Download as pdf or txt
Download as pdf or txt
You are on page 1of 5

Proceedings of the 3rd

RSI International Conference on Robotics and Mechatronics


October 7-9, 2015, Tehran, Iran

Control of a Two-link (Rigid-Flexible) Manipulator

Mustafa Turki Hussein Mohammed Najeh Nemah


Department of Mechanical Engineering Al-Furat Al-Awsat Technical University
University of Babylon Engineering Technical College
Babylon, Iraq Najaf, Iraq
Email: eng.mustafa.turky@uobabylon.edu.iq Email: mohammed.najeh85@yahoo.com

Abstract—This paper investigates the motion control of two with the tip output which is introduced in [3]. In order to
link (rigid-flexible) manipulator through analysis, computer sim- deal with the exact tracking problem many researchers have
ulation. An explicit nonlinear dynamic model is developed using been proposed the non-causal controllers for the purpose of
assumed mode method. The balance of moments and shear forces trajectory tracking.
are represented by using the boundary conditions. The dynamic
model is linearized around the operating point of the system. The goal of this study is to present an approach to the
The linear feedback plus model based feed-forward control is control of two link rigid-flexible manipulators moving along
developed as well, based on three different measurements. The predefined paths, based on joints angles and tip deflection
joint angles and the tip deflection are used in this study. The feedbacks. Spectral approximation methods for modelling of
control inputs are generated using linear dynamic model. The rigid flexible manipulators are studied in [5]. The Lagrange
simulation results of proposed control are compared with those of
PD control, in order to show the efficiency of the developed model
equation is approximated then an experimental setup is done
and the proposed controller. As a future work; the proposed to validate the derived mathematical model. Various control
controller will be tested using two link (rigid-flexible) test-rig. strategies have been proposed for rigid-flexible manipulators
as reviewed in [6].The performance of joint proportional,
I. I NTRODUCTION derivative (PD) controller using of a non-model based is
demonstrated also in [6]. Trajectory tracking and various
The modeling and motion control of flexible link manipu- implementation issues are also discussed in the research. A
lators have been studied interestingly by many researchers in controller design used a quantitative feedback theory for two
the last two decades. Several reasons made the field attractive link rigid-flexible manipulators is presented in [7]. The end
for the researchers such as: low weight to payload ratio, point deflection is compensated using piezoelectric transducer.
power consumption, high-speed manipulation and increased An adaptive control based on motors and transducer actuators
productivity [1] . On the other hand, the main problem related using a nonlinear dynamic model is presented in [8]. The
to flexible manipulator in operation is mainly due to the stability of the proposed controller is studied using lyponov
requirement of having a stable closed-loop system [4]. In order theory.
to have less control problems the designers of conventional
robots have designed the manipulators more rigid using bulky In this paper the nonlinear mathematical model is derived
structures. The desirable objectives to be achieved for flexible using Lagrange method and assumed mode method to express
manipulators are exact, high-speed manipulation. This means the deflection in the second link. The dynamic model is
precise and stable control of flexible manipulator end effector; linearized around the operating point of the system for the
which requires the elastic deflection effects due to the flexi- sake of controller design. The state space model form is
bility of the arms to be included in the dynamic equations of used in this work for simulation. The controller proposed
motion. In general, the analysis and design of flexible robot in this work consists of an output feedback controller plus
control is more complicated in comparison to rigid robots. model based feed-forward controller. The desired trajectory is
In flexible link manipulators, more difficulty will arise when designed to test the efficiency proposed controller. The paper is
trying to follow a specific tip position trajectory by applying organized as follows: the mathematical model derivation will
the torque to the joints actuators. be presented, in the next section (II) . The proposed controller
model will be detailed later in section (III). Then the case
Several control strategies have been presented by many study parameters and simulations will be discussed in details
researchers in order to solve the control problem of flexible (IV). Finally, some conclusion regarding this work addressed
link robot. The end-point vibration compensation problem in the last section.
through the output redefinition to reduce tracking errors is
addressed in [2]. The authors in [2] have been shown that II. M ATHEMATICAL M ODEL
the nonlinear flexible-link system is not globally linearizable,
however the system is locally input-state feedback linearizable. The dynamic model of two link (rigid-flexible) robot shown
The control objective is not just to drive the manipulator to in Figure (1) is outlined in this section. The motion of the robot
a specific point and stabilize the vibrations. The controller is assumed as a rotation on a horizontal plane. The first link
should concern about the specified path to follow. For a single is rigid and the joint variable is 𝜃1 (𝑡). The second link of the
link flexible arm, an optimal control approach is candidate robot is flexible. The variables related to the flexible link are
to succeed [3]. The non-minimum phase property of flexible 𝑤(𝑥, 𝑡) and 𝜃2 (𝑡), which are the deformation of the flexible
robot has not addressed in pseudo link concept associated link and the joint’s angle respectively. The axis 𝑂 − 𝑥, of the

