Sie sind auf Seite 1von 7

13th World Congress in Mechanism and Machine Science, Guanajuato, Mxico, 19-25 June, 2011 A12_364

1
A Fully-Isotropic Parallel Orientation Mechanism
Chin-Hsing Kuo
*
Jian S. Dai


National Taiwan University of Science and Technology Kings College London
Taipei, Taiwan London, United Kingdom
AbstractA new three-degree-of-freedom fully-
isotropic parallel orientation mechanism is presented in
this paper. The output link (end-effector) is connected to
the base by one spherical joint and by another three in-
parallel legs, respectively. In each leg, one revolute joint
attached to the base is selected as the active joint. It
shows that, as a fact of being a fully-isotropic parallel
mechanism, actuating any of the active joints can make
the end-effector rotating about the axis of the actuated
joint with a same speed of the joints at any instant. In
contrast to the existing fully-decoupled parallel
orientation mechanisms, the mechanism proposed here is
non-overconstrained and contains no redundant
actuations and homokinetic joints. The inverse and
direct kinematics problems of this mechanism are
analyzed. Meanwhile, the geometric reasoning of the
position kinematics solutions is demonstrated.
Keywords: decoupled motion, fully decoupled, isotropy, parallel
wrist mechanism, spherical parallel manipulator
I. Introduction
Parallel orientation mechanism, also known as the
spherical parallel manipulator or parallel wrist
mechanism, is a mechanism that, based on in-parallel
actuations, generates a purely rotational motion with
respect to the fixed coordinate over its output link. It has
been widely applied into various industrial or medical
environments such as antenna orienting, surgical
manipulation, robotic wrists, etc.
Generally, a kinematically non-redundant parallel
orientation mechanism can be constructed in two
different ways. For the first type, an n-degree-of-
freedom (DOF) parallel orientation mechanism is made
up of the base, an output link (or so-called the end-
effector in parallel manipulators), and n serial chains (or
so-called the legs in parallel manipulators) that in-
parallel connect the end-effector to the base. Through the
structural constraints provided by the n serial chains, the
output link is prohibited from moving in any direction so
it can only rotate about a fixed point in space. Normally,
one joint in each chain is selected to be actuated so that
the n-DOF pure rotation can be controlled by the n serial
chains together. Based on this concept, many parallel
orientation mechanisms have been proposed and
*
chkuo717@mail.ntust.edu.tw

