Proceedings of 2014 IEEE

International Conference on Mechatronics and Automation

August 3 - 6, Tianjin, China

Dynamics Analysis of a Novel Over-constrained

Three-DOF Parallel Manipulator
Bin Li and Yangmin Li Weimin Ge, Xinhua Zhao and Yuwei Yang
Faculty of Science and Technology Tianjin Key Laboratory of Control Theory
University of Macau and Application in Complicated Industrial
Taipa, Macau,Chian System, Tianjin 300384, China
[email protected], [email protected] [email protected]

Abstract -In this paper, a novel over-constrained three forces/torques to follow a desired end effector trajectory.
degree-of-freedom (DOF) spatial parallel manipulator (SPM) is Inverse dynamics is used in a wide range of controllers known
proposed. The architecture of the SPM is comprised of a moving as computed torque controllers, in optimum trajectory
platform attached to a base through two revolute- prismatic- planning, and also in manipulator design (selection of the
universal jointed serial linkages and one spherical-prismatic-
actuators, structure design, and so on). Several approaches
revolute jointed serial linkage. The prismatic joints are
considered to be actively actuated. Kinematics and dynamics of have been proposed including the Newton–Euler laws, the
the SPM are studied systematically. Firstly, both of the inverse principle of virtual work and the Lagrangian formulation.
and forward displacements are derived by analytic formulae. Among these methods, the principle of virtual work is the
Then, the velocities and the accelerations of the moving platform most efficient method for the dynamic analysis of parallel
and each limb are obtained. Finally, dynamics of the mechanism manipulators. Different researchers [11–14] present the
is derived based on the principal of the virtual work. The results dynamic analysis of parallel manipulators based on this
are illustrated by numerical examples. principle.
In this paper, a novel over-constrained 3-DOF SPM 2-
Index Terms – Over-constrained parallel mechanism, Kinematics,
RPU&SPR with one degree of translational freedom and two
Dynamics.
degrees of rotational freedom is proposed. The architecture of
I. INTRODUCTION the SPM is comprised of a moving platform attached to a base
through two revolute- prismatic-universal jointed serial
Compared with 6-DOF parallel manipulators, lower- linkages and one spherical-prismatic-revolute jointed serial
mobility mechanisms possess such merits as simpler linkage.Kinematics and dynamics of the SPM are studied
mechanical design, lower manufacturing cost, and larger systematically. Firstly, both of the inverse and forward
workspace, in addition to the inherent advantages of the displacements are derived by analytic formulae. Then, the
general parallel manipulators in terms of high accuracy, high velocities and the accelerations of the moving platform and
stiffness, high velocity, high dynamic performance, big load to each limb are obtained. Furthermore, dynamics of the
weight ratio, low moving inertia and little accumulation of mechanism is derived based on the principal of the virtual
positional errors. A lower-mobility SPM with 3-DOF is the work. Finally, the results are illustrated by numerical
focus of the current trend in the research community, and examples.
various forms of 3-DOF SPM have been presented.
Hunt [1] presented a 3-RPS parallel manipulator, which II. STRUCTURE DESCRIPTION
has two rotational and one translational DOFs, up to now this
mechanism has been widely studied and used in practice.
Sokolov [2] studied the dynamics of the mechanism. Liu [3] Moving platform
developed a type of new 3-DOF SPM with three nonidentical Complex universal joint Revolute joint
chains, the moving platform has two translational and a
rotational freedom, both the inverse and forward kinematics Prismatic joint
problems and singularity are described. Zhang [4] studied a Prismatic joint
novel 3-DOF SPM used for machining applications, the Revolute joint
structure of the mechanism is comprised of three active limbs Spherical joint
Prismatic joint
and one passive limb aiming to achieve a greater stiffness and
more effective 3-DOF motions. Xie [5] proposed a decoupled
3-DOF parallel tool head without parasitic motions, and the
kinematic optimization of the tool head was carried out. Base platform
Besides these new 3-DOF SPMs mentioned above, some other
Revolute joint
3-DOF architectures can be found in the literature [6-10].
Dynamic analysis plays an important role in predicting Fig. 1. CAD model of the proposed 3-DOF SPM 2RPU & SPR
the behavior of mechanical systems. The inverse dynamic
model of a robot manipulator provides the input actuator

