of hybrid
intelligent control schemes for the trajectory tracking
andvibrationcontrolofaflexiblejointmanipulator.To
study the effectiveness of the controllers, a collocated
proportionalderivative (PD)type Fuzzy Logic
Controller (FLC) is first developed for the tip angular
position control of a flexible joint manipulator. This is
then extended to incorporate a noncollocated Fuzzy
Logic Controller, a noncollocated proportional
integralderivative (PID) and an inputshaping scheme
for the vibration reduction of the flexible joint system.
The positive zerovibrationderivativederivative
(ZVDD) shaper is designed based on the properties of
thesystem.Theimplementationresultsoftheresponse
of the flexible joint manipulator with the controllers
are presented in time and frequency domains. The
performances of the hybrid control schemes are
examinedintermsofinputtrackingcapability,levelof
vibration reduction and time response specifications.
Finally, a comparative assessment of the control
techniquesispresentedanddiscussed.
Keywords Elastic Joint, Vibration Control, Fuzzy Logic,
InputShaping,CollocatedandNonCollocatedControl

1.Introduction
Currently,elasticjointmanipulatorshavereceivedagreat
deal of attention due to their light weight, high
manoeuvrability, flexibility, high power efficiency and
large number of applications. However, controlling such
systems still faces numerous challenges that need to be
addressed. The control issue of the flexible joint is to
designthecontrollersothatthelinkoftherobotcantrack
aprescribedtrajectorypreciselywithminimumvibration
to the link. In order to achieve these objectives, various
methods using different techniques have been proposed.
Yim [1], Oh and Lee [2] proposed an adaptive output
feedbackcontrollerbasedonabacksteppingdesign.This
technique is proposed in order to deal with parametric
uncertainty in a flexible joint. The relevant work also
being done by Ghorbel et al. [3]. Lin and Yuan [4] and
Spong et al. [5] introduced a nonlinear control approach
using the feedback linearization technique and the
integralmanifoldtechniquerespectively.Arobustcontrol
design was reported by Tomei [6] by using simple PD
1 Mohd Ashraf Ahmad, Mohd Zaidi Mohd Tumari and Ahmad Nor Kasruddin Nasir:
Composite Fuzzy Logic Control Approach to a Flexible Joint Manipulator
www.intechopen.com
ARTICLE
www.intechopen.com
Int J Adv Robotic Sy, 2013, Vol. 10, 58:2013
control and Yeon and Park [7] by applying robust H
control. An openloop optimal control approach for
generating the optimal trajectory of the elastic robotic
armsinpointtopointmotionisproposedbyKorayemet
al. [8]. This method is based on Pontryagins minimum
principle which considers the mobility of the elastic
manipulator.

Inrecentyears,toolsofcomputationalintelligencesuch
as artificial neural networks and fuzzy logic controllers
have been credited in various applications as powerful
tools capable of providing robust controllers for
mathematicallyilldefinedsystems.Thishasledtorecent
advancesintheareaofintelligentcontrol[9,10].Various
neural network models have been applied in the control
of flexible joint manipulators which have led to
satisfactory performances [11]. H. Chaoui et al. [12] used
a sliding mode control approach that learns the systems
dynamics through a feedforward neural network. A
timedelay neurofuzzy network was suggested in [13],
where a linear observer was used to estimate the joint
velocitysignalsandeliminatedtheneedtomeasurethem
explicitly. Subudhi et al. [14] presented a hybrid
architecturecomposedofaneuralnetworktocontrolthe
slow dynamic subsystem and an H to control the fast
subsystem. A feedback linearization technique using a
TakagiSugeno neurofuzzy engine was adopted in [15].
LinandChen[16]proposeacombinedrigidmodelbased
the computed torque and fuzzy control of flexiblejoint
manipulators.

As to another aspect, an acceptable system performance

with reduced vibration that accounts for system changes
can be achieved by developing a hybrid control scheme
thatcatersfortherigidbodymotionandvibrationofthe
system independently. This can be realized by utilizing
controlstrategiesconsistingofeithernoncollocatedwith
collocated feedback controllers or feedforward with
feedbackcontrollers.Ineachcase,theformercanbeused
for vibration suppression and the latter for the input
tracking of a flexible manipulator. A hybrid collocated
and noncollocated controller has been widely proposed
forthecontrolofflexiblestructures[17,18,19].Theworks
haveshownthatthecontrolstructuregivesasatisfactory
system response with significant vibration reduction as
compared to a response with a collocated controller. A
feedback control with a feedforward control for
regulating the position of a flexible structure has been
proposed previously [20, 21]. A control law partitioning
schemewhichusestheendpointsensingdevicehasalso
been reported [22]. The scheme uses an endpoint
position signal in an outer loop controller to control the
flexible modes, whereas the inner loop controls the rigid
body motion independently of the flexible dynamics of
themanipulator.However,tothebestofourknowledge,
thereisstillnoresearchwhichadoptsintelligentschemes
especiallythefuzzylogicapproachforbothcollocated
and noncollocated feedback schemes. Moreover, there is
still no article focusing on a comparison study between
thenoncollocatedcontrolandfeedforwardcontrolofan
elasticjointmanipulator.

