31 multi-degree of freedom vibration model for a 3-dof hybrid robot
TRANSCRIPT
Multi-degree of Freedom Vibration Model for a 3-DOF Hybrid Robot
Yuan Yun and Yangmin Li
Abstract— In recent years, many applications in precisionengineering require a careful isolation of the instrument fromthe vibration sources by adopting active vibration isolationsystem to achieve a very low remaining vibration level especiallyfor the very low frequency under 10Hz vibration signals. Inthis paper, based on our previous research experiences in thesystematical modeling and study of parallel robots, a hybridrobot is described and the vibration model is given by usingLagrange’s Equations. Then a dynamic analysis by using Kane’smethod is presented to verify the result. Finally, numerical sim-ulations on the dynamic characteristics and natural frequencyare presented.
I. INTRODUCTION
Nowadays, a careful isolation of the instrument from the
vibration sources is required in many precision engineering
applications, which is realized by adopting active vibration
isolation system to achieve a very low remaining vibration
level especially for the very low frequency under 10Hz
vibrations.
Parallel manipulators can offer the advantages of high
stiffness, low inertia, and high speed capability which have
been intensively investigated and evaluated by industry and
institutions in recent years. And some designers adopt the
flexure hinges instead of conventional mechanism joints
since the backlash and frictions in the conventional joints in-
fluence the performances of parallel mechanisms remarkably.
Micro/nano positioning manipulators and active vibration
control devices are increasingly being made of parallel
manipulators due to their characteristics of high precision.
Although the adoption of flexure hinges in the parallel
mechanism systems increases the high precision, short stroke
actuators with nanometer scale level precision result in a very
small workspace.
Therefore, a lot of designers focus their attention on multi-
degree of freedom hybrid manipulators or developing wide
range flexure hinges. A spatial compliant 3-DOF parallel
robot with SMA pseudo-elastic flexure hinges was presented
in [1], which has a workspace larger than 200 × 200 × 60mm3 and resolution is better than 1µm. A parallel structure
for macro-micro systems was proposed in [2]. In this new
design, the macro-motion (DC motor) and micro motion
are connected by a parallel structure, the two motions are
coupled under one compliant mechanism framework. At the
same time, a kind of dual parallel mechanism was developed
This work is supported by the research committee of University of Macauunder Grant no. UL016/08-Y2/EME/LYM01/FST and Macao Science andTechnology Development Fund under Grant no. 016/2008/A1
Yuan Yun and Yangmin Li are with the Department of ElectromechanicalEngineering, Faculty of Science and Technology, University of Macau,Av. Padre Toms Pereira, Taipa, Macau, China, [email protected],[email protected]
[3], called a 6-PSS parallel mechanism and a 6-SPS one,
which is integrated with wide-range flexure hinges as passive
joints to ensure the large workspace of the whole system and
high precision motion. A XYZ-flexure parallel mechanism is
proposed with large displacement and decoupled kinematics
structure [4], which has a large motion range beyond of
1mm.
Our research is concerned with the development of a
system that can achieve three high accurate translational
positioning or rough positioning with a 3-DOF active vi-
bration isolation, which attenuates the vibration transmission
above some corner frequencies, to protect the payload from
jitters induced by the various disturbance sources. In this
paper, based on our previous research experiences in the
systematical modeling and study of parallel robots, a novel
dual 3-DOF parallel robot with flexure hinges will be de-
scribed. A vibration model of the parallel manipulator will be
presented which will be verified by Kane’s dynamics. Finally,
numerical simulations about the dynamic characteristics and
natural frequency will be presented.
II. SYSTEM DESCRIPTION
The designed manipulator is a 3-DOF dual parallel mecha-
nism combining a 3-PUU parallel mechanism with a spatial
3-UPU one. In the 3-PUU parallel structure, the prismatic
actuators provide three translational macro motions with
micron level accuracy and cubic centimeter workspace. At
the same time, the micro motion is provided by a spatial 3-
UPU structure which can increase the accuracy of the whole
system to the nanometer level or provide a strict acceleration
level for payload placed on the moving platform. 3-PUU and
3-UPU kinematical structures with conventional mechanical
joints can be arranged to achieve only translational motions
with some certain geometric conditions satisfied.
