You are on page 1of 15

ARTICLE

International Journal of Advanced Robotic Systems

Chattering-Free Neuro-Sliding
Mode Control of 2-DOF Planar
Parallel Manipulators
Regular Paper

Tien Dung Le1, Hee-Jun Kang2,* and Young-Soo Suh2


1 Graduate School of Electrical Engineering, University of Ulsan, Ulsan, South Korea
2 School of Electrical Engineering, University of Ulsan, Ulsan, South Korea
* Corresponding author E-mail: hjkang@ulsan.ac.kr

Received 20 Aug 2012; Accepted 16 Nov 2012

DOI: 10.5772/55102

© 2013 Le 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 This paper proposes a novel chattering free Keywords Parallel Manipulators, Neural Network,
neuro‐sliding mode controller for the trajectory Sliding Mode Control, Chattering‐Free, Uncertainty
tracking control of two degrees of freedom (DOF) Compensator
parallel manipulators which have a complicated
dynamic model, including modelling uncertainties,
frictional uncertainties and external disturbances. A 1. Introduction
feedforward neural network (NN) is combined with an
Parallel manipulators are closed‐loop kinematic chain
error estimator to completely compensate the large
mechanisms which have such advantages as high
nonlinear uncertainties and external disturbances of
accuracy, high stiffness, high payload capability and low
the parallel manipulators. The online weight tuning
moving inertia, etc. They are widely used in numerous
algorithms of the NN and the structure of the error applications, such as humanoid robots, simulators,
estimator are derived with the strict theoretical medical robots and micro‐robots, and they are becoming
stability proof of the Lyapunov theorem. The upper increasingly popular in the machine‐tool industry [1].
bound of uncertainties and the upper bound of the Compared to serial manipulators, the dynamic model of
approximation errors are not required to be known in parallel manipulators is significantly complicated by the
advance in order to guarantee the stability of the presence of multiple closed‐loop chains and singularities.
closed‐loop system. The example simulation results As a result, the control of parallel manipulators needs
show the effectiveness of the proposed control strategy more advanced control techniques than that of serial
for the tracking control of a 2‐DOF parallel manipulators.
manipulator. It results in its being chattering‐free, very
small tracking errors and its robustness against The motion control of parallel manipulators has attracted
uncertainties and external disturbances. many researchers in studying its potential performance.

www.intechopen.com Int J Adv


Tien Dung Robotic Sy,
Le, Hee-Jun Kang2013,
andVol. 10, 22:2013
Young-Soo Suh: 1
Chattering-Free Neuro-Sliding Mode Control of 2-DOF Planar wParallel Manipulators
A proportional derivative (PD) controller [2], a nonlinear eliminate chattering while achieving small tracking errors
PD controller [3] and an adaptive switching learning PD becomes an important topic.
control method (ASL‐PD) [4] were proposed for the
motion control of parallel manipulators. All of these In recent years, intelligent methods such as neural
controllers are simple and easy to implement but they do networks and fuzzy logic systems have been successfully
not perform well because the full dynamics of the robots applied in order to universally approximate the unknown
are not considered and compensated for. In [5‐7], the dynamics and uncertainties of robotic manipulators.
synchronization controllers were presented for parallel Many important adaptive NN‐based and fuzzy logic‐
manipulators. This kind of controller ‐ also only based on based SMC schemes have been proposed in which the
the parallel manipulator’s kinematics ‐ can improve the discontinuous control of the conventional SMC is
performances of the trajectory tracking further, but it replaced by a NN or fuzzy compensators. These
requires more complicated computation. Some other controllers consist of an equivalent continuous feedback
advanced controllers were proposed, such as the control component and a component derived from the
computed torque controller [8‐11], the model‐based intelligent compensators for the compensation of
iterative learning controller [12] and the adaptive uncertainties and external disturbances. If the
controller [13]. These approaches are based on having full uncertainties and external disturbances are sufficiently
knowledge of the dynamic model of the robot and require compensated, there is no need to use the discontinuous
heavy computational power. However, it is impossible to feedback control law to achieve the sliding mode and,
obtain a precise dynamic model of the parallel therefore, the chattering phenomenon can be eliminated.
manipulators, due to the presence of multiple closed‐loop For example, in [16] and [22], two similar adaptive fuzzy
chains, singularities, structured and unstructured sliding mode controllers for serial robotic manipulators
uncertainties and external disturbances. Hence, there is a were proposed in which an adaptive single‐input single‐
need for control strategies for parallel manipulators with output (SISO) fuzzy logic system (FLS) or SISO radial
robustness, adaptive capability, fast convergence and a basic function networks (RBFNs) were used to
simple structure. approximate the discontinuous part of the control signal.
These controllers can eliminate the chattering
Sliding mode control (SMC) has received much attention phenomenon and they do not need knowledge of the
in the last few decades as a useful and powerful, robust upper boundary of the uncertainties. However, the
control method in overcoming uncertainties, bounded typical SISO, FLS or SISO RBFN cannot approximate
external disturbances and unpredictable parameter precisely the complicated, highly nonlinear uncertainties
variations [14, 15]. The theory of SMC has been and external disturbances of robotic manipulators.
successfully applied to serial manipulators [14, 16, 17] Therefore, the bounds of the approximation errors have
and parallel manipulators [18‐21]. The main characteristic to be known for the chosen control gains of an auxiliary
of SMC is the inclusiveness of a discontinuous control controller for enhancing the stability of the control
input which drives the control system towards a sliding system. In another paper, two kinds of adaptive SMC
surface S(x,t)=0 whereby the sliding mode happens along schemes for a serial robotic manipulator using a fuzzy
this surface. For handling large uncertainties (such as in compensator were proposed [23]. In these SMC schemes,
the case of parallel manipulators) and external the decomposition of the uncertainties’ function is
disturbances, a large gain of the discontinuous control introduced and the properties of the uncertainties and the
input must be applied. For choosing the value of this dynamics of the serial manipulators are considered.
gain, the upper bound of the uncertainties has to be However, these schemes are still complicated and the
known in advance. It is, however, not easy to estimate number of fuzzy rules of each FLS is big. Moreover, in
this bound of the uncertainties. In addition, a large [24], an adaptive fuzzy SMC for affine nonlinear systems
switching gain is undesirable for the increased chance of was developed and successfully applied to control a four‐
input chattering, which leads to the high wear of the bar linkage system. A FLS is combined with a switching
mechanism and the excitation of un‐modelled high‐ control to compensate the large uncertainties of the
frequency dynamics. The most common approach for control system. However, this control method can only be
chattering reduction is to define a boundary layer around applied to SISO dynamic systems and it seems that the
the sliding surface and then use a continuous chattering phenomenon still exists in their experimental
approximation of the switching control input within the results. Furthermore, an adaptive control of robot
boundary layer [14]. This approach can reduce the manipulators with the NN‐based compensation of
chattering effectively, but there is a tradeoff between frictional uncertainties was proposed in [25]. The friction
asymptotic tracking and the elimination of chattering for dynamic which is usually the cause of performance
the width of the boundary layer. A thicker boundary degradation at low velocities in the motion control of
layer would reduce the chattering but make the tracking robotic manipulators is successfully modelled and
error bigger, and vice versa. Therefore, when applying compensated for. However, the control scheme did not
the SMC for parallel manipulators, the issue of how to consider the compensation of any modelling errors and

