6th annual IEEE Conference on Automation Science and TuA2.

2
Engineering
Marriott Eaton Centre Hotel
Toronto, Ontario, Canada, August 21-24, 2010

Global Kinematic Model Generation for n-DOF Reconfigurable

Machinery Structure
A. M. Djuric1, R. Al Saidi2, W. ElMaraghy3

Abstract—Automated model generation and solution for connectors. In [3], autonomous modules (CONRO) are
motion planning and re-planning of automated systems will introduced with the capability to be reconfigured in different
play an important role in the future of reconfigurable shapes, such as snakes or hexapods. Many modular and self-
manufacturing systems. Multi-DOF kinematic structure was reconfigurable robots are investigated in, [4-8]. These robots
generated for any combination of either rotational or
have the ability to change their number of links, form closed
translational type of joints, using a novel methodology
presented in this paper. This model is named n-DOF Global chains and, open chains, and form single or multiple
Kinematic Model (n-GKM). All possible kinematic structures branches. All the discussed modular self-reconfigurable
are considered in 3D space, which is divided into eight robots need special physical features (docking) to connect
subspaces and three planes. The planes are perpendicular to and release the modules. Furthermore, different applications
the x, y, and z axis of the kinematic structure’s base frame. The are executed with a central control to direct and shape the
intersections of the three planes divide 3D space into eight
final required structure, which enables the robot to perform a
subspaces. The kinematic structure generated using the
proposed methodology belongs to any of the three surfaces or specified task. Accordingly, D-H parameters have not been
to any of the eight subspaces. This kinematic structure can be introduced to describe the kinematic and dynamic models of
implemented for any robotic system or CNC machine. A the resulting configuration. For space explorations purposes,
methodology for model generation is presented and a conceptual design of a reconfigurable robot has been
demonstrated with 2-DOF Global Kinematic Model (2-GKM) developed [9]. The authors proposed a non-modular
case study.
reconfigurable robot with changing D-H parameters, such as
the sliding link length and rotary angle. A modular and
I. INTRODUCTION
reconfigurable robot for industrial purposes has been

T oday’s manufacturing environment is dominated by

market change and global competition. Manufacturing
success and survival are characterized by low cost and high
introduced [10]. The PROFACTOR GmbH has presented a
modular and reconfigurable robot with power cube
(Mechatronical Components) modules. These modules are
quality to meet market and customer demands. designed to be identical and self-contained with actuation,
Reconfigurable manufacturing systems are identified as a memory, mechanical, electrical and embedded
means of improving the production outcome by quickly programming. In [11], the authors derived the kinematic and
adjusting its production capacity and functionality in dynamic models of reconfigurable robots using D-H
response to sudden market changes and needs, [1]. The goal parameters for different sets of joints, links and grippers
here is to ensure that individual machinery systems, such as modules. Furthermore, a library of modules is formed from
robots and CNC machines, are capable of reconfiguration for which any module can be called with its associated
different applications. A variety of modular self- kinematic and dynamic models. In [12], a modular and
reconfigurable robots have been investigated. SUPERBOT a reconfigurable robot design is introduced with modular
robotic system presented in [2], is a deployable robot and joints and links. The proposed design introduces zero links,
build of modules. The modules have to connect and offset to increase the robot’s dexterity and maximizes the
disconnect each other to form a new chain loop needed to reachability. This link design methodology is producing
perform a new task. The modules (units) of the modular and singularity. A Task-based configuration optimization based
reconfigurable robots are independent units with their own on a genetic algorithm was used to solve a pre-defined set of
actuation, power, CPU, memory and physical hardware modules for specific kinematic configuration [13]. In [14],
the authors present a unified reconfigurable kinematic model
Manuscript received March 01, 2010. and solver for 6R industrial robots. The reconfigurable
A. M. Djuric is with the Intelligent Manufacturing Systems Centre,
kinematic model includes six rotational joints, each with two
University of Windsor, 401 Sunset Avenue, Windsor, Ontario N9B 3P4,
Canada ([email protected]). different positive directions controlled by reconfigurable
R. Al Saidi is with the Industrial and Manufacturing Systems parameters K i as a function of the robot’s twist angles. This
Engineering Department, University of Windsor, 401 Sunset Avenue,
Windsor, Ontario N9B 3P4, Canada ([email protected]). model does not include the translational type of joints, which
Dr. Waguih ElMaraghy, Professor & Head, Dept. of Industrial and limits the model’s capability to represent only the 6R
Manufacturing Systems Engineering, Director of Intelligent Manufacturing industrial robots.
Systems Centre, 401 Sunset Avenue, Windsor, Ontario N9B 3P4, Canada,
(corresponding author) phone: (519) 253-3000 Ext. 2607 Direct: 5037; To improve the capability of the previously developed
fax(519) 971-3656; e-mail: [email protected]. model, a novel general n-DOF model was developed. This