978-1-4799-3979-4/14/$31.00 ©2014 IEEE 828

 III. INVERSE AND FORWARD POSITION ANALYSES
c3
A. Inverse position analysis
Inverse position analysis of the 2-RPU&SPR SPM
involves the determination of the limb lengths given the pose
of the moving platform.
In the B-xyz, the position vector r=(x y z)T of A can be
q3w3 expressed by the loop closure equation
r = bi + qi wi − ai , i = 1,2,3 (2)
q1w1 q2 w2 where ai = Rai0 , qi and wi are the length and unit vector of limb
i, a10 = a20 = a1 ( 0 −1 0) , a30 = a2 ( 0 1 0) and b1 = b1 ( − 1 0 0 ) ,
T T T

b2 = b2 (1 0 0 ) , b3 = b3 ( 0 1 0 ) are the position vectors of Ai

T T

and Bi measured in A-uvw and B-xyz with ai and bi being the

c1
distance of AAi and BBi , here, a1 = a2 = a3 = a , b1 = b2 = b .
Note that for the RPU limb the constraint imposed by the
c2 revolute joint restricts both wi and bi to be normal to the unit
vector ci of the revolute joint axis. Thus, taking the dot
product with ci on both sides of Eq.(2), leads to
Fig. 2. Schematic model of the mechanism
( r + ai ) ci = 0 , i = 1, 2
T
(3)
where c1 = c 2 = ( 0 1 0 ) .
T
As shown in Fig. 1, the proposed over-constrained 3-
DOF SPM 2-RPU & SPR consists of a base platform and a Similarly, for the SPR limb, the constraint imposed by
moving platform connected by two identical RPU limbs and the revolute joint restricts both w3 and a3 to be normal to the
one SPR limb. Here, R, U and S represent the revolute, unit vector c3 of the revolute joint axis. Thus, taking the dot
universal and spherical joints, and the underlined P denotes product with ci on both sides of Eq.(2), leads to
the actuated prismatic joint. Place the reference frame B-xyz
( r − bi ) ci = 0 , i = 3
T
attached to the base and the moving frame A-uvw attached to (4)
the moving platform with B and A being the origins located at where c3 = Rc30 , c30 = (1 0 0 ) .
T

the midpoint of lines B1B2 and A1A3 with the x and u axes
According to the geometric constraint of the mechanism,
being parallel to B1B2 and A1A3, and the z and w axes being
axis of the universal joint R2 can just only moves in the plane
normal to the base platform and the moving platform. Here, Bi
Bxz as shown in Fig. 2, so another equation can be formulated
(i=1,2) and A3 are the intersection of the axes of the revolute
uy = 0 (5)
joints and actuated prismatic joints, B3 is the center of the
spherical joint, A1 and A2 are the centers of the universal Based on Eqs.(3), (4), and (5), we can obtain
joints, respectively. φ =0 (6)
For the proposed mechanism, the two identical RPU y = a cos (ψ ) (7)
limbs are in the same plane of Bxy and axes of the two
x = z tan (θ ) (8)
revolute joints are parallel to y axis as shown in Fig. 2. It
should be pointed out that, two universal joints of RPU limbs Here, φ , y , and x represent the parasitic motions of the
are shared and connected on the moving platform. The axis of moving platform. The parasitic motions lead to complex
revolute joint of the SPR limb is parallel to u axis. kinematics, require real-time compensation, increase difficulty
The orientation matrix of the A-uvw with respect to the B- in calibration, and even result in damage to the SPM. For a
xyz can be formulated by three Euler angles ψ , φ and θ given set of unconstrained variables (ψ θ z ) , the parasitic
satisfying x-z-y conventions: motion of the mechanism can be calculated through Eqs.(6),
(7), and (8). Then, the limb length can be calculated as
ª cθ cφ sψ sθ − cψ cθ sφ cψ sθ + sψ cθ sφ º
R = «« sφ cψ cφ − sψ cφ »
» qi = r − bi +ai , i =1,2,3 (9)
¬« − sθ cφ sψ cθ + cψ sθ sφ cψ cθ − sψ sθ sφ ¼»
(1) B. Forward position analysis
ªu x vx wx º
The forward position analysis of the mechanism is
= ««u y vy wy »» = [ u v w] concerned with the determination of the moving platform pose
«¬ u z vz wz »¼ given the limb lengths. Firstly, taking Euclidean norm on both
sides of Eq.(9), and implementing necessary addition and
where u, v and w are the unit vectors of the frame A-uvw with subtraction, yields
respect to the frame B-xyz, “s” and “c” here denote sin and cos 2 2
functions. q12 = ¬ª z tan (θ ) − a sin (ψ ) sin (θ ) + b ¼º + ª¬ z − a sin (ψ ) cos (θ ) º¼ (10)