2 Int J Adv Robotic Sy, 2013, Vol. 10, 22:2013 www.intechopen.com


external disturbances of the robotic control system. On closed‐loop system. The structure of the NN and the
the other hand, a neuro‐sliding mode control of robotic error estimator, as well as the online learning
manipulators was presented in [26]. The controller used algorithm of the NN in the proposed controller of
two parallel NNs ‐ one for learning the continuous this paper, are simple and easy to implement but
equivalent control and the other for the discontinuous lead to the acquisition of good results.
control computation. Although the experimental studies
of the paper showed good results, there is a lack of The rest of the paper is organized as follows. In section 2,
mathematical proof of stability. In addition, it is difficult the dynamic model of the 2‐DOF planar parallel
to implement this controller with robotic manipulators, manipulators is formulated in the active joint space based
which have a complicated structure due to the large size on Lagrange‐DʹAlembert and the virtual work approach.
of NNs and the enormous number of calculations. In section 3, the architecture of the feedforward NN
which is used in the proposed controller is presented. The
Unlike serial manipulators, the dynamic behaviour of proposed neuro‐sliding mode controller for the trajectory
parallel manipulators is strongly nonlinear due to the tracking of 2‐DOF planar parallel manipulators is
highly dynamic coupling between the links. In addition, presented in section 4. In section 5, simulations of
the number of links in parallel manipulators is often trajectory tracking are carried out in order to verify the
double or triple the number of links in serial effectiveness of the proposed controller. Finally, several
manipulators with the same number of degrees of important remarks are concluded in section 6.
freedom. Therefore, the modelling errors and
uncertainties in parallel manipulators are significantly 2. Dynamic model of 2‐DOF planar parallel manipulators
larger and more complicated than in the case of serial
manipulators. This makes it difficult to apply the above In this section, we develop a dynamic model for a class of
mentioned method which as proposed for serial 2‐DOF planar parallel manipulators acting on a
manipulators to parallel ones. In this paper, we propose a horizontal plane and a reference frame is established in
novel SMC method for a tracking controller of 2‐DOF the workspace, as depicted in Figure 1. This kind of
parallel manipulators. First, the dynamic model of the 2‐ parallel manipulator consists of two active joints and
DOF parallel manipulators is analysed to build a three passive joints. The active joints are actuated by
continuous equivalent control. Next, a feedforward NN is actuators while the passive joints are free to move.
combined with an error estimator to sufficiently
y Active joints
compensate the modelling errors, uncertainties and
external disturbances of the parallel manipulator system. E(x,y) Passive joints
The weights of the NN are adapted online and the
proposed controller can guarantee the stability of the l12 l22
 p2
closed‐loop system, overcome the chattering problem and
improve the robustness of the control system. Compared
 p1 P2
with the existing controllers, the advantages of the P1
control strategy proposed in this paper are twofold: l21
l11 a1
1. By using a NN combined with an error estimator, the a2
x
large model uncertainties and external disturbances
A1 l0 A2
in the parallel manipulators’ control system can be
sufficiently compensated for. This is different from Figure 1. The 2‐DOF planar parallel manipulator.
the existing methods using only a NN or a FLS for
compensating the perturbation and ignoring the By a = (a1,a2)T and p = (p1,p2)T we denote the
approximation errors or higher‐order terms of corresponding active joint vector and passive joint vector,
Taylor’s series expansion. This feature means that the respectively; X = (x,y)T as the Cartesian coordinates; and
proposed control method is suitable for application E(x,y) as the end‐effector of the parallel manipulators. The
to the tracking control of parallel manipulators link lengths of the parallel manipulators are l11 = A1P1, l12 =
which have a complicated dynamic model and huge P1E, l21 = A2P2, l22 = P2E and l0 = A1A2.
uncertainties.
2. The control strategy does not need to know the The 2‐DOF planar parallel manipulators have one closed‐
upper bound of any uncertainties and also does not loop. Suppose that this closed‐loop is virtually cut at the
need to know the bound of any approximation common point E and forms an equivalent tree‐structure
errors. The condition of the chosen control gains in open kinematic chain, including two independent 2‐DOF
the proposed controller is simple (it just needs a serial manipulators, as shown in Figure 2. It is assumed
positive parameter and it does not need to know any that in this open kinematic chain all of the passive joints
threshold) but can guarantee the stability of the are actuated by virtual actuators. Suppose that the

www.intechopen.com Tien Dung Le, Hee-Jun Kang and Young-Soo Suh: 3


Chattering-Free Neuro-Sliding Mode Control of 2-DOF Planar wParallel Manipulators
equivalent tree‐structure makes the same motion as the 1  2
Li     2  i cos( ai   pi )aipi   ipi2  (3)
original closed‐chain without force or moment interaction 2  i ai 
at the virtually cut joint.
in which i = mi1r2i1 + Izi1 + mi2l2i1, i = mi2li1ri2, and i = mi2r2i2
The dynamic model of each planar 2‐DOF serial + Izi2 correspond to the dynamic parameters, i = 1,2.
manipulator in the tree‐structure is given by Lagrangeʹs
equation [27]: Substituting the Lagrange function (3) into Lagrange’s
equation (1), we have the dynamic equations of each
d  Li  Li serial chain of a tree‐structure as:
   τ i  Fi , i  1,2 (1)
dt  θ i  θi
  C (θ ,θ )θ  F  τ , i  1,2
M i (θi )θ (4)
i i i i i i i

where Li is the Lagrangian function; i = (ai,  ) is the pi T


where Mi  22 is the inertia matrix and Ci  22 is the
joint vector; i = (ai, pi)T is the joint torque vector; and Fi =
Coriolis and centrifugal force matrix, which are defined as:
(fai, fpi)T is the vector of the active and passive joint friction
torques of the ith serial chain.  i  i capi 
M i (θi )   , (5)
y Active joints   i capi i 

Virtually
cut Passive joints
 0 i sapipi 
l22 Ci (θi ,θ i )    (6)
l12   i sapiai 0 
 p2

 p1 P2 where the symbols capi and sapi are, respectively, defined


P1
as: capi = cos(ai ‐ pi), sapi = sin(ai ‐ pi), i = 1,2.
l21
l11 a1 Combining the dynamic models of two open serial chains
a2
x together, the dynamic model of the equivalent tree‐
A1 l0 structure mechanism is obtained as:
A2

