Journal of Science & Technology 127 (2018) 022-028

Dynamic Model of Flexible Link Manipulators with Translational and

Rotational Joints
Bien Xuan Duong1,*, My Anh Chu1, Khoi Bui Phan2
1
Military Technical Academy, No. 236, Hoang Quoc Viet, Cau Giay, Hanoi, Viet Nam
2
Hanoi University of Science and Technology, No. 1, Dai Co Viet, Hai Ba Trung, Hanoi, Viet Nam
Received: August 10, 2017; Accepted: June 25, 2018

Abstract
In this paper, two nonlinear dynamic models of flexible manipulator with translational and rotational joints are
formulated based on Finite Element Method (FEM). Dynamic equations of manipulators are derived by using
Lagrange equations. The new model TR (Translational-Rotational) is developed based on single flexible link
manipulator with only rotational joint. The difference between flexible manipulators which have only
rotational joints and others with translational joint is presented in RT model. Dynamic behaviors of flexible
manipulators are analyzed through values of joints variable and elastic displacements at the end-effector
point. The numerical simulation results are calculated by using MATLAB/SIMULINK toolboxes. The results of
this study play an important role in modeling generalized planar flexible two-link robot, in designing the
control system and basing on select the suitable structure robot with the same request.
Keywords: Dynamic model, flexible link, translational joint, elastic displacements

1. Introduction1 using assumed modes method (AMM). Other authors

have focused on the flexible manipulator with a link
Years ago, a number of researches focused on
slides through a translational joint with a
the flexible manipulator. Most of the investigations
simultaneous rotational motion (R-T robot). Pan et al
on the dynamic modeling of robot manipulators with
[4] presented a model R-T with FEM method. The
elastic arms have been confined to manipulators with
result is differential algebraic equations which are
only revolute joints. Combining such systems with
solved by using Newmark method. Yuh and Young
translational joints enables these robots to perform
[5] proposed the partial differential equations with R-
manipulation tasks in a much larger workspace, more
T system by using AMM. Al-Bedoor and Khulief [6]
flexibility and more applications. The flexible robot
presented a general dynamic model for R-T robot
arm constructed with couples of rotational–
based on FEM and Lagrange approach. They defined
translational joints (R-T/T-R joints) challenge the
a concept which is translational element. The stiffness
modeling and analysis for the robot. The two main
of translational element is changed. The translational
methods for dynamic modeling of flexible link
joint variable is distance from origin coordinate
manipulators are the FEM and assumed mode method
system to translational element. The number of
(AMM). Although the AMM has been used widely to
element is small because it is hard challenge to build
model and analyze flexible links manipulators but
and solve differential equations. Khadem [7] studied
most of studied have only considered the rotational
a three-dimensional flexible n-degree of freedom
joints. The FEM is a general method and it can be
manipulator having both revolute and translational
applied to the manipulators with complexly shaped
joint. A novel approach is presented using the
links. This technique is used of this paper.
perturbation method. The dynamic equations are
Few authors have studied the manipulator with derived using the Jourdain’s principle and the Gibbs-
only translational joint. Wang and Duo Wei [1] Appell notation. Korayem [8] also presented a
presented a single flexible robot arm with systematic algorithm capable of deriving equations of
translational joint. Dynamic model analysis is based motion of N-flexible link manipulators with revolute-
on a Galerkin approximation with time dependent translational joints by using recursive Gibbs-Appell
basis functions. They also proposed a feedback formulation and AMM.
control law in [2]. Kwon and Book [3] present a
In this work, two nonlinear dynamic models of
single link robot which is described and modeled by
flexible manipulator with translational and rotational
joints are presented based on using FEM and
*
Lagrange equations. The flexible link is assumed
Corresponding author: Tel.: (+84) 166.7193.567 Euler-Bernoulli beam. The first model is shown in
Email: xuanbien82@yahoo.com
22
 Journal of Science & Technology 127 (2018) 022-028

fig. 1 (called T-R model). The second model is 1  

presented in fig. 2 (called R-T model). Flexible link ( )
C Q, Q .Q = M ( Q ) Q − 
2  Q
QT .M.Q 

