Anda di halaman 1dari 6

A Generalized Method for Solving the Kinematics of 3 D.O.F.

Reconfigurable Manipulators
H. Valsamosa, V. Moulianitisa,b, N. Aspragathosa
a
Department of Mechanical & Aeronautics Engineering, University of Patras, Rio, Patras,26500, Greece
b
Department of Product and System Design Engineering, University of the Aegean, Syros, Greece

Abstract

A generalized method for solving the kinematic problems of 3 d.o.f. reconfigurable manipulators is
presented. The method is based on the product of exponentials formula (PoE) and can be used for reconfigurable
manipulators structured with the use of pseudo joints, active twisting joints and rigid links. A software built in
MATLAB allowing the user to structure 3 d.o.f. reconfigurable manipulators with arbitrary anatomies that can
automatically provide solutions to the direct and inverse kinematic problems of the designed robot, based on the
aforementioned method, is also presented. Results are also presented for one case study with seven pseudo joints
and 3 joints.

Keywords: Reconfigurable robots, kinematics

1. Introduction methods for reconfigurable articulated robots where


developed: analytical and numerical or iterative
As manufacturing is shifting from the flexible to methods. Analytical methods [4,5,6] provide an
the reconfigurable paradigm, the design of industrial accurate solution to the inverse kinematics problem
manipulators and robotic workcells should adopt requiring low computational time once the final set
reconfiguration. Although reconfigurable of equations for the joint variables is defined.
manipulators offer great advantages due to their However, they are structure dependant, meaning that
innate capability for altering their anatomy [1] still one can only derive an analytical solution for a
most designs for articulated robotic systems single manipulator or a series of manipulators with
implementing this paradigm, present structure similar structural characteristics, while additionally
similarities to currently used industrial articulated they require extensive algebraic calculations in order
manipulators. Most systems follow Piepers to derive the final set of equations to be solved for
suggestion of having either three consecutive joint the joint variables. However, their rapid calculation
axes intersecting at a common point or three of a solution makes them ideal for on-line control
consecutive parallel axes (Pieper, 1968). In addition, and optimization algorithms.
their geometry is somewhat bounded to have the Numerical or iterative methods [7,8,9,10] on the
angles formed by consecutive joint axes set to either other hand, have a more general approach and can
900 or 00, to reduce the kinematic complexity [2]. provide a solution for a wider range of manipulators,
In the relevant literature the kinematics for fixed regardless of their structural configuration. However,
anatomy articulated systems are addressed by two iterative and AI based methods can usually derive
different approaches, from which the relative the closest solution which reduces the required
accuracy, while they also suffer from convergence different matter. The inverse kinematic problem is
problems which may result in an inability of non-linear, directly dependant of the manipulators
producing any solutions at all. Computational configuration (posture) and anatomy (structure). This
methods also require a significant amount of time in causes the inverse kinematics problem to be very
order to calculate a solution, compared with the difficult, and in some cases impossible, to be solved
direct solving of analytical methods. Therefore, these analytically, especially for reconfigurable robots
methods can be a very helpful tool in deriving the whose structure is changeable [15]. Current practice
kinematic solutions for manipulators but only when in solving the inverse kinematics of robotic
used off-line. manipulators involves the use of iterative methods as
Reconfigurable articulated systems on the other an effort to reduce the required effort in order to
hand present far more a challenge regarding their reach a solution, as these methods present a more
inverse kinematics since their anatomies are variable, general approach. However, their increased
requiring a new analysis for each new structural requirements in computational time and power,
configuration. Therefore in most cases, numerical especially when considering optimization problems
methods are favored in order to address the inverse inhibits their usage in on-line algorithms, while at
kinematics of such systems [11, 12, 13]. This is also the same time greatly increases the total amount of
one of the reasons why reconfigurable articulated time and power required for the whole procedure.
systems resemble, in terms of structure, their fixed A generalized analytical method for solving the
anatomy counterparts. inverse kinematics of a reconfigurable manipulator
In this work, a generalized method for obtaining could therefore help reduce the required
analytical expressions and solutions for the inverse computational time and power required by the
kinematics of reconfigurable 3 d.o.f. RRR relative numerical ones making it ideal for offline
(rotational) type manipulators is presented. The main and online optimization, design and control
objective of the method is to provide a means to processes.
develop closed form solutions for the reconfigurable Pseudo joints were presented as a means to
manipulators kinematic problems that are able to facilitate the rapid and with limited effort
track the changes in its anatomy during requirements anatomical reconfiguration of
reconfiguration. The proposed method is presented articulated modular robotic manipulators [14]. One
by determining closed form kinematics expression of the main advantages of using pseudo joints for
for a 3 d.o.f. reconfigurable manipulator with structuring articulated reconfigurable robots is that a
arbitrary anatomies. An automated software system parametric solution to their kinematic problems,
developed allowing the user to construct a having as parameters the changeable kinematic
reconfigurable manipulator using active joints, rigid features of the robot is made feasible. This
links and pseudo joints [14] in a completely arbitrary parametric solution allows the derivation of a set of
sequence which is capable of solving both the decoupled equations solvable for the joint angles of
forward and inverse kinematics of the constructed manipulator consisting of a given number of active
manipulator is also presented. Results are presented joints, pseudo-joints and rigid links for every
for three case study reconfigurable manipulators. possible anatomy it can achieve. This allows for a
significant reduction in the complexity as well as in
2. A generalized parametric analytical method for the time and computational power requirements for
solving the inverse kinematics of 3 d.o.f. solving the optimization problem.
reconfigurable manipulators Using this ability provided by the pseudo joints
it is possible to create a generalized procedure
One of the key elements for robot design is the leading to a parametrical solution of the inverse
development of solutions to their kinematics, since kinematics problems of 3 d.o.f. reconfigurable
these are the cornerstones upon which every possible manipulators, structured with the usage of pseudo
aspect of them is based, whether it is their design, joints. The PoE [6] is used for deriving the
performance optimization and analysis, kinematic parametric solution to the inverse kinematics
and dynamical analysis and optimization, control etc. problem of the manipulator. It is preferred over other
Although the direct kinematics of any given methods such as D-H [4,5], since it is relatively
manipulator whether reconfigurable or not are simpler due to the fact that only two coordinate
straightforward and in every respect analytically frames are used and together with the geometric
solvable, the inverse kinematics are an altogether significance of the twists i the produced
Fig. 1. A general reconfigurable rotational robot.