Figure 2. Tree‐structure system of the cutting joint.   C θ  F  τ


M tθ (7)
t t t

As the parallel manipulator operates on a horizontal in which  = (a,p)T  4 is the vector of the joint angles;
planar plane, the Lagrangian functions only contain the t = (a,p)T  4 is the joint vector; and Ft = (Fa,Fp)T  4 is
kinetic energy of the mechanism: the friction torque vector of the equivalent tree‐structure
open chain system. a = (a1,a2)T and p = (p1,p2)T = (0,0)T
1 1 are the input torque vectors of the active joints and
Li  m ( x 2  y i21 )  I zi1ai2 
2 i1 i1 2 passive joints respectively. Fa = (fa1,fa2)T and Fp = (fp1,fp2)T =
(0,0)T are the friction force vectors of the active joints and
1 1 passive joints respectively. Here, we assume that the
 mi 2 ( x i22  y i22 )  I zi 2pi2 (2)
2 2 effect of the friction force on the passive joints is much
smaller than that on the active joints. Thus, in order to
where mi1 and mi2 are the masses of the links of the serial simplify the dynamic model, only the disturbances on the
chain i; Izi1, Izi2 are the inertia tensors of the links of the active joints are considered.
serial chain i (i=1,2).
The inertia matrix Mt  44 and the Coriolis and
Letting ri1 and ri2 be the distance from the joints to the centrifugal force matrix Ct  44 in (7) are expressed by:
centre of mass for each link, we have:
 1 0 1c ap1 0 
x i1  ri1 sin  aiai ,  
 0 2 0  2 c ap 2 
Mt   , (8)
y i1  ri1 cos  aiai ,  1c ap1 0 1 0 
 0  2 c ap 2 0 2 
 
x i 2  li1 sin  aiai  ri 2 sin  pipi ,
 0 0 1sap1p1 0 
 
y i 2  li1 cos aiai  ri 2 cos pipi .  0 0 0  2 sap 2p 2 
Ct  
  (9)
  1sap1 a1 0 0 0 
By substituting the above equations into (2), the  0   2 sap 2a 2 0 0 
Lagrangian function becomes:  

4 Int J Adv Robotic Sy, 2013, Vol. 10, 22:2013 www.intechopen.com


Next, the loop constraints are taken into account using a robot will never be known. If the modelling errors
Jacobian matrix. From D’Alembert’s principle and the caused by the uncertainties are bounded, we can
principle of virtual work, the active joint torques a and express the actual dynamics by combining the
the generalized torques t satisfy the following equation estimated dynamics with the modelling errors as the
[28]: following equation:

τ a  ΨT τt (10)   C θ  F  τ
M aθ (17)
a a a a a

where  = /a  42 is the Jacobian matrix of all the ˆ  M and C  C


where M a  M ˆ  C are the actual
a a a a a
joints with respect to the active joints. We have:  = [I, J]T dynamic parameters of the parallel manipulators; and
where I  22 is identity matrix and J = p/a  22 is Ma and Ca are the bounded modelling errors.
computed from the loop constraint equations:
Consider the external disturbances which affect the active
1
θp  h   h  joints; we define the vector of unknown uncertainties and
J      (11) external disturbances as:
θa  θp   θa 
  C θ  F  d(t )
τ a  M aθ (18)
a a a a
 hx  l11ca1  l12c p1  l0  l21ca 2  l22 c p 2 
h   (12)
 hy   l11sa1  l12 sp1  l21sa 2  l22 sp 2  where d(t)  2 is the vector of the external
disturbances.
where the symbols cai, sai and cpi, spi are, respectively,
defined as: cai = cosai, sai = sinai, cpi = cospi, spi = sinpi, i = From (17) and (18), we obtain the actual dynamic
1,2. equation of the 2‐DOF planar parallel manipulators in the
active joint space:
Taking the constraint (10) into the tree‐structure by
multiplying both sides of equation (7) by  T, we M̂ aθ ˆ θ  τ  τ
  C (19)
a a a a a
obtain:
The dynamic equation (19) is very useful for the analysis,
  ΨT C θ  ΨT F  ΨT τ
ΨT M t θ (13) simulation study and control design of the 2‐DOF planar
t t t
parallel manipulators.
In addition, we have the following relationships:
Although the dynamic model (19) of the 2‐DOF
θ 
θ  θ (14) parallel manipulators in the active joint space has a
θa a similar form to the model (4) of the serial
manipulators, its dynamic behaviour is much more
  Ψθ
θ    Ψθ
 (15)
a a complicated and strongly nonlinear due to the highly
dynamic coupling between the links. We can see that
By substituting (10), (14) and (15) into (13), we obtain the
the nominal parameter matrices in equation (16) are
dynamic model of the 2‐DOF planar parallel
much more complicated in comparison with the
manipulators in the active joints space:
dynamic model of serial manipulators. In addition, the
ˆ θ  F  τ
  C
M̂ aθ (16) uncertainties  a in the dynamic model (18) are huge
a a a a a
and highly nonlinear due to the large number of links,
where M̂a  ΨT Mt Ψ  22 is the inertia matrix and the loop constraints, the friction and the variation of
  Ψ TC Ψ  22 is the Coriolis and
Ĉa  ΨT Mt Ψ the parameters. Thus, it is difficult to reuse the existing
t
centrifugal matrix. control strategies which were proposed for serial
manipulators in the literature for improving the
The dynamic model (16) has the following properties, performance of the tracking control of the parallel
which are proved in [29]: manipulators.

Property 1: M̂ a is symmetric and positive definite. 3. The feedforward neural network architecture

The NN used in this paper has the structure indicated in


 ˆ is skew‐symmetric.
Property 2: M̂ a  2C a Figure 3. The architecture of the NN includes the input
layer, the hidden layer and the output layer. The NN has
Because of the presence of the highly nonlinear 2 outputs corresponding to the 2 active joints of the
uncertainties, the exact dynamic model of the parallel parallel manipulators considered.

www.intechopen.com Tien Dung Le, Hee-Jun Kang and Young-Soo Suh: 5


Chattering-Free Neuro-Sliding Mode Control of 2-DOF Planar wParallel Manipulators
net i Gi The outputs of the NN can be represented in a vector
vij  form:
x1 wik
y  WT G( x,V ) 2 (27)
 y1
x2 T
 where G  G1 , G2 ,..., GN   R N h 1 .
 h 
 y2
x3 
4. The proposed controller

4.1 Traditional Sliding Mode Controller


Ni
xN i Nh Let da = (da1,da2)T be the desired state vector and e = a ‐
 da be the tracking error vector of the parallel
manipulators. The first step in the design of the sliding
mode control for the system (19) is to design the sliding
surface function as:
Figure 3. Structure of the feedforward neural network.