978-1-4673-7234-3/15/$31.00 ©2015 IEEE 720

Authorized licensed use limited to: Ritsumeikan University. Downloaded on April 20,2023 at 10:08:34 UTC from IEEE Xplore. Restrictions apply.
𝑥2 The dynamic model is derived using Lagrangian technique
in conjunction with assumed mode method [12]. The model
𝑦2 𝑥 , 𝑡) consists of nonlinear partial differential equations, which can
𝑤( be obtained using some approximation. The equations of mo-
tion is derived using the Lagrangian technique, the constrained
motion of the rigid-flexible can be described as:
𝑀 (𝑞)¨
𝑞 + 𝐻(𝑞, 𝑞)
˙ 𝑞˙ + 𝐾𝑞 + 𝐷𝑞˙ = 𝑢.

𝑙2
(5)

𝑦1 𝜃2 Where 𝑞 = [𝜃1 𝜃2 𝛿1 𝛿2 ]T is the generalized displacement


𝑌 𝑥1 vector. The matrices of the
[ ] dynamic model can be partitioned
𝑙1 𝑀𝜃𝜃 𝑀𝜃𝛿
as: 𝑀 (𝑞) = ,
𝑀𝛿𝜃 𝑀𝛿𝛿
𝜃1 [ ] [ ]
𝑂 𝐻𝜃𝜃 𝐻𝜃𝛿 0 0
𝑋
𝐻(𝑞, 𝑞)
˙ = , 𝐷= ,
𝐻𝛿𝜃 𝐻𝛿𝛿 0 𝐷𝛿𝛿
[ ] [ ]
Fig. 1. Robot with two rigid-flexible links. 0 0 𝜏
𝐾= , and 𝑢 = .
0 𝐾𝛿𝛿 0
rotating frame 𝑂 − 𝑥𝑦 passes through the end point of the link Here the indexes 𝜃𝜃 , 𝜃𝛿 and 𝛿𝛿 are the terms from
whose transverse deflection 𝑤(𝑥, 𝑡) is expressed with respect the matrices corresponding to rigid body, rigid coupling with
to this rotating frame, hence the position for a point along the flexible body and flexible body, respectively. 𝑀, 𝐻 and 𝐾 are
second link is: the inertia, coriolis and stiffness 4 × 4 matrices respectively.
[ ] [ ] [ ][ ]
𝑋 𝑐1 𝑐12 −𝑠12 𝑙2 The control input represents joints torques of the robots
= 𝑙1 + . (1) 𝜏 = [𝜏1 𝜏2 ]T ; in which 𝜏1 and 𝜏2 are the rigid and flexible
𝑌 𝑠1 𝑠12 𝑐12 𝑤(𝑥, 𝑡)
joint torques respectively.
In (1) 𝑠 = 𝑠𝑖𝑛 , 𝑐 = 𝑐𝑜𝑠 , while subscripts 1 and 12 are
denote 𝜃1 and 𝜃1 + 𝜃2 respectively. In order to include the low A. Dynamic Model Linearization
frequency modes, the flexible link is considered as a beam of
To use the obtained nonlinear model, Taylor series is uti-
length 𝑙2 , uniform density 𝜌, and constant elastic properties
lized to linearize the nonlinear terms about the operating point
