Instantaneous center of rotation and

singularities of planar parallel manipulators

H. R. Mohammadi Daniali
Department of Mechanical Engineering, Faculty of Engineering, Mazandaran University,
Babol, Mazandaran, Iran
E-mail: mohammadi@nit.ac.ir

Abstract With regard to planar parallel manipulators, a general classification of singularities into
three groups is given. The classification scheme relies on the properties of instantaneous centers of
rotation. This method is very fast and can easily be applied to the manipulators under study. The
method is applied to a planar three-degrees-of-freedom parallel manipulator and all its singular
configurations are found.

Keywords planar; parallel manipulator; singularity; instantaneous center of rotation

Introduction
Parallel manipulators consist of multiple branches acting on a common payload
platform. They have superiorities over serial ones, which include greater stiffness,
improved accuracy and dynamic characteristics, higher payload/weight ratio and
higher operating speeds. These advantages stem from multiple support and the fact
that all the motors are fixed to the base. However, near singular configurations, all
manipulators experience poor performances, and parallel manipulators are not
exempt form this rule. Therefore, they lose their advantages over serial manipu-
lators at these configurations.
A manipulator singularity occurs at the coincidence of different direct or inverse
kinematic solutions. The latter are understood here as the computation of the values
of driving-joint variables from given Cartesian variables, while the former (direct
kinematics) are defined as the computation of the values of the Cartesian variables
from given driving-joint variables. Algebraically, singularity amounts to a rank defi-
ciency of the Jacobian matrices; geometrically, singularity is observed whenever the
manipulator gains some additional, uncontrollable degrees of freedom (dof), or loses
some dof. Similarly, the force transmission performance of a parallel manipulator is
very poor near singular configurations.
The concept of singularity has been extensively studied in connection with serial
manipulators [13]. As regards manipulators with kinematic loops, the literature is
more limited [410]. Litvin et al. [2] used coordinate transformation matrices to
locate singular configurations, while Merlet [5] proposed a method based on
Grassmann line geometry. Zlatanov et al. [6, 7] classified singularities via
motionspace and velocityspace models. Notash [8] found singular configurations
based on the concept of screw theory. However, the classical method to locate
singular configurations relies on the properties of the Jacobian matrices of the
manipulator [9].

International Journal of Mechanical Engineering Education 33/3

252 H. R. Mohammadi Daniali

All these methods have the same drawback: they rely on cumbersome math-
ematical or geometrical computations. Here, we introduce a fast method of finding
singular configurations of planar parallel manipulators, based on the properties of
instantaneous centers of rotation. Moreover, the method is easy to implement.

Instantaneous center of rotation

The instantaneous center of rotation (ICR) is defined as the instantaneous location
of a pair of coincident points of two rigid bodies, the absolute velocities of which
are equal. In the other words, one rigid body can rotate about the ICR, relative to
the other one. The ICR is defined between any two rigid bodies that have relative
planar motion. Therefore, there are three ICRs between three rigid bodies in rela-
tive motion. They are related by the following well known theorem.

Theorem 1: the ArnholdKennedy theorem

The three instant centers of rotation shared by three rigid bodies in relative motion
to one another all lie on the same line.

There are three types of planar motion between two rigid bodies, namely, transla-
tion, rotation and general motion. In translation, the relative velocity of all points
are parallel, thus, the ICR lies on a line perpendicular to the velocity vector at infin-
ity. In rotation, the ICR is trivial, but any relative general motion can be regarded
as a pure rotation about the ICR. Therefore, as a direct consequence of the
ArnholdKennedy theorem, one can propose theorem 2.

Theorem 2
There is no relative motion between two rigid bodies in the absence of any ICR
between them.

Singularity analysis
In serial manipulators, singularities occur whenever one actuator does not produce
any motion of the end effector (EE). In parallel manipulators, in addition to the fore-
going, we have another type of singularity, in which the EE cannot resist forces or
torques in one or more directions, even if all the actuators are locked. Thus, we clas-
sify singularities into three groups:
(1) The first type of singularity occurs when one actuator does not produce any
motion of the EE. This type of singularity consists of a point or a set of points
where different branches of the inverse kinematic problem meet. It is supposed
that each motor moves the EE in one direction. Therefore, considering the
motion of only one actuator, the EE is fixed and the Cartesian velocity vector
related to the motion of that actuator cannot be produced. In this case one can
find all ICRs while the EE is fixed. Thus, all the rigid bodies, including the
actuator under study, move except the EE, i.e., that actuator cannot move the
EE.