jian.dai@kcl.ac.uk
extensively studied. For example, Kong and Gosselin [1,
2] proposed the type synthesis of 3-DOF spherical
parallel manipulators by using screw theory [1, 2]. Fang
[3] also addressed a screw-theoretic-based method for the
type synthesis of the 3-DOF spherical parallel
manipulator with same legs. Karouia and Herv [4]
synthesized a group of 3-DOF spherical parallel
mechanisms having asymmetric structures. Hess-Coelho
[5] summarized a list of possible structures for parallel
wrists and suggested a qualitative procedure for
evaluating the wrist mechanisms. Many other parallel
orientation mechanisms using this concept have been
exhaustively studied [6-16]. Further, some special
parallel orientation mechanisms have been built by using
parallelograms as the serial chains [17, 18].
The second type is based on the previous concept but
one additional joint is employed to completely constrain
the motion of the output link. The added joint, normally
a universal or spherical joint, articulate the output link to
the base such that the available motion space of the
output link is completely defined by the joint. The n
serial chains, each containing one actuated joint, are then
structured in a way that does contribute other motion
constraints to the output link. The most well-known
parallel orientation mechanism for this type may be the
3-DOF 3SPS/S (three spherical-prismatic-spherical-joint
legs and one additional spherical joint connecting the
end-effector and the base) parallel manipulator. Some
literatures have studied this mechanism exhaustively [19-
21]. Some other 3-DOF parallel orientation mechanisms
using a constrained joint can be found in [5]. In addition,
there have been few parallel orientation mechanisms with
two DOFs synthesized by using this concept [22, 23].
A robotic manipulator is said to be at an isotropic
configuration while its actuation motions and forces can
be best reflected onto the end-effector at this
configuration. In kinematics, when the manipulator is at
an isotropic configuration, there exists a one-to-one
correspondence between the velocity spaces of the
actuated joints and the end-effector. Further, the
manipulator is said to be fully-isotropic if it is isotropic
throughout the entire workspace [24, 25]. There have
been some isotropic parallel wrist mechanism proposed
in literatures. In 2005, Gogu presented a family of fully-
isotropic over-constrained 2-DOF parallel wrist
mechanisms [26]. He further proposed two classes of 3-
DOF parallel wrist mechanisms with fully-isotropic
configurations achieved by introducing the actuation
13th World Congress in Mechanism and Machine Science, Guanajuato, Mxico, 19-25 June, 2011 A12_364
2
redandancy [27] or the homokinetic joints [15].
Basically, depending on the mobility of the mechanism,
the above fully-isotropic parallel orientation mechanisms
are structured by using two or three constrained legs
without passive constrained joint. The motion space of
the end-effector is cooperatively constrained by the legs.
A new 3-DOF fully-isotrpic parallel orientation
mechanism is put forward in this paper. The mechanism
is non-overconstrained and without actuation redundancy
and homokinetic joints. The proposed parallel
manipulator consists of three active legs and one passive
spherical joint that connects the end-effector to the base.
In each leg, a relovute joint (R-joint) mounted on the
base is selected for actuation. Under a specific joint
configuration, the three rotational DOFs of the end-
effector is fully decoupled. The angular velocity of either
actuated R-joint will make the end-effector rotating about
the same axis of this R-joint with a same speed. In what
follows, the structure and geometric arrangement of this
mechanism will be introduced. The solving of inverse
and direct kinematics problems will be provided. And,
the geometric reasoning of the inverse and direct
kinematics problems will be discussed.
II. Description of the Mechanism
The proposed fully-isotropic parallel wrist mechanism
is as shown in Fig. 1. The end-effector is connected to
the base by one spherical joint (S) and by another three
in-parallel legs, respectively. Each leg possesses the
same topological structure, which is composed of, as read
from base to the end-effector, a revolute joint (R), a
universal joint (U), and three consecutive prismatic joints
(P). According to the Grbler-Kutzbach criterion, the
mobility of the mechanism, m, can be calculated as:
3 ) 1 ( = + =
i
f j n m , (1)
where = 6 is the degrees of freedom of the space, n = 14
is the number of links, j = 16 is the number of joints, and
21 =
i
f is the total degrees of freedom of the joints.
As a result, the mobility of the mechanism is three, and
this mechanism is non-overconstrained. Besides, it can
be quickly realized that the mechanism can output a 3-
DOF spherical motion at the end-effector which can
rotate about the center of the S-joint only.
For analyzing the kinematics of the mechanism, two
Cartesian coordinate systems A(x, y, z) and B(u, v, w) are
attached to the base and end-effector, respectively. As
shown in Fig. 1, in order to simplify the analysis, we
assume that the origin of the fixed coordinate A is located
at the center of the S-joint O, and the origin of the
moving coordinate B is originally coincident with O.
Since the end-effector can only rotate about the center of
the S-joint, the origin of the moving coordinate will be
always pivoted at O. Further, we assume that the moving
y
z
x
passive joint end-effector active joint
u
v
w
O
Fig. 1. A 3-DOF fully-isotropic parallel orientation mechanism
coordinate is coincident with the fixed coordinate at the
initial position, i.e., the u-, v-, and w-axis are initially
pointing at the x-, y-, and z-directions, respectively.
Furthermore, for achieving the fully-isotropic
condition, the mechanism is initially configured in a way
that meets the following geometric and actuation
conditions:
i) The axes of the three R-joints are coincident with the
x-, y-, z-axis, respectively.
ii) In each leg, the axis of the R-joint passes through the
center of the U-joint.
iii) In each leg, the two axes of the U-joint are
perpendicular to the axis of the R-joint of this leg.
iv) In each leg, the three P-joints can be arbitrarily
deposed provided that all the joint directions are not
parallel to each other.
v) In each leg, the R-joint is selected as the actuated
joint, whereas all the others are passive.
The above geometric arrangements are as shown in Fig. 1.
Note that conditions i) and ii) remain valid during the full
cycle of motion but condition iii) may be destroyed after
the mechanism has an infinitesimal displacement.
In accordance with the above geometric conditions, we
further assemble the mechanism at a special initial
configuration for simplifying the analysis. Without
violating condition iii), we align the two axes of the U-
joint in each leg with the two axial directions other than
the R-joints. For example, if the R-joint is directed at
the x-axis, the two axes of the U-joint in this leg are
13th World Congress in Mechanism and Machine Science, Guanajuato, Mxico, 19-25 June, 2011 A12_364
3
initially aligned with the y- and z-directions, respectively.
Also, without violating condition iv), we place the three
P-joints in each leg pointing at the three axial directions
of the fixed coordinate, respectively. Since the
kinematics of a prismatic joint is independent of the joint
position, the articulation order of the three allocated joint
orientation can be freely altered. Accordingly, a feasible
joint configuration is defined as illustrated by the
structural graph in Fig. 2. In this graph, the characters
adjacent to the line segments represent the type of the
joints whereas the subscripts of the characters denote the
orientation(s) of the joint at the initial mechanism
configuration. The corresponding mechanism is shown
in Fig. 1.
Fig. 2. Structural graph of the mechanism in Fig. 1
III. Geometry Analysis
In this chapter, the fully-isotropic condition of this
mechanism is examined via the geometry analysis of the
mechanism. We will show that the mechanism is fully-
isotropic, that is, actuating any of the active R-joints can
make the end-effector rotating about the axis of this joint
with a same speed of the joints at any instant.
Since the three legs have the same topology and are
configured by a similar way, here we only analyze the leg
mounted at the x-axis for illustration. The analysis of the
other two legs can be done by following the same logic.
First of all, Fig. 3 depicts the geometry of the leg
mounted on the x-axis. In this figure, A
x
indicates the
point at which the U-joint locates. E is a point on the
end-effector at which the three P-joints attached on the
end-effector intersect. a
x
is the position vector measured
from point O to A
x
, whereas e is the position vector
measured from point O to E. d
1,x
, d
2,x
, and d
3,x
are the
position vectors representing the three consecutive P-
joints in the leg, where d
1,x
is measured from A
x
to a point
R
x
, d
2,x
is measured from R
x
to a point B
x
, and d
3,x
is
measured from E to B
x
. d
3,y
and d
3,z
denote two P-joints
of the other two legs which are attached to the end-
effector and are both measured outward from E. u
2,x
is
the unit vector of first axis of the U-joint connected with
the R-joint, whereas u
3,x
is the unit vector of the second
axis of the U-joint connected with the first P-joint (as
read from base).
Ax
is the plane parallel to yz-plane and
containing point A
x
. Lying on
Ax
, y and z are two unit
vectors passing through A
x
and pointing at the y- and z-
directions, respectively.
x
is the angle between z and
d
2,x
, which represents the angular displacement of the
actuated joint.
z
x