(6) ( )
slides through the base of translational joint which
rotates at the fixed point. Both motion on horizontal Structural damping is ignored in this paper.
plane and ignore the effect of gravity and structure Mass matrix M and stiffness matrix K are
damping. The boundary conditions of both are calculated by using FEM theory in each detail case,
differently. The differences between flexible respectively. Size of matrices M, K and C is
manipulators which have only rotational joints and (2n + 4)  (2n + 4) and size of F ( t ) and Q(t ) is
others with translational joint are mentioned. (2n + 4) 1 .
Furthermore, T-R model has not been mentioned yet.
2. Dynamic modeling 2.2. The T-R configurations

2.1. General equations of motion The T-R model is shown as fig. 1. The
translational joint moves along its axis. The
In both model, the coordinate system XOY is translational joint variable d ( t ) is distance from
the fixed frame. The translational joint variable d ( t ) origin point of fixed coordinate to position of
is driven by force FT ( t ) . The rotational joint variable translational joint. Coordinate system X1O1Y1 is
q ( t ) is driven by torque  ( t ) . Both joints are attached to end point of link 1. Coordinate system
X 2O2Y2 is attached to first point of link 2. Link 1
assumed rigid. Flexible link is divided n elements.
Elements are assumed interconnected at certain points with length L1 is assumed rigid and link 2 with length
which are known as nodes. Each element has two L2 is assumed flexibility. Symbol mt is mass of
nodes. Each node of element j has 2 elastic payload on the end-effector point. Coordinate r01 of
displacement variables which are flexural ( u2 j −1 ) and end point of link 1 on XOY is computed as
d ( t ) . Coordinate r2 j of element j on
T
slope displacements ( u 2 j ). Fundamentally, the r01 =  L1
dynamic equations motion relies on the Lagrange
X 2O2Y2 can be found as
equations with Lagrange function L = T − P are
r2 j = ( j − 1) le + x j wj ( x j , t ) ;0  x j  le . Where,
T
given by [9]
T T
d  L   L  le =
L2
 = F (t )
length of each element is and
  − (1) n
dt  Q ( t )   Q ( t ) 
wj ( x j , t ) = N j ( x j ) Q j
(t ) is total elastic
Where, T and P are kinetic and potential energy of
system. External generalized forces vector is defined displacement of element. The vector of shape
as function N j ( x j ) is presented in [9]. Coordinate of
F ( t ) =  F1 ( t ) F2 ( t ) 0 . . 0 0 
T
(2) element j on X1O1Y1 is r21 j = T21.r2 j . Where,
Specific generalized coordinate vector is given by cos q ( t ) − sin q ( t ) 
T21 =   is transformation matrix
Q ( t ) =  q1 ( t ) q2 ( t ) u1 . . u2 n +1 u2 n + 2   sin q ( t ) cos q ( t ) 
T
(3)
Elastic displacements vector of element j is from X 2O2Y2 to X1O1Y1 .

( t ) = u2 j −1
T
Q j
u2 j u2 j +1 u2 j + 2  (4)

In T-R case, driving forces and joints variables are

F1 ( t ) = FT (t ) ; F2 (t ) =  (t ) ; q1 (t ) = d (t ) ; q2 (t ) = q (t )
Kinetic and potential are T = T T − R and P = PT − R .
In R-T case, respectively, F1 (t ) =  (t ); F2 (t ) = FT (t ) ,
q1 (t ) = q(t ); q2 (t ) = d (t ) and T = T R −T , P = P R −T .
When kinetic and potential energy are known, it is
possible to express Lagrange equations as shown

( )
M ( Q ) Q + C Q,Q Q + KQ = F ( t ) (5)
Coriolis force are calculated as Fig. 1. T-R model

