Jacobi An
Jacobi An
Jacobi An
(5.50)
Substituting in Eq. (5.49
, j - 1 + "R-1 Zi-16, = 0 - + "R- Z-1
(5.51)
It is easily concluded from Eq. (5.51) that the rotary joint i contributes to the
angular velocities of all links from linki to n. Further, it is to be noted that, both,
the angular velocities of the manipulator links and the joint velocity are relative to
the base frame {0}.
I tis thus found from the above account that (i) linear velocities of all
manipulator links are obtained using Eq. (5.46), and (ii) angular velocities of all
the links are obtained using Eqs. (5.48) and (5.51).
From the results of the previous section, it is observed that the Cartesian velocities
(linear as well as angular) of the end-effector are
linearly related to the joint
velocities. The relationship between infinitesimal (differential) joint motions with
infinitesimal (differential) changes in end-effector position (and orientation) is
investigated now. As these changes take place in infinitesimal (differential) time,
We are really looking for a mapping between instantaneous end-effector velocity,
i n Cartesian space) to instantaneous joint velocities (in
joint space). This
mapping between differential changes is linear and can be expressed as
V 0) = Jq)D4 (5.52)
where V0) 6x 1 Cartesian velocity vector. [Eq. (5.15)].
=
End-effector
Joint End-effector
Jacobian Jg)
Ve
velocities velocity
Joint-link parameters
Joxi
(5.55)
- Joxi
jov
Jou. and angula
the component, k of linear velocity
where jvki and joki represents and i= 1, 2, 3,
contributed by joint i withk =x, y, orz
velocity, respectively, depending on
wn
The contribution
One simple method of computation of
Jaco
velocity of
is
end-effector Joi 4,.
Manipulator Diftorential Motion and SMaties3
Consider the prismatic joint 1, a whown in Pig. 5.8. The link i translates aleong
with joint (sCalar) velocity d. To find the " column J(4) of the
a
L. axis locked in position.
Lacobian matrix J(q), assume that all joints except joint i are
Z
Z-1
Joint
)
to base
The vector PL is the transformation of this unit vector ü with respect
trame {0). The homogeneous transformation matrix from frame {0} to frame
(i-1) is
(5.58)
T,(41) T4-)= °T
The vector P is obtained as
P-= R- (5.59)
where R , is the 3 x3 orientation sub-matrix (the rotation matrix) of 7j-1
Note that third column of the rotation matrix
P is the "RJ
162 Robotics and Control
5.5.3 The Rotary Joint Jacobian
Next, consider the joint i to be a rotary
joint with angular velocit gular velocity
6, ah
-1axis, as shown in Fig. 5.9. With all joints, except joint i, locked inab
ne rotary joint rotates all the distal links from link i to linkn at t i
velocity @. For a rotary joint, the angular velocity will be given angul
by
Link i-
O-1
i-1P
'Pi-1 On
P
P
Z
0)
Fig. 5.9 Motion of
end-effecter generated by rotary joint
a
The angular
velocity of link i also produces a linear
due to the rotation of
all the distal links velocity at the end-eftedu
origin O,-I of frame {i-1}. along with the end-effector
defined by a position vector From O-, the end-effector aboutx
frame, frame
(5.60), the linear velocity Pr as shown in Fig, 5.9. From Eqs. (O. {
generated by o, is
It follows from Eqs. (5.60) and
v,
= o,xP, (P-xP,)®, (5.61)
For P
computing J; in Eq. (5.62), we
P is obtained from Eq. (5.59), whileneed
a n d P.. The vecto
vectors Pi-
the
vector'P, is found from ig.
Fig
( 5 . 9 ) a s
be written as
forany value ofn. The
above
bove
=
o o 11.1This*
can
then
vector equation
vector equation
Manipulator Differential Motion and Statics
P, =°T(g)..T,) 0,1-T(41)..7410,
P°7,0, -T,0, (5.63)
matrix
3xl vector.
Nsjcalcoordinates as
is obtained by determining all the elements using
The manipulator Jacobian
5.57).
Bos. (5.5 (5.59), (5.62) and (5.63). In summary, it is:
Jacobian of a
rates.
164 Robotics and Control
In earlier it has been shown that a manipulator m e
chapters,
Similarly, 6-DOF,at e
have
6-DOF to locate the end-effector in an arbitrary location.
necessary for moving the end-effector in arbitrary direction with-D)
Similarly.
speed. an arbitar
The above discussion is restricted to
manipulator with
Nevertheless, there are manipulators with less than 6-DOF for exactl exactly 6-D
For example, to many annl
manipulate objects lying within a cylindrical wordion
4-DOF SCARA robot described in
that only four end-effector
Examples 3.6 can be used. It can
velocities, v,4 Uy4» Vz4» and o4 can be e
sh
varying four joint velocities at any given configuration.
the control
velocity vector is V, =[v, V,4 V4 l Hence, the c.
and Jacobian J(q) is a 4x artesian
x44 ma
matrix with
10, 6,d, 0,1 and =[ê, è, d^ 0,1.