This paper presents an investigation into the

developmentofhybridintelligentcontrolschemesforthe
trajectory tracking of the tip angular position and the
vibration control of a flexible joint manipulator. Control
strategies are investigated based on a collocated PDtype
FLCwithnoncollocatedcontrolandacollocatedPDtype
FLC with a feedforward scheme. For noncollocated
control, a deflection angle is fed back through a fuzzy
logic control configuration. In order to evaluate the
performance of the noncollocated fuzzy logic scheme, it
will be compared with the noncollocated PID scheme.
Forthefeedforwardscheme,aninputshapingstrategyis
utilizedforreducingadeflectioneffect.Thepositivezero
vibrationderivativederivative (ZVDD) shaper is
designed based on the properties of the flexible joint
manipulator.Inthisstudy,wewillprovideacomparative
assessment of the performances of the composite
strategies which are evaluated in terms of input tracking
capability, level of vibration reduction in the frequency
domain and time response specifications. The
implementation environment is developed within
Simulink and Matlab for the evaluation of the
performanceofthecontrolschemes.Therestofthepaper
is structured as follows: Section 2 provides a brief
description of the singlelink flexible joint manipulator
system considered in this study. Section 3 describes the
modelling of the system derived using EulerLagrange
formulation.ThecompositecollocatedPDtypeFLCwith
a Fuzzy Logic Control, PID control and inputshaping
schemearedescribedinSection4.Thesimulationresults
and a comparative assessment are presented in Section 5
andthepaperisconcludedinSection6.
2.TheFlexibleJointManipulatorSystem
In this study, the joint flexibility of the robot is
representedbyalinearspring,asshowninFigure1.The
rotary flexible joint module provided by the Quanser
system [23] includes a servomotor in the high gear ratio
configurations and a free arm attached to two identical
springs. The springs are mounted to an aluminium
chassis which is driven by the servomotor. Rotating the
base of the arm causes the entire arm to oscillate due to
the joint flexibility introduced by the springs. An optical
encoder attached to the shaft of the DC motor is used to
measure the angular position of the shaft (t). The joint
deflection (t) is measured by an encoder located at the
motorendofthearm.

In Figure 2, is the tip angular position and is the

deflection angle of the flexible link. The base of the FJM
which determines the tip angular position of the FJM is
2 Int J Adv Robotic Sy, 2013, Vol. 10, 58:2013 www.intechopen.com
drivenbyaservomotor.Thedeflectionofthelinkwillbe
determined by the flexibility of the spring as their
intrinsicphysicalcharacteristics.

Figure1.Rotaryflexiblejointmanipulator

Figure2.Descriptionoftheflexiblejointmanipulatorsystem
3.TheModellingoftheFlexibleJointManipulator
Thissectionprovidesabriefdescriptiononthemodelling
oftheFJMsystemasabasisofasimulationenvironment
for the development and assessment of the proposed
control techniques. The EulerLagrange formulation is
considered in characterizing the dynamic behaviour of
thesystem.

The linear model of the uncontrolled system can be

represented in a statespace form [23] as shown in
equation(1),thatis:

,
x Ax Bu
y Cx
= +
=

(1)
where the vector
T
x u o u o ( =

and the matrices A,

BandCaregivenby:

2
2
0 0 1 0
0 0 0 1
0 0
0 0
0 0 , 1 0 0 0 .
stiff m g t m g eq m
eq eq m
stiff eq arm m g t m g eq m
eq arm eq m
m g t g m g t g
eq m eq m
K K K K B R
A
J J R
K (J J ) K K K B R
J J J R
K K K K
B C
J R J R
(
(
(
(
+
(
=
(
(
(
+ +
(
(

(
( ( = =

(

(2)
In equation (1), the input u is the input voltage of the
servomotor
m
V which determines the FJM base
movement.Inthisstudy,thevaluesoftheparametersare
definedinTable1.