978-1-4244-5449-5/10/$26.00
978-1-4244-5448-8/10/$26.00 ©2010
©2010 IEEE
Crown 804
model includes rotational and translational joints, with six For the n-GKM the D-H parameters are presented in
possible z-axis (axis of rotations and/or joints translations). Table1. The twist angle αi has only five different values,
This general model can be used to represent any open which keeps perpendicularity between joints coordinate
kinematic configuration machinery. frames.
TABLE I
II. RESEARCH OBJECTIVE D-H PARAMETER OF THE N-GKM MODEL
Current machines (CNC, robots, etc.) have physical i di θi ai αi
limitations regarding their configurations and capabilities. 1 R1 d DH 1 + T1 d 1 R1θ 1 + T1θ DH 1 a1 0 D ,±180 D ,±90 D
They are built to do certain tasks, which they can do with 2 R 2 d DH 2 + T2 d 2 R2θ 2 + T2θ DH 2 a2 0 D ,±180 D ,±90 D
their predefined fixed configurations. For example: 5-DOF
" " " " "
(3R-2T) CNC machine have three rotational and two Rn d DHn + Tn d n Rnθ n + Tnθ DHn
n an 0 D ,±180 D ,±90 D
translations joints with fixed coordinate frames (D-H
parameters), which cannot be automatically changed to any
other configuration. The structure of some machines can be All D-H parameters presented in Table 1 are not fixed
changed only by physically replacing their joints or links. values. They represent and satisfy properties of all possible
Those limitations reflect the machine’s path, workspace, machinery kinematic structures. This means that each joint
inertia, torque, power concept, etc, and they cannot be used has six different positive directions of rotations and/or
for the future Reconfigurable Manufacturing Systems translations. Any joint’s vector zi−1 can be placed in positive
(RMS). New global trends in industry require more powerful and/or negative direction of x , y , and z axis of the
machines with reconfigurable properties that can be used in Cartesian coordinate frame. This is expressed in equations
a highly changeable production environment. All these (1)-(2).
mechanically constructed machines are presented as open
Rotational Joints: Ri = 1 and Ti = 0 (1)
kinematic chains, named Multi-DOF Machinery. Those
types of machines will have the properties of as many Translational Joints: Ri = 0 and Ti = 1 (2)
different machines as possible, which will allow them to be Ri and Ti are used to control the selection of joint type
used for different applications. Analysis of many different
(rotational and/or translational). The orthogonality between
kinematic structures [14], reveals the similarities between
many different machines. This leads us to develop a model the joint’s coordinate frames is defined with twist angles α i .
that can represent the properties of many different machines. Their sinus and cosines are defined as the joint’s
reconfigurable parameters ( KSi & KCi ) and expressed in
equations (3) &- (4).
III. THE DEVELOPMENT OF N-DOF GLOBAL KINEMATIC K Si = sin αi (3)
MODEL
KCi = cos αi
(4)
A devotement of the n-DOF Global Kinematic Model (n- To build a reconfigurable joint, we need to include all six
GKM) is much needed for supporting any kinematic different positive directions of rotations and/or translations.
configuration presented in Figure 1, and possible redundant The procedure will start from the first coordinate frame by
kinematic structures that are intended to support more than defining the orientation of the vector, Z0 . Because we have
6-DOF.
six combinations of vector- Z 0 , we will start from the first
RRT .....
RRR one named Z 01 , see Figure 2. The selection of vector Z 01 ,
Reconfigurable
Parameters Ki
3 DOF can be combined with four more orientations of vectors X 0
Reconfigurable RTRR
RRRT
Parameters Ki
Reconfigurable
Parameters Ksi
.....
and Y0 . They are: X 011 , Y011 , X 012 , Y012 , X 013 , Y013 , X 014 ,
and Kci
RR
TR Reconfigurable
Parameters Ki
Y014 .
RT
TT
UNIFIED
4 DOF y 12
0 x 013
2 DOF Global
d1 d1
Kinematic y 13
0 x 14
0
Model d1
θ1
11 12
z 10 y 0 θ1 x 0 θ1 d1
TRRRR θ 1 x 11
0
z 10 z 10 z 10 y 14
0
RTTRT
..... Fig. 2. First, four combinations of Joint1
Reconfigurable
Parameters Ki
Reconfigurable Parameters Ki