𝐸𝐼. A common model of flexible-link robots which is based
of the robot system. The operating point is chosen when the
on the Euler-Bernoulli beam theory [9] is used. The flexible
inputs of the system are zero: 𝜏 = [0 0]T =⇒ 𝑞 = [0 0 0 0]T .It
link is assumed to be a thin beam, pinned at the end of the
has been noticed from the simulation for the linearized model
rigid link with a rigid hub, and only the lateral vibration of
that the coriolis terms 𝐻 in (3) can be neglected.Then, by
the beam is considered. For the flexible link, the equation due
keeping only the linear terms, the equations of motion can be
to the flexibility effect of order 𝑛 [10]is given by:
written as:
𝑛
∑ 𝑀 𝑞¨ + 𝐾𝑞 + 𝐷𝑞˙ = 𝑢. (6)
𝑤(𝑥, 𝑡) = 𝜙𝑖 (𝑥) 𝛿𝑖 (𝑡). (2)
𝑖=1 The displacement vector and the input vector are changed
to 𝑞 = [Δ𝜃1 Δ𝜃2 Δ𝛿1 Δ𝛿2 ]T , 𝑢 = [Δ𝜏1 Δ𝜏2 0 0]T . The
The time part separation of the the solution 𝛿(𝑡) is included measurement vector 𝑦𝑚 is specified as:
in the dynamic equation of motion. The assumed mode method ⎡ ⎤
[9] is used to describe the mode shapes function 𝜙𝑖 (𝑡) of the 1 0 0 0
⎢ 0 ⎥
flexible link. In order to simplify the solution problem; the 𝑦𝑚 = 𝐶1 𝑞 = ⎣0 1 0 ⎦ 𝑞. (7)
first two modes (𝑛 = 2) of vibration are considered for the 0 0 𝜙1 (𝑙2 ) 𝜙2 (𝑙2 )
modeling of flexible link [11]. The mode shape equation is:
𝑥 𝑥 B. State Space Equations
𝜙𝑖 (𝑥) = 𝐶𝑖 (𝑐𝑜𝑠ℎ(𝜆𝑖 ) − 𝑐𝑜𝑠(𝜆𝑖 )
𝑙2 𝑙2
𝑥 𝑥 (3) The matrix differential equation (6) is represented in a
−𝛾𝑖 (𝑠𝑖𝑛ℎ(𝜆𝑖 ) − 𝑠𝑖𝑛(𝜆𝑖 ))), state-space form as:
𝑙2 𝑙2
and 𝛾𝑖 = 𝑠𝑖𝑛(𝜆𝑐𝑜𝑠(𝜆𝑖 )+𝑐𝑜𝑠ℎ(𝜆𝑖 )
. Here 𝜆𝑖 are calculated accord- 𝑧˙ = 𝐴𝑧 + 𝐵𝑢 (8)
𝑖 )+𝑠𝑖𝑛ℎ(𝜆𝑖 )
ing to the boundary conditions (pinned-payload) beam, 𝐶𝑖 𝑦 = 𝐶𝑧.
is
∫ 1 the normalized coefficient which calculated according to [ ]
𝜙 2
(𝑥)𝑑𝑥 = 1. The fundamental frequencies of the link are 04×4 𝐼4×4
0 𝑖 With 𝐴 = ,
calculated according to: −𝑀 −1 × 𝐾 −𝑀 −1 × 𝐷
√ [ ]
𝜆𝑖 𝐸𝐼2 04×2
𝑤𝑖 = ( 2 ) × . 𝐵= , and
𝑙2 𝑚2
(4) 𝑀 −1 × 𝐺