x
w
u
v
E
O
d
1,x
d
3,y
d
3,z
e
a
x
u
2,x
u
3,x
z'
A
x
plane
Ax
// yz-plane
and containing A
x
y'
B
x
y
d
2,x
R
x
d
3,x
Fig. 3. The geometry of the leg mounted on the x-axis
Owing to the specific joint configuration, there are
some geometric constraints pertinent to these position
vectors. First, due to the geometric relationship between
the R-joint and the U-joint, point A
x
will be a fixed point
in space and d
2,x
will always lie on
Ax
. Further, since the
three consecutive P-joints are configured with a
orthogonal pose, vector d
1,x
, will be always perpendicular
to d
2,x
and so do d
2,x
to d
3,x
. Therefore, R
x
and B
x
will be
the two perpendicular foots between the three vectors.
Furthermore, thanks to the intentionally-defined moving
coordinate and the orthogonal condition among all the P-
joints in the mechanism, the following conditions will be
satisfied at any instant:
x , 3
// d u , (2a)
x y , 1 ,
//
3
// d d v , (2b)
x z , 2 ,
//
3
// d d w , (2c)
where u, v, and w are the unit vectors along the u-, v-,
and w-axes of the moving coordinate and symbol //
indicates that the vectors in the both sides of the symbol
are parallel. Particularly, it is noticed that in Eq. (2b) the
axial direction of the first P-joint is parallel the v-axis.
We further inspect this geometric relationship by together
considering the u
2,x
on plane
Ax
. Referring to Fig. 4,
u
, 1 x
d is the unit vector of d
1,x
. Since
u
, 1 x
d is always
coincident with the second axis of the U-joint (i.e.,
x x , 3
u
, 1
u d = in Fig. 4), it will be always perpendicular to
u
2,x
. Recall that u
2,x
always lies on
Ax
. Therefore,
provided that
u
, 1 x
d is not perpendicular to
Ax
, the
projective vector of
u
, 1 x
d on
Ax
, k
Ax
, will be always
13th World Congress in Mechanism and Machine Science, Guanajuato, Mxico, 19-25 June, 2011 A12_364
4
perpendicular to u
2,x
. Since
u
, 1 x
d is parallel to the v-axis
of the moving coordinate and
Ax
is parallel to the yz-
plane of the fixed coordinate, it turns out that the
projection of vector v on the yz-plane will be always
perpendicular to the projection of u
2,x
on the yz-plane.
Further, from Fig. 4, we can observe that the angle
between k
Ax
and y,
yz
v
u , is identical to that between u
2,x
and z,
x
.
Fig. 4. Geometrical interpretation of the input angle
The above geometric condition shows an interesting
input-output relationship between the actuated joint and
the end-effector. First, it proves that the projection of the
v-axis of the moving coordinate on the yz-plane will be
always perpendicular to the unit vector of the first axis of
the U-joint on the x-axis. In the other word, the
projective displacement of the v-axis on the yz-plane will
be completely defined by the displacement of the R-joint
at the x-axis. Further, the amount of this projective
displacement (
yz
v
u ) will be equal to the amount of the
angular displacement of the actuated joint (
x
). Therefore,
by following the same deduction above, we can find
other similar input-output relationships in the other two
legs. Consequently, a one-to-one correspondence
between the displacement spaces of the actuated joints
and the end-effector can be expressed in a matrix form as
(
(