RRRRRR(RRKM )
5 DOF The second combination of Z0 and it’s X 0 and Y0 are
RTRTRT ..... presented in Figure 3. This includes new vector Z 02 and its
6 DOF
Fig. 1. Process of Unification of reconfigurable Models four combinations: X 021 , Y021 , X 022 , Y022 , X 023 , Y023 , X 024 ,

978-1-4244-5448-8/10/$26.00 ©2010 Crown 805

Y024 . Similarly, all other possible combinations of different Definition 2:
Z0 vector and their X 0 and Y0 vectors are shown in Similarly, we defined links reconfigurable parameter, RL ,
Figures 3-7. which is simply the changeable link length, and can be any
real number, Equation 6.
y 023 x024
x 21
y 22
x 23
y024 RL ∈ Re (6)
0 0 0

y 21 d1 θ1 Definition 3:
0 d1 x 022 d1 θ1 d1
θ1 z 02 θ1 z 2 z 02 2 For any joint to be reconfigurable the following condition
0 z 0
must be satisfied: The number of different motions should be
Fig. 3. Second, four combinations of Joint1
minimum two, Equation 7.
y 031 x031 x 032
d1
y 034 min(RJ ) = 2 (7)
z 3
0
z 03 z 03
From those three definitions we can have a clear
θ1 y 032 θ1 x 033 z 03 θ1
d1 d1 θ1 34 description of the reconfigurable robot.
y 33 x 0 d1
0 Definition 4:
Fig. 4. Third, four combinations of Joint1 The robot is reconfigurable iff it has a minimum one
x 41
0 z 4
0 z 04 z 04 y 044 z 04 reconfigurable joint or link.
d1 Definition 5:
d1 θ1 θ1 y 43
0 θ1 θ1
41
n-DOF Global Kinematic Model (n-GKM) is a
y x 42
d1 x044 d1
0
y 042
0 reconfigurable model because it has all reconfigurable joints
x043
and links. Its joints satisfy the maximum number of
Fig. 5. Fourth, four combinations of Joint1
reconfigurations ( RJ = 12 ).
z 5
0 θ1 z 05 θ 1 z 05 z 05 The n-GKM model, Figure 9, starts from the base frame
y 51
d1 x 52
d1 θ1 θ1
d1
0 0 d1 x0 , y0 , z0 which represents the coordinate frame of the first
x 51
0
y 52
0 x 53
0 y 054
y053 x 054 joint, and has six possible frames for the second joint
Fig. 6. Fifth, four combinations of Joint1 presented with coordinate frame x1 , y1 , z1 . From second joint
x 061 d1 d1 d1 y064 coordinate frame x1 , y1 , z1 , again has six different
z 06 θ 1 y061 θ1 x 062 θ1 θ1 combinations for joint three coordinate frame x2 , y2 , z2 , and
z 06 z 06 z 06
so on up to the flange frame xn , yn , zn .
d1 y062 y 063 x 063 x 064
"
Fig. 7. Sixth, four combinations of Joint1 x3 y 3 z 3

x3 y 3 z 3 x3 y 3 z 3
This will produce a reconfigurable joint to have 24 "
"

different possible coordinate frames. The graphical "

x2 y2 z2
representation of a single reconfigurable joint is given in
Figure 8.
x2 y2 z2 Joint 3 x2 y 2 z 2
z 05 4 "
z 0 " x3 y 3 z 3
θ1
d1 x1 y 1 z 1
"
z 06 z 03 Joint 4
"