Symbol Quantity Value

Rm ArmatureResistance() 2.6
Km MotorBackEMFConstant(V.s/rad) 0.00767
Kt MotorTorqueConstant(N.m/A) 0.00767
Jlink TotalArmInertia(kg.m
2
) 0.0035
Jeq EquivalentInertia(kg.m
2
) 0.0026
Kg Highgearratio 14:5
Kstiff JointStiffness(N.m/rad) 1.2485
Beq EquivalentViscousDamping(N.m.s/rad) 0.004
g GearboxEfficiency 0.9
m MotorEfficiency 0.69
Table1.SystemParameterofFJM
4.ControlAlgorithm
Inthissection,controlschemesfortherigidbodymotion
control and vibration suppression of a flexible joint
manipulator are proposed.Initially, acollocated PDtype
FLC controller is designed. Then, a noncollocated fuzzy
logic, noncollocated PID and inputshaping schemes are
incorporated in the closedloop system for the control of
vibrationofthesystem.
4.1PDtypefuzzylogiccontroller(PDFLC)
A PDtype FLC utilizing tip angular position error and a
derivative of tip angular position error is developed to
control the rigid body motion of the system. The hybrid
fuzzy control system proposed in this work is shown in
Figure 3, where ( ) r t , ( ) t u and ( ) t o are the desired angle,
tip angular position and deflection angle of the FJM,
whereask1,k2andk3arescalingfactorsfortwoinputsand
oneoutputoftheFLCusedwiththenormalizeduniverse
of discourse for the fuzzy membership functions. For a
PDtype FLC, triangular membership functions are
chosen for the tip angle error, tip angle error rate and
inputvoltagewith50%overlap.Normalizeduniversesof
discourse are used for both the tip angle error and its
error rate and input voltage. Scaling factors k1 and k2 are
3 Mohd Ashraf Ahmad, Mohd Zaidi Mohd Tumari and Ahmad Nor Kasruddin Nasir:
Composite Fuzzy Logic Control Approach to a Flexible Joint Manipulator
www.intechopen.com
choseninsuchawayastoconvertthetwoinputswithin
the universe of discourse and activate the rule base
effectively,whereask3isselectedsuchthatitactivatesthe
system to generate the desired output. Initially, all these
scaling factors are chosen based on trial and error. To
construct a rule base, the tip angle error, tip angle error
rate and input voltage are partitioned into five primary
fuzzysetsas:

TipangleerrorE={NMNSZEPSPM},
TipangleerrorrateV={NMNSZEPSPM},
VoltageU={NMNSZEPSPM},

where E, V and U are the universes of discourse for tip

angle error, tip angle error rate and input voltage,
respectively. The nth rule of the rule base for the FLC,
with the angle error and angle error rate as inputs, is
givenby:

Rn:IF( e isEi)AND( e isVj)THEN(uisUk),

where,Rn,n=1,2,Nmax,isthenthfuzzyrule,Ei,Vjand
Ukfori,j,k=1,2,,5,aretheprimaryfuzzysets.

A PDtype FLC was designed with 11 rules as a closed

loop component of the control strategy for maintaining
the angular position of the FJM. The rule base was
extracted based on the underdamped system response
and is shown in Table 2. The three scaling factors, k1, k2
and k3 were chosen heuristically to achieve a satisfactory
set of time domain parameters. These values were
recordedask1=0.552,k2=0.073andk3=985.

Figure3.PDtypefuzzylogiccontrolstructure

No. Rules
1. If( e isNM)and( e isZE)then(uisPM)
2. If( e isNS)and( e isZE)then(uisPS)
3. If( e isNS)and( e isPS)then(uisZE)
4. If( e isZE)and( e isNM)then(uisPM)
5. If( e isZE)and( e isNS)then(uisPS)
6. If( e isZE)and( e isZE)then(uisZE)
7. If( e isZE)and( e isPS)then(uisNS)
8. If( e isZE)and( e isPM)then(uisNM)
9. If( e isPS)and( e isNS)then(uisZE)
10. If( e isPS)and( e isZE)then(uisNS)
11. If( e isPM)and( e isZE)then(uisNM)
Table2.LinguisticRulesofthePDtypeFuzzyLogicController
4.2PDtypefuzzylogicwithanoncollocatedfuzzylogic
controller(PDFLCFLC)
AcombinationofPDtypefuzzylogicandanoncollocated
fuzzylogiccontrolschemeforthecontrolofthetipangular
position and the vibration suppression of the system
respectively is presentedin thissection. The use ofa non
collocated control system, where the deflection angle of the
flexible joint manipulator is controlled by measuring its
angle,canbeappliedtoimprovetheoverallperformanceas
more reliable output measurement is obtained. The control
structurecomprisestwofeedbackloops:(1)Thehubangleas
aninputtothePDtypeFLCfortipangularpositioncontrol.
(2) The deflection angle as an input to a separate non
collocatedcontrollawforvibrationcontrol.Thesetwoloops
are then summed together to give a torque input to the
system. A block diagram of the composite fuzzy logic
controlschemeisshowninFigure4.