(
(

=
(
(
(

z
y
x
xy
u
zx
w
yz
v
o
o
o
u
u
u
1 0 0
0 1 0
0 0 1
, (3)
where ) , , (
xy
u
zx
w
yz
v
u u u represents the projective
displacements of the v-, w-, and u-axes on the yz-, zx-,
and xy-planes and (
x
,
y
,
z
) indicates the angular
displacements of the actuated joint on the x-, y-, and z-
axes. Differentiating Eq. (3) with respect to time, the
following velocity equation is derived:
q J = , (4)
where
T
] , , [
z y x
e e e = is the velocity vector of the end-
effector expressed in the fixed coordinate,
T
] , , [
z y x
o o o = q expresses the joint velocities of the
three R-joints mounted on the x-, y-, and z-axes, and J is
the Jacobian matrix which is a 3 3 identity matrix. It
turns out that this mechanism is fully-isotropic and the
input joint velocities can be equally reflected onto the
velocity space of the end-effector expressed in the fixed
coordinate.
IV. Inverse Kinematics
For the inverse kinematics problem, the location of the
end-effector with respect to the fixed coordinate is given
and the angular displacements of the R-joints,
x
,
y
,
z
,
are to be found. Because the end-effector can only rotate
about the S-joint, the location of the end-effector is
completely defined by its rotation matrix. Therefore, let
us consider the rotation matrix,
B
A
R , which transforms
the fixed coordinate A to the moving coordinate B by the
direction cosines of the unit vectors u, v, and w; that is
(
(