721

Authorized licensed use limited to: Ritsumeikan University. Downloaded on April 20,2023 at 10:08:34 UTC from IEEE Xplore. Restrictions apply.
[ ]
𝐶 = 𝐶1 03×4 . The control law of equation (10) will lead to a full
compensation of the rigid body interactions, the desired joint
In (8) the state vector is written as: 𝑧 = acceleration calculation will be explained later. The control
[Δ𝜃1 Δ𝜃2 Δ𝛿1 Δ𝛿2 Δ𝜃˙1 Δ𝜃˙2 Δ𝛿˙1 Δ𝛿˙2 ]T . Matrix 𝐺 is law schematic diagram is shown in figure (3).
a constant matrix related to the input signals which is chosen
in this work as: In order to obtain an improved joint error; a full dynamic
[ ] has been taken into account (𝐶𝑇 𝐹 ). By inverting the state
1 0 0 𝜙´1 (0) space dynamic model (8) at the joint level a new control
𝐺= .
0 1 0 𝜙´2 (0) law can be computed. Substituting the 𝜃 = 𝜃𝑑 in (8), the
state vector can be integrated using zero initial conditions to
The state space form is used later for the simulation of estimate 𝛿𝑑 and 𝛿˙𝑑 off-line. Then the control law in this case
the robot system. Then the feedback can be changed to be can be rewritten as:
appropriate with the control type.
𝑢𝑓 (𝑡) = 𝑀𝜃𝜃 𝜃¨𝑑 + 𝑀𝜃𝛿 𝛿¨𝑑 + 𝑢𝑃 𝐷 (𝑡). (11)
III. C ONTROLLER D ESIGN
Since the zero dynamics of the system exponentially stable
The computed torque method based on PD feedback con- [4]; a bounded solution for the deformation variables is pro-
trol plus model based feed-forward signals is used in this work duced. The schematic diagram for the control law shown in
to control the motion of the robot system. With the addition of figure (4).
a feed-forward term tracking accuracy can be improved and the
compensation of dynamic non-linearity and interactions will be 𝜃¨𝑑
𝑀𝜃𝜃
relaxed [13]. Different alternatives are possible, based either on
a part or on the whole available model. Two computed torque
𝜃𝑑 ,𝜃˙𝑑 State Space 𝛿¨𝑑
methods are presented in this section, the compensation of 𝑀𝜃𝛿
the rigid body interactions is included in the first one. In the Model
second the full dynamic model is used in order to improve the + +
Desired + PD + Robot
joint error. Controller 𝑢𝑃 𝐷 𝑢𝑓 Plant 𝑦𝑚
Value
-
The proportional-derivative (PD) control law for tracking
𝜃 , 𝜃˙
joint trajectory error [2] is:
𝑢𝑃 𝐷 (𝑡) = 𝐾𝑃 (𝜃𝑑 (𝑡) − 𝜃(𝑡)) + 𝐾𝐷 (𝜃˙𝑑 (𝑡) − 𝜃(𝑡)).
˙ (9) Fig. 4. PD plus feed-forward controller includes full dynamic model.

Positive definite diagonal 𝐾𝑃 and 𝐾𝐷 matrices is required to


make the system exponentially stable. The description of the IV. S IMULATION R ESULTS
control law is shown in figure (2). The parameters used in this work for the system illustrated
in section II are:
Desired + PD Robot
Controller 𝑢𝑃 𝐷 Plant 𝑦𝑚 𝑙1 = 0.3𝑚 𝑙2 = 0.7𝑚
-
Value
𝐴1 = 0.0039𝑚2 𝐴2 = 0.00005𝑚2
𝜃 , 𝜃˙
𝑚1 = 0.3𝑘𝑔 𝑚2 = 0.3𝑘𝑔
Fig. 2. PD controller block diagram. 𝜌1 = 1780𝑘𝑔/𝑚3 𝜌2 = 7700𝑘𝑔/𝑚3
𝑚ℎ𝑢𝑏1 = 0.5𝑘𝑔 𝑚ℎ𝑢𝑏2 = 0.5𝑘𝑔
The performance of the system can be improved by using 𝐸𝐼1 = 94.3𝑁.𝑚2 𝐸𝐼2 = 54.6𝑁.𝑚2
feed-forward terms to the control law presented in (9). Based
on the rigid terms of the dynamic model the control law (𝐶𝑇 𝑅) Incorporating the parameters given above using equation
can be modified as: (3) the mode shape parameters can be calculated as:

𝑢𝑟 (𝑡) = 𝑀𝜃𝜃 ∣𝛿=0 𝜃¨𝑑 + 𝐻𝜃𝜃 ∣𝛿=0 + 𝑢𝑃 𝐷 (𝑡). (10)


𝜙1 (𝑙2 ) = −1.446 𝜙2 (𝑙2 ) = 1.369
𝜙´1 (0) = 5.74 𝜙´2 (0) = 11.64
𝑀𝜃𝜃 𝑤1 = 4.716 𝑟𝑎𝑑.𝑠𝑒𝑐 𝑤2 = 14.395 𝑟𝑎𝑑.𝑠𝑒𝑐,

here 𝑤1 and 𝑤2 are the first and the second fundamental


𝐻𝜃𝜃 frequency of the flexible link (4). The linearized dynamic
+ + equations matrices obtained using the parameters above are
Desired + PD + Robot as follow:
⎡ ⎤
Value
-
Controller 𝑢𝑃 𝐷 𝑢𝑟 Plant 𝑦𝑚 1.14 0.498 0.336 0.195
𝜃 , 𝜃˙
⎢0.498 0.336 0 0 ⎥
𝑀 =⎢ ⎣0.336
⎥,
0 1 0 ⎦
Fig. 3. PD plus feed-forward controller block diagram. 0.195 0 0 1