x3 y 3 z 3 x2 y 2 z 2 x2 y 2 z 2
z 10 2
"
z 0 "
" Joint 2
Fig. 8. Reconfigurable model of Joint1
x1 y1 z1 x1 y1 z1
A reconfigurable joint model includes 6R and 6T different
types of motion, which is the maximum number of motions x0 y 0 z0 Joint 1
that can be produced in 3D space. The following five
" "
definitions are developed for proper use of the model.
x1 y1 z1 x1 y1 z1
Definition 1:
The degree of the joint’s reconfigurability, RJ , can be "
x1 y1 z1
between 2 and 12. This parameter defines the level of the
joint’s reconfigurable capabilities, Equation 5. Fig. 9. Kinematic Structure of the n-GKM Model
2 ≤ RJ ≤ 12 (5) The total number of all possible kinematic structures C n ,

978-1-4244-5448-8/10/$26.00 ©2010 Crown 806

is expressed in Equation 8. Each combination can support
either rotational or translational joint types. The 2-GKM kinematic model is graphically presented in
C n = 48 ( 48 − 1) n −1 (8) Figure 11. Each joint is reconfigurable and in between any
The kinematic structures generated from the n-GKM can two joints all links have changeable values. The number of
be in eleven different parts of the 3D spaces. Eight are three- different kinematic structures can be calculated using the
dimensional and three are planer kinematic structures. See equation (10) for n=2.
Figure 10. This is important for solving the inverse C2 = 48(48 − 1)2−1 = 2256 (10)
kinematic problem. z26
z 25
z24
Plane 2 z12 z 23
z 25 2 z 25
z
z26 z24 2
z26 z24
Octant 2 Plane 1 End-effector z12 z 3
θ2 z15 z12 z 23
2 2 4 2
z d2 z z
z 25 2 1 2

Octant 1 z26 z24 z 25

1 z 16 z13 z 5 z26 z24
z z 23 2
z 25 2
z 25 6 4 1
6 4 z 22 6 4 z 11 z 12 z 2 z 2 z 2 z 23 z 25
z z z z z 22
1
2 2
1
2 2
z12 z 23 z26 z24
z 2 z 23 θ 2 z15 z 2 z 23 z 25 z 22
Octant 4
5 1

Octant 3 z 2
2 d2 z14 z 2
2 z26 z24 θ2 z 1 z 2 z 23
z14 z 22
Plane 3 z12 z 23
d2
Octant 6 z 16 z13 z 22
z 25
Octant 5 z 11 z 12 z2
6
z24 z 25
z 16 z13 z 5
2
Joint 2 z12 z26 z24 z 11 z12 z2
6
z24
z 25
z 23 z 05 z12
z 22 z12 z 23 z 23
z26 z24 2 z 25 z 22
z
z 04 2
z26 z24
z12 z 23 θ1
z 22 z12 z 23
d1 z 2

Octant 7 Octant 8 z 25
2

z26 z24
z 06 z 03
z 25
z12 z 23 Joint 1 z26 z24
Fig. 10. Eight octants and three planes in 3D Space 6
z 25
4
z 22
6
z 25
4 z 1 z 1
2 z 23
z z z z 0 2
2 2
θ2 z15 2 2 z 0 z 25 z 2
2
z1
2 z 23 d 2 z14
1
z
2 z 23 z 25 z26 z24
z 22 z 22 z26 z24 θ2 z15
The kinematics of the n-GKM can be calculated using the z 16 z13 z12 z 23
z 1
2
z 22
z 3
2d2 z14
z 25 z 22
multiplication of the all homogeneous matrices from the z26 z24 z 11 z 12
z26
z 25
z24 z 25
z 16 z13 z 5
2
z12 θ2 z15 6
z24
z 23 z 12 z2
base frame to the flange frame. The homogeneous z 22
z6
z 25
z 4
z12
z 2
z 23 d 2 z14 z26
z12
z24 z 11
z12 z 23
2 2 2 z 23 z 5
z 22
transformation matrix for the n-DOF Global Kinematic z12
z 22
z 23
z 25
z 16 z13 z 5
2
z 22
z26
2
z24
6
z26 z24 z 11 z12 z2 z24 z12
Model (n-GKM) is shown in equations (9). z12 z12 z 23
z 22
z 23
z 23 z 25
z 22 z 22
z26 z24