If manipulator has less 6-DOF (n < 6), it has n
a
is
the singularity. required to track a trajectory that pas
lose
At a singular
configuration, the manipulator loses
freedom. The singular degr
r e e s .
of configurations
the location end-effectorin th are
one or
morc
classified into two catego
d-effector in the e classified hase
categoriesDu
categories
ore asedn
workspace.
Manipulator Differential Motion and Statics
L2
0
5.7.1
Computation of Singularities
Computation of internal singularities can be carried out by analyzing the rank
the Jacobian matrix. The Jacobian matrix loses its rank and becomes
ditioned at values of joint variablesq at which its deteminant vanishes, that
IJl=0 (5.67)
The
C Solutions of Eq. (5.67) give the singular configurations of a
manipulator.
totodnipulators
simplify the that have 3-DOF arm and 3-DOF spherical wrists, it is
possible
Wiy
two separate problem of singularity computation by dividing the problem into
problems:
166 Robotics and Control
[J Ji2]
JJa J2
As the singularities are typical of a configuration and are not
(5.68
1J22l=0 (5.72
Note that this form of Jacobian cannot be used to relate joint velocities ane
end-effector velocities. It is useful only for singularity computations.
ZA
6s 0
Arm end
End-effector
Z3 Zs
of
is the
5W t -s'
n x 1 vector of infinitesimal end-effector displace.
where emenm
x S o , ) caused by the endpoint force . In the above, it is assumed that.
joints of the mechanism are frictionless and the joint torques are the net tor he
e
that balance the endpoint force.
From Eq. (5.52). the Jacobian J relates infinitesimal joint displacement
infinitesimal end-effector displacement p as:
(582
gives
Substituting Eq. (5.82) into Eq. (5.81) and rearranging
SW=(f-Jq))'Sq (5.83
According to the principle of virtual work the manipulator mechanismis
static equilibrium, if and only if, the net virtual work is zero for arbitrary virtu
displacements, that i1s.
SW =0 (5.84
Substituing it in Eq. (5.83), leads to the result
T Jq) (5.85
Equation (5.85) states that the transpose of the Jacobian matrix transforms the
end-effector torque to the corresponding jointtorques.
sOLVED EXAMPLES
Example 5.1 Angular velocity of a frame
A moving frame {1} is represented by the following rotation matrixR, where ais
the angle of rotation of the frame {1} with respect to the base frame. If a s1
function
oftime,
find the angular velocity of {1}.frame
Ca-Sa 01
OR, =Sa Ca
o (5.86
Solution
The given rotation
matrix is the fundamental rotation
frame {1} about z-axis of frame {0), see Eq. (2.54). From
matrix
Eq. (5.26).
S)= R()R (0)
[-dSa -aCa 0 Ca Sa 01
or S)= Ca -Sa 0-Sa Ca 0
0 ol 0
0 - 0]
or
S) = [à 0 0 (5.8
0 0 0
Manipulator Differential Motion and Statics
(5.88)
ular velocity
of the frame about z-axis.
aole 5.2
ENample5 . 2 Velocity of RR-manipulator
h ulate thethe velocity of the tip of the two-link, planar, RR-manipulator arm
5.13.
in Fig.
hON7 is
The frame assignments and joint-link parameters identification
dOut as discussed in Chapter 3 and are shown in Fig. 5.13 and Table 5.1.
of origin ofthis
ene (2) is attached to the end of manipulator and the velocity with
respect to
neis to be determined. This velocity can be expressed
- well as with respect to frame {0). Assume length of each link is 1,
2) as
that 1s
L =L2= 1
matrices are:
The link transformation
C-5 0 C1
S2 C 0 S2
S (5.89)
T0 0 10 0 0 1 0
o 0 0 1J 0 0 1J
c S -C-5C 0 G+CC -S,5]
ad7,=H+SC G%-S,S2 0 S+GS2+SC|
0 0
0 0 0
Y2
L2 Z2
Yo X1
L
Xo Z
Zo
Fig. 5.13 A two-link, RR planar manipulator
Manipulator Differential Motion and Statics
-S-Ci% S,S-CC 0
-S-5G-G
-5,5+C,C -5C2-CS,
0
0
0
G-5,5,+C
0
5.94)
0 0 0
[-GS-SC -CC+S,5, 0 -CS -SC;]
a'7G-S,5 -C.S-5C 0 CC-S,5,
0 0
(5.95)
0
0 0 0
The position vector of endpoint of link 2 is origin of frame {2), that is
D =[0 0 0 11
Substituting these gives,
-S,6,-S2(®, +0,
C+Ca(0, +0, 0
(5.96)
(5.98)
Example 5.3 Jacobian of articulated arm
Determine the manipulator Jacobian matrix for the 3-DOF articulated arm
discussed in Example 3.3 and shown in Fig. 5.14.