= =
z z z
y y y
x x x
B
A
w v u
w v u
w v u
R ] , , [ w v u .
As shown in Fig. 4, we have learnt that the angular
displacement,
x
, is fully dependent on the projective
vector, k
Ax
. It implies that as long as k
Ax
is determined
the angular displacement
x
can be obtained accordingly.
Therefore, from the geometrical interpretation in Fig. 4,
the angle between k
Ax
and y can be calculated by the
equation below provided that k
Ax
is a non-null projective
vector of
u
, 1 x
d on
Ax
:
|
|
.
|

\
|
'
'
=

y k
y k
cos
1
Ax
Ax
yz
v
u . (5)
Since v d k =
u
, 1
//
x Ax
and k
Ax
is on the yz-plane, we can
readily realize
T
] , , 0 [
z y Ax
v v = k . Then, introducing the
facts of
T
] 0 , 1 , 0 [ = ' y and
yz
v
x
u o = into Eq. (5), the input
angle rotating about the x-axis,
x
, can be obtained as
|
|
|
.
|

\
|
+
=

2 2
1
cos
z y
y
x
v v
v
o . (6)
Similarly, by applying this analysis procedure for the
other legs, the other two input angles rotating about the y-
and z-axes can be derived as
13th World Congress in Mechanism and Machine Science, Guanajuato, Mxico, 19-25 June, 2011 A12_364
5
|
|
.
|

\
|
+
=

2 2
1
cos
z x
z
y
w w
w
o (7)
and
|
|
|
.
|

\
|
+
=

2 2
1
cos
y x
x
z
u u
u
o . (8)
Hence, when the rotation matrix
B
A
R is given, there
are two solutions in each of Eqs. (6), (7) and (8).
Therefore, for a given location of the end-effector, there
are two corresponding configurations in each leg. It
therefore concludes that there are eight possible solutions
for the inverse kinematics problem for this mechanism.
In addition to the input angles, the link lengths of the
three P-joints in each leg are worth to be computed.
According to Fig. 3, the loop equation of the leg mounted
on the x-axis can be written as
x x x x
a e d d d = +
, 3 , 2 , 1
. (9)
Note that a
x
is a given geometric parameter and e can be
obtained when the rotation matrix
B
A
R and an arbitrary
point E are defined. Furthermore, due to the geometric
constraint, the following equations are held in any
configuration:
v d
x x
d
, 1 , 1
= , (10)
w d
x x
d
, 2 , 2
= , (11)
u d
x x
d
, 3 , 3
= . (12)
where d
1,x
, d
2,x
, and d
3,x
are the representative lengths of
vectors d
1,x
, d
2,x
, and d
3,x
and the unit vectors u, v, and w
have been given by the rotation matrix. Substituting Eqs.
(10), (11), and (12) into (9) and expressing Eq. (9) in
three scalar equations, we can conclude a system of three
linear equations with three unknown: d
1
, d
2
, and d
3
. As a
result, the representative lengths of the three P-joints can
are solved.
V. Direct Kinematics
For the direct kinematics problem, the angular
displacements of the R-joints are given and the location
of the end-effector with respect to the fixed coordinate is
to be found. Since the end-effector only possesses a 3-
DOF rotational motion centered at the S-joint, the direct
kinematics problem is reduced to the solving of the
rotation matrix,
B
A
R .
A. Algebraic solution
The algebraic solution of the rotation matrix can be
figured out based on the special geometrical relationship
in the mechanism. Referring to Fig. 3, we know that
based on the defined coordinate systems d
1,x
is always
parallel to the v-axis of the moving coordinate. Also,
subject to the structural constraint of the U-joint, d
1,x
is
always perpendicular to u
2,x
. Considering the above two
geometric constraints together, we can readily realize that
all possible solutions of d
1,x
will form a set of vectors
lying on the plane that passes A
x
and is normal to u
2,x
.
For illustrating this geometry, Fig. 5 is provided. In this
figure, plane
Dx
is defined by its normal vector u
2,x
and
point A
x
, and
u
, 1 x
d , which is to be found, should locate on