ªcos( Riθ i + Tiθ DHi ) − K Ci sin( Riθ i + Tiθ DHi )

z12 z 23
z 22
« sin( R θ + T θ ) K Ci cos( Riθ i + Tiθ DHi ) Fig. 11. Kinematic Structure of the 2-GKM Model
i −1
Ai = « i i i DHi
« 0 K Si A. Forward Kinematics
« i = 1, 2 , ! , n (9)
¬ 0 0
0
K Si sin( Riθ i + Tiθ DHi ) ai cos( Riθ i + Tiθ DHi ) º
The two homogeneous transformation matrices A1 and
» 1
− K Si cos( Riθ i + Tiθ DHi ) ai sin( Riθ i + Tiθ DHi ) » A2 for the 2-GKM model are generated from the general
K Ci Ri d DHi + Ti d i »
» homogeneous matrix (9). By multiplying these two matrices
0 1 ¼ the arm transformation matrix can be calculated. See
Equation 11:
ªn x s x a x p x º
IV. CASE STUDY: 2-GKM MODEL DEVELOPMENT AND «n s a p y »»
SOLUTIONS
A1 A2 = 0A2 = « y
0 1 y y
(11)
«nz s z a z p z »
The D-H parameters of the 2-GKM kinematic model are « »
¬0 0 0 1¼
presented in Table II.
TABLE II Equation (11) represents a forward kinematic solution for
D-H PARAMETER OF THE 2-GKM MODEL the 2-GKM model. The first 3× 3 matrix is a rotational
i di θi ai αi matrix, which represents the projections of the normal,
R1 d DH 1 + T1 d 1 R1θ 1 + T1θ DH 1 a1 0 ,±180 D ,±90 D
D sliding and approach vectors. The position of the end-
1
0 D ,±180 D ,±90 D
effecter is expressed with 3× 1 matrix. The elements of the
2 R 2 d DH 2 + T2 d 2 R 2θ 2 + T2θ DH 2 a2
transformation matrix are a function of the 2-GKM model’s
D-H parameters.
The reconfigurable parameters of the 2-GKM kinematic
model are presented in Table III. B. Inverse Kinematics
TABLE III From the Equation (11) the solution for the 2-GKM
D-H RECONFIGURABLE PARAMETERS
model’s inverse kinematics can be found.
D
KC1 = cos(±180D ;0) = ±1
K S1 = sin(±90 ) = ±1
θ1 = R1 {A tan 2[ sin( R1θ1 + T1θ DH1 ),
K S 2 = sin(±90D ) = ±1 KC 2 = cos(±180D ;0) = ±1
cos( R1θ1 + T1θ DH1 )] − T1θ DH1} (12)

978-1-4244-5448-8/10/$26.00 ©2010 Crown 807

 θ 2 = R 2 {K c1 [ A tan 2( n y , n x ) − ( R1θ 1 + T1θ DH 1 )] can be seen from Equations (12-18).
− T2θ DH 2 }∗ config
(13) § p ·
q 1 = θ1 = A tan 2 ¨ − x ¸ (17)
The config is used to define different inverse kinematic ¨ py ¸
© ¹
solutions for the 2-DOF kinematic structure, see [15].
§n ·
q 2 = θ 2 = A tan 2 ¨¨ z ¸¸ (18)
d 1 = T1 {p z − K s1 a 2 sin( R2θ 2 + T2θ DH 2 ) − R2 d DH 1 © bz ¹
K c1 a 2 sin[( R1θ1 + T1θ DH 1 ) + K c1 ( R2θ 2 + T2θ DH 2 )] Similarly, three more cases are done. See Figure 13, 14
− and 15.
K s21 cos( R1θ1 + T1θ DH 1 ) (14) d2
K c1 a1 sin( R1θ 1 + T1θ DH 1 ) − p y ½°
− ¾ Z2
K s21 cos( R1θ1 + T1θ DH 1 ) °¿
Z1

Y1
X1 X2
Y2
d DH1
­ a sin[( R1θ 1 + T1θ DH 1 ) + K c1 ( R2θ 2 + T2θ DH 2 )] Link di θi ai αi
d 2 = T2 ® 2 Z0
θ1
K s1 cos( R1θ 1 + T1θ DH 1 ) 1 d DH 1 0 -90
¯ 1 2 d2 θDH2 = 0 0 0
a1 sin( R1θ 1 + T1θ DH 1 ) − p y ½° (15) Y0
+ − R2 d DH 2 ¾
K s1 cos( R1θ 1 + T1θ DH 1 ) °¿
X0