The input layer: The input vector of the NN is denoted by: s  e  Λe  θ a  (θda  Λe )  θ a  θ ar (28)

T where  = diag(1,2), 1 and 2 are positive constants


x   x1 , x2 ,..., xNi  (20)
which determine the motion feature in the sliding surface;
and θ ar  θ da  Λe is defined as the reference velocity
where Ni represents the number of components of the
vector.
input vector.

In the second step, a control law a  2 is designed such


The hidden layer: By denoting the number of neurons in
that the system state trajectories are driven to the sliding
the hidden layer as Nh, the weight matrix connecting the
surface and kept on the sliding surface. The reaching
input and the hidden layers is expressed by:
condition is expressed by [14]:

V   v1 , v2 ,..., vNh   
Ni N h
, 1 d 2
s   i si , i=1,2 (29)
2 dt i
 
T
vi  vi1 , vi 2 ,..., viN  Ni , i  1, N h (21)
i where i is a strictly positive constant. Equation (29)
indicates that the energy of s should decay so long as s is
The inputs and outputs of the hidden layer are, not zero.
respectively, presented as:

N In general, the control input a consists of two


neti   j i1 vij x j , (22) components:

Gi  g(neti ), i  1, Nh . (23) τ a  τ eq  τ sw (30)

The transfer function in the hidden layer is used as a where the first term eq  2 is the equivalent control which
sigmoid function: keeps the trajectory of the system state on the sliding
surface; and the second term sw  2 is the discontinuous
1  exp(  z) control which drives the system states toward the sliding
g( z )  (24)
1  exp(  z) surface when they are deviating from this surface.

The output layer: The weight matrix connecting the hidden The equivalent control is considered for the nominal
and output layers is expressed by: system in the absence of the uncertainties and external
disturbances:
 
T
W  w1 , w2 ,..., wN  N h 2 ,
h
ˆ θ
τ eq  M  ˆ  (31)
a ar  C aθar

wi   wi1 , wi 2    , i  1, N h2
(25)
The discontinuous control is designed as:
The outputs of the NN are expressed by:
τ sw  Ksign(s) (32)
Nh
yk   w G, k  1,2 (26)
i 1 ik i where K = diag(k1,k2), k1 and k2 are positive constants.

6 Int J Adv Robotic Sy, 2013, Vol. 10, 22:2013 www.intechopen.com


Now, by substituting (31) and (32) into (30), the the decay of the energy of s ‐ as long as s  0 ‐ is
conventional sliding mode controller for the parallel guaranteed and the reaching condition (29) is satisfied.
manipulators’ dynamic model (19) is presented by:
4.2 Chattering Free Neuro‐Sliding Mode Controller
ˆ θ ˆ 
τa  M a ar  C aθar  Ksign(s) (33)
The sign function in the discontinuous control term of the
SMC (33) leads to the chattering problem. Therefore, in
It has been proven in [19] that by considering the
this section we propose a new controller in which the
Lyapunov function candidate as:
discontinuous control is replaced by a NN combined with
1 Tˆ an error estimator. The connection weights of the NN are
V s M as (34)
2 adapted online and the structure of the error estimator fest
is designed with the strict theoretical stability proof of the
and choosing the switching gain matrix K such that Lyapunov theorem. The structure of the overall system is
ki   ai , where  ai is the upper boundary of presented in Figure 4.
bound bound
 ai , the overall system is asymptotically stable. Thus,

τ NN
s

θda , θ da τa θa , θ a
s
s  e  Λe T
fest
s τ eq
 

θ ˆ θ Cˆ θ


τeq  M
da a ar a ar

Figure 4. Block diagram of the proposed controller.

The proposed controller is expressed by the following V*); N  2 is the bounded reconstruction error due to the
equation: inadequate number of neurons in the hidden layer of the
NN.
ˆ θ
τa  M  ˆ 
a ar  Caθar  fNN  fest  Ts (35)
For convenience, equation (36) is rewritten as:
where fNN  2 is the output of the NN whose structure is
described in section 3 for the online approximation of the  W
τ a  fNN  W*T G ˆ ε
 TG (37)
N
unknown vector a  2. Since the output of the NN is
not able to approximate a accurately, the error  and W
where the symbols G * , Ĝ , G  are, respectively,
estimator fest  2 in (56) is used to attenuate the * *
defined as G  G( x , V )  Nh, G ˆ  G( x , Vˆ )  Nh,
approximation errors. The term Ts is used in enhancing ˆ  Nh and W
  G*  G   W*  Wˆ  Nh2.
G
the robustness of the control system. Moreover, T =
diag(T1, T2) is a diagonal positive definite matrix in which  for a given x can be
The Taylor series expansion of G
T1 and T2 are positive constants.
written as follows:
The inputs of the NN are chosen as the errors and
T G    G VT 
derivative of errors: x   e1 , e1 , e2 , e 2  4 . The NN is 1  1

    VT 
G2   G
used to approximate the unknown uncertainties a
online. The approximation error is denoted as:

G

 


2



 ˆ  Ο( V
V*  V ˆ T x)

 
G    
Nh  G VT 
ˆ T G( x,V
τ a  fNN  W*T G( x,V* )  W ˆ )ε (36)
ˆ
|V  V  Nh |V  Vˆ
N

where W*  Nh2 and V*  4Nh are the optimal values of  GV V ˆ T x)


 T x  Ο( V (38)
the weight matrices W and V; Ŵ  Nh2 and V̂  4Nh
are the estimates of the optimal weight matrices (W* and

www.intechopen.com Tien Dung Le, Hee-Jun Kang and Young-Soo Suh: 7


Chattering-Free Neuro-Sliding Mode Control of 2-DOF Planar wParallel Manipulators
where: The derivative of V1 is computed as follows:

 
T 1
 G GN  V 1  s T M ˆ s  sT M
ˆ s  sT M ˆ s (46)
GV   1 , G2 , , N h  Nh a a a
h
 2
T T
 V x V x VT x |V  Vˆ
Properties 1 and 2 in section 2 give us:
ˆ 
 V V
V * 4 N h
ˆ s  sT M
s T M ˆ s (47)
a a
ˆ T x)  Nh is a vector of higher‐order terms and it is
Ο( V
assumed to be bounded. ˆ s  2sT C
sT M ˆ s (48)
a a

By substituting (38) into the approximation error By substituting (47) and (48) into (46) we obtain:
equation (37), we have:
V 1  sT C ˆ s 
ˆ sM
τ a  fNN *T
 W GV V ˆ  W*T Ο( V
 TG
Tx  W ˆ T x)  ε  a a 