Fortipangularpositioncontrol,thePDtypeFLCstrategy
developedintheprevioussectionisadoptedwhereasfor
the vibration control loop, the deflection angle feedback
through a noncollocated fuzzy logic control scheme is
utilized. In designing the noncollocated fuzzy logic
control,basictriangularandtrapezoidalformsarechosen
fortheinputandoutput membershipfunctions.Figure5
shows the membership functions of the fuzzy logic
controllerforvibrationcontrol.ItconsistsofNegativeBig
(NB), Negative Small (NS), Zero (Z), Positive Small (PS)
and Positive Big (PB), as shown in the diagram. The
universesofdiscoursesofthedeflectionangle,deflection
angle rate and input voltage are from 1.2 to 1.2 rad,
0.015to0.015rad/sand447to447Vrespectively.

Table 3 lists the generated linguistic rules for vibration

control.Therulesaredesignedbasedontheconditionofthe
deflectionangleandthedeflectionanglerate,asillustrated
inFigure6.Considerthatthejointofthemanipulatorrotates
in an anticlockwisedirectionand thatthelink deflects ina
clockwise direction. As illustrated in Figure 6(a), under this
conditionandintuitivelythetorqueshouldbeappliedin
a clockwise directionin ordertocompensatethe deflection.
In this case, the relation between the input voltage and the
torqueperinertiaisshowninequation(3):

m m g t g
m
V K K
R
(3)

Figure4.PDtypeFLCwithanoncollocatedfuzzylogiccontrol
structure
4 Int J Adv Robotic Sy, 2013, Vol. 10, 58:2013 www.intechopen.com

Figure5.Membershipfunctionsoftheinputandoutputsignals
Deflectionangle
rate
Deflectionangle

PB PS Z NS NB

PB PB PB PB NB NB
PS PB PS PS NS NB
Z PB PS Z NS NB
NS PB PS NS NS NB
NB PB PB NB NB NB
Table3.Linguisticrulesfornoncollocatedfuzzylogiccontrol

Figure6.Rulesgenerationbasedonthemotioncondition
Meanwhile,ifthejointrotatesinaclockwisedirection,as
shown in Figure 6(b), and the link deflects in an anti
clockwise direction, the torque should be imposed in an
anticlockwise direction to suppress the deflection
motion.Inthecasewherethereisnodeflection,notorque
shouldbeapplied.Furthermore,theproposedfuzzylogic
control adopts the wellknown Mamdani minmax
inferenceandcentreofarea(COA)methods.
4.3PDtypefuzzylogicwithanoncollocatedPIDController
(PDFLCPID)
AcombinationofaPDtypeFLCandnoncollocatedPID
control scheme for the control of the rigid body motion
and vibration suppression of the system is presented in
this section. The control structure of the PDFLCPID is
not very different from the PDFLCFLC and a block
diagramofthecontrolschemeisshowninFigure7.

For the rigid body motion control, the PDtype FLC

control strategy developed in the previous section is
adopted whereas, for the vibration control loop, the
deflection angle feedback through a PID control scheme
is utilized. The PID controller parameters were tuned
using the Signal Constraint blockset of MATLAB via a
closedlooptechnique.Atthismoment,thePIDcontroller
is optimized by considering the following desired
specifications:

Overshoot10%
Settlingtime0.75sec
Risetime0.55sec

The initial parameters for PID controllers must be

specifiedbeforetheSignalConstraintblocksetexecutesthe
tuning and optimizing process. The initial controller
parameters can be obtained either by trial and error or
else given as a default by the Signal Constraint blockset.
FromtheresultsofaSignalConstrainttuningprocess,the
valueswererecordedaskp=602.5128,ki=43.8895andkd
=19.7696.