Fig. 13. RT configuration

The inverse kinematic solution includes translational and For the case in Figure 13 the reconfigurable parameters
rotational types of joints for the 2-GKM model. This must be are as follows: R1 = 1 , T1 = 0 , R2 = 0 , T2 = 1 , K S1 = −1 ,
expressed as generalized coordinates and their mathematical
expression is given in the following Equation: K S 2 = −1 , KC1 = 0 , KC2 = 0 . The inverse kinematic
qi = Riθi + Ti di (16) solution for the case in Figure 13 is shown in Equations (19-
20).
V. MODEL EVALUATION § ny ·
q1 = θ1 = A tan 2¨¨ ¸
¸ (19)
The model evaluation was done by separately solving © nx ¹
different single configurations, with only one positive py
direction for each joint and only one type of joint. For q2 = d2 = (20)
example, the kinematic structure with Rotational, Rotational sin θ1
D D
(RR) joints ( α1 = 90 and α 2 = 0 ), is shown in Figure 12.
The inverse kinematics is solved for all four cases and the
results are compared with general inverse kinematic
solution.
di θi ai αi
d1
d DH 2 θ2

di θi ai αi
d DH 1 θ1 Fig. 14. TR configuration
d DH 2 θ2 For the case in Figure 14 the reconfigurable parameters
are as follows: R1 = 0 , T1 = 1 , R2 = 1 , T2 = 0 , K S1 = −1 ,
K S 2 = −1 , KC1 = 0 , KC2 = 0 . The inverse kinematic solution
Fig. 12. RR configuration
for the case in Figure 14 is shown in Equations (21-22).
For the case in Figure 12, the reconfigurable parameters q 1 = d1 = p z (21)
are as following: R1 = 1 , T1 = 0 , R2 = 1 , T2 = 0 , K S1 = −1 , §n ·
q 2 = θ 2 = A tan 2¨¨ z ¸¸ (22)
K S 2 = −1 , KC1 = 0 , KC2 = 0 . In the general model we © bz ¹
defined the values for the configuration parameters, and the
general solution automatically generated the solution of the For the case in Figure 15 the reconfigurable parameters
selected case. The results of the general model provide the are as follows: R1 = 0 , T1 = 1 , R2 = 0 , T2 = 1 , K S1 = −1 ,
same result as the solution of the single selected case, which K S 2 = −1 , KC1 = 0 , KC2 = 0 .