23
 Journal of Science & Technology 127 (2018) 022-028

Coordinate r02 j of element j on XOY is  m11 m12 

Matrix M rr = 
m22 
has elements which are
r02 j = r01 + r21 j . Coordinate r0E of end point of  m21
flexible link 2 on XOY can be computed as computed as

 L1 + L2 cos q ( t ) − u2 n +1 sin q ( t )  m11 = m2 .le ; m12 = m21 ;

r0TE− R =  
 d ( t ) + L2 sin q ( t ) + u2 n +1 cos q ( t )
(7)  sin ( q ) le u2 j − sin ( q ) le u2 j + 2 
1  
If assumed that robot with all of links are rigid, m12 = − m2 le  −12 cos ( q ) jle + 6sin ( q ) u2 j −1 
coordinate r0TE−_Rrigid on XOY is given as
12  
 +6sin ( q ) u2 j +1 + 6 cos ( q ) le 
r0TE−_ Rrigid =  L1 + L2 cos q ( t ) d ( t ) + L2 sin q ( t )
T
(8)  210le2 j ( j − 1) + le2 (2u22 j − 3u2 j u2 j + 2 (14) 
 
1  +2u22 j + 2 ) + 22le (u2 j −1u2 j − u2 j +1u2 j + 2 ) 
The kinematic energy of rigid link 1 can be computed m22 = m2 le  
 +13le (u2 j u2 j +1 − u2 j −1u2 j + 2 ) + 70le 
2
1 210
as T1T − R = m1 .r01T .r01 . Where, symbol m1 is the  
 +78(u2 j −1 + u2 j +1 ) + 54u2 j −1u2 j +1
2 2
2 
mass of first link. Kinetic energy of element j is
determined as
m m14 m15 m16 
1 le  r02 j   r02 j 
T
M rf = M Tfr =  13
m26 
Matrix has
T2 j =  m2 
T −R
   dx j  m23 m24 m25
2 0  t   t  (9) elements which are computed as
T −R
T2 j = Q jg
2
(
1 T −R T
)
( t ) M j Q jg ( t )
T −R T −R
1 1
m13 = m15 =
m2 le cos q; m14 = m2le2 cos q;
Where, m2 is mass per meter of link 2. Generalized 2 12
elastic displacement vector of element j is given as 1
m16 = −m14 ; m23 = m2le2 (10 j − 7);
20
Q jg ( t ) =  d ( t ) q ( t ) u2 j −1 u2 j u2 j +1 u2 j + 2 
T −R T
(15)
(10) 1 1
m24 = m2le (5 j − 3); m25 =
3
m2le (10 j − 3);
2

60 20
T −R
Each element of inertial mass matrix M j can be 1
m26 = − m2 le3 (5 j − 2);
computed as 60
T
 r02 j   r02 j 
M j ( s, e ) =  m2 
le
T −R Total elastic kinetic energy of link 2 is yielded as
   dx j ;
0
 Q js   Q je  (11)
( )
n
1
TeT − R =  T2Tj− R = QT − R ( t ) MTe − R QT − R ( t )
T
s, e = 1, 2,.., 6 2
(16)
j =1
th th
Where, Q js and Q je are the s , e element of
The mass matrix MTe − R is constituted from matrices
QTjg− R (t ) vector. It can be shown that MTj − R is of the
of elements follow FEM theory, respectively. The
form
generalized coordinate vector of system is defined as
T −R  M rr M rf 
Mj =   (12) QT − R ( t ) =  d ( t ) q ( t ) u1 . . u2 n +1 u2 n + 2 
T
M fr M base  (17)
Where, matrix Mbase is the base mass matrix which is Kinetic energy of payload is
given as
T T − R = mt . ( r0TE− R ) .r0TE− R and kinetic energy of
1 T

 13 11 9 13 2 P
 35 m2 le m2 le2
m2 le − m2le  2
210 70 420 system is determined as
 