Figure7.PDtypeFLCwithanoncollocatedPIDcontrol
structure
4.4PDtypefuzzylogicwithinputshaping(PDFLCIS)
A control structure for the control of the rigid body
motionanddeflectionanglereductionoftheflexiblejoint
manipulator based on a PDtype FLC and inputshaping
scheme is proposed in this section. The positive input
shapers are proposed and designed based on the
propertiesofthesystem.Inthisstudy,theinputshaping
control scheme is developed using a ZeroVibration
-400 -300 -200 -100 0 100 200 300 400
0
0.5
1
Input voltage (V)
NB NS Z PS PB
D
e
g
r
e
e

o
f

m
e
m
b
e
r
s
h
i
p
-0.015 -0.01 -0.005 0 0.005 0.01 0.015
0
0.5
1
Deflection angle rate (rad/s)
D
e
g
r
e
e

o
f

m
e
m
b
e
r
s
h
i
p
NB NS Z PS PB
1 0.8 0.6 0.4 0.2 0 -0.2 -0.4 -0.6 -0.8 -1
0
0.5
1
Deflection angle (rad)
NB NS Z PS PB
D
e
g
r
e
e

o
f

m
e
m
b
e
r
s
h
i
p
5 Mohd Ashraf Ahmad, Mohd Zaidi Mohd Tumari and Ahmad Nor Kasruddin Nasir:
Composite Fuzzy Logic Control Approach to a Flexible Joint Manipulator
www.intechopen.com
DerivativeDerivative (ZVDD) inputshaping technique
[21]. Previous experimental studies with a flexible
manipulator have shown that significant vibration
reduction and robustness is achieved using a ZVDD
technique[24].AblockdiagramofthePDtypeFLCwith
aninputshapingcontroltechniqueisshowninFigure8.

Theinputshapingmethodinvolvesconvolvingadesired
command with a sequence of impulses known as input
shapers. The design objectives are to determine the
amplitudeandtimelocationoftheimpulsesbasedonthe
natural frequencies and damping ratios of the system.
Thepositiveinputshapershavebeenusedinmostinput
shaping schemes. The requirement of positive amplitude
for the impulses seeks to avoid the problem of large
amplitudeimpulses.Inthiscase,eachindividualimpulse
must be less than one in order to satisfy the unity
magnitude constraint. In addition, the robustness of the
inputshaper to errors in the natural frequencies of the
systemcanbeincreasedbysolvingthederivativesofthe
system vibration equation. This yields a positive ZVDD
shaperwithaparameteras:
t1=0,t2=
d

,t3=
2
d

,t4=
3
d

1
2 3
1
1 3 3
A
H H H

,
2
2 3
3
1 3 3
H
A
H H H

2
3
2 3
3
1 3 3
H
A
H H H

,
3
4
2 3
1 3 3
H
A
H H H

(4)
where:
2
1
H e

,
n
and represent the natural frequency and damping
ratiorespectively.Fortheimpulses,tjandAjarethetime
locationandamplitudeofimpulsejrespectively.

Figure8.PDtypeFLCwithaninputshapingcontrolstructure
5.ImplementationandResults
In this section, the proposed control schemes are
implemented and tested within the simulation
environment of the flexible joint manipulator and the
corresponding results are presented. The manipulator is
required to follow a trajectory of 50. System responses
namelythetipangularpositionandthedeflectionangle
areobserved.Toinvestigatethevibrationofthesystemin
the frequency domain, the power spectral density (PSD)
of the deflection angle response is obtained. The
performances of the control schemes are assessed in
terms of vibration suppression, trajectory tracking and
time response specifications. Finally, a comparative
assessment of the performance of the control schemes is
presentedanddiscussed.

(a)Tipangularposition

(b)Deflectionangle

(c)Powerspectraldensity
Figure9.ResponseoftheflexiblejointmanipulatorusingPD
FLC
2
1
n d
0 0.5 1.0 1.5 2.0 2.5 3.0
0
10
20
30
40
50
60
Time (s)
T
i
p

a
n
g
u
l
a
r

p
o
s
i
t
i
o
n

(
d
e
g
)


0 0.5 1.0 1.5 2.0 2.5 3.0
-20
-15
-10
-5
0
5
10
15
20
Time (s)
D
e
f
l
e
c
t
i
o
n

a
n
g
l
e

(
d
e
g
)


0 2 4 6 8 10 12 14 16 18
-100
-80
-60
-40
-20
0
20
40
Frequency (Hz)
P
o
w
e
r

S
p
e
c
t
r
a
l

D
e
n
s
i
t
y

(
d
B
)


8.83 Hz
2.94 Hz
14.52 Hz
6 Int J Adv Robotic Sy, 2013, Vol. 10, 58:2013 www.intechopen.com
Figure 9 shows the responses of the flexible joint
manipulator to the reference input trajectory using the
PDFLCinatimedomainandafrequencydomain(PSD).
These results were considered as the system response
under rigid body motion control and will be used to
evaluate the performance of the noncollocated fuzzy
logic control and inputshaping scheme. The steadystate
tip angular trajectory of 50 for the flexible joint
manipulator was achieved within the rise and settling
times and an overshoot of 0.222 s, 0.565 s and 1.78 %
respectively. It is noted that the manipulator reaches the
required position within 1 s, with little overshoot.
However,anoticeableamountofvibrationoccursduring
the movement of the manipulator. It is noted from the
deflectionangleresponsethatthevibrationofthesystem
settles within 3 s with a maximum residual of 15.
Moreover, and from the PSD of the deflection angle
response,thevibrationsattheflexiblejointaredominated
by the first three vibration modes, which are obtained as
2.94Hz,8.83Hzand14.52Hzwithamagnitudeof32.38
dB,12.08dBand26.52dBrespectively.