829
 2
q22 = ª¬ z tan (θ ) − a sin (ψ ) sin (θ ) − b º¼ + ª¬ z − a sin (ψ ) cos (θ ) º¼ (11)
2
qi = wi r + ( ai × wi ) ⋅ Ȧ i = 1,2,3
(22)
2 When the manipulator is away from singularities, Eq.
q32 = ¬ª z tan (θ ) + a sin (ψ ) sin (θ ) º¼ + (22) can be written in the matrix form
(12)
2 2
qa = J a X
¬ª 2a cos (ψ ) − b3 º¼ + ª¬ z + a sin (ψ ) cos (θ ) º¼
(23)
1) If q1 ≠ q2 , implementing subtraction and addition for Eq.(10) ª w1T ( a1 × w1 ) º»
T
ª q1 º «
ª 
r º
and Eq.(11), yields where, qa = «« q 2 »» , X = « » , J a = « w2T ( a2 × w2 )T »» .
z = f1c tan (θ ) + a sin (ψ ) cos (θ ) (13) ¬ Ȧ¼ «
«¬ q3 »¼ «wT ( a3 × w3 ) »¼
T

2z2 4az sin (ψ ) ¬ 3

+ 2a 2 sin (ψ ) + 2b 2
2
q12 + q22 = − (14) Differentiating Eqs.(6), (7) and (8) with respect to time,
cos (θ )
2
cos (θ )
respectively, yields
q12 − q22 φ = 0 (24)
where f1 = .
4b y = −a sin (ψ )ψ (25)
Then, substituting Eq.(13) into Eqs. (14) and (12) gives
x = z tan (θ ) +z sec 2 (θ )θ (26)
2 f12
sin (θ ) = ± 2 (15) Note that differentiating the orientation matrix R with
q1 + q22 − 2b 2
respect to time yields
f12 sin (ψ ) R = S ( ω ) R
q32 = + 4af1 − 4ab3 cos (ψ ) + 4a 2 + b32 (16) (27)
sin (θ )
2
sin (θ )
ª0 −ω z ω y º
2) If q1 = q2 , implementing subtraction for Eq.(10) and « »
where S (ω ) = «ωz 0 −ω x »
Eq.(11), yields « −ω ω
¬ y 0 »¼
θ =0 (17) x

Then, Ȧ = (ωx ω y ω z ) can be expressed by ψ and θ

T
Eqs. (10), (12) can be rewritten as

z = q12 − b 2 + a sin (ψ ) (18) though Eq. (27)

ªcosθ 0º
2
q32 = ª¬ 2a cos (ψ ) − b3 º¼ + ª¬ z + a sin (ψ ) º¼
2
ªψ º
(19) Ȧ = ««0 1 »» « » (28)
θ
Then, substituting Eq.(18) into Eq. (19) gives «¬ − sin θ 0 »¼ ¬ ¼
q32 = 4a 2 + b32 + f 22 + 4af 2 sin (ψ ) − 4ab3 cos (ψ ) (20)
Let x = (ψ θ z ) , the following relation can be
T