 11 m l 2 1 13
−
1 3
T T − R = T1T − R + TeT − R + TPT − R
3 2
m2le m2le m2le
 210 2 e 105 420 140 
M base =  
( )
1 (18)
T T − R = QT − R ( t ) M T − R QT − R ( t )
T
 9 ml 13 13 (13) 2 
11
m l 2
m l − m l
 70 2 e 420
2 e
35
2 e
210
2 e
 2
 13 1 11 1  T −R
− m2 le2 − m2 le3 − m2le2 m2le3 Matrix M is total mass matrix. The gravity effects
 420 140 210 105 can be ignored as the robot movement is confined to
the horizontal plane. Defining E and I are Young’s
modulus and area moment of inertia link 2. Elastic

24
 Journal of Science & Technology 127 (2018) 022-028

potential energy of element j with stiffness matrix

K base which is presented as [9] is given as

1 le   w j ( x j , t ) 
2
2

P T −R
=  EI   dx j
x 2j
j
2 0   (19)
1
PjT − R = QTj (t )K Tj − R Q j (t )
2
Stiffness matrix of element j is defined as

0 0 0 
K Tj − R = 0 0 0  (20)
0 0 K base 

Where, Fig. 2. R-T model

It is noteworthy to mention that value of k is the
 12 EI 6 EI 12 EI 6 EI 
 l3 −
le2 
number of position element k . So, it must be take raw
le2 le3
 e  value of k in computing process. Coordinate r j on
 6 EI 4 EI 6 EI 2 EI 
 2 −  X 2O2Y2 of element j is
le le le2 le 
=
rj = ( j − 1) le − L2 + d ( t ) + x j wj ( x j , t ) and on
K base (21) T
 12 EI 6 EI 12 EI 6 EI 
− 3 − − 2 
 le le2 le3 le  XOY is r0 j = r01 + T10 .T12 .rj . Transformation matrix
 6 EI 2 EI 6 EI 4 EI 
 2 −  T01 between XOY and X1O1Y1 is
 le le le2 le 
cos q ( t ) − sin q ( t ) 
T10 =   . Transformation matrix
Total elastic potential energy of system is yielded as  sin q ( t ) cos q ( t ) 
T12
( Q ) ( t ) KT − R QT − R (t )
n between
1 T −R T X1O1Y1 and X 2O2Y2 is
PT − R =  Pj = (22)
j =1 2 cos  0 − sin  0 
T12 = 
cos  0 
. Coordinate of end point of
Total stiffness matrix K T − R is constituted from  sin  0
is r01 =  L1 .cos q (t ) L1 .sin(t )  .
T
matrices of elements follow FEM theory similar link 1 on XOY
MT − R matrix, respectively. Coordinate of end-effector is given as
2.3. The R-T configurations   d ( t ) + L1  .cos q ( t ) − u2 n +1 .sin q ( t ) 
The R-T model is shown as fig. 2. Flexible link r0RE−T =   (23)
  d ( t ) + L1  .sin q ( t ) + u2 n +1 .cos q ( t ) 
2 is divided n elements. The position of elements k
continuously changes through translational joints.
If assumed that link 2 is rigid, coordinate r0RE−_Trigid on
The angle 0 is the angle between O1 X 1 and O2 X 2 .
XOY is given as
The translational joint variable d ( t ) is the distance
from element k at the origin coordinate system to the   d ( t ) + L1  .cos q ( t ) 
r0RE−_Trigid =   (24)
end point of link 2. The value of k depends on time.   d ( t ) + L1  .sin q ( t ) 
The position of other elements is changed, too. Total
length of flexible link 2 is L2 . Length of each element Kinetic and potential energy of R-T system are
L2 calculated by using the same method in T-R case
is le = . Element k is inside the hub of the
n
translational joint and we have contact T R −T = TeR −T + TPR −T
formulas k.le = L2 - d (t ) . Effect of the length of hub
(
1 R −T
) (t ) M Q R −T ( t ) + mt . ( r0RE−T ) .r(25)
1
T T
T R −T = Q R −T
e
R −T
0E
is ignored. 2 2

25
 Journal of Science & Technology 127 (2018) 022-028

generalized coordinate Q ( t ) is rebuilt after each

( Q ) ( t ) K R −T Q R −T ( t )
n
1 R −T T
P R −T =  PjR −T = (26)
j =1 2 loop too because of changing value k variable.