Dx
.
Fig. 5. Geometry interpretation for the direct kinematics problem
Therefore, when the actuated joint angle,
x
, is given,
the vector u
2,x
can be derived as
T
, 2
] cos , sin , 0 [
x x x
o o = u . (13)
Further, because k
Ax
is a vector lying
Ax
and being
perpendicular to u
2,x
, it can be expressed as
T
] sin , cos , 0 [
x a x a Ax
o o = k , (14)
where |
a
| represents the length of the vector. Since k
Ax
is the projective vector of
u
, 1 x
d on plane
Ax
,
u
, 1 x
d can be
written as
T u
, 1
] sin , cos , [
x a x a x
a o o = d . (15)
where a is a variable defining the x-component of
u
, 1 x
d .
Because
u
, 1 x
d is a unit vector, Eq. (15) should follow the
identity equation of 1
2 2
= + a
a
.
Similarly, we can carry out the above analysis in the
other two legs and obtain two equations in terms of
u
, 1 y
d
13th World Congress in Mechanism and Machine Science, Guanajuato, Mxico, 19-25 June, 2011 A12_364
6
and
u
, 1 z
d (which express the vector of the first P-joint in
the leg amounted on the y- and z-axes, respectively) and
two associated identity equations. Therefore, having
known the fact of ) , , ( ) , , (
u
, 1
u
, 1
u
, 1 y x z
d d d w v u = , we obtain
T
] , sin , cos [ c
z c z c
o o = u , (16)
T
] sin , cos , [
x a x a
a o o = v , (17)
T
] cos , , sin [
y b y b
b o o = w , (18)
subject to the three identity equations below:
0 1
2 2
= + a
a
, (19)
0 1
2 2
= + b
b
, (20)
0 1
2 2
= + c
c
. (21)
Further, since u, v, and w are three orthogonal vectors,
the orthogonal conditions 0 = = = w v w u v u should
be satisfied. Accordingly, based on Eqs. (16), (17), and
(18), the following three equations are obtained:
0
13 12 11
= + +
a c a c
c e e a e , (22)
0
23 22 21
= + +
b c c b
c e b e e , (23)
0
33 32 31
= + +
b a a b
e b e a e . (24)
where e
ij
, i, j = 1 to 3, are the coefficients given by
x
,
y
,

z
. Equations (19) through (24) form a system of six
second-degree polynomials in six unknowns: a, b, c,
a
,

b
, and
c
. When the six unknown are solved, the
elements in the rotation matrix can be computed through
Eqs. (16), (17), and (18). Although the fundamental
algebra theorem indicates that this system has at most
64 2
6
= solutions, a 3-homogeneous formulation

of the
system shows that the maximum number of solutions of
the homogeneous system can be reduced to 16. While
there will be half of solutions conflicting the right-hand
rules between u, v, and w, we can conclude that the direct
kinematics problem of this mechanism has at most 8
possible solutions.
B. Geometric reasoning
The algebraic solution of the direct kinematics for this
mechanism can be realized by a geometrical reasoning.
In Fig. 3, we showed that the v-axis of the moving
coordinate must be parallel to d
1,x
. In Fig. 5, we learnt
that d
1,x
must lie on a plane whose normal vector u
2,x
is