Using flexure hinges at all joints, the parallel platform
consists of a mobile platform, a fixed base, and three limbs
with identical kinematic structure. Each limb connects the
mobile platform to the fixed base by one prismatic actuator,
one flexure universal (U) hinge and a piezoelectric (PZT)
ceramics actuator followed by another U joint in sequence
as shown in Fig. 1, where the first U joint is fixed at the
prismatic actuator and the second one connects to the moving
platform actuated by a PZT to offer the micro motions
with merits involving smooth motion, high accuracy, and
fast response, etc. Piezoelectric actuators are adopted as the
prismatic actuators to provide the macro motion for the
mechanism and three groups of guiding-rail and gliding-
block are adopted as the components. The flexure universal
hinge is a slender shaft configuration with very high torsional
2009 IEEE/ASME International Conference on Advanced Intelligent MechatronicsSuntec Convention and Exhibition CenterSingapore, July 14-17, 2009
978-1-4244-2853-3/09/$25.00 ©2009 IEEE 1474
Authorized licensed use limited to: Universidade de Macau. Downloaded on November 25, 2009 at 04:32 from IEEE Xplore. Restrictions apply.
stiffness which is adopted as passive joint to ensure the large
workspace of the whole system and high precision motion.
Fig. 1. The 3-DOF 3-PUU/3-UPU dual parallel manipulator.
III. DYNAMIC MODEL
Since the macro motion is adopted for the rough posi-
tioning, the kinematic analysis is necessary for the 3-PUU
structure. The details have been published in [5]. In this
section, the vibration model using Lagrange’s equations will
be described first for the 3-UPU structrue, and then Kane’s
dynamics modeling will be built up to verify the results.
A. Kinematic Analysis
In this case, the prismatic actuators for macro motion are
self-locked when the micro motion is available.
1) Coordinate System: As shown in Fig. 2, let L=[l1 l2l3]
T be the vector of the three PZT actuated length variables
and the vector roo′=[x y z]T of the reference point o′ be
the position of the moving platform. In addition, let bi be
the vector−−→oBi and mi be the vector
−−→o′Mi. The mass of
moving platform is M . The mass of each strut is ml. C∗
i
(i=1, · · · , 3) is the center of mass for the ith limb. Let
a limb fix right-handed coordinate system with origin C∗
i
(i=1, · · · , 3) located at the center of mass for the ith limb,
with axis directions determined by an orthonormal set of
unit vectors cij (j=1,· · · ,3). The overhat indicates unit length,
the index i corresponds to the ith limb, and the index j
distinguishes the three vectors. ci3 is along the ith limb,
toward the ith flexure hinge which connects the moving
platform. ci2 is perpendicular to the vector
−−→oBi when the
moving platform is in the home position, and ci1 is in the
direction ci2× c
i3. Let a reference frame sj attach to the fixed
platform at the center o with the s1 toward the point B1 and
s3 vertical the fixed platform. Fix a coordinate system fj to
the moving platform at the center o’ with f1 toward the point
Mi and f3 vertical the moving platform. ℜ is a three-order
identity transformation matrix from fj coordinate system
to sj . ℜci is the transformation matrix from cj coordinate
system to sj .
In order to determine the angles of flexure hinges directly,
an initial coordinate system is set on origin C∗
i with axis
directions determined by an orthonormal set of unit vectors
cij0 when the moving platform is in the home position. The
axis directions are located coincident with the corresponding
limb fixed coordinate system cij when the moving platform is
Fig. 2. Coordinate system of the 3-UPU parallel platform
in the initial position. Let the orientation of the cij coordinate
system, relative to the cij0, be described by consecutive
positive rotations qi1 about the c
i10 and qi
2 about the moved
two-axis. The rotation matrix is:
ℜci0 =[
ci10 c
i20 c
i30
]−1 [
ci1 c
i2 c
i3
](1)
It is assumed that the small-angle approximations hold for
angles qik, the rotation matrix:
ℜci0 =
1 0 qi
2
0 1 −qi1
−qi2 qi
1 1
(2)
2) Generalized Velocities for the System: Define general-
ized velocities u for the system as the time rate of change
of the generalized coordinates of q=[x y z]T in the inertial
reference frame:
u =[
x y z]T
(3)
An intermediate reference frame is introduced previously
to permit describing the angular velocity of each limb to
the initial limb position. Designate the unit vectors of these
intermediate reference frames by mij . The expressions for
the angular velocities of the reference frame of each upper
arm with respect to initial reference frame of upper arm are:
ωC∗
i= −qi
1ci10 − qi
2mi2 (4)
The position vectors are:
roo′ = lici3 + bi − mi
rmi= lic
i3 + bi, rC∗
i=
li
2ci3 + bi
Differentiating Eq. (5) with respect to time:
vo′ = lici3 + liωC∗
i× c
i3 (5)
vmi= vo′ , vC∗
i=
lici3 + liωC∗
i× c
i3
2=
vo′
2
Dot-multiplying both sides of Eq. (5) by ci3:
(ci3)
T vo′ = li (i = 1, 2, 3) (6)
which can be assembled into a matrix form:[
l1 l2 l3]T
= A · vo′ (7)
1475
Authorized licensed use limited to: Universidade de Macau. Downloaded on November 25, 2009 at 04:32 from IEEE Xplore. Restrictions apply.