where f 2 = q − b . 2
1
2
calculated by Eqs.(24), (25) , (26) and (28)
Lastly, given a set of limb lengths ( q1 q2 q3 ) , we can X = J r x (29)
solve for θ and ψ by using corresponding equations as ª0 z sec θ 2
tan θ º
mentioned above. Then the orientation matrix R can be « »
« −a sinψ 0 0 »
generated, leading to the solution of x, y and z by using «0 0 1 »
Eqs.(8), (7) and (18). where J r = « ».
«cosθ 0 0 »
For the proposed 2-RPU&SPR SPM in this paper, both «0 1 0 »
the inverse and the forward position analyses of the « »
mechanism can be calculated directly though analytical «¬ − sin θ 0 0 »¼
method, and the parasitic motions of the moving platform can Substituting Eq.(29) into Eq. (23) gives
be expressed in explicit way, which is extremely significant qa = J b x (30)
for the possible practical application of the mechanism. where J b = Jα J r is a 3h3 matrix.
C. Velocity analysis When the manipulator is away from singularities
Differentiating Eq.(2) with respect to time yields x = J b−1qa (31)
r = qi wi + Ȧi × qi wi − Ȧ × ai i = 1, 2,3
(21)
Then, substituting Eq.(31) into Eq. (29) gives
where qi is the velocity of the ith linear actuator, and
X = J r J b−1qa (32)
Ȧi represents the three-dimensional angular velocity of
link Ai Bi , “ͪ” denotes the cross product between vectors, D. Acceleration analysis
Differentiating Eq.(23) with respect to time, yields
and Ȧ = (ωx ω y ω z ) represents the three-
T
r = ( x z )
T
y
a Pi = ª« įiT º A + V T H V , i = 1,2,3
T

¬
( ai × įi ) »¼ i (33)
dimensional linear and angular velocity of the moving
platform, respectively. ª0 −δ z δy º
1 ª − įˆ įˆi2aˆi º
2
« »
The passive variable Ȧi can be eliminated by dot where H i = « i 2 » , į = «δ z
ˆ 0 −δ x » ,
multiplying both sides of Eq. (21) with wi , which gives qi ¬« −aˆi įˆi ˆ ˆ 2
qi aˆi įi + aˆi įi aˆi ¼» « −δ »
¬ y δx 0 ¼

830
a Pi is the acceleration of the actuated prismatic joint of the ith platforms, as shown in Fig.3. Given the distance between
centres of mass of the active part and point Ak on the active
limb. A = ( a İ ) is the acceleration of the moving platform,
T

platform is duk (k = 1, 2,3) . The linear / angular velocity and

a and İ are the linear acceleration and the angular
acceleration of centres of mass of active levers can be
acceleration of the moving platform. represented as vuk , auk , Ȧuk and İuk respectively. The
Differentiating Eq.(29) with respect to time, yields
counterparts of follower part from above can be
x + x T hx
A = J r  (34) represented vdk , adk , Ȧdk and İdk respectively. And the active
x = (ψ θ 
z ) and h = [ h1 h2 h3 h4 h5 h6 ] is the
T T lever and the follower rod respectively have the same angular
where, 
velocity and angular acceleration which equals to each
Hessian matrix of this mechanism. branched chain velocity and angular acceleration of the driver.
§0 0 0 · § − k cosψ 0 0·
¨ ¸ ¨ ¸
h1 = ¨ 0 2 z sec 2 θ tan θ sec 2 θ ¸ , h2 = ¨ 0 0 0¸ ,
¨0 ¸ ¨ zi vuk
© sec 2 θ 0 ¹ © 0 0 0 ¸¹
§0 0 0· § − sinψ 0 0· § − cosψ 0 0·
¨ ¸ ¨ ¸ ¨ ¸
h3 = h5 = ¨ 0 0 0 ¸ , h4 = ¨ 0 0 0 ¸ , h6 = ¨ 0 0 0¸ .
¨0 0 0¸ ¨ 0 0 0 ¸¹ ¨ 0 0 0 ¸¹ Active part
© ¹ © ©
Given the values of x = (ψ θ z ) and 
x = (ψ θ 
z)
T T
vdk
the acceleration of the mechanism can be calculated by
Eqs.(33), (34).
II. DYNAMICS ANALYSIS Lower part
o
A. Velocity and acceleration analysis of limbs xi
According to the Eq.(21), the velocity of point Ai , v Ai Fig. 3. Diagram of the limb
i=1,2,3 can be expressed as
v Ai = qi wi + Ȧi × qi wi i = 1, 2,3 (35) The linear velocity of centres of mass of follower levers
for RPU branches and SPR branch can be expressed as
In order to solve the angular velocity of link Ai Bi , Ȧi , follows
taking the cross product with wi on both sides of Eq.(35),
v dk = − qdk wˆ k J ω k X
(41)
leads to
wi × v Ai = qi Ȧi − qi wi ( Ȧi ⋅ wi ) i = 1, 2,3 (36) The velocity of follower levers for each branch can be
calculated though Eqs. (35), (37) and (41),
Note that, Ȧi ⋅ wi = 0 for RPU limbs, Eq.(36) can be
rewritten as § v dk · § −qdk wˆ k J ω k ·§ v · §v·
¨ ¸=¨ ¸¨ ¸ = J dk ¨ ¸ (42)
Ȧi = J ωi X i = 1,2 (37) © Ȧlk ¹ © J ωk ¹© Ȧ¹ © Ȧ¹
1    The linear acceleration of center of mass of follower levers
where Jωi = [ wi − wi ai ] . can be equaled as follow,
qi
adk = İlk × wk qdk + Ȧlk × (Ȧlk × qlk )qdk (43)
Differentiating Eq.(37) with respect to time, yields
1 The linear velocity of centres of mass of active levers for RPU
İi = ( w i × v Ai + wi × v Ai − qi Ȧi ) i = 1, 2 (38) branches and SPR branch can be expressed as follows
qi
§v· §v·
where İi is the angular acceleration velocity of link Ai Bi . (
vuk = − ( qi − quk ) wˆ k J ω k ¨ ¸ + wk wkT )
− wk wkT aˆ k ¨ ¸ (44)
© Ȧ¹ © Ȧ¹
The angular velocity of SPR limb can be expressed as
The velocity of active levers for each branch can be obtained
Ȧi = J ωi X i = 3 (39)
by Eq.(44)
1    § − ( qk − quk ) wˆ k J ω k + ·
where Jωi = ª wi − wi ai + qi wi wiT º¼ .
qi ¬ § vuk · ¨ ¸§ v · §v·
Differentiating Eq.(39) with respect to time, yields © Ȧlk ¹ ¨¨
( T
)
¨ ¸ = ¨ wk wk − wk wk aˆ k ¸ ¨ ¸ = J uk ¨ ¸
T