978-1-4244-5448-8/10/$26.00 ©2010 Crown 808

 [6] G. S. Chirikjian, “Kinematics of a Metamorphic System,” Proceedings
of the 1994 IEEE International Conference on Robotics and
Automation, pages 449-455, 1994.
[7] L. Kelmar and P. K. KHhosla, ”Automatic Generation of Kinematics
for a Reconfigurable Modular Manipulator System”, J. Robotic
Systems, vol. 7, no. 4, pp. 599-619, Aug. 1990.
[8] M. Yim, D. G. Duff, and K. D. Roufas,”POLYBOT: a Modular
di θi ai αi Reconfigurable Robot”, 2000.
[9] F. Aghili, K. Parsa,” A Reconfigurable Robot with Lockable
d1 Cylindrical Joints,” IEEE Transactions on Robotics, vol. 25, No. 4,
d2
August 2009.
[10] T. Strasser, M.N. Rooker and G. Ebenhofer, ”Distributed Control
Concept for a 6-DOF Reconfigurable Robot Arm”, Innovation
production machines and systems: Fourth I*Proms Virtual
International Conference, 1st-14th July, 2008.
Fig. 15. TT configuration [11] Z. M. Bi, W.J. Ahang, I.-M. Chen, S.Y.T. Lang,” Automated
Generation of the D-H Parameters for Configuration Design of
The inverse kinematic solution for the case in Figure 15 is Modular Manipulators,” Robotics and Computer-Integrated
shown in Equations (23-24). Manufacturing 23 (2007) 553-562.
q1 = d1 = p z (23) [12] Z. Li, W. Melek, C.M.Clark,” Development and Characterization of a
Modular and Reconfigurable Robot,” CARV 2007, Toronto, Canada
q2 = d2 = py (24) [13] S. Tabandeh, C. Clark , W. Melek, “Task-based Configuration
Optimization of Modular and Reconfigurable Robots using a Multi-
solution Inverse Kinematics Solver”, CARV 2007, Toronto, Canada
[14] A. M. Djuric and W. H. ElMaraghy, “Generalized Reconfigurable 6-
VI. CONCLUSIONS AND FUTURE WORK Joint Robot Modeling”, Transactions of the CSME, Vol. 30, No. 4, pp.
Multi-DOF kinematic structure was generated for any 533-565, 2006.
[15] A. M. Djuric and J. R. Urbanic, “A Methodology for Defining the
combination of either rotational or translational type of Functional Space (Work Window) for a Machine Configuration”, 3nd
joints. A novel methodology presented in this paper was International Conference on Changeable, Agile, Reconfigurable and
used to develop an n-DOF Global Kinematic Model (n- Virtual Production, Munich, October 5th-7th, 2009.
[16] A. M. Djuric and W. H. ElMaraghy, “Automatic Separation Method
GKM). All possible kinematic structures are considered in for Generation of Reconfigurable 6R Robot Dynamics Equations”,
3D, space which is divided in eight subspaces and three International Journal of Advanced Manufacturing Technology,
planes. The planes are perpendicular to the x, y, and z axis of Springer London, (ISSN 0268-3768), July 27, 2009.
the kinematic structure’s base frame. The intersections of the [17] I-Ming Chen and J.W. Burdick, ”Enumerating the Non-Isomorphic
Assembly Configurations of Modular Robotic Systems”, Proceedings
three planes divide the 3D space into eight subspaces. The of IEEE/RSJ International Conference on Intelligent Robots and
kinematic structure generated using this methodology can Systems, (IROS), pp 1985-1992, Yokohama, Japan, July, 1993.
belong to any of the three surfaces or to any of the eight
subspaces. This kinematic structure can be implemented for VII. NOMENCLATURE
any robotic system or CNC machine. A methodology for i −1
Ai - Homogeneous matrices
model generation and solution was presented and
demonstrated with an example of 2-DOF Global Kinematic di - Link offset along previous z to the common normal
Model (2-GKM). Model evaluation was done using four θi - Joint angle about z, from old x to new x
single kinematic structures (RR, RT, TR and TT) and the
results are confirmed. ai - Link length of the common normal
The solution of the inverse and forward dynamics for the αi - Twist angle about common normal, from old z axis to
n-GKM will be generated using the automatic separation new z axis
method (ASM) [16]. Furthermore, the dynamics of the
Ri - Joints types control parameters
motors will be added for future control purposes.
Ti - Joints types control parameters
REFERENCES K Si - Twist angles sinus control parameter
[1] H.A.ElMaraghy, “Changeable and Reconfigurable Manufacturing
Systems. New York: Springer Verlag, 2009. K Ci - Twist angles cosines control parameter
[2] B. Salem, M. Moll, W-M. Shen, “SUPERBOT: A Deployable, Multi- config - Configuration parameter
Functional, and Modular Self-Reconfigurable Robotic Systems,”
Proceedings of the 2006 IEEE/RSJ International Conference on qi - Generalized coordinates
Intelligent Robots and Systems, pp. 3636-3641
[3] W-M. Shen, B. Salem, Peter Will,”Hormones for Self-Reconfigurable RJ - The degree of joint’s Reconfigurability
Robotics,” Intelligent Autonomous System 6, pages 918-925, 2000.
[4] S. Hirose,”Biologically Inspired Robots: Snake-Like Locomotors and RL - The degree of link’s Reconfigurability
Manipulators,” Oxford University Press, New York, 1993.
[5] T. Fukuda and Y. Kawauchi,”Cellular Robotic System (CEBOT) as C n -Number of all possible different kinematic structures
one of the realization of Self-Organizing Intelligent Universal
Manipulator,” Proceedings of the 1990 IEEE International Conference
nz - Projection of the end-effecters’ normal vector
on Robotics and Automation, pages 662-667, 1990.
bz - Projection of the end-effecters’ sliding vector

978-1-4244-5448-8/10/$26.00 ©2010 Crown 809