representation of the manipulator kinematics is a


superior alternative to the D-H parameters.

Additionally it presents a lower complexity products of the exponentials of the pseudo joints that
during the determination of the inverse kinematics exist between the robots base and joint 1 (a),
solution because the Paden Kahan subproblems are between joints 1 and 2 (b) between joints 2 and 3 (c)
used i.e. the inverse kinematics are broken down to and between joint 3 and the end effector tool (d),
appropriate simpler subproblems with known whose variables v are geometrical characteristics of
solutions for each joint. the robot and are considered known. gst(0)
In order to demonstrate the procedure a 3 d.o.f. represents the position and orientation of the tool
manipulator is considered, structured using 3 frame relative to the base frame in the manipulators
twisting joints (1,2,3), and an arbitrary number of reference position illustrated in fig.1. and gd
pseudo joints and links along the robots lattice in 4 represents the forward kinematics map of the
successive sub-structures as illustrated in fig. 1 at its manipulator, as a function of the joint variables
reference position, i.e. where the joint variables all (vector ) and the variable geometrical parameters
have a value of 00 and the pseudo joint variables are (pseudo joint variables, vector v) of the
either 00 or +900 or -900. manipulator. This map is the input to the inverse
The procedure begins with the manipulators kinematics problem as it represents the position and
forward kinematics map [6]. orientation of the end effector for a set of joint
n1
a j j
n2
b j j
n3
c j j
variables and parameters.
e
j 1
e1 1 e
j 1
e2 2 e
j 1
As a first step, the exponential matrices of the joints
and pseudo joints are isolated in eq. 1.
(1) n1 n2 n3
a j j b j j c j j
e e1 1 e e2 2 e
n4
d j j
e3 3
e gst 0 gd ,v
j 1 j 1 j 1 j 1
(2)
n4
d j j

Where e i , i=1,2,3 are the exponential matrices for
e3 3
e gd ,v gst 1
0 g1
j 1
the three twisting joints whose variables are 1,2,3,
n1 n2 n3 n4
a j j b j j c j j d j j
e
j 1
,e
j 1
,e
j 1
,e
j 1
are the Since the exponential matrices of the pseudo joints
are a priori known eq. 2 can be rewritten as
n2 n3 n2
b j j
e
b j j c j j
e 1 1
e e 2 2
e e3 3
Since q lies on 1 and is known, eq. 8
j 1 j 1 j 1
(3)
n1
a j j
n4
d j j
becomes:
e g1 e g2
n2
e2 2 p1 q1 g 2 p q (9)
j 1 j 1 b j j
e
1 j
e
Let p be a point on the axis of the third joint, j 1

applying both sides of eq. 3 to it yields:


Where
n2 n3
b j j c j j
e1 1 e e2 2 e p g 2 p (4) 1
b j j
1
b j j
j 1 j 1 e
j n2
e1 1 q e
j n2
q q1 (10)

since p lies on 3 and therefore it stands that


n3 Taking the norms of both sides of eq. 9 and keeping
c j j
e 3 3
p p . The product ej 1
is known in mind that rigid transformations preserve distances,
yields:
since it is composed of the exponential matrices of
pseudo joints only, and its multiplication with p e2 2 p1 q1 g2 p q (11)
transforms it to a new point p1
Eq. 11 represents the rotation of a point to a given
n3
c j j
e
distance around an existing axis and is easily
p p1 (5) solvable for the variable of the second joint [6].
j 1
Solving the above subproblem yields a set of two
possible solutions for 2.
Therefore eq. 4 becomes: Knowing the value of 2, eq. 6 can be rewritten as:

e1 1 p2 g2 p
n2
b j j
e1 1 e
(12)
e2 2 p1 g 2 p (6)
j 1
Where
Let q be a point that lies on the axis of joint 1.
n2
b j j
e
Subtracting q from both parts of eq. 6 yields:
e2 2 p1 p 2 (13)
n2 j 1
b j j
e1 1 e e2 2 p1 q g 2 p q (7)
j 1 Since both the exponential matrices of the pseudo
joints and the second joint are known.
However, eq. 7 can be rewritten as: Eq. 12 represents the rotation of a point around a
given axis, and is solvable for the value 1 of the first
joint variable. Knowing both 1 and 2, eq. 3 can be
n2
1
e11 e e2 2 p1 e j e11 q g 2 p q (8)
b j j b j
rewritten as:
j 1 j n2
Table 1
Geometrical Parameters for the structure shown in fig. 3.