International Journal of Mechanical Engineering Education 33/3

Planar parallel manipulators 253

(2) The second type of singularity occurring only in parallel manipulators consists
of a point or a set of points where different branches of the direct kinematic
problem meet. This can easily be proven whenever one can find the ICR of the
EE with respect to the base, even if the actuators are locked. In this case the
EE can rotate about the ICR relative to the base. Thus, the EE can move in one
or more directions and cannot resist a force or a torque in those directions,
while the actuators are locked. Therefore, the EE gains some additional uncon-
trollable dof.
(3) The third type of singularity occurs whenever we have both types of the above
singularities simultaneously. Under a singularity of this type, configurations
arise in which the EE can undergo finite motion even if the actuators are locked,
or, equivalently, it cannot resist forces or torques in one or more directions even
if the actuators are locked. Further, a finite motion of the actuators produces
no motion of the EE and some of the Cartesian velocity vectors cannot be
produced.

Planar three-dof parallel manipulators

In this section the three types of singularities discussed above are investigated for a
planar three-dof parallel manipulator as depicted in Fig. 1. The manipulator consists
of a platform supported by three legs, each of which has one revolute joint. The dof
of the manipulator is determined by means of the ChebyshevGrublerKutzbach
formula [11] for planar manipulators, as follows:

Fig. 1 Planar three-dof parallel manipulator.

International Journal of Mechanical Engineering Education 33/3

254 H. R. Mohammadi Daniali

k
f = 3(n - 1 - k ) + fi
i =1

in which n and fi are the number of links and the dof of the ith joint, respectively,
and k is the number of joints. Therefore, the dof of the manipulator under study is:
f = 3(8 - 1 - 9) + 9 = 3
Thus, the three fixed motors move the EE in the plane.

The first type of singularity

This type of singularity occurs whenever one of the actuator does not move the EE.
Considering one of the legs of the manipulator as depicted in Fig. 2, one can find
ICRs as follows. The ICR between links one and two is shown as (1,2). Point (2,3)
is the ICR between links two and three, while point (3,4) is the ICR between links
three and four, which is the EE of the manipulator. According to theorem 1, the ICR
between links two and four, namely point (2,4), lies on the line joining (2,3) and
(3,4). Similarly, the ICR between links one and three lies on the line joining (1,2)
and (2,3). If this leg is fully extended or folded, the EE remains fixed and ICRs (1,3)
and (3,4) coincide, while the motor actuating link two is moving. Therefore, link
two rotates about ICR (1,2) relative to the base, while link three rotates about ICR
(1,3) relative to the fixed EE. In this case the corresponding motor cannot move the
EE, thus the EE cannot move in the direction of the extended or the folded legs and
the manipulator loses one degree of freedom (Figs 3 and 4).

Fig. 2 One leg of the manipulator.

International Journal of Mechanical Engineering Education 33/3

Planar parallel manipulators 255

Fig. 3 Example of the first type of singularity in which one leg is fully extended.

Fig. 4 Example of the first type of singularity in which one leg is folded.

International Journal of Mechanical Engineering Education 33/3

256 H. R. Mohammadi Daniali

Fig. 5 Example of the second type of singularity in which links three, six and eight
intersect at a point.

The second type of singularity

This type of singularity occurs whenever the EE can move relative to the base, even
if all the actuators are locked. If the motors are locked, links two, five and seven are
fixed to the base and ICRs (1,3), (1,6) and (1,8) are equivalent to those of (2,3), (5,6)
and (7,8), respectively. According to theorem 1, ICR (1,4) lies on the line joining
(1,3) and (3,4). Moreover, it is located on the line joining (1,6) and (4,6). Therefore,
ICR (1,4) is the intersection of these two lines. Similarly, ICR (1,4) is the intersec-
tion of lines joining (1,3) to (3,4) and (1,8) to (4,8). We have this type of singular-
ity in the case where two points coincide. If this is the case, the EE can rotate about
that point, while the actuators are locked.
Two different cases for which we have this type of singularity can be identified.
The first occurs when three lines bearing links three, six and eight intersect at a
common point. This point is the ICR of link four with respect to the base, which
means the EE can rotate freely about that point relative to the base, even if the
actuators are locked. Thus, the manipulator gains one rotational dof. Moreover, any
torque applied to the EE cannot be balanced by the actuators (Fig. 5).
The second case in which we have this type of singularity occurs whenever the
three lines bearing links three, six and eight are parallel. Therefore, they intersect at
infinity. Thus, the EE can rotate about that point relative to the base, i.e., the EE can
translate along u even if the actuators are locked, and the manipulator gains one
translational degree of freedom along u. Moreover, the EE cannot balance forces
along this direction (Fig. 6).