The tip angular position, deflection angle and power

spectral density responses of the flexible joint
manipulator using PDFLCFLC, PDFLCPID and PD
FLCIS are shown in Figure 10. It is noted that the
proposed control schemes are capable of reducing the
system vibration while maintaining the trajectory
tracking performance of the manipulator. A similar tip
angular position, deflection angle and power spectral
density of the deflection angle responses were observed
as compared to the PDFLC.The steadystate tip angular
trajectory of noncollocated fuzzy logic was achieved
withintheriseandsettlingtimesandovershootof0.215s,
0.613 s and 9.68 %, respectively and 0.222 s, 0.501 s, and
9.40 %, respectively for PID control.. Meanwhile, with
PDFLCIS, the manipulator reached the rise, settling
times and overshoot of 0.469 s, 0.857 s and 0.16 %
respectively.ItisnotablethatthesettlingtimeofPDFLC
FLC is much faster as compared to the case of PDFLC
PID and PDFLCIS. However, the percentage overshoot
of PDFLCIS is much lower than both cases of PDFLC
with noncollocated control. Besides this, the results also
demonstrate a significant amount of deflection angle
reduction at the tip angle of the manipulator with the
proposed composite controllers. The vibration of the
systemusingPDFLCFLCandPDFLCPIDsettlewithin
1 s, which is much faster as compared to PDFLCIS.
However, in terms of the maximum residual of the
deflection angle, the PDFLCIS produces almost a four
fold improvement as compared to the PDFLC scheme.
This result is slightly higher than with both cases of PD
FLCPID and PDFLCFLC, with only a twofold
improvement over the PDFLC scheme. In terms of the
power spectral density of the deflection angle response,
the magnitudes of vibrations using PDFLCFLC, PD
FLCPID and PDFLCIS were reduced to [1.87, 41.95,
48.25] dB, [2.55, 33.53, 45.55] dB and [17.59, 25.28,
36.69] dB respectively for the first three modes of
vibration. It is noted that both the PDFLCPID and PD
FLCFLC schemes produce a higher level of vibration
reductionascomparedtoPDFLCIS.

(a)Tipangularposition

(b)Deflectionangle

(c)Powerspectraldensity
Figure10.Responseoftheflexiblejointmanipulatorusing
compositePDFLCFLC,PDFLCPIDandPDFLCIS
0 0.5 1.0 1.5 2.0 2.5 3.0
0
10
20
30
40
50
60
Time (s)
T
i
p

a
n
g
u
l
a
r

p
o
s
i
t
i
o
n

(
d
e
g
)


PD-FLC-FLC
PD-FLC-PID
PD-FLC-IS
0 0.5 1.0 1.5 2.0 2.5 3.0
-15
-10
-5
0
5
10
Time (s)
D
e
f
l
e
c
t
i
o
n

a
n
g
l
e

(
d
e
g
)


PD-FLC-FLC
PD-FLC-PID
PD-FLC-IS
0 2 4 6 8 10 12 14 16 18
-100
-80
-60
-40
-20
0
20
40
Frequency (Hz)
P
o
w
e
r

S
p
e
c
t
r
a
l

D
e
n
s
i
t
y

(
d
B
)


PD-FLC-FLC
PD-FLC-PID
PD-FLC-IS
2.94 Hz
8.83 Hz
14.52 Hz
7 Mohd Ashraf Ahmad, Mohd Zaidi Mohd Tumari and Ahmad Nor Kasruddin Nasir:
Composite Fuzzy Logic Control Approach to a Flexible Joint Manipulator
www.intechopen.com
Table 4 and Figure 11 summarize the magnitude of the
vibration and the level of the vibration reduction of
deflection, respectively, for the proposed composite control
schemes.Itisseenthatahighperformanceinthereduction
of the vibration of the system is achieved using both non
collocatedfuzzylogicandPIDcontrolascomparedtofeed
forward control based on an inputshaping technique. This
isobserved and comparedtothe PDFLC for the first three
modes of vibration. In addition, as demonstrated in the tip
angulartrajectoryresponse,aslightlyfasterresponsewitha
higher overshoot is obtained using both noncollocated
fuzzy logic and PID control scheme as compared to the
inputshaping technique. Comparisons of the specifications
of the tip angular trajectory responses are summarized in
Table5.Inaddition,andasdemonstratedinthetipangular
trajectoryresponsewithPDFLCFLCandPDFLCPID,the
minimumphasebehaviourofthemanipulatorisunaffected.
Intermsofthedesigncomplexity,theimplementationofthe
inputshaper technique is much easier as compared to both
noncollocatedcontrolsasalargeamountofdesigneffortis
required in order to determine the controller parameters,
such as the range of the membership function in the fuzzy
logic design and kp, ki and kd gains in the PID design.
However, the inputshaper strategy is very sensitive to the
system parameter variation and could not compensate for
anyeffectofdisturbanceascomparedtothenoncollocated
control.