N

ˆ sM
 sT C ˆ θ ˆ  
ˆ TG V
W Tx  W ˆ δ
 TG (39)  a a a  M aθar 
V

where: ˆ sτ C
 sT C ˆ θ  τ  M
ˆ θ  (49)
 a a a a a a ar 

 TG V
δW ˆ T x)  ε
 T x  W *T Ο( V (40)
V N Now, substituting the proposed controller (35) into (49)
we obtain:
Next, we design an error estimator fest to estimate and
compensate for the error vector   2. The online V 1  sT fNN  τ a  fest  Ts  (50)
learning algorithms of the NN and the structure of the
error estimator fest are derived in the next section From (39) and (50), we have:
following the Lyapunov method.
ˆ TG V
V 1  sT   W ˆ  δ  f  Ts 
 TG
Tx  W (51)
4.3 Stability analysis  V est 

According to the Lyapunov stability analysis, the system From (41) and (51), we have the derivative of the
is stable if the Lyapunov function candidate is positive Lyapunov candidate function:
definite and its derivative is negative semidefinite.
V  V 1  V 2  V 3  V 4
Consider the Lyapunov function candidate:
ˆ TG V
 sT   W  Tx  W ˆ δ
 TG   Ts  
 V 
V  V1  V2  V3  V4 (41)

where:  T 1W
tr W  
ˆ  tr V ˆ  δ
 T 1V T ξ 1f

est (52)

1 Tˆ
V1  s M as (42) As V is desired to be at least negative semidefinite, we
2
choose the online update laws for the NN and design the
estimator as follows:
V2 
1
2
  T 1W
tr W 
 (43)
ˆ   Gs
ˆ T
W (53)

V3 
1
2
tr V 
 T  1V

 (44) 
 ˆ
V̂  ‐  x G v Ws 
T
(54)

1 T 1
V4  δ ξ δ (45) fest  ξs (55)
2

in which  and  are the positive learning rates;  is a fest  ξ  sdt  Γ (56)
diagonal positive constant matrix of the error estimator
(56); and δ  δ  fest is the estimated error. where equation (56) is derived from equation (55).
Moreover, the constant matrix  is eliminated from the
Obviously, V1, V2, V3 and V4 are positive definite integration in (56) since the recursive estimation algorithm
functions. Therefore, V is a positive definite function. can recover it. The constant vector  is chosen as zero.

8 Int J Adv Robotic Sy, 2013, Vol. 10, 22:2013 www.intechopen.com


-4
Next, by substituting (53)‐(56) into (52), we have: x 10

ˆ TG V
V  sT  W ˆ δ
 TG
Tx  W   Ts   5

Tracking error X‐direction [m]


 V 
0


ˆ T  tr  V
    T
T
 T Gs
tr W T ˆ
 x G v Ws δ s
  -5

ˆ TG V
 sT   W  Tx  W ˆ δ
 TG   Ts  
 V  -10

 sT W ˆ  sT W
 TG ˆ TG V T s
Tx  δ 0 1 2 3 4 5 6 7 8 9 10
V Time [s]

(a)
 sT Ts  0 (57)
-3
x 10
In (57), since T is a diagonal positive definite matrix, 1
V  0 only when s = 0. Therefore, we can see that the

Tracking error Y‐direction [m]


0.5
control system is asymptotically stable with respect to s. 0
This means that:
-0.5

lim s  lim  e  e   0 (58) -1


t  t  -1.5

Or, equivalently: -2
0 1 2 3 4 5 6 7 8 9 10
Time [s]
 lim e  0  lim θa  θda
t  t  (b)
 (59)
tlim e  0  lim θ a  θ da Conventional SMC using BLM
 t 
Adaptive Fuzzy SMC
Proposed controller
Thus, it is proved that, with the proposed controller (35), the
actual active joint positions converge on their desired values. Figure 6. Tracking errors of the end‐effector in case 1: (a) X–
direction and (b) Y‐direction.
5. Simulation study
In practice, it is difficult to measure the distances from the
Simulation studies were conducted on Matlab‐Simulink and joint to the centre of mass and the inertia tensors of the
the mechanical model of the 2‐DOF planar parallel links. As such, we conducted the simulations with
manipulator was built in SimMechanics toolbox. The link different parameters, both in the mechanical model of the
parameters in the mechanical model are set as follows: l11 = robot and in the controllers, as follows:
l21 = 0.102(m), l12 = l22 = 0.18(m) and l0 = 0.132(m) are the link
rˆi1  0.9ri1 and rˆi 2  0.9ri 2 , i = 1,2 (60)
lengths; r11 = 0.05(m), r21 = 0.055(m), r12 = r22 = 0.09(m) are the
distances from the joint to the centre of mass of the links; m11
where rˆi1 and rˆi 2 were used for calculating M̂ a and Ĉ a
= 0.8(kg), m21 = 0.78(kg), m12 = 1.17(kg), m22 = 1.2(kg) are the
in the controllers.
masses of the links; Iz11 = 0.0027(kg.m2), Iz21 = 0.0031(kg.m2),
Iz12 = Iz22 = 0.0013(kg.m2) are the inertia tensors of the links.
The friction models of the system, including the viscous
friction and the Coulomb friction torques, are defined as
follows:
0.22
A0

0.2

fai  Fci sign(ai )  Fviai , i  1,2 (61)


0.18

where the coefficients are chosen as Fc1 = Fc2 = 0.3 and Fv1 =
Y[m]

0.16

Fv2 = 2.
0.14

0.12 The simulations were carried out with respect to those


0.1
situations when the end‐effector of the parallel
0 0.02 0.04 0.06 0.08 0.1 0.12
manipulator is driven to track a circular trajectory on the
X [m] XY plane under different initial conditions. To illustrate
Desired trajectory
the improvement in performance, the proposed controller
Conventional SMC using BLM method (35) is compared with two other controllers:
Adaptive Fuzzy SMC
Proposed controller + A conventional SMC using the boundary layer method
Figure 5. Results of circular trajectory tracking control in case 1. (BLM) which was proposed in [14]:

www.intechopen.com Tien Dung Le, Hee-Jun Kang and Young-Soo Suh: 9


Chattering-Free Neuro-Sliding Mode Control of 2-DOF Planar wParallel Manipulators
a ar  C aθar  Ksat  s Φ 
ˆ θ
τa  M  ˆ  (62)

Control input active joint 1 [Nm]


10
(c)
where K is a diagonal positive matrix of switching gains
chosen as K = diag{10, 10}; sat(s/) = [sat(s1/1), 5
sat(s2/2)]T is a vector of saturation functions which is
defined in [14]; and 1 and 2 are the boundary layer 0
thicknesses, chosen as 1 = 2 = 0.15.
-5
0 1 2 3 4 5 6 7 8 9 10
(a)
Control input active joint 1 [Nm]