¸© Ȧ¹ © Ȧ¹
(45)
J ωk ¸
İi = 1 qi [(Ȧi × wi ) × v + wi × a + (Ȧi × wi ) × (Ȧ × ai ) + © ¹
Ȧi × ( İ × ai + Ȧ × (Ȧ × ai )) + qi (Ȧi × wi ) wiT Ȧ + i = 3 (40) The linear acceleration of center of mass of active levers can
v Ai wi wiT Ȧ + q j wi (Ȧi × wi )T Ȧ + qi wi wiT İ − v Ai Ȧi ] be equaled as follow,
Each limb of 2-RPU&SPR mechanism can be divided auk = İlk × wk ( qk − quk ) +
into two parts. The active part quk (k = 1,2,3) which is the upper (46)
Ȧlk × (Ȧlk × wk ) ( qk − quk ) + wk alk + 2(Ȧlk × w k )vlk
part of each limb connected with platform. The follower part
qdk which is the lower part of each limb connected with fixed B. Driving force calculation

831
 Given the gravity, inertia force and the moment of inertia
for active platform, active lever and follower lever as GA, fA, tA,
Guk, fuk, tuk, Gdk, fdk, tdk respectively, the Inertia matrix of
platform, active and follower lever related to coordinates of
A k k
each center of mass can be expressed as IA, Iuk, Idk. And each
body fixed coordinate of center of mass the lever relative to
the fixed coordinate system {B} can be expressed as a rotation
matrix,
RlkB = ( Rk wk × Rk wk ) (47)
The moment of inertia for each part of this mechanism related
to the fixed coordinate system can be expressed as

t A,uk ,lk = − B I A,uk ,lk İ A, uk ,lk − ȦA,uk ,lk × ( B

I A, uk ,lk ȦA,uk ,lk ) (48)
According to virtual work principle, yields (a) Positions of the mechanism

