Modern Applied Science March, 2009

Kinematics and Dynamics of a Master Maipulator

Yechu Hu
College of Mechanical and Electronic Engineering
Tianjin Polytechnic University
Tianjin 300160, China
E-mail: hyc820815@yahoo.cn
Abstract
I analyze the kinematics and Dynamics of Phantom Premium 1.5 made and sold by SenSable Technologies Inc., which
is widely-used as the master manipulator in telerobotic systems. The forward kinematics is studied using
Denavit-Hartenberg method, the manipulator Jacobian is also presented. The dynamic equations incorporating frictional
effects of Premium 1.5 are derived.
Keywords: Master manipulator, Kinematics analysis, Dynamic model, Frictional effects
1. Introduction
The first telerobotic system was developed by Raymond C. Goertz in 1940s to let an operator handle radioactive
materials behind a shielded wall, at National Argonne Laboratory in the US. In the past decades, teleoperation has found
applications in many areas including space technologies, underwater explorations and assistance, surgery and
rehabilitations, nuclear/toxic material handling and waste disposal, military/firefighting operation, mining. Recently, the
applications of teleoperation systems have been extended to training, education, entertainment, and virtual reality areas
as well.
As can be seen in figure 1, a teleoperation system generally has five components: operator, master, control system, slave,
and environment. The main function of the system can be explained as follows: a control command is sent through the
control system to the slave to make the remote manipulator perform a task as desired; to prevent damage, to reduce task
completion time and to enhance performance, contact interaction from the remote site has to be transmitted to the
operator.
As shown in figure 1, throughout an interaction, the master mechanism must perform the dual task of position
measurement and force display. The kinematics especially the forward kinematics is of great importance for the control
of the telerobotic system. While, when dealing with the dynamics of robotic manipulators, frictional effects is often
neglected. This paper mainly discusses the forward kinematics using Denavit-Hartenberg method, and derives the
dynamic equations incorporating frictional effects in the joints.
2. Kinematics analysis
Figure 2 is the photo and structural sketch of a typical master manipulator (PHANToM Premium 1.5), which can
provide the operator with 3 DOF motion and three dimensional force feedbacks. As shown in figure 2(b), Premium 1.5
consists of three rotational joints, zero configuration of Premium 1.5 is shown in figure 2(b), spatial frame and tool
frame are superposed in zero configuration. Figure 3 depicts the side and top views of the configuration of Premium 1.5.
And, the reference frames are shown in figure 4, the D-H parameter is listed in table 1, where
ai represents the distance from Zi to Zi+1 measured along Xi,
i represents the angle from Zi to Zi+1 measured about Xi,
di represents the distance from Xi-1 to Xi along Zi,
i represents the angle from Xi-1 to Xi measured about Zi.
The general transformation matrix 1 comes directly from the D-H parameters (Craig, 2005 & Murray, 1998).
cos i sin i 0 ai 1
sin cos cos i cos i 1 sin i 1 sin i 1di
i 1 i 1
i T =
i
(1)
sin i sin i 1 cos i sin i 1 cos i 1 cos i 1di

0 0 0 1

155
Vol. 3, No. 3 Modern Applied Science
w
And the final transformation matrix can be written in equation 2 with some mediated matrixes from 0 T to
3
t T omitted.

R( ) p( )
s
T =0w T10T21T32Tt 3T = (2)
1
t
0
Where
cos 1 sin 1 sin 3 sin 1 cos 3

R( ) = 0 cos 3 sin 3 (3)
sin cos 1 sin 3 cos 1 cos 3
1

sin 1 (l1 cos 2 + l2 sin 3 )

p( ) = l2 + l1 sin 2 l2 cos 3 (4)
l + cos (l cos + l sin )
1 1 1 2 2 3
Except the forward kinematics of the manipulator, the velocity relationship between the rotational joints and the end
effector is often concerned. So, the spatial and body manipulator Jacobian is written here, moreover, they can be used to
describe the relationship between the end point wrench and the joint torque.

l1 l1 sin 1 sin 2 sin 1 (l2 + l1 sin 2 )