Controller Mode1 Mode2 Mode3

Magnitudeof
vibration(dB)
PDFLCFLC 1.87 41.95 48.25
PDFLCPID 2.55 33.53 45.55
PDFLCIS 17.59 25.28 36.69
Table4.Magnitudeofthevibrationandlevelofvibration
reductionofthedeflectionangle
Controller Settlingtime(s) Risetime(s) Overshoot(%)
PDFLCFLC 0.613 0.215 9.68
PDFLCPID 0.501 0.222 9.40
PDFLCIS 0.857 0.469 0.16
Table5.Timeresponsespecificationsoftipangularposition

Figure11.Levelofthevibrationreductionofthedeflectionangle
6.Conclusions
The development of composite fuzzy logic schemes for
the trajectory tracking and vibration suppression of a
flexiblejointmanipulatorhasbeenpresented.Thecontrol
schemes have been developed based on a PDtype FLC
with noncollocated fuzzy logic control, a PDtype FLC
withnoncollocatedPIDcontrolandaPDtypeFLCwith
an inputshaping scheme. The proposed composite
control schemes have been implemented and tested on a
flexible joint manipulator. The performances of the
control schemes have been evaluated in terms of input
tracking capability and vibration suppression for the
resonance modes of the manipulator. An acceptable
performance in input tracking and vibration control has
been achieved with the proposed control strategies. A
comparative assessment of the control schemes has
shown that the PDFLC with noncollocated strategies
performs better than the PDFLC with a feedforward
technique in respect of the deflection angle reduction of
the elastic joint. Moreover, in terms of the speed of
responses, PDFLC with noncollocated PID and FLC
results in a faster settling time response with high
overshoot as compared to PDFLC with a feedforward
scheme. The work thus developed and reported in this
paper forms the basis of the design and development of
composite control schemes for the input tracking and
vibration suppression of multidegree of freedom robotic
armswithflexiblelinksandjoints.
7.Acknowledgments
This work was supported by the Faculty of Electrical &
Electronics Engineering, University of Malaysia Pahang,
and especially the Control & Instrumentation (COINS)
ResearchGroup.
8.References
[1] Yim, W. (2001) Adaptive Control of a Flexible Joint
Manipulator. Proc. 2001 IEEE, International Robotics
&Automation,Seoul,Korea,pp.34413446.
[2] Oh,J.H.andLee,J.S.(1997)ControlofFlexibleJoint
Robot System by Backstepping Design Approach.
Proc.ofIEEEInternationalConferenceonRobotics&
Automation,pp.34353440.
[3] Ghorbel, F., Hung, J.Y. and Spong, M.W. (1989)
Adaptive Control of Flexible Joint Manipulators.
ControlSystemsMagazine.9:913.
[4] Lin,L.C.andYuan,K.(2007)ControlofFlexibleJoint
Robots via External LinearizationApproach. Journal
ofRoboticSystems.1(1):122.
[5] Spong, M. W., Khorasani, K. and Kokotovic, P. V.
(1987) An Integral Manifold Approach to the
Feedback Control of Flexible Joint Robots. IEEE
JournalofRoboticsandAutomation.3(4):291300.
[6] Tomei, P. (1991) A Simple PD Controller for Robots
with Elastic Joints. IEEE Trans. on Automatic
Control.36(10):12081213.
0 5 10 15 20 25 30 35
1
2
3
Level of vibration reduction (dB)
M
o
d
e