where A =[
(c1
3) (c2
3) (c3
3)]T
.
Cross-multiplying both sides of Eq. (5) by ci3:
ωC∗
i=
ci3 × vo′
li(8)
The linearized accelerations are:
ao′ = (2liωC∗
i+ liεC∗
i) × c
i3 + liωC∗
i× (ωC∗
i× c
i3) + lic
i3
aC∗
i=
ao′
2(9)
Cross-multiplying both sides of Eq. (30) by ci3, the angular
acceleration of each limb can be obtained by:
εC∗
i=
ci3 × ao′ − 2liωC∗
i− li(ωC∗
i· c
i3)(ωC∗
i× c
i3)
li(10)
B. Vibration Model
In this paper, the vibrational model is established by
using Lagrange’s equation which is a well known tool for
establishing equations of motion of discrete systems. The key
point of Lagrange’s equations is kinetic energy, which can
be formulated favorably with respect to a moving coordinate
system as well.
1) Kinetic Energy of the System: For the moving platform,
the kinetic energy can be given by:
EkM =1
2M(x2 + y2 + z2) (11)
For each limb, assume rotations about a fixed axis, the kinetic
energy can be given by:
Ekli =1
2ml‖vC∗
i‖2 +
1
2ICi
ω2C∗
i(12)
where ICi= 1
3mll
2i and according to Eq. (4), the kinetic
energy of the whole system can be derived by:
Ek = EkM+
3∑
i=1
Ekli =1
2M(x2+y2+z2)+
3∑
i=1
(1
2ml‖vC∗
i‖2
+1
2ICi
((qi1)
2 + (qi2)
2)) = (1
2M +
3
8ml)(x
2 + y2 + z2)
+
3∑
i=1
1
6mll
2i ((q
i1)
2 + (qi2)
2)) (13)
2) Potential Energy of the System: The potential energy
of moving platform is given by:
EpM = Mgz (14)
Since the system is investigated in tiny vibration environ-
ment, according to the Eq. (5), we can obtain:
ci3 =
1
li
x − xbi+ xmi
y − ybi+ ymi
z − zbi+ zmi
(15)
Let
[ci10 c
i20 c
i30
]−1
=
ri11 ri
12 ri13
ri21 ri
22 ri23
ri31 ri
32 ri33
(16)
According to Eq. (1) and (2), the consecutive positive rota-
tions qi1 and qi
2 about the moved two-axis are:
1
li
ri11 ri
12 ri13
ri21 ri
22 ri23
ri31 ri
32 ri33
x − xbi
+ xmi
y − ybi+ ymi
z − zbi+ zmi
=
qi2
−qi1
1
qi1 = H1i −
ri21
lix −
ri22
liy −
ri23
liz
qi2 = H2i +
ri11
lix +
ri12
liy +
ri13
liz (17)
where
H1i =ri21
li(xbi
− xmi) +
ri22
li(ybi
− ymi) +
ri23
li(zbi
− zmi)
H2i =ri11
li(xmi
− xbi) +
ri12
li(ymi
− ybi) +
ri13
li(zmi
− zbi)
For each limb, the potential energy can be written as:
Epli = mlgz − zbi
+ zmi
2+ ki
1(qi1)
2 + ki2(q
i2)
2 (18)
Hence, the potential energy of the whole system can be given
by:
Ep = Mgz +
3∑
i=1
(mlgz − zbi
+ zmi
2+ ki
1(qi1)
2 + ki2(q
i2)
2)
(19)
3) Lagrange’s Equations: The Lagrange’s equations are:
d
dt(∂Ek
∂uj
) −∂Ek
∂qj
+∂Ep
∂qj
= FDj +3∑
i=1
F ai
j (20)
where
d
dt(∂Ek
∂x) = (M +
3
4ml)x +
3∑
i=1
1
3ml(((r
i21)
2 + (ri11)
2)x
+(ri22r
i21 + ri
12ri11)y + (ri
23ri21 + ri
13ri11)z)
d
dt(∂Ek
∂y) = (M +