10

(c)

Control input active joint 2 [Nm]


10
5

(d)
5
0

-5 0

0 1 2 3 4 5 6 7 8 9 10
Time [s]
-5
(a) 0 1 2 3 4 5 6 7 8 9 10
Time [s]

(b)
Control input active joint 2 [Nm]

10
5
(d)

Control input active joint 2 [Nm]


Control input active joint 1 [Nm ]

3.6
5 4.5
3.5
4
3.4
0
3.5 3.3

3 3.2
-5
0 1 2 3 4 5 6 7 8 9 10 3.1
2.5
Time [s] 3
2
(b) 4.5 4.6 4.7 5.9 6 6.1 6.2
Time [s] Time [s]

(c) (d)
Control input active joint 2 [Nm]
Control input active joint 1 [Nm]

5 3.6
Figure 8. Case 1 when applying the adaptive fuzzy SMC: (a)
4.5 3.5 Control input of the active joint 1; (b) Control input of the active
3.4 joint 2; (c) Enlargement of the localized region in (a); (d)
4
3.3
Enlargement of the localized region in (b).
3.5
3.2
The parameters in the proposed controller are set as: T =
3
3.1 diag{10, 10}; the number of neurons in the hidden layer of
2.5 3 the NN is 10; the learning rates are chosen as  = 10‐3 and
4.5 4.6
Time [s]
4.7 5.9 6 6.1
Time [s]
6.2
 = 210‐3; the initial weight matrices are W(0) = 0.001 
I410 and V(0) = 0.001  I102; the constant matrix in the
(c) (d)
error estimator (50) is chosen as  = diag{5000, 5000}.
Figure 7. Case 1 when applying the conventional SMC using
BLM: (a) Control input of the active joint 1; (b) Control input of
In simulations, the 2‐DOF parallel manipulator is
the active joint 2; (c) Enlargement of the localized region in (a);
(d) Enlargement of the localized region in (b). disturbed by step external disturbance forces da1(t) = [2,
0]T at t = 2.5s (on active joint 1) and periodic external
+ An adaptive fuzzy sliding mode controller which was disturbance forces da2(t) = [0, 2sin(1.7t)]T at t = 5s (on
proposed in [22]: active joint 2).

ˆ θ
τa  M  ˆ  (63)
a ar  C aθar  As  K fuzzy First, in case 1, we conduct the simulation when the initial
position of the end‐effector E(x,y) of the parallel
where A is the diagonal positive parameter matrix chosen manipulator lies on the top of the reference circular
as A = diag{10, 10}; and Kfuzzy = [kfuzzy1, kfuzzy2]T in which each trajectory A0(0.066, 0.21). The centre coordinates of the
kfuzzyi is estimated by an individual SISO fuzzy system (i = reference circular trajectory are (0.066, 0.16) and the
1, 2). radius is 0.05. The end‐effector is driven to track the

10 Int J Adv Robotic Sy, 2013, Vol. 10, 22:2013 www.intechopen.com


circular trajectory 5 times over 10 seconds. Figure 5 shows
the results of the tracking control of the 2‐DOF parallel 0.22 A0
manipulator in case 1.
0.2
Control input active joint 1 [Nm]

10 0.18

(c)

Y [m]
0.16
5

0.14
0

0.12

-5
0 1 2 3 4 5 6 7 8 9 10 0.1
0 0.02 0.04 0.06 0.08 0.1 0.12
Time [s]
X[ ]
(a) Desired trajectory
Conventional SMC using BLM method
Adaptive Fuzzy SMC
10 Proposed controller
Control input active joint 2 [Nm]

Figure 10. Results of the circular trajectory tracking control in


(d)
5 case 2.

-3
0 x 10
1
Tracking error X‐direction [m]

0
-5
-1
0 1 2 3 4 5 6 7 8 9 10
Time [s] -2

(b) -3
Conventional SMC using BLM
-4 Adaptive Fuzzy SMC

5 -5 Proposed controller
3.6
Control input active joint 2 [Nm]
Control input active joint 1 [Nm]

-6
4.5 3.5 0 1 2 3 4 5 6 7 8 9 10
Time [s]
4 3.4
(a)
3.5 3.3

3 3.2 -3
x 10

2.5 3.1 1
Tracking error Y‐direction [m]

0
2 3
4.5 4.6 4.7 5.9 6 6.1 6.2 -1
Time [s] Time [s]
-2
(c) (d) -3
Conventional SMC using BLM
Figure 9. Case 1 when applying the proposed controller: (a) -4 Adaptive Fuzzy SMC

Control input of the active joint 1; (b) Control input of the active -5 Proposed controller

joint 2; (c) Enlargement of the localized region in (a); (d) -6


0 1 2 3 4 5 6 7 8 9 10
Enlargement of the localized region in (b). Time [s]

(b)
Figure 6 shows the tracking errors of the end‐effector in
Figure 11. Tracking errors of the end‐effector in case 2: (a) X–
the X‐direction and in the Y‐direction in case 1. As can
direction and (b) Y‐direction.
be seen from the figure, the tracking errors caused by
the adaptive fuzzy SMC are a little bit smaller than the The control inputs of the active joints 1 and 2 in case 1
errors associated with the conventional SMC using of the conventional SMC using BLM are shown in
BLM. In particular, the proposed controller brings about Figure 7. From the enlargements of the localized
the smallest tracking errors (almost converging on zero) regions, it can clearly be seen that the chattering
compared with the conventional SMC using BLM and phenomenon remains. If we increase the boundary
the adaptive fuzzy SMC. In addition, it can be seen that layer thickness or decrease the switching gains for
the large model uncertainties and external disturbances reducing greater reduction of the chattering, the
are greatly compensated for using the proposed tracking errors will be increased and the robustness of
controller. the system will not be guaranteed.

www.intechopen.com Tien Dung Le, Hee-Jun Kang and Young-Soo Suh: 11


Chattering-Free Neuro-Sliding Mode Control of 2-DOF Planar wParallel Manipulators
10

Control input active joint 1 [Nm]


0 (c)
Sliding surface 1

5
-0.5

-1 Conventional SMC using BLM


0
Adaptive Fuzzy SMC
Proposed controller
-1.5
0 1 2 3 4 5 6 7 8 9 10 -5
Ti [ ] 0 1 2 3 4 5 6 7 8 9 10
(a) Time [s]
0.2
(a)
0

-0.2
Sliding surface 2

Control input active joint 2 [Nm]


10
-0.4

-0.6
Conventional SMC using BLM (d)
-0.8 5
Adaptive Fuzzy SMC
-1 Proposed controller

-1.2 0
0 1 2 3 4 5 6 7 8 9 10
Time [s]

(b) -5
Figure 12. Sliding surfaces of (a) active joint 1 and (b) active joint 0 2 4 6 8 10
Time [s]
2 in case 2.
(b)
10
Control input active joint 1 [Nm]

(c)
2.8

Control input active joint 2 [Nm]


Control input active joint 1 [Nm]

5.4
5

5.2 2.6

0 5 2.4

4.8 2.2
-5
0 1 2 3 4 5 6 7 8 9 10 4.6
Time [s] 2

(a) 4.4
4.1 4.2 4.3 4.4 4.5 6.2 6.25 6.3
10 Time [s] Time [s]
Control input active joint 2 [Nm]

(c) (d)
(d)
5 Figure 14. Case 2 with the adaptive fuzzy SMC: (a) Control input
of the active joint 1; (b) Control input of the active joint 2; (c)
0
Enlargement of the localized region in (a); (d) Enlargement of the
localized region in (b).