Based on partitioning the six variables into three groups, (a,


a
), (b,
b
),
and (c, c), the 3-homogeneous Bezout number can be derived as 16.
defined by the input angle
x
. With these two geometric
constraints, we can understand that the v-axis must lie on
a plane, say
v
, which should pass origin O and should be
perpendicular to u
2,x
. In the same logic, the u- and w-
axes must respectively lie on another two planes, say
w
and
u
, that are determined by origin O and by u
2,y
and
u
2,z
(the two vectors defined by
y
and
z
), respectively.
Accordingly, the direct kinematics problem of this
mechanism is mapped onto a geometric problem, that is:
Given three planes (
u
,
v
, and
w
) with common
intersection point O, identify three vectors, one from each
plane (say u on
u
, v on
v
, and w on
w
), such that
i) all identified vectors pass through point O;
ii) all identified vectors are perpendicular to each other;
and
iii) all identified vectors obey the right-hand rule together
(i.e., w v u = ).
Note that the above three conditions should be satisfied
simultaneously. Apparently, this geometric problem is
basically the searching of three, one from each plane,
intersecting orthogonal lines. When the lines are found,
each line will correspond two solutions for u, v, or w (i.e.,
two vectors pointing at opposite directions in the same
line). Figure 6 demonstrates such a line-searching
problem. As shown, in a general, nonsigular geometry,
we can find that there are at most two lines on each plane,
which are lines L
u1
and L
u2
on
u
, L
v1
and L
v2
on
v
, and
L
w1
and L
w2
on
w
. Lines (L
u1
, L
v1
, L
w1
) and (L
u2
, L
v2
, L
w2
)
forms two set of orthogonal lines, each of which
corresponding to one solution space for accommodating
the three vectors, u, v, and w. Therefore, there will be
16 2
2
1
2
1
2
1
= C C C possible combinations for (u, v, w).
However, only 8 of the sixteen possible solutions are
feasible due o the right-hand rule between u, v, and w.
This geometric validation proves the algebra solution as
we derived above.