3
4ml)y +
3∑
i=1
1
3ml((r
i21r
i22 + ri
11ri12)x
+((ri22)
2 + (ri12)
2)y + (ri23r
i22 + ri
13ri12)z)
d
dt(∂Ek
∂z) = (M +
3
4ml)z +
3∑
i=1
1
3ml((r
i21r
i23 + ri
11ri13)x
+(ri22r
i23 + ri
12ri13)y + ((ri
23)2 + (ri
13)2)z)
∂Ek
∂x= 0,
∂Ek
∂y= 0,
∂Ek
∂z= 0
∂Ep
∂x=
3∑
i=1
(2ki1(−
r21
li)(H1i −
r21
lix −
r22
liy −
r23
liz)
+2ki2(
r11
li)(H2i +
r11
lix +
r12
liy +
r13
liz))
∂Ep
∂y=
3∑
i=1
(2ki1(−
r22
li)(H1i −
r21
lix −
r22
liy −
r23
liz)
1476
Authorized licensed use limited to: Universidade de Macau. Downloaded on November 25, 2009 at 04:32 from IEEE Xplore. Restrictions apply.
+2ki2(
r12
li)(H2i +
r11
lix +
r12
liy +
r13
liz))
∂Ep
∂z= Mg +
3mlg
2+
3∑
i=1
(2ki1(−
r23
li)(H1i −
r21
lix−
r22
liy
−r23
liz) + 2ki
2(r13
li)(H2i +
r11
lix +
r12
liy +
r13
liz)) (21)
The final forward dynamic equation of this parallel multi-
body system can be written as:
M · u + K · q = Q (22)
where M, K and Q are the mass matrix, stiffness matrix and
force vector of the system, among which
M = (M +3
4ml)I +
1
3ml
3∑
i=1
Mi
Mi =
(ri21)
2 + (ri11)
2 ri22r
i21 + ri
12ri11 ri
23ri21 + ri
13ri11
ri21r
i22 + ri
11ri12 (ri
22)2 + (ri
12)2 ri
23ri22 + ri
13ri12
ri21r
i23 + ri
11ri13 ri
22ri23 + ri
12ri13 (ri
23)2 + (ri
13)2
and I is the identity matrix. In this case, the pertinent flexure
hinge stiffness has the relationship of
ki1 = k
j2 = k (i = 1, 2, 3, j = 1, 2, 3).
Hence, the stiffness matrix of the whole system is:
K = 2k
3∑i=1
r2
21+r2
11
l2i
3∑i=1
r21r22+r11r12
l2i
3∑i=1
r21r23+r11r13
l2i
3∑i=1
r21r22+r11r12
l2i
3∑i=1
r2
22+r2
12
l2i
3∑i=1
r22r23+r12r13
l2i
3∑i=1
r21r23+r11r13
l2i
3∑i=1
r22r23+r12r13
l2i
3∑i=1
r2
23+r2
13
l2i
Let Fai be the force exerted by the ith PZT actuator on each
limb at the mass center of the strut. The direct disturbance
forces FD is acting on the center of moving platform. The
generalized external force matrix is:
Q = FD+3∑
i=1
Fai+
2k3∑
i=1
( r21
liH1i −
r11
liH2i)
2k3∑
i=1
( r22
liH1i −
r12
liH2i)
2k3∑
i=1
( r23
liH1i −
r13
liH2i) − Mg − 3mlg
2
Let Fa be the matrix of driving forces Fi given by:
Fa =[
F1 F2 F3
]T(23)
where
Fai = Fici3 (24)
The fundamental natural frequency can be derived by the
global stiffness matrix and mass matrix of the system. First,
the eigenvalues λi ≥ 0 (i = 1, 2, 3) can be given by:
|K − Mλ2| = 0 (25)
The fundamental natural frequency fi is:
fi =λ
2π(26)
C. Kane’s method
In Kane’s method, the dynamics equations are concise
since unwanted reaction forces and torques do not appear
in the final form. The key point underlying Kane’s method
is the derivation of partial velocities.