-5 Figures 8 and 9 show the control inputs of the active


0 1 2 3 4 5
Time [s]
6 7 8 9 10
joints 1 and 2 in case 1 using the adaptive fuzzy SMC and
the proposed controller, respectively. Compared with the
(b)
5.8
conventional SMC using BLM, it can be seen that the
Control input active joint 2 [Nm]

chattering is removed. However, the proposed controller


Control input active joint 1 [Nm]

2.8

5.6
2.6
can completely compensate for the uncertainties and any
external disturbances; hence, the tracking errors in the
5.4 2.4 case of using the proposed controller are almost reduced
5.2 2.2 to zero, as shown in Figure 6.

2
5 Next, a simulation is carried out to investigate the control
6.2 6.25 6.3
performance when the end‐effector E(x,y) of the parallel
4.1 4.2 4.3 4.4 4.5
Time [s] Time [s] manipulator does not lie on the reference circle. In this
(c) (d) case, we can demonstrate the convergence of the tracking
Figure 13. Case 2 with the conventional SMC using BLM: (a) errors and sliding functions. Figure 10 shows the
Control input of the active joint 1; (b) Control input of the active comparison of the trajectory tracking among the
joint 2; (c) Enlargement of the localized region in (a); (d) conventional SMC using BLM, the adaptive fuzzy SMC
Enlargement of the localized region in (b). and the proposed controller. The centre coordinates of the

12 Int J Adv Robotic Sy, 2013, Vol. 10, 22:2013 www.intechopen.com


reference circular trajectory are (0.066, 0.16) and the using the proposed controller converge on the smallest
radius is 0.05. The initial position of the end‐effector values (almost equal to zero).
E(x,y) of the parallel manipulator is A0 (0.071,0.215). The
values of the parameters in the controllers for case 2 are Figures 13‐15 show the control inputs of the active joints 1
set to be the same as with case 1 of the simulations. and 2 in case 2 for the conventional SMC using BLM, the
adaptive fuzzy SMC and the proposed controller,
respectively. From the enlargements of the localized
Control input active joint 1 [Nm]

10
(c) regions, it can clearly be seen that the chattering
phenomenon remains in the case of the conventional
5
SMC using BLM but that it is eliminated in the cases of
the adaptive fuzzy SMC and the proposed controller.
0
It can be concluded from the above‐mentioned simulation
results that the proposed controller is highly efficient in
-5
0 1 2 3 4 5 6 7 8 9 10 the control of the 2‐DOF planar parallel manipulators.
Time [s]

(a) 6. Conclusions

In this paper, we have presented a novel chattering‐free


Control input active joint 2 [Nm]

10
neuron sliding mode controller for tracking the control of
(d) 2‐DOF parallel manipulators. The proposed controller is
5
based on the combination of three ingredients. The first
ingredient is the equivalent control, which is derived
0
from the dynamic model of the 2‐DOF parallel
manipulators. The second one is a feedforward NN used
-5
to adaptively learn the large nonlinear uncertainties and
0 1 2 3 4 5 6 7 8 9 10
Time [s] external disturbances of the parallel manipulators. The
(b) final part is an error estimator for compensating the
approximation errors of the NN and the higher‐order
5.6
2.8
terms in Taylor series expansion. The online weight
tuning algorithms of the NN and the structure of the
Control input active joint 1 [Nm]

Control input active joint 2 [Nm]

5.4
2.6 error estimator are derived with the strict theoretical
stability proof of the Lyapunov theorem. The connection
5.2
2.4 weights of the NN can be adapted online during the
5
trajectory tracking control of the parallel manipulators
2.2
without any offline training phase. The main advantages
4.8 2
of the proposed controller in comparison with the
existing SMC methods are as follows: (1) The proposed
4.1 4.2 4.3 4.4 4.5 6.2 6.25 6.3 controller can completely compensate the large nonlinear
Time [s] Time [s]
uncertainties and external disturbances of parallel
(c) (d)
manipulators. (2) The proposed control strategy does not
Figure 15. Case 2 with the proposed controller: (a) Control input need to know either the upper bounds of any
of the active joint 1; (b) Control input of the active joint 2; (c) uncertainties or the bound of any approximation errors.
Enlargement of the localized region in (a); (d) Enlargement of
Its structure is simple, easy to implement and yet leads to
the localized region in (b).
the acquisition of good results. These advantages mean
The results of the tracking errors of the end‐effector in that the proposed controller is suitable in application to
case 2 on the X‐direction and the Y‐direction ‐ which are those tracking control parallel manipulators which have a
shown in Figure 11 ‐ prove the excellence of the control complicated dynamic model and huge uncertainties.
performance of the proposed controller in comparison
with the conventional SMC using BLM and the adaptive Simulation results demonstrated the effectiveness of the
fuzzy SMC. It can be seen that the proposed controller proposed controller in the trajectory tracking control of a
brings about the smallest and the fastest convergence of 2‐DOF parallel manipulator. It has been shown that the
tracking errors. proposed controller brings about the smallest tracking
errors compared with the conventional SMC using BLM
Figure 12 shows the comparison of the convergence of the and the adaptive fuzzy SMC. The chattering in the control
sliding functions among all three controllers. As can be inputs is eliminated and the control system is highly
seen from the figure, the sliding functions in the case robust against uncertainties and external disturbances.