Zo
Ls
L2
Xa
2 93
62 B3
Yo
Xo
Fig. 5.14 3-DOF articulated manipulator arm
Solution Each column of Jacobian matrix is computed separately and all the
coluruare combined to form the total Jacobian matrix. The joint displacements
Robotics and Control
(5.100
o ol
The end-effector position vector (for i = 1 andn = 3) is determined fron
Eq. (5.63).
P,0,-°T,O,
or
P=°7-°T
0
obtained as
following the similar steps for joint 2 and joint 3, J, and J, are
[-C,(L4S2 + LS2)]
-S,(LS2 +L1S,)
J LC2 t C (5.103)
S
-C
0
[-L,C,S23]|
-LSS23
J3 LC23 (5.104)
S
C
matrix for the articulated arm
Combining the three columns, the total Jacobian
0 C+ C2
P=0 P=S P=|S, +S2 (5.106)
The unit
Vectors along the revolute joint axes are
(5.107)
Substituting from Eqs. (5.106) and
-S S2 -S12]
C +Ci2 C2
0 0
J (5.109
--h2-Si2] (5.110
LG+Ci2 C2
It is instructive to note that Eq. (5.91) and Eq. (5.96) of Example 5.2, when
written in the form
(5.11
lead to the same result for J as in Eq. (5.109) from which J' of Eq. (5.110
follows.
-S1-S2 -2= S
S2S2 (5.112
J1c+C22Aent
The determinant vanishes for non-zero link
Ci2
Note that the Jacobian determinant is independent of first joint variabie
lengths
whenever
of first jo
e =0 or 6, = T
itheron
Hence, singular configurations occur when
the outer
arm-endpoint is located he
boundary (6, 0) or on the inner boundary (0,= r) of the rea
=
(and the rank becomes one); (ii) the tip velocity components U, and
independent, and (ii) the two links are collinear, as in Fig. 5.10. v,
(b) Joint velocities The joint velocities in terms of ies
(5.113)
The inverse o f J i s
C Sy2
UT S,l-C-C2 S+S (5.114)
velocities are
Thus. the joint
v,Ci2 +ySi2 (5.115)
S2
-,(C+C2)+0,(S1+ Sh2) (5.116)
S2
Atregions near singular points, the denominator in Eqs. (5.115) and
the
that is, to move the endpoint in the vicinity of the
116) is almost zero,
velocities are required.
singularities excessively large joint
The joint velocity profiles for tracking specified trajectory
a at specified speed
the angles that correspond to the endpoint
7.)can be obtained by finding
and 6,) substituting these into Eqs.
position on the trajectory, that is, 6,0)
5.115) and and plotting
(5.116) joint velocities with respect to time.
the
arm configuration
Example 5.5 Singularities of articulated
discussed
Determine the singularities of the 3-DOF articulated arm configuration
n Example 5.3.
olution A close examination of the Jacobian of the articulated arm,
of the Jacobian are linearly
t4 (5.105) indicates that only three of the six rowsthree
ihGependent. It is also known that the arm has only degrees of freedom, that
3x 3 block
S,ilrequires only three variables to be controlled. Hence, the upper
u acobian is only significant and worth considering. The significant Jacobian is
P: Py
= 0
O S 0
S 0-C
T 0 10 0 (5.119
LO 0 0
-S%0 C2 01
C2 0 S2 0
'T7=
T01 00 (5.120
Lo 0 0 1
C -S3 0 0
S3 C3
0 (5.12
0 0 0 1
The rotational transformation matrices for the
three links of the wrist are g ve
by the upper-right-3 x3 submatrix of the respective transformation matriCes
Eqs. (5.119), (5.120) and (5.121), as
[C 0 S1
R =S 0 -C (5.122
Lo I 0
'R=C 0 S2 (5.123
01
Manipulator Differential Motion and Statics
C -S 01
C 0 (5.124)
incobian for the 3-DOF RPY wrists is computed using Eq. (5.62) as all
The Ja
three
are revolute
joints joints.