1) Linearized Partial Velocities: Since the velocity of
mass center of moving platform can also be obtained by:
vo′ = x · f1 + y · f2 + z · f3 = x · s1 + y · s2 + z · s3 (27)
the partial velocities and partial angular velocities are formed
by inspection of the relevant velocity vectors:
vo′,u1= s1, vo′,u2
= s2, vo′,u3= s3
vC∗
i,u1
=s1
2, vC∗
i,u2
=s2
2, vC∗
i,u3
=s3
2, vmi,uj
= vo′,uj
ωC∗
i,u1
=ci3 × s1
li, ωC∗
i,u2
=ci3 × s2
li, ωC∗
i,u3
=ci3 × s3
li(28)
2) Generalized Active Forces: Let Fmi and Mmi repre-
sent the force and moment exerted by the ith limb on the
moving platform, at the ith flexure hinge. Since Fmi is a
noncontributing force, it can be ignored in the analysis. The
total moment Mm due to the three flexure hinges connecting
the moving platform is:
Mm =
3∑
i=1
Mmi = k
3∑
i=1
(qi1c
i10 + qi
2mi2) (29)
where ki1 and ki
2 is pertinent flexure hinge stiffness. Let
FD and MD represent the disturbance force and moment
acting on the mass center o′ of moving platform. The moving
platform’s contribution to the set of generalized active forces,
for the jth generalized velocity, is:
Qm,uj= vo′,uj
· (Mg + FD) (30)
The forces and moments acting on the ith limb are due to
the moving platform (through the ith upper flexure hinge),
gravity of each limb, the PZT actuator and the moments
through the lower flexure hinge. The contributing loads of
the limbs are as follows:
QC∗
i,uj
= vC∗
i,uj
·mlg+(li ·ci3)uj
·Fai +ωC∗
i ,uj·(Mli−Mmi)
Mli = −k(qi1c
i10 + qi
2mi2) (31)
where (li · ci3)uj
= (vo′ − liωC∗
i× c
i3)uj
= (vo′ − (ci3 × vo′) × c
i3)uj
= ((vo′ · ci3)c
i3)uj
= (sj · ci3)c
i3
3) Generalized Inertia Forces: The contribution Q∗
m,ujto
the generalized inertia forces for the jth generalized velocity
is:
Q∗
m,uj= vo′,uj
· (−Mao′) (32)
The contribution Q∗
C∗
i,uj
to the generalized inertia forces
due to the ith limb for the jth generalized velocity is:
Q∗
C∗
i,uj
= vC∗
i,uj
· (−muaC∗
i) + ωC∗
i ,uj· (−IC∗
iεC∗
i
−ωC∗
i× IC∗
iωC∗
i) (33)
1477
Authorized licensed use limited to: Universidade de Macau. Downloaded on November 25, 2009 at 04:32 from IEEE Xplore. Restrictions apply.