www.intechopen.com Tien Dung Le, Hee-Jun Kang and Young-Soo Suh: 13


Chattering-Free Neuro-Sliding Mode Control of 2-DOF Planar wParallel Manipulators
7. Acknowledgements [12] Abdellatif H. and Heimann B. (2010) Advanced
Model‐Based Control of a 6‐DOF Hexapod Robot: A
The authors would like to express their thanks for Case Study. Mechatronics, IEEE/ASME Transactions
financial support from the Korean Ministry of Knowledge on, vol. 15, pp. 269‐279.
Economy under the Human Resources Development [13] Xiaocong Z., Guoliang T., Bin Y. and Jian C. (2009)
Program for Convergence Robot Specialists. (NIPA‐2012‐ Integrated Direct/Indirect Adaptive Robust Posture
H1502‐12‐1002) Trajectory Tracking Control of a Parallel Manipulator
Driven by Pneumatic Muscles. Control Systems
8. References
Technology, IEEE Transactions on, vol. 17, pp. 576‐
[1] Merlet J. P. (2006) Parallel Robots. Second ed.: Springer. 588.
[2] Ghorbel F. H., Chetelat O., Gunawardana R. and [14] Slotine J. J. E. and Li W. (1991) Applied Nonlinear
Longchamp R. (2000) Modeling and set point control Control: Prentice‐Hall.
of closed‐chain mechanisms: theory and experiment. [15] Young K. D., Utkin V. I. and Ozguner U. (1999) A
Control Systems Technology, IEEE Transactions on, control engineerʹs guide to sliding mode control.
vol. 8, pp. 801‐815. Control Systems Technology, IEEE Transactions on,
[3] Ouyang P. R., Zhang W. J. and Wu F. X. (2002) vol. 7, pp. 328‐342.
Nonlinear PD control for trajectory tracking with [16] Sadati N. and Ghadami R. (2008) Adaptive multi‐
consideration of the design for control methodology. model sliding mode control of robotic manipulators
Robotics and Automation, Proceedings. ICRA ʹ02. using soft computing. Neurocomputing, vol. 71, pp.
IEEE International Conference on, pp. 4126‐4131 . 2702‐2710.
[4] Ouyang P. R., Zhang W. J. and Gupta M. M. (2006) An [17] Zeinali M. and Notash L. (2010) Adaptive sliding
adaptive switching learning control method for mode control with uncertainty estimator for robot
trajectory tracking of robot manipulators. manipulators. Mechanism and Machine Theory, vol.
Mechatronics, vol. 16, pp. 51‐61. 45, pp. 80‐90.
[5] Lu R., Mills J. K. and Dong S. (2008) Trajectory [18] Kim N.‐I., Lee C.‐W. and Chang P.‐H. (1998) Sliding
Tracking Control for a 3‐DOF Planar Parallel mode control with perturbation estimation:
Manipulator Using the Convex Synchronized application to motion control of parallel manipulator.
Control Method. Control Systems Technology, IEEE Control Engineering Practice, vol. 6, pp. 1321‐1330.
Transactions on, vol. 16, pp. 613‐623. [19] Qi Z., McInroy J. and Jafari F. (2007) Trajectory
[6] Yuxin S., Dong S., Lu R. and Mills J. K. (2006) Tracking with Parallel Robots Using Low Chattering,
Integration of saturated PI synchronous control and Fuzzy Sliding Mode Controller. Journal of Intelligent
PD feedback for control of parallel manipulators. & Robotic Systems, vol. 48, pp. 333‐356.
Robotics, IEEE Transactions on, vol. 22, pp. 202‐207. [20] Yang Z., Huang T., Xu X. and Cooper J. (2007)
[7] Weiwei S., Shuang C., Yaoxin Z. and Yanyang L. Variable structure control of high‐speed parallel
(2009) Active Joint Synchronization Control for a 2‐ manipulator considering the mechatronics coupling
DOF Redundantly Actuated Parallel Manipulator. model. The International Journal of Advanced
Control Systems Technology, IEEE Transactions on, Manufacturing Technology, vol. 34, pp. 1037‐1051.
vol. 17, pp. 416‐423. [21] Shang W.‐w., S. Cong and S.‐l. Jiang (2010) Dynamic
[8] Yang Z., Wu J. and Mei J. (2007) Motor‐mechanism model based nonlinear tracking control of a planar
dynamic model based neural network optimized parallel manipulator. Nonlinear Dynamics, vol. 60,
computed torque control of a high speed parallel pp. 597‐606.
manipulator. Mechatronics, vol. 17, pp. 381‐390. [22] Yuzheng G. and Peng‐Yung W. (2003) An adaptive
[9] Paccot F., Andreff N. and Martinet P. (2009) A Review fuzzy sliding mode controller for robotic
on the Dynamic Control of Parallel Kinematic manipulators. Systems, Man and Cybernetics, Part A:
Machines: Theory and Experiments. The International Systems and Humans, IEEE Transactions on, vol. 33,
Journal of Robotics Research, vol. 28, pp. 395‐416. pp. 149‐159.
[10] Müller A. and Hufnagel T. (2012) Model‐based [23] Byung Kook Y. and Woon Chul H. (2000) Adaptive
control of redundantly actuated parallel control of robot manipulator using fuzzy
manipulators in redundant coordinates. Robotics and compensator. Fuzzy Systems, IEEE Transactions on,
Autonomous Systems, vol. 60, pp. 563‐571. vol. 8, pp. 186‐199.
[11] Le T. D., Kang H.‐J., Suh Y.‐S. and Ro Y.‐S (2012) An [24] Chih‐Lyang H. and Chia‐Ying K. (2001) A stable
Online Self Gain Tuning Method Using Neural adaptive fuzzy sliding‐mode control for affine
Networks for Nonlinear PD Computed Torque nonlinear systems with application to four‐bar
Controller of a 2‐dof Parallel Manipulator, linkage systems. Fuzzy Systems, IEEE Transactions
Neurocomputing, on, vol. 9, pp. 238‐252.
http://dx.doi.org/10.1016/j.neucom.2012.01.047.

14 Int J Adv Robotic Sy, 2013, Vol. 10, 22:2013 www.intechopen.com


[25] Ciliz M. K. (2005) Adaptive control of robot [28] Nakamura Y. and Yamane K. (2000) Dynamics
manipulators with neural network based computation of structure‐varying kinematic chains
compensation of frictional uncertainties. Robotica, and its application to human figures. Robotics and
vol. 23, pp. 159‐167. Automation, IEEE Transactions on, vol. 16, pp. 124‐
[26] Ertugrul M. and Kaynak O. (2000) Neuro sliding 134.
mode control of robotic manipulators. Mechatronics, [29] Hui C., Yiu‐Kuen Y. and Zexiang L. (2003) Dynamics
vol. 10, pp. 239‐263. and control of redundantly actuated parallel
[27] Murray R. M., Zexiang L. and Sastry S. S. (1994) A manipulators. Mechatronics, IEEE/ASME
Mathematical Introduction to Robotic Manipulation. Transactions on, vol. 8, pp. 483‐491.
CRC Press, Boca Raton.

www.intechopen.com Tien Dung Le, Hee-Jun Kang and Young-Soo Suh: 15


Chattering-Free Neuro-Sliding Mode Control of 2-DOF Planar wParallel Manipulators

You might also like