Component Orientation of Initial


Rotation axis setting
value
Pseudo joint 1 Z 00
Pseudo joint 2 Z 900
Rotational Joint 1 X
Pseudo joint 3 Z 00
Pseudo joint 4 Y -900
Fig. 2 The developed system environment.
Rotational Joint 2 Z
Pseudo joint 5 Y 900
Pseudo joint 6 Y 00
Rotational Joint 3 X
Pseudo joint 7 Z 900
structuring a reconfigurable robot using twisting
joints, pseudo joints and rigid links and for solving
its kinematic problems was built in MATLAB (Fig.
2). The system includes three basic modules of the
software:
Create a structure: A GUI for building robot
structures. The structure can be saved and loaded at
any time (fig.1).
Forward kinematics solver: A GUI that produces
Fig. 3 The reconfigurable robot structured in Table 1. the position and orientation of the tool point of a
1 1
structured robot by inserting the joint and pseudo
c j j b j j
e3 3 e e2 2 e e11 g 2 g3 (14) joint angles variables values.
j n3 j n2 Inverse kinematics solver: A GUI which solved
Let v be a point that does not lie on 3. Applying the inverse kinematic problem of the manipulator
both sides of eq. 14 to it yields: analytically. The solver returns the values for the
regular joint angles as the pseudo joint angles values
are known. The input to the solver is the known
e3 3 v g3 v (15)
position and orientation of the tool point.
The geometrical parameters for the structure
Eq. 15 is similar to eq. 12 and is also solvable for 3. shown in fig 3 are shown in Table 1. The twisting
At the end of the procedure, a set of two possible joints are defined by the orientation of their
solutions for the three joint angles [1, 2, 3] is rotational axis relative to the global coordinate
derived for the given current anatomy of the system (robots base) and the length of origin
manipulator. As this anatomy is reconfigured to a between two components. The pseudo joints which
new one, via the resetting of the pseudo joints in the can be considered as rotational joints are fully
robots lattice, the procedure is repeated in exactly defined by five parameters:
the same way, requiring only the calculation of the The axis of rotation of the pseudo joint as it is
exponentials of the pseudo joints, for their new referred in the second column of Table 1.
settings.
The value of the setting of the relative pseudo
joint as shown in the third column of Table 1. This
3. A system for solving the inverse kinematics.
value setting allows for the manipulator to achieve
Application.
different anatomies.
The lengths of the subcomponents of the pseudo
Using the presented method, a software system for
joints (i.e. their base and rotating connective surface) and Computer Integrated Manufacturing 17, pp. 199-
and the orientation of their directional axes, as these 213, 2001
are situated for the given setting of each pseudo joint [2] Pieper D., The Kinematics of Manipulators Under
angle. Computer Control, PhD Thesis, Stanford University,
1968
A number of test runs, for the presented case study, [3] Cai M-S., Rovetta A., Motion Errors Analysis and a
showed confirmed that both the presented method Contribution to Numerical Inverse Kinematics for
and software functioned correctly. During these runs Robot Manipulators, Meccanica 24, pp. 107-112, 1989
arbitrary values for the joint and pseudo joints angles [4] Craig J.J., Introduction to Robotics, Mechanics and
were used to form the forward kinematics mapping, Control, Addison Wesley Publishing, 1989
which was in turn used as an input to the inverse [5] Yoshikawa S., Foundations of Robotics Analysis and
kinematics solver. The returned joint variable Control, The MIT Press, 1990
solutions where then compared to the original ones [6] Murray R. M., Li Z., Sastry S.S., A Mathematical
for reference. Results clearly showed the Introduction to Robotic Manipulation, CRC Press,
New York, 1994
effectiveness of the method and the software system. [7] Orin D.E., Oh Y.S., Determination of Joint Positions
In order to verify the correct application of the From Limb Segment Constraints in Robotic Systems,
system further, additional tests were conducted for Proc. Of the Fifth World Congress on Theory of
arbitrary structured robots and for arbitrary joint Machines and Mechanisms, 1979
variables. The system managed to return the correct [8] Bestaoui Y., An Unconstrained Optimization Approach
solution in all cases, proving the validity of the to the Resolution of the Inverse Kinematics Problem of
method presented for solving the inverse kinematics. Redundant and Nonredundant Robot Manipulators,
Robotics and Autonomous Systems 7, pp.37-45, 1991
[9] Nearchou A., Solving the Inverse Kinematics Problem
4. Conclusions
of Redundant Robots Operating in Complex
Environments via a Modified Genetic Algorithm,
In this paper a generalized method for solving the Mechanism and Machine Theory 33 (3), pp.273-292,
kinematic problems of 3 d.o.f. RRR type 1988
reconfigurable manipulators is presented. The [10] Hasan A.T., Hamouda A.M.S., Ismail N., Al-Assadi
method is based on the product of exponentials H.M.A.A., An Adaptive-Learning Algorithm to Solve
formula (PoE) and can be used for reconfigurable the Inverse Kinematics Problem of a 6 D.O.F. Robot
manipulators structured with the use of pseudo Manipulator, Advances In Engineering Software 37,
joints, active rotating joints and rigid links. pp.432-438, 2006
A system developed in MATLAB is presented, [11] Chen I-M., Gao Y., Configuration independent
Kineatics for modular robots, Proc. Of the 1996 IEEE
allowing the user to structure 3 d.o.f. reconfigurable Int. Conference on Robotics and Automation,
manipulators with arbitrary anatomies that can Minneapolis Minnesota
automatically provide solutions to the direct and [12] Chen I-M., Gao Y., Inverse Kinematics for Modular
inverse kinematic problems of the designed robot, Robots, Proc. Of the 1998 IEEE Int. Conference on
based on the aforementioned method. Robotics and Automation, Leuven, Belgium
An illustrative example of a structure with seven [13] Kelmar L., Khosla P., Automatic Generation of
pseudo joints and three joints is presented. Kinematics for a Reconfigurable Modular Manipulator
Future work includes the extension of the method for System, Journal of Robotic Systems 7 (4), pp. 599-
6 d.o.f. robots satisfying Piepers principle with a 619, 1990
[14] Valsamos H., Aspragathos N.A., Design of a versatile
random number of pseudo joints. In addition, the use passive connector for reconfigurable robotic
of translational joints will be included in the method. manipulators with articulated anatomies and their
kinematic analysis, 2007 I*PROMS Virtual
Acknowledgements Conference, 2007
[15] Chen I-M., Gao Y., Closed-Form Inverse Kinematics
University of Patras is partner of the EU-funded Solver for Reconfigurable Robots, IEEE Int.
FP6 Innovative Production Machines and Systems Conference on Robotics and Automation, Seoul,
(I*PROMS) Network of Excellence. Korea, pp. 2395-2400, 2001

References

[1] Chen I-M, Rapid Response Manufacturing through a


Rapidly Reconfigurable Robotic Workcell. Robotics

Anda mungkin juga menyukai