Where, TeR −T is the elastic kinetic and TPR −T is kinetic 4. Simulation results
R −T
energy of payload. Mass matrix M e and stiffness Simulation specifications of two flexible models
R −T
matrix K can be calculated by using FEM theory are given by Table. 1.
following mass and stiffness matrices of all elements. Table 1. Parameters of dynamic models
Mass matrix of element j with mass per meter m2 is
Property T-R model R-T model
given as MTj − R at recipe (12). It noted that Mbase is Length of link 1 (m) L1=0.1 L1=0.2
the same but others elements are difference. They are Mass of link 1 (kg) m1=1.4 m1=0.9
computed as Flexible link parameters
Length of flexible
m11 = f (m, le, j , L, d (t ), Q j (t )); L2=0.3 L2=0.8
link (m)
m2 le ( le u2 j − le u2 j + 2 + 6u2 j −1 + 6u2 j +1 )
1 Width (m) b=0.03 b=0.01
m12 = −
12 Thickness (m) h=0.003 h=0.002
Number of element n=5
m2 le (10 jle + 10d ( t ) + 10 L1 − 10 L2 − 7le )
1
m13 = Cross section area
20 A=9.10-5 A=2.10-5
(m2) (A=b.h)
m2 le2 ( 5 jle + 5d ( t ) + 5L1 − 5L2 − 3le )
1
m14 = (27) Mass density (kg/m3) =7850
60
Mass per meter
0.7 0.157
m2 le (10 jle + 10d ( t ) + 10 L1 − 10 L2 − 3le )
1
m15 = (kg/m) (m2=.A)
20 Young’s modulus
E=2.1010
m16 = − m2 le2 ( 5 jle + 5d ( t ) + 5L1 − 5L2 − 2le )
1 (N/m2)
60 Inertial moment of
I=b.h3/12=1.67.10-12
m22 = m.le ; m23 = m24 = m25 = m26 = 0; cross section (m4)
Mass of payload (g) mt=100 mt=20
R −T
Matrices M base and K Rj −T have format as Initial value at t=0
0 d0=0.4
T −R T −R (m)
M ,K .
base j β0 0
3. Boundary conditions Simulation time (s) T=10

In T-R case, rotational joint of link 2 is
constrained so that elastic displacements of first node
of element 1 on link 2 can be zero. Thus variables
u1 , u2 are zero. The rows and columns 3th, 4th of
matrices F ( t ) , Q ( t ) vectors
M, K , C and are
eliminated as presented in FEM theory.
In R-T case, the displacements of element k are
zero because assumed that the translational joint hub
is treated as rigid. The rows and columns
( 2k − 1) ; ( 2k ) of matrices M, K , C and F (t ) , Q (t )
th th
vectors are eliminated and values of these are
changed, too. It is noteworthy to mention that value
of k depends on time. This boundary condition is
clearly different point between flexible link
manipulator with only rotational joints and flexible
link manipulator with combine translational joint and
rotational joint. So now, size of matrices M , K,C is
( 2n + 2)  ( 2n + 2) and ( 2n + 2) 1 is size of F ( t )
and Q ( t ) vector. The k variable is continuously
updated for each time step in solving process. Vector
Dynamic equations (5) for both models are
solved by using MATLAB/SIMULINK toolboxes
with boundary conditions. The simulation results of
T-R model are shown in fig. 3, fig. 4 and fig. 5.
Driving force and torque of joints are built using
Bang-Bang rule which are shown fig. 3. Driving time
is 0.5(s). Dynamic behaviors of system are described
through values of joint variables and elastic
displacements in fig. 4 and fig. 5. Maximum values of
translational and rotational joint are 0.075(m) and
0.24(rad). These values are shown in fig. 4. Elastic
displacements at the end-effector point are presented
in fig. 5. Maximum values of these displacements are
0.14(m) and 0.25(rad) at 0.6(s). These displacements
are fast reduced after that with 0.02(m) and 0.035
(rad). The simulation results of R-T model are shown
from fig. 6 to fig. 8. The Bang-Bang rule in fig. 6 is
used for driving torque and force of joints. Fig. 7
shows values of joint variables. It note that value of
rotational joint continuously increases while
maximum value of translational joint increases from
initial value 0.4(m) to 0.51(m) at 0.5(s). Values
different of rotational joint between both models can