o
f

f
r
e
q
u
e
n
c
i
e
s


PD-FLC-FLC
PD-FLC-PID
PD-FLC-IS
8 Int J Adv Robotic Sy, 2013, Vol. 10, 58:2013 www.intechopen.com
[7] Yeon, J. S. and Park, J. H. (2008) Practical Robust
Control for Flexible Joint Robot Manipulators. Proc.
2008 IEEE International Conference on Robotic and
Automation,Pasadena,CA,USA,pp.33773382.
[8] Korayem, M.H., Nohooji, H.R. and Nikoobin, A.
(2011) Path planning of mobile elastic robotic arms
byindirectapproachofoptimalcontrol.International
JournalofAdvanceRoboticSystems.8(1):1020.
[9] Ahmad, M.A., Raja Ismail, R.M.T., Ramli, M.S.,
Zawawi, M.A., Hambali, N. and Abd Ghani, N.M.
(2009) Vibration Control of Flexible Joint
Manipulator using Input Shaping with PDtype
Fuzzy Logic Control. Proceedings of IEEE
International Symposium on Industrial Electronics,
Seoul,pp.11841189.
[10] Ahmad,M.A.,Nasir,A.N.K.,Hambali,N.andIshak,
H. (2008) Vibration and Input Tracking Control of
Flexible Manipulator using Hybrid Fuzzy Logic
Controller. Proceedings of the IEEE International
Conference of Mechatronics and Automation,
Kagawa,pp.593598.
[11] Chaoui, H., Gueaieb, W., Yagoub, M. and Sicard, P.
(2006) Hybrid neural fuzzy sliding mode control of
flexiblejoint manipulators with unknown dynamics.
Proceedings of the 32nd Annual Conference of the
IEEE Industrial Electronics Society (IECON2006),
pp.40824087.
[12] Chaoui, H., Sicard, P. and Lakhsasi, A. (2004)
Reference model supervisory loop for neural
network based adaptive control of a flexible joint
with hard nonlinearities. IEEE Canadian Conference
on Electrical and Computer Engineering. 4: 2029
2034.
[13] Hui, D., Fuchun, S. and Zengqi, S. (2002) Observer
based adaptive controller design of flexible
manipulators using timedelay neurofuzzy
networks. J. Intell. Robot. Syst.: Theory and
Applications.34(4):453466.
[14] Subudhi, B. and Morris, A. (2006) Singular
perturbation based neuroh innity control scheme
for a manipulator with exible links and joints.
Robotica.24(2):151161.
[15] Park, C.W. (2004) Robust stable fuzzy control via
fuzzy modeling and feedback linearization with its
applications to controlling uncertain singlelink
flexible joint manipulators. J. Intell. Robot. Syst.:
TheoryandApplications.39(2):131147.
[16] Lin, L. C. and Chen, C. C. (1995) Rigid modelbased
fuzzy control of flexiblejoint manipulators. J. of
IntelligentandRoboticSystems.13:107126.
[17] Tokhi, M.O. and Azad, A.K.M. (1996) Control of
flexible manipulator systems. Proceedings IMechEI:
Journal of Systems and Control Engineering. 210:
113130.
[18] Ahmad, M.A., Suid, M.H., Ramli, M.S., Zawawi,
M.A., Ismail, R.M.T.R. (2010) PD Fuzzy Logic with
noncollocated PID approach for vibrationcontrol of
flexible joint manipulator. Proceedings of 6th
International Colloquium on Signal Processing and
ItsApplications,Mallaca,Malaysia,pp.15.
[19] Ahmad, M.A. (2008) Vibration and Input Tracking
Control of Flexible Manipulator Using LQR with
NonCollocated PID Controller. Proceedings of the
2
nd
UKSIM European Symposium on Computer
Modeling and Simulation, Liverpool, United
Kingdom,pp.4045.
[20] Shchuka,A.andGoldenberg,A.A.(1989)Tipcontrol
of a singlelink flexible arm using feedforward
technique.MechanicalMachineTheory.24:439455.
[21] Mohamed,Z.andAhmad,M.A.(2008)HybridInput
ShapingandFeedbackControlSchemesofaFlexible
Robot Manipulator. Proceedings of the 17
th
World
Congress The International Federation of Automatic
Control,Seoul,Korea,pp.1171411719.
[22] Rattan, K.S., Feliu, V. and Brown, H.B. (1990) Tip
position control of flexible arms. Proceedings of the
IEEEConferenceonDecisionandControl,Honolulu,
pp.18031808.
[23] Quanser Student Handout, Rotary Flexible Joint
Module.http://www.quanser.com.
[24]Mohamed,Z.andTokhi,M.O.(2002)Vibrationcontrol
of a singlelink flexible manipulator using command
shaping techniques. Proceedings IMechEI: Journal of
SystemsandControlEngineering.216:191210.

9 Mohd Ashraf Ahmad, Mohd Zaidi Mohd Tumari and Ahmad Nor Kasruddin Nasir:
Composite Fuzzy Logic Control Approach to a Flexible Joint Manipulator
www.intechopen.com