v

w
O u
w
v
L
w1
L
w2
L
u2
L
u1
L
v1
L
v2
Fig. 6. Geometrical reasoning for the direct kinematics problem
13th World Congress in Mechanism and Machine Science, Guanajuato, Mxico, 19-25 June, 2011 A12_364
7
VI. Conclusions
A fully-isotropic parallel orientation mechanism has
been introduced in this paper. The proposed 3-DOF
parallel manipulator is non-overconstrained, and it
possesses a one-to-one correspondence between the
input-output velocities that actuating any of the active
joints can make the end-effector rotating about the axis of
the actuated joint with a same speed of the joints at any
instant. For studying this mechanism, the geometry
analysis has been carried out and the fully-isotropic
condition is proven accordingly. Furthermore, the
inverse and direct kinematics problems were studied. For
the inverse kinematics analysis, we found that there are at
most 8 possible solutions. For the direct kinematics
problem, it has shown that there are at most 16 possible
solutions. The geometric reasoning for the position
kinematics has been furnished for validating the algebraic
solutions.
References
[1] Kong, X. and Gosselin, C.M. Type synthesis of three-degree-of-
freedom spherical parallel manipulators, The International Journal
of Robotics Research, 23(3):237-245, 2004.
[2] Kong, X. and Gosselin, C.M. Type synthesis of 3-dof spherical
parallel manipulators based on screw theory, ASME Transactions,
Journal of Mechanical Design, 126(1):101-126, 2004.
[3] Fang, Y. and Tsai, L.-W. Structure synthesis of a class of 3-dof
rotational parallel manipulators, IEEE Transactions on Robotics
and Automation, 20(1):117-121, 2004.
[4] Karouia, M. and Herv, J.M. Asymmetrical 3-dof spherical parallel
mechanisms, European Journal of Mechanics - A/Solids, 24(1):47-
57, 2005.
[5] Hess-Coelho, T.A. Topological synthesis of a parallel wrist
mechanism, ASME Transactions, Journal of Mechanical Design,
128(1):230-235, 2006.
[6] Gosselin, C.M. and Lavoie, E. On the kinematic design of spherical
three-degree-of-freedom parallel manipulators The International
Journal of Robotics Research, 12(4):394-402 1993.
[7] Karouia, M. and Herv, J.M., 2000, A three-dof tripod for
generating spherical rotation, in Lenari, J., Stanii, M. M., Eds,
Advances in robot kinematics, London, Kluwer Academic, pp. 395-
402.
[8] Di Gregorio, R. Kinematics of a new spherical parallel manipulator
with three equal legs: The 3-urc wrist, Journal of Robotic Systems,
18(5):213-219, 2001.
[9] Di Gregorio, R. A new parallel wrist using only revolute pairs: The
3-ruu wrist, Robotica, 19(3):305-309, 2001.
[10] Di Gregorio, R. A new family of spherical parallel manipulators,
Robotica, 20(4):353-358, 2002.
[11] Di Gregorio, R. Kinematics of the 3-UPU wrist, Mechanism and
Machine Theory, 38(3):253-263, 2003.
[12] Di Gregorio, R. The 3-RRS wrist: A new, simple and non-
overconstrained spherical parallel manipulator, ASME
Transactions, Journal of Mechanical Design, 126(5):850-855, 2004.
[13] Di Gregorio, R. Kinematics of the 3-RSR wrist, IEEE Transactions
on Robotics, 20(4):750-753, 2004.
[14] Karouia, M. and Herv, J.M. Non-overconstrained 3-dof spherical
parallel manipulators of type: 3-RCC, 3-CCR, 3-CRC, Robotica,
24(1):85-94, 2006.
[15] Gogu, G. Fully-isotropic three-degree-of-freedom parallel wrists, In
Proceedings of IEEE International Conference on Robotics and
Automation, Roma, Italy, 10-14 April 2007.
[16] Enferadi, J. and Tootoonchi, A.A. A novel spherical parallel
manipulator: Forward position problem, singularity analysis, and
isotropy design, Robotica, 27(5):663-676, 2009.
[17] Baumann, R., Maeder, W., Glauser, D., and Clavel, R. The
pantoscope: A spherical remote-center-of-motion parallel
manipulator for force reflection, In Proceedings of IEEE
International Conference on Robotics and Automation, pp. 718-723,
Albuquerque, New Mexico, USA, 20-25 April 1997.
[18] Vischer, P. and Clavel, R. Argos: A novel 3-dof parallel wrist
mechanism, The International Journal of Robotics Research,
19(1):5-11, 2000.
[19] Innocenti, C. and Parenti-Castelli, V. Echelon form solution of
direct kinematics for the general fully-parallel spherical wrist,
Mechanism and Machine Theory, 28(4):553-561, 1993.
[20] Wohlhart, K. Displacement analysis of the general spherical
stewart platform, Mechanism and Machine Theory, 29(4):581-589,
1994.
[21] Huang, Z. and Yao, Y.L. A new closed-form kinematics of the
generalized 3-dof spherical parallel manipulator, Robotica,
17(5):475-485, 1999.
[22] Carricato, M. and Parenti-Castelli, V. A novel fully decoupled two-
degrees-of-freedom parallel wrist, The International Journal of
Robotics Research, 23(6):661-667, 2004.
[23] Gallardo, J., Rodrguez, R., Caudillo, M., and Rico, J.M. A family
of spherical parallel manipulators with two legs, Mechanism and
Machine Theory, 43(2):201-216, 2008.
[24] Carricato, M. and Parenti-Castelli, V. Singularity-free fully-
isotropic translational parallel mechanisms, International Journal of
Robotics Research, 21(2):161-174, 2002.
[25] Gogu, G. Structural synthesis of fully-isotropic translational
parallel robots via theory of linear transformations, European
Journal of Mechanics - A/Solids, 23(6):1021-1039, 2004.
[26] Gogu, G. Fully-isotropic over-constrained parallel wrists with two
degrees of freedom, In Proceedings of IEEE International
Conference on Robotics and Automation, pp. 4014- 4019,
Barcelona, Spain, 18-22 April 2005.
[27] Gogu, G. Fully-isotropic redundantly-actuated parallel wrists with
three degrees of freedom, In Proceedings of the ASME
International Design Engineering Technical Conferences &
Computers and Information in Engineering Conference,
DETC2007-34237. Las Vegas, Nevada, USA, 4-7 September 2007.