0 l1 cos 2 l1 (cos 1 cos 2 )
0 l1 cos 1 sin 2 cos 1 (l2 + l1 sin 2 )
J s ( ) = (5)
0 0 cos 1
1 0 0

0 0 sin 1

l1 cos 2 + l2 sin 3 0 0

0 l1 cos( 2 3 ) 0
0 l1 sin( 2 3 ) l2
J b ( ) = (6)
0 0 1
cos 3 0 0

sin 3 0 0
3. Dynamic Model of the Master Manipulator
M. C. Cavusoglu (Cavusoglu, 2001) identified the mechanical structure of Premium 1.5 into seven segments A through
G shown on figure 5. Note that the spatial frame used in dynamic calculations is centered at the intersection point of
three axises of rotational joints.
And he calculated the kinetic and potential energy of each segment in the spatial frame, then wrote the dynamic
equation of the system using Lagrange method.
M ( )&& + C ( , & )& + N ( ) = (7)

In equation 7, M ( )&& is inertial force, C ( , & )& is Coriolis and centrifugal force, N ( ) is gravitational force, is
the vector of joint torque, each element of the matrix of M, C, and N is listed in Cavusoglus paper (Cavusoglu, 2001).
Note that the inertial force, Coriolis and centrifugal force and gravitational force are highly non-linear, making it
difficult for dynamic parameter identification and controller algorithm design (McJunkin, 2007).
So, equation 7 could be linearly parameterized as follows:
d = Yd ( , & , &&) d (8)

where Y is the regressor matrix and is the vector 12 dynamic parameters defined as:

156
Modern Applied Science March, 2009

y12 = (1 + cos 23 )&&1 2sin 23&1&3

y = (1 cos 2 )&& + 2sin 2 & &
13 3 1 3 1 3

y15 = (1 + cos 22 )&&1 2sin 22 &1&2

y = (1 cos 2 )&& + 2sin 2 & &
16 2 1 2 1 2

y17 = &&1 (9a)

y18 = (1 + cos 22 )&&1 2sin 22 &1&2
y19 = (1 cos 23 )&&1 + 2sin 23&1&3
y10 = 2 cos 2 sin 3&&1 2sin 2 sin 3&1&2
+ 2 cos cos & &
2 3 1 3

y24 = &&2
y25 = sin 22 &12
y26 = sin 22 &12
y28 = 2&&2 + sin 22 &12 (9b)
y210 = sin(2 3 )&&3 + sin 2 sin 3&12
+ cos( )& 2
2 3 3

y211 = cos 2

y31 = &&3
y32 = sin 23&12
y33 = sin 23&12
y = 2&& sin 2 & 2
39 3 3 1 (9c)
y310 = sin(2 3 )&&2 cos 2 cos & 3 1
2

+ cos( )&2
2 3 2

y312 = sin 3
and
1 = I axx + I efxx (10a)

1
2 =
2
( I ayy + I efyy ) (10b)

1
3 =
2
( I azz + I efzz ) (10c)

4 = I bxx + I cdxx (10d)

1
5 =
2
( Ibyy + I cdyy ) (10e)

1
6 = ( I bzz + I cdzz ) (10f)
2
7 = I gyy (10g)

1 1
8 = ma l12 + mb l12 (10h)
2 8
1 1
9 = ma l22 + mb l32 (10i)
8 2
1 1
10 = ma l1l2 + mb l1l3 (10j)
2 2

157
Vol. 3, No. 3 Modern Applied Science

1
11 = ma gl1 + mb gl1 mcd gl4 (10k)
2
1
12 = ma gl2 + mb gl3 mef gl5 (10l)
2
The physical and geometric parameters Iaxx, Iayy, Iazz, Ibxx, Ibyy, Ibzz, Icdxx, Icdyy, Icdzz, Iefxx, Iefyy, Iefzz, Igyy, ma, mb, mcd, mef, l1, l2,
l3, l4, l5 can be found in the literature (Cavusoglu, 2001), while g is the acceleration gravity.
4. Inclusion of Frictional Effects
Friction effect in the joints of Premium 1.5 is neglected in Equation 7 and 8, as a matter of fact, the friction effect
should not be ignored. Here I use Coulomb friction plus viscous damping (Armstrong, 1994 & Olsson, 1998) to include
friction effect in the dynamic model of the device (Tahmasebi, 2005).
f = fc sgn(& ) + fv & (11)
or
& &
f 1 fc1 sgn(1 ) + fv11
f = f 2 = fc 2 sgn(&2 ) + fv 2 &2 = Y f f (12)
& &
f 3 fc 3 sgn(3 ) + fv 33
where
sgn(&1 ) 0 0 &1 0 0