26
 Journal of Science & Technology 127 (2018) 022-028

be explained by effect of higher nonlinearity in R-T

model, difference of boundary condition and
direction of motion of flexible link when value of
translational joint increase over time. Elastic
displacements at the end-effector point are shown in
fig. 8. Values of these displacements are smaller than
T-R case.

Fig. 6. Driving torque and force at joints in R-T

model

Fig. 3. Driving force and torque at joints of T-R

model

Fig. 7. Values of joint variables in R-T model

Fig. 4. Values of joint variables in T-R model

Fig. 8. Value of elastic displacement at the end-

Fig. 5. Value of elastic displacements at the end-
effector point in T-R model
effector point in R-T model
5. Conclusion
Mathematical model of two flexible
manipulators with translational and rotational joints
has been presented based upon finite element method
and Lagrange equations. T-R model is the new model
which is developed from single flexible link
manipulator with only rotational joint. R-T model
shows difference points between translational joint
and rotational joint. These difference points are

27
 Journal of Science & Technology 127 (2018) 022-028
boundary conditions. The final models derived for the
flexible robot are nonlinear and complex. Simulation
results show that the response of flexible robots is
unstable especial without effect of structure damping
and needs to control. Moreover, the effects of
structure damping and varying payload on dynamic
characteristic of models have been studied and
discussed. The results in this paper are helpful and
important in development of modeling generalized
planar flexible two-link robot which combines
translational and rotational joints and designing
control system.
References
[1] P. K. C Wang and Jin Duo Wei, Vibration in a
moving flexible robot arm, Journal of Sound and
vibration. 116 (1987) 149-160.
[2] P. K. C Wang and Jin Duo Wei, Feedback control of
vibrations in a moving flexible robot arm with rotary
and prismatic joints, Proceedings of the IEEE
International Conference on Robotics and
Automation, Raleigh, North Carolina, March (1987)
1683-1689.
[3] D. S. Kwon and W. J. Book, A time-domain inverse
dynamic tracking control of a single link flexible
manipulator, Journal of Dynamic Systems,
Measurement and Control. 116 (1994) 193–200.
[4] Pan. Y. C, Scott and R. A. Ulsoy, Dynamic modeling
and simulation of flexible robots with translational
joints, J. Mech. Design. 112 (1990) 307–314.
28
[5] Yuh. J and Young. T, Dynamic modeling of an
axially moving beam in rotation: simulation and
experiment, Trans. ASME J. Dyn. Syst. Meas.
Control. 113 (1991) 34–40.
[6] Al-Bedoor. B. O and Khulief. Y. A, General planar
dynamics of a sliding flexible link, Sound and
Vibration. 206 (1997) 641–661.
[7] S. E. Khadem and A. A. Pirmohammadi, Analytical
development of dynamic equations of motion for a
three-dimensional flexible manipulator with revolute
and prismatic joints, IEEE Trans. Syst. Man Cybern.
B Cybern. 33 (2003) 237–249.
[8] M. H. Korayem, A. M. Shafei and S. F. Dehkordi,
Systematic modeling of a chain of N-flexible link
manipulators connected by revolute–prismatic joints
using recursive Gibbs-Appell formulation, Springer-
Verlag, Berlin Heidelberg, (2014).
[9] M. O. Tokhi and A. K. M. Azad. Flexible robot
manipulators-Modeling, simulation and control, The
Institution of Engineering and Technology. London,
UK (2008).