§v· 3 § §v · § v ··
FpT vl + Fall ¨ ¸ + ¦ ¨¨ Fdk ¨ dk ¸ + Fuk ¨ uk ¸ ¸¸ = 0 (49)
Ȧ
© ¹ k =1 © Ȧ
© dk ¹ © Ȧuk ¹ ¹
Whereas, Fp = ( Fp1 Fp 2 F p 3 ) is the driving force provided
T

by P pair on the three branches of mechanism, and

Fall = ( FAT + f AT + G TA TAT + t TA ) , Fdk = ( f dkT + Gdk
T T
t dk ) ,
(
Fuk = fukT + Guk
T T
t uk )
.
By the Eq.(15), (33) and (36)
§ v · §v · § v · §v · §v·
vl = J a ¨ ¸ , ¨ uk ¸ = J uk ¨ ¸ , ¨ dk ¸ = J dk ¨ ¸ (50)
Ȧ Ȧ
© ¹ © lk ¹ Ȧ Ȧ
© ¹ © lk ¹ © Ȧ¹
Substituting the Eq.(50) into Eq.(49), (b) Displacement of platform

§v· §v· 3 ª §v· § v ·º

FpT J a ¨ ¸ + Fall ¨ ¸ + ¦ « Fdk J dk ¨ ¸ + Fuk J uk ¨ ¸ » = 0 (51)
© Ȧ¹ © Ȧ ¹ k =1 ¬ © Ȧ¹ © Ȧ ¹¼
Substituting the Eq.(16) into Eq.(51),
3
FpT J l J o−1 + Fall + ¦ ( Fdk J dk + Fuk J uk ) = 0 (52)
k =1

Drawing driving force expression by arranging Eq.(52) ,

3
( ) ( ) ( )
T T
Fp = − J 3−×13 J oT FallT − J 3−×13 J oT ¦ J dk
T
FdkT + J uk
T
FukT (53)
k =1

II. NUMERICAL EXAMPLE

Set the structural parameters of the active platform and
fixed platforms for m=1.0mΔn=0.8mΔk=0.3m. Set mass of
(c) Lengths of limbs
active platform and each branch for mA=4.67kg, mu1=1.01kgΔ
mu2=1.01kg Δ mu3=1.08kg, md1=1.30kg Δ md2=1.30kg Δ
A
md3=1.40kg. Set IA=diag(0.297, 0.281, 0.155),
1 2 3
Iu1= Iu2=diag(0.073, 0.073, 0.00007), Iu3=diag(0.088, 0.088,
1 2 3
0.00007), Id1= Id2=diag(0.1, 0.1, 0.00028), Id3=diag(0.127,
0.1, 0.0003). Given initial , , ZA as output pose parameters of
o o o 2 o 2
active platform are 0 , 0 , 1.5m, accelerations are 3 /s , 2 /s ,
2
0.02m/s .
The displacement of the active platform, driving limb
length, velocity, acceleration and driving force of each
branches versus time shown in Fig.2 are obtained by Matlab
simulated mathematical model of the dynamics.