International Journal of Mechanical Engineering Education 33/3

Planar parallel manipulators 257

Fig. 6 Example of the second type of singularity in which links three, six and eight are
parallel.

Fig. 7 Example of the third type of singularity.

The third type of singularity

This type of singularity occurs whenever both types of the foregoing singularities
occur simultaneously. We have this type of singularity whenever the three lines
bearing links three, six and eight are either parallel or concurrent at a common point
and at least one leg is fully extended or folded (Figs 7 and 8). At these configura-

International Journal of Mechanical Engineering Education 33/3

258 H. R. Mohammadi Daniali

Fig. 8 Example of the third type of singularity.

tions the motion of at least one actuator does not produce any motion of the EE.
Therefore, the manipulator loses one or more dof. Further, the EE can move freely
in one or more directions even if all the actuators are locked and some forces or
torque applied to it cannot be balanced by the actuators. Thus, the manipulator gains
one or more uncontrollable dof.

Conclusions
A classification of the singularities of parallel manipulators into three groups has
been given. The three types of singularities, which have different kinematic inter-
pretations, were identified for a planar parallel manipulator. The simplicity and the
robustness of the method make it possible to implement it for all planar parallel
manipulators.

References
[1] K. Sugimoto, J. Duffy and K. H. Hunt, Special configurations of spatial mechanisms and robot
arms, Mechanism and Machine Theory, 17(2) (1982), 119132.
[2] F. L. Litvin, Z. Yi, V. Parenti-Castelli and C. Innocenti, Singularities, configurations, and dis-
placement functions for manipulators, Int. J. Robotics Research, 5(2) (1986), 5265.
[3] T. Shamir, The singularities of redundant robot arms, Int. J. Robotics Research, 9(1) (1990),
113121.
[4] M. G. Mohamed, Instantaneous Kinematics and Joint Displacement Analysis of Fully-Parallel
Robotic Devices (Doctoral Dissertation, University of Florida, Gainesville, 1983).
[5] J. P. Merlet, Singular configurations of parallel manipulators and Grassmann geometry, Int. J.
Robotics Research, 8(5) (1989), 4556.

International Journal of Mechanical Engineering Education 33/3

Planar parallel manipulators 259

[6] D. Zlatanov, R. G. Fenton and B. Benhabib, Singularity analysis of mechanisms and robots via a
motion-space model of the instantaneous kinematics, in Proceedings of the IEEE International
Conference on Robotics and Automation (San Diego, 1994), pp. 980985.
[7] D. Zlatanov, R. G. Fenton and B. Benhabib, Singularity analysis of mechanisms and robots via a
velocity-equation model of the instantaneous kinematics, in Proceedings of the IEEE International
Conference on Robotics and Automation (San Diego, 1994), pp. 986991.
[8] L. Notash, Uncertainty configurations of parallel manipulators, Mechanism and Machine Theory,
33(12) (1998), 123138.
[9] H. R. Mohammadi Daniali, P. J. Zsombor-Murray and J. Angeles, Singularity analysis of planar
parallel manipulators, Mechanism and Machine Theory, 30(5) (1995), 665678.
[10] K. Kozak, I. Ebert-Uphoff, P. A. Voglewede and W. Singhose, Concept paper: On the significance
on the lowest linearized natural frequency of a parallel manipulator as a performance measure for
concurrent design, in C. M. Gosselin and I. Ebert-Uphoff (eds), Proceedings of the Workshop on
Fundumental Issues and Future Research Directions for Parallel Mechanisms and Manipulators
(Quebec, 2002).
[11] J. Angeles, Rational Kinematics (Springer, New York, 1988).

International Journal of Mechanical Engineering Education 33/3