722

Authorized licensed use limited to: Ritsumeikan University. Downloaded on April 20,2023 at 10:08:34 UTC from IEEE Xplore. Restrictions apply.
[ ]
878.03 0 the system. The initial motion of the joint angles are due to
𝐾𝛿𝛿 = , and 𝐷𝛿𝛿 =
[ ] 0 8180.6 initial deflection of the second flexible link; still the controller
0.07𝑤1 0 compensates the initial deflection of the link well very fast.
.
0 0.03𝑤2
(a)

(a) 100
100

𝜃1 [deg]
𝜃𝑑 [deg]

50
50

0
0 0 2 4 6 8 10
0 1 2 3 4 5

(b) (b)
80
1
𝜃˙𝑑 [rad/sec]

60

𝜃2 [deg]
0.5
40
0
20
−0.5
0 1 2 3 4 5
0
0 2 4 6 8 10
(c)
2
(c)
𝜃¨𝑑 [rad/sec2 ]

1
Tip-deflection [m] 0.1
0
0
−1
−0.1
−2
0 1 2 3 4 5
−0.2
Time[sec]
0 2 4 6 8 10
Time[sec]
Fig. 5. The desired joint angles and derivatives, joint1, joint2.

The desired output trajectory used in this work is a higher


order polynomials are some times used for path segments, this Fig. 6. Controlled behavior of the robot system, 𝑃 𝐷 Controller,
polynomial is quintic polynomial: 𝑃 𝐷+rigid terms feed-forward controller, 𝑃 𝐷+full dynamic
model feed-forward controller.
𝜃(𝑡) = 𝑎0 + 𝑎1 𝑡 + 𝑎2 𝑡2 + 𝑎3 𝑡3 + 𝑎4 𝑡4 + 𝑎5 𝑡5 . (12)
The joints error obtained in the simulation are reported in
The constraints specify a linear set of six equations with six figure(7); for 𝑃 𝐷, 𝐶𝑇 𝑅 and 𝐶𝑇 𝐹 in (a, b and c) respectively.
unknowns were solved by using MatLab software the solution It can be concluded from these figures that the controlled
is provided from robotics toolbox [14]. The initial and final output follow the reference signal very quickly; and the
conditions for the desired joints trajectories are controller compensates the vibration of the flexible link ef-
fectively without any abnormal behavior. The maximum value
𝜃1𝑑 (0) = 0 𝜃2𝑑 (0) = 0 of the error was approximately 15∘ with 𝑃 𝐷 controller; with
computed torque controller 𝑢𝑓 (equation 11) the suppression
𝜃1𝑑 (𝑇 ) = 𝜋2 𝜃2𝑑 (𝑇 ) = 𝜋3
of the vibration is improved and control signal damp out the
vibration faster than 𝑃 𝐷 controller and the controller 𝑢𝑟 in
and the total time 𝑇 is 3 seconds. The desired joint angles equation (10).
and their derivatives are shown in figure 5 (a, b, c). Those
desired variables have been used in the reference side of the The torque inputs for the three controllers used in the
control laws. The initial condition of the deflection are set to simulations are shown in figure (8). The control input has
𝑧 = [0 0 0.05 0.01 0 0 0 0]. The diagonal PD control gains the same order in all cases, feed-forward terms relaxed the
used here are: 𝐾𝑃 =[11.5 6] and 𝐾𝐷 =[2 0.8]. torques in correspondence to the acceleration discontinuities
in equation (11).
The output joint variables and second link tip deflection of
the controlled motion for robot system are shown in figure (6).
V. C ONCLUSIONS
The joint angles for the first and second links are presented in
figure 6 (a,b) receptively. The Tip deflection of the link for the A simulation of two-link planar rigid flexible robot has
different types of control laws is shown in figure 6 (c). It can be been presented. The nonlinear dynamic model is derived; a
noticed from these sets of figures that the best behavior of the linearized version of the dynamic model is also given. The
system was under the computed torque controller (described problem of trajectory tracking control is studied in the work.
by equation (11)) which includes the full dynamic model of For the sake of comparison three different control algorithms

723

Authorized licensed use limited to: Ritsumeikan University. Downloaded on April 20,2023 at 10:08:34 UTC from IEEE Xplore. Restrictions apply.
(a) (a)
5
10