(d) Velocities of limbs

832
 FST11-LYM), and the Natural Science Foundation of Tianjin,
China (12JCYBJC12200, 13JCYBJC17700).

REFERENCES
[1] K. H. Hunt, “Structural kinematics of in-parallel-actuated robot arms,”
ASME Trans., J. Mech. Trans. Autom. Des. 105(1983) 705–712.
[2] A. Sokolov, P. Xirouchakis, “Dynamics analysis of a 3-DOF parallel
manipulator with R–P–S joint structure,” Mechanism and Machine
Theory 42 (2007) 541–557.
[3] X. Liu, J. Kim, A new spatial three-DoF parallel manipulator with high
rotational capability, IEEE/ASME Translations on Mechatronics 10(5)
(2005) 502–512.
[4] D. Zhang, Z. Bi, B. Li, Design and kinetostatic analysis of a new parallel
manipulator, Robotics and Computer-Integrated Manufacturing 25 (2009)
782–791.
(e) Accelerations of limbs [5] F. Xie, X. Liu, J. Wang, A 3-DOF parallel manufacturing module and its
kinematic optimization, Robotics and Computer-Integrated
Manufacturing 28 (2012) 334–343.
[6] E. Cuan-Urquizo, E. Rodriguez-Leal, Kinematic analysis of the 3-CUP
parallel mechanism, Robotics and Computer-Integrated Manufacturing 29
(2013) 382-395.
[7] J. A. Carretero, R. P. Podhorodeski, M. A. Nahon, Kinematic analysis and
optimization of a new three degree-of-freedom spatial parallel
manipulator, ASME Trans., J. Mech. Des. 122(2000) 17–24.
[8] Y. Li, Q. Xu, Kinematic analysis of a 3-PRS parallel manipulator,
Robotics and Computer-Integrated Manufacturing 23 (2007) 395-408.
[9] J. Dai, Z. Huang, H. Lipkin, Mobility of overconstrained parallel
mechanisms, ASME, J. Mech. Des.,128(1) (2006) 220–229.
[10]R. D. Gregorio, The 3-RRS wrist: a new, simple and non-overconstrained
spherical parallel manipulator, ASME Trans., J. Mech. Des. 126(2004)
850–855.
[11] L.W. Tsai, Solving the Inverse Dynamics of a Stewart-Gough
Manipulator by the Principle of Virtual Work, ASME Trans., J. Mech.
Des. 122(2000) 3-9.
[12] C.D. Zhang, S.M. Song, An efficient method for inverse dynamics of
(f) Driving forces of limbs manipulators based on the virtual work principle, J. Rob. Syst.10 (5)
Fig. 3. The kinematics and dynamics graph of the parallel mechanism (1993) 605–627.
[13] S. Staicu, D. Zhang, R. Rugescu, Dynamic modelling of a 3-DOF
parallel manipulator using recursive matrix relations. Robotica 24(1)
(2006) 125–130.
IV. CONCLUSIONS [14] X.-J. Liu, X. Tang, J. Wang, HANA: a novel spatial parallel manipulator
with one rotational and two translational degrees of freedom. Robotica
This paper proposes a new over-constrained 3-DOF SPM
23(2) (2005) 257–270.
2-RPU & SPR with one translational and two rotational DOFs. [15] Z. Geng, L. S. Haynes, J.D. Lee, Carroll, R.L.: On the dynamic model
Both the inverse and the forward position analyses are and kinematics analysis of a class of Stewart platforms. Robot. Auton.
addressed, and the analytical solutions are obtained for these Syst.9 (1992) 237–254.
two problems, unlike most parallel robots, the proposed SPM [16] B. Dasgupta, T. S. Mruthyunjaya, A Newton–Euler formulation for the
inverse dynamics of the Stewart platform manipulator. Mech. Mach.
has only one unique real answer to the invers and forward Theory 34(1998) 711–725.
kinematics. Therefore, both the path planning and the control [17] J. Wang, X.-J. Liu, Analysis of a novel cylindrical 3-DoF parallel robot.
problems of the mechanism are very simple. Additionally, the Robot. Auton. Syst.42 (2003) 31–46.
velocities and the accelerations of the moving platform and [18] S. Staicu, D.C. Carp-Ciocardia, Dynamic analysis of Clavel’s Delta
each limb are obtained. Finally, dynamics of the mechanism is parallel robot. In: Proceedings of the IEEE International Conference on
Robotics & Automation, Taipei, Taiwan (2003) pp. 4116–4121.
derived based on the principal of the virtual work. The results [19] K. Miller, R. Clavel, The Lagrange-Based Model of Delta-4 Robot
are illustrated by numerical examples. Based on the Dynamics, Roboter systeme, (8)1992 49–54.
kinematics and dynamics analyses of the mechanism, the [20] X. Z. Zheng, R. Nakashima, T. Yoshikawa, On dynamic control of
stiffness performance analysis and the kinematic optimization finger sliding and object motion in manipulation with multi fingered
hands, IEEE Trans. Robot. Automat., (16) 2000 469–481.
of the mechanism will be investigated in the future work.
[21] V. Parenti-Castelli, R. Di Gregorio, A new algorithm based on two
extra-sensors for real-time computation of the actual configuration of
generalized Stewart–Gough manipulator. J. Mech. Des. 122(2000) 294–
ACKNOWLEDGMENT 298.
This work was supported by the National Natural Science
Foundation of China (NSFC) (Grant No. 51205289 and
51275353), Macao Science and Technology Development
Fund (108/2012/A3), Research Committee of University of
Macau (MYRG183(Y1-L3)FST11-LYM, MYRG203(Y1-L4)-

833