where IC∗
iis the inertia matrix of each limb with respect to
its mass center C∗
i in global coordinate system sj . Let I′C∗
i
be the inertia matrix of each limb with respect to its mass
center C∗
i in the coordinate system cij . Hence,
IC∗
i=
[ci1 c
i2 c
i3
]· I′C∗
i·[
ci1 c
i2 c
i3
]T
4) Dynamical Equations: The holonomic generalized ac-
tive force for the jth (j = 1, · · · , 3) generalized velocity
is:
Fj = Qm,uj+
3∑
i=1
QC∗
i,uj
(34)
Likewise, the contribution to the set of holonomic general-
ized inertia force is:
F∗
j = Q∗
m,uj+
3∑
i=1
Q∗
C∗
i,uj
(35)
Kane’s dynamical equations are:
F∗
j + Fj = 0(j = 1, 2, 3) (36)
Eq. (36) can be expanded as:
vo′,uj·(Mg+FD−Mao′)+
3∑
i=1
((li ·ci3)uj
·Fai +vC∗
i,uj
·ml(g
−aC∗
i)+ωC∗
i,uj
·(Mli−Mmi−IC∗
iεC∗
i−ωC∗
i×IC∗
iωC∗
i)) = 0
The equations for the inverse dynamics can be arranged to
the form:
D · Fa = H (37)
where
D =
s1 · c1
3 s1 · c2
3 s1 · c3
3
s2 · c1
3 s2 · c2
3 s2 · c3
3
s3 · c1
3 s3 · c2
3 s3 · c3
3
(38)
H =
−vo′,u1· (Mg + FD − Mao′)
−3∑
i=1
(vC∗
i,u1
· ml(g − aC∗
i) + ωC∗
i ,u1
·(Mli − Mmi − IC∗
iεC∗
i− ωC∗
i× IC∗
iωC∗
i))
−vo′,u2· (Mg + FD − Mao′)
−3∑
i=1
(vC∗
i,u2
· ml(g − aC∗
i) + ωC∗
i,u2
·(Mli − Mmi − IC∗
iεC∗
i− ωC∗
i× IC∗
iωC∗
i))
−vo′,u3· (Mg + FD − Mao′)
−3∑
i=1
(vC∗
i,u3
· ml(g − aC∗
i) + ωC∗
i,u3
·(Mli − Mmi − IC∗
iεC∗
i− ωC∗
i× IC∗
iωC∗
i))
(39)
Here we noticed that the forward dynamic result is easy to
be obtained by using Lagrange’s equation. And the Kane’s
method is a convenient and rapid approach to get the inverse
dynamic analytical solution, even though they all lead to the
same model.
Fig. 3. Displacements and accelerations of the moving platform (a).
IV. NUMERICAL SIMULATION
Due to the requirement of the cubic centimeter scale large
workspace, the parameters of the parallel mechanism are
shown in Table I. As the well elastic nature of the selected
material beryllium bronze of wide-range flexure hinges, the
workspace of micro motion only depends on the limits
of PZT ceramics. Considering the different characteristics
between flexure hinge and conventional mechanical joint, the
parasitic rotations of moving platform have to be calculated
and limited in 5mrad in this paper. The maximal usable
inscribed workspace of micro motion, when the inputs of
PZT motors are zero and self-locked at the initial position,
is a column with a radius of 66.6µm and 38.6µm height
by using PRB model considering the strokes of selected
PZT ceramics. For the details of the theoretical derivation
of macro/micro usable workspace, please refer to the [5] of
our previous work.
TABLE I
GEOMETRIC AND MATERIAL PARAMETERS
Item Value
Radius of moving platform r 20mm
Radius of fixed platform R 90mm
Radius of flexure hinge rf 0.9mm
Length of flexure hinge lf 10mm
Radius of strut rs 6.35mm
Length of strut ls 120mm
Modulus of elasticity of flexure hinge Ef 130GPa
Modulus of elasticity of strut Es 70GPa
Stroke of PZT ceramic Qc 100µm
Stroke of PZT motor Qm 10mm
A. Dynamic Simulation
In order to confirm the dynamic model, the moving
platform is moving along the z axis with the displacements
and accelerations given by Fig. 3. In this state, it is obvious
that the actuators should output the same forces as shown in
Fig. 4. The results solved by Lagrange’s equations are almost
the same with those obtained by Kane’s method.
Then another motion is simulated as following curves
shown in Fig. 5 and the three input forces of actuators
with respect to time are shown in Fig. 6. Since the mass
center of moving platform has movement along each axis,
1478
Authorized licensed use limited to: Universidade de Macau. Downloaded on November 25, 2009 at 04:32 from IEEE Xplore. Restrictions apply.
Fig. 4. Driving forces of the three actuators solved by Lagrange’s equationsand Kane’s method (a).
Fig. 5. Displacements and accelerations of the moving platform (b).
the three driving forces appear different in amplitude, phase
and period.