That is
he
(5.125)
(5.126)
Po
where from Eq. (5.59)
Pa-Ra=lo 1 o (5.127)
lo o i
ad from Eq. (5.63) and Eqs. (5.119), (5.120) and (5.121)
°P=°7, 05--°T,O,
-C,S,Cg +S,S CS,S3 +S,C3 GC 01 f0 fi o o 0To
-S,S,C-CS, SSS-CC SCa o I0 0 0
P 0 0 o o
CC3 -C2S3 S2 0
0 0 loo o 1
(5.128)
Thus.
ro
, =Llo
(5.129)
J=|
Sumilarly, for i 2,
180 Robotics and Control
0 S TO S
P =°R,=|S 0 -C =-C (5.13
(5.131
0
0
and [Rx P|
S 5.132
-C
0
and for i =3
-CS S GCO sGC]
P=°R i -C S,C|SC(5.13
Ca 0
(5.13
CC2
S,C2
From Eqs. (5.129), (5.132) and
(S5.135) the Jacobian for the RPY wrist
0 0
0
0
J=UJ J]= 0 S
0 (5.136
CGC2
0
-C S,C2
S2
Manipulator Differential Motion and Statics
hat the first three rows of this Jacobian for the wrist are zero. This means
Notethatthe
wrist is
have no linear velocities. This is expected because a
wrist
t h at links
th e w r i s t
ientation only.
Ised
f o ro or
of the RPY wrist are determined using the last three rows
ingularities
in Eq.
Jacobian
136). The pseudo-Jacobian is
S
(5.137)
J'=0-C SC2
0 S
or in
will when inverse of J' cannot be found,
occur
The wrist singularities
determinant ofJ
becomes ill conditioned. This occurs whenever
sher wordsJ
the conditions of singularity are found from
nishes. Thus,
[o S CGC
(5.138)
U=0 -C SC2=0
0 S
S(SC-0)+ CC0 +C) =0
C2 =0
(5.139)
La
d2,d-
Solu
on In order to determine the Jacobian of the arm,
first its forward
mo0 in terms of transformation matrices is required.
ematics model
T4cS
o frames are assigned as per
Adetermine the transformation matrices,
SOthm 3.1 and these are shown in Fig. 5.16. The joint-link parameters
18Robotics and Control
L3
d2
Z Z2
obtained there from are tabulated in Table 5.2 and the three transformatian
matrices obtained from these are given in Eqs. (5.141), (5.142) and (5.143),
CO CaSa
-90°
0
0 0
C 0
d
90° L 0 S3 0
C
G0-S, 0
T S0 -1
C
0
(5.14)
O 00
10 0 01
0 10 0
T2=0 0 1
d2
(5.142
[C 0 S
L,C
S 0-C LS3
T=0 1 0 (5.143
L0 0 0
From the individual ts o
transformation matrices following P
transformation matrices are computed:
Manipulator Differential Motion and Statics
C0-5,-S,4,]
S0 C Cd (5.144)
0
0 0 0
GC-S G% LGC-S4
SC C SS L45G+G5.145)
7-°7,7,=| -S 0 C3 -LS3
0 0 0 1
The link 1 is a revolute link and Jacobian for link 1 is computed using
Eq. (5.62) as:
=x (5.146)
Po
where from Eq. (5.59)
1 0 07TO7 0
Po= Ro = 0 1 0o (5.147)
L0 0 1
and from Eq. (5.63)
[LGC-S4] 0 [LCC-Sd
-7 7 L3SC +Cd 5C+a|(5.148)
(5.148)
0 -LgSs LS3
1 0
Thus,
ro LC,C-Sd2 [-LS,C, -Cd21
LCCS,d,
0 x LSC3 +CGd2
J -LSs (5.149)
(5.150)
where
(5.151)
0 -1 0
184 Roboties and Control
Thus
-S,]
C
0 5.15
0
For the revolute joint 3, the Jacobian is computed from
Pa 5.153
where
C0ST
0
-S
P2=R =|S (5.15
0 -1
and
P=°T, 0,-°T, O,
LCGC-Sd,1 C 0-S -Sd2 [LCGC
2 LSC +Cd S0 C Cd2 LgSC
or P -LgS3 0 -1 -LS
(5.15
1 Jlo 0 0
Substituting Eqs. (5.154) and (5.155) in Eq. (5.153) gives
-S1 SLC,C]-L,CGS,]
C1 LgS,C3 -LS1S
J3 = oJl-Ls -LC (5.156
-S -S
C C
0
binins
The Jacobian for the manipulator is, therefore, obtained by co
Eqs. (5.149), (5.152) and (5.156).
[-LS,C3Cd2 -S -LgCS1
LCC3-S,d2 C -LS,S 5.157
0 -LC3
J=| 0 -S
0 C
1
Manipulator Differential Motion and Statics 185
Car
For
determination of singularities, consider first three rows ofJ in Eq. (5.157)
pseudo,
JacobianJ':
the
as
-LSC3-d,C -5-LC,S1
J=| L,GC3-S,d2 C
C LS,S3 (5.158)
-L,C3
determinant ofJ is
The