𝑃 𝐷 [N.m]
𝑃 𝐷 [deg]

0
0
−10

−20
−5
0 2 4 6 8 10 0 2 4 6 8 10
(b) (b)
5

𝐶𝑇 𝑅 [N.m]
0
𝐶𝑇 𝑅 [deg]

0
−10

−20 −5
0 2 4 6 8 10 0 2 4 6 8 10
(c) (c)
5

𝐶𝑇 𝐹 [N.m]
0
𝐶𝑇 𝐹 [deg]

0
−10

−20 −5
0 2 4 6 8 10 0 2 4 6 8 10
Time[sec] Time[sec]

Fig. 7. Joints error 𝜃𝑑 − 𝜃𝑖 for the controlled behavior of robot system, Fig. 8. Controlled joint torques, joint1, joint2.
joint1, joint2.

[6] A. S. Yigit, Modelling and control of rigid-flexible manipulators, Flexible


are used in order to test the behavior of the system under the Robot Manipulators: Modelling, simulation and control, Institution of
derivation conditions. Based on the simulation for the three Technology and Engineering, chapter 17, 2008.
controllers; the end effector behavior is changed. The presented [7] M. Rafeeyan and M. Mirshamsi, QFT Control of a Two-Link Rigid-
Flexible Manipulator 4th IFAC Symposium on System, Structure and
results in this contribution can be used as a basis for experi- Control, pp 274-281, 2010.
mental study and comparison with advanced control strategy. [8] V. Bottega, R. Pergher and J. Fonseca, Simultaneous Control and
Further developments is required for the feedback laws that use Piezoelectric Insert Optimization for Manipulators with Flexible Link,
measures of the arm deflection, and of nonlinear controllers Journal of the Brazilian Society of Mechanical Sciences and Engineering,
designed directly for tracking end-effector trajectories. vol.31 no.2, pp. 105-116, 2009.
[9] F. Y. Wang and Y. Gao, Advanced Studies of Flexible Robotic Manipu-
lators, World Scientific, 2003.
R EFERENCES [10] J. Knani, Dynamic modelling of flexible robotic mechanisms and adap-
tive robust control of trajectory computer simulation Part I, Applied
[1] G. G. Rigatos, Control of Robotic Systems with Flexible Components Mathematical Modelling, Vol. 26, pp. 1113-1124, 2002.
Using Hermite Polynomial-Based Neural Networks, Robot Manipulators
[11] C. H. Chang, Mechanics of Elastic Structures with Inclined Members,
New Achievements , InTech, ISBN: 978-953-307-090-2, 2010. Available
Lecture Notes in Applied and Computational Mechanics, Springer: Berlin
from: http://www.intechopen.com/books/robotmanipulators- new-
Heidelberg, 2005.
achievements/control-of-robotic-systems-with-flexible-components-
using-hermitepolynomial- based-neural-networks [12] A. De Luca, Flexible Robots, Encyclopedia of Systems and Control,
Springer-Verlag, London, 2014.
[2] M. O. Tokhi, and A. K. M. Azad, Flexible Robot Manipulators: Mod-
elling, Simulation and Control, Institution of Technology and Engineer- [13] S. Moberg, J. Ohr and S. Gunnarsson, A Benchmark Problem for
ing, 2005. Robust Feedback Control of a Flexible Manipulator, IEEE Transactions
on Control Systems Technology, vol.17, no.6, pp.1398-1405, 2009.
[3] D. Rush, D. Robinett Clark, E. Richard, F. John, G. Gordon, G. David,
and S. Dennis, Flexible Robot Dynamics and Control, IFAC, International [14] P. I. Corke, Robotics, Vision and Control, Springer, 2011.
Series on systems and Engineering, 19, 2002.
[4] A. De Luca, Robots with Flexible Links: Modeling and
Control, outlined, Roma University, Department of informatics
and systematic (DIS), Roma, 2003. online avialable:
http://www.prisma.unina.it/cira03/TA DeLuca2.pdf
[5] L. Chen and H. Deng, Model Reduction of Rigid-Flexible Manipulators
with Experimental Validation, Advanced Materials Research, Vol. 655-
657, pp. 1101-1107, 2013.

724

Authorized licensed use limited to: Ritsumeikan University. Downloaded on April 20,2023 at 10:08:34 UTC from IEEE Xplore. Restrictions apply.

You might also like