B. Natural Frequency
The distributions of three fundamental natural frequencies
are simulated when the mass center of moving platform is
located in z = 121.2435mm and moving in a plane with a
radius of 66.6µm as shown in Fig. 7 according to the Eq.
(26). The fist-order natural frequency is about 9.5811Hz, and
the second-order and third order natural frequency is about
16.478Hz.
V. CONCLUSIONS AND FUTURE WORKS
In this paper, a novel dual 3-DOF parallel robot with flex-
ure hinges has been presented. This system can achieve three
high accurate translational positioning or rough positioning
with a 3-DOF active vibration isolation and excitation func-
tion to the payload placed on the moving platform. Based
on our previous research experiences in the systematical
modeling and study of parallel robots, a vibrational model
Fig. 6. Driving forces of the three actuators solved by Lagrange’s equationsand Kane’s method (b).
Fig. 7. The three fundamental natural frequency.
of the parallel manipulator has been presented which has
been verified by Kane’s dynamics equation. Finally, some
simulation results by MATLAB are shown to verify the
correctness of the formulae derivation and natural frequency
are presented. The investigations of this paper are expected
to make contributions to the research on active vibration
isolation based on parallel manipulators. After this work,
some control modeling for active vibration control will
be given and optimal control algorithm will be discussed.
Finally, a prototype of the parallel robot will be developed,
and the experimental investigation will be carried out to
verify the vibration control performance of the prototype.
REFERENCES
[1] J. Hesselbach, A. Raatz, J. Wrege, and S.Soetebier, “Design and anal-ysis of a macro parallel robot with flexure hinges for micro assemblytasks,” in Proc. of 35th International Symposium on Robotics, Paris,France, 2004, No.TU14-041fp.
[2] P. R. Ouyang, “Hybrid intelligent machine systems: Design, Modelingand Control,” Ph.D. Thesis, University of Saskatchewan, Canada,2005.
[3] W. Dong, L. N. Sun, and Z. J. Du, “Design of a precision compliantparallel positioner driven by dual piezoelectric actuators,” Sensors and
Actuators A: Physical, Vol. 135, pp. 250-256, 2007.[4] X. Tang and I. M. Chen, “A large-displacement 3-DOF flexure
parallel mechanism with decoupled kinematics structure,” in IEEE/RSJ
International Conference on Intelligent Robots and Systems, Beijing,China, pp. 1668-1673, 2006.
[5] Y. Yun and Y. Li, “Performance Analysis and Optimization of a NovelLarge Displacement 3-DOF Parallel Manipulator”, in Proc. of IEEE
International Conference on Robotics and Biomimetics, Bangkok,Thailand, pp. 246-251, 2008.
[6] R. D. Hampton and G. S. Beech, “A Kane’s Dynamics Model for theActive Rack Isolation System”, NASA/TMł2001C2110063, MarshallSpace Flight Center, AL, 2001.
[7] Y. Yun, Y. Li and Q. Xu, “Active vibration control on A 3-DOF parallelplatform based on Kanes dynamic method”, in Proc. of SICE Annual
Conference, Chofu, Tokyo, Japan, pp. 2783-2788, 2008.[8] L. W. Tsai and S. Joshi, “Kinematics analysis of 3-DOF position
mechanisms for use in hybrid kinematic machines,” ASME Journalof Mechanical Design, Vol. 124, No. 2, pp. 245C253, 2002.
[9] Y. Li and Q. Xu, “Kinematic analysis and dynamic control of a 3-PUUparallel manipulator for cardiopulmonary resuscitation,” Proc. of 12thInt. Conf. on Advanced Robotics, pp. 344C351, 2005.
[10] M. S. Whorton, H. Buschek and A. J. Calise, “Homotopy Algorithmfor Fixed Order Mixed H2/H∞ Design,” Journal of Guidance,
Control, and Dynamics, Vol. 19, No. 6, pp. 1262C1269, 1996.[11] M. Whorton, “Robust Control for Microgravity Vibration Isolation Us-
ing Fixed-order, Mixed H2/µ Design,” AIAA Guidance, Navigation,
and Control Conference and Exhibit, Austin, Texas, Aug. 11-14, 2003.
1479
Authorized licensed use limited to: Universidade de Macau. Downloaded on November 25, 2009 at 04:32 from IEEE Xplore. Restrictions apply.