Y f = 0 &
sgn(2 ) 0 0 &2 0 (13)

0 0 sgn(&3 ) 0 0 &3
T
f = fc1 fc 2 fc 3 fv1 fv 2 fv 3 (14)

Yf is the regressor matrix, f is 6 dimensional parameter vector.

Including the friction effect of the device and combing equation 8 with 12, I could write
= Y ( , & , &&) (15)

where
Y = Yd MY f (16)
T
= d M f (17)

Equation 15 is named the dynamic model of Premium 1.5 with joint friction. Each of the element of the parameter
vector is identified as shown in table 2.
5. Conclusion
In telerobotic systems, the kinematics especially the forward kinematics is of great importance for the control of
teleoperaton, frictional effects is often neglected in dynamic equations. This paper mainly discusses the forward
kinematics using D-H method, the manipulator Jacobian is also presented. And I derive the dynamic equations
incorporating frictional effects in the joints.
References
Armstrong Brian, Dupont Pierre, Canudas de Wit C. ( 1994). A Survey of Models, Analysis Tools and Compensation
Methods for the Control of Machines with Friction[J]. Automatica, 30(7): 1083-1138.
Cavusoglu M C, Feygin D. ( 2001). Kinematics and Dynamics of Phantom(TM) Model 1.5 Haptic Interface. [R].
Berkeley: University of California at Berkeley, Electronics Research Laboratory Memo M01/15.
Craig J J. Introduction to Robotics. (2005). Mechanics and Control (3rd Editon)[M]. Upper Saddle River:
Addison-Wesley.
Haptic Device. [A]. Proceedings of the 2005 IEEE Conference on Control Applications[C]. Toronto, Canada: IEEE,
1251-1256.
McJunkin S T. (2007). Transparency Improvement for Haptic Interfaces. [D]. Houston: Rice University.
Murray R M, Li Z X, Sastry S S. (1994). A Mathematical Introduction to Robotic Manipukation. [M]. Boca Raton:

158
Modern Applied Science March, 2009

CRC Press.
Olsson H, Astrom K J, Canudas de Wit C, et al. (1998). Friction Models and Friction Compensation[J]. European
Journal of Control, 4(3): 176-195.
Tahmasebi A M, Taati B, Mobasser F, et al. (2005). Dynamic Parameter Identification and Analysis of a
PHANToM(TM)

Table 1. D-H parameter of Premium 1.5

Frame i-1 ai-1 i di

0 -/2 0 -/2 0
1 0 -l1 1 l2
2 /2 0 2 0
3 0 l1 -/2-2+3 0
tool -/2 l2 /2 0
-4
Table 2. Dynamic Parameter of Premium 1.5(10 )
1 2 3 4 5 6
7.5784 0.3154 3.3572 12.0490 5.5095 0.2981
7 8 9 10 11 12
11.8700 6.1644 0.8572 4.5693 -160 -739
13 14 15 16 17 18
707 251 248 -57 -35 -5

l1
3
y
Position Command y

Con. 2
Ope. Mas. Sla. Env. l2
1
Sys. x
x
T
Tip Force S z z

Figure 1. Diagram of the master-slave teleoperation system Figure 2. Photo and structural sketch of Premium1.5
zt
/2+2-3
z3 xt
x3

l2
z1
x2
l1
x 2
y z z2
3 x1
T

1
2 l2
2 x z0
3 xs
T z
y y
x zs, x0
x
S z S z
l1

Figure 3. Configuration of Premium 1.5 Figure 4. Reference frames of Premium1.5

159
Vol. 3, No. 3 Modern Applied Science

l1
y
E
3
x
D F C z
S

2 B l3
A
G 1 l2

Figure 5. Segments of Premium 1.5

160