CHAPTER FIVE

DIFFERENTIAL MOTIONS AND

VELOCITIES

ECE-141 Introduction to Robotics 1

 Introduction to Robotics: Analysis, Control, Applications, Third Edition.
Saeed B. Niku.
© 2020 John Wiley & Sons Ltd. Publishe

This solution manual may not be copied, posted, made available to students,
placed on any outlet, any other electronic media system, and the Internet
without prior expressed consent of the copyright owner.
This material may not be used except in conjunction with the “Introduction to
Robotics: Analysis, Control, Applications”

ECE-141 Introduction to Robotics 2

Introduction

 Differential motions, divided by differential time intervals,

constitute velocity relationships.
 We cover :
 Differential motions of frames
 Differential motions of robots
 Relationship between the above

ECE-141 Introduction to Robotics 3

Differential Motions of a 2-DOF Mechanism

 A mechanism and its velocity diagram

v B v A  v B / A
. . .
l1 1 [ to l1 ]  l2 (1   2 )[ to l2 ]
. . . . . .
 l1 1 sin 1i  l1 1 cos 1 j  l2 (1   2 ) sin(1   2 )i  l2 (1   2 ) cos(1   2 ) j
y

B
y

l2
VB/A

2
l1 VA
A VB

o 1 x x

(a) (b)
ECE-141 Introduction to Robotics 4
Differential Motions

 In matrix form we get:

 vBx   .
  l1 sin 1  l2 sin(1   2 )  l2 sin(1   2 )   1 
    .
 vBy  l
 1 cos   l cos(   ) l cos(   )
2 
2 
1 2 1 2 2 1

 Differential motions are the product of the Jacobian and

the joint differential motions

ECE-141 Introduction to Robotics 5

Differential Motions

 The same relationship can be derived from direct

differentiation:
 xB l1 cos 1  l2 cos(1   2 )

 yB l1 sin 1  l2 sin(1   2 )
dxB  l1 sin 1d1  l2 sin(1   2 )(d1  d 2 )

dyB l1 cos 1d1  l2 cos(1   2 )(d1  d 2 )
 dxB    l1 sin 1  l2 sin(1   2 )  l2 sin(1   2 )   d1 
 dy   l cos   l cos(   ) l cos(   )   d 
 B  1 1 2 1 2 2 1 2   2

Differential Jacobian Differential

motion of B ECE-141 Introduction to Robotics motion of joints 6
The meaning of a Jacobian

 A representation of the configuration of the mechanism

Resulting
motion
B

Re
m sulti
ot ng
io
A n
B
1

Resultin
motion
g
o o o
A B A
1 1

ECE-141 Introduction to Robotics 7

What is the Jacobian
 For an equation in the form of
yi  fi ( x1 , x2 , x3 ,..., x j )
 The derivatives are:
 f1 f1 f1
 y 
 1 x 1 x  x   x 2  L   xj
x j
 1 2

 f 2 f 2 f 2
 y2   x1   x2  L   xj
 x1 x2 x j
M

 f i f i f i
 y 
 i x 1 x  x   x2  L   xj
 1 2 x j
ECE-141 Introduction to Robotics 8
Jacobian
 This can be rearranged into:

 f1 f1 f1 

  y1   x    x1 
  x2 x j  
 1 
   f 2  
  y2   K K K    x2   fi 
   x1   or  yi      x j 
   M    x j 
 M    M
   fi fi   
  yi   x
 x1
 x j   j 

ECE-141 Introduction to Robotics 9

Robot’s Jacobian
 For a robot, we get:

 dx     d1 
 dy     d 
     2
 dz   Robot   d3 
     or  D   J  D 
  x   Jacobian   d 4 
 y    d5 
    
 z    d 6 

ECE-141 Introduction to Robotics 10

Differential versus Large-Scale Motions

 Large scale motions happen once; differential motions

happen continuously.
 The final consequences are different.

l

B C

l
l A 

ECE-141 Introduction to Robotics 11

Differential Motions of a Frame versus a Robot

 A frame may move by any means, including as a result of

the motions of a robot arm when the frame is attached to
the end effector.
 We will study the differential motions of a frame and of a
robot separately, then combine them together.

z z
z
o' Hand differential
o n' o'
a' motions a' o n'
a n a n

Joint differential bot

Ro
motions

x y x y x
y
ECE-141 Introduction to Robotics 12
Differential Motions of a Frame

 Differential motions of a frame can be divided into the

following:
 Differential translations
 Differential rotations
 Differential transformations (combinations of translations
and rotations)

ECE-141 Introduction to Robotics 13

Differential Translations

 The translation of a frame at differential values or

Trans(dx,dy,dz). This means the frame has moved a
differential amount along the x-, y-, and z-axes.

ECE-141 Introduction to Robotics 14

Differential Rotations about Reference Axes

 Small rotations of the frame, represented by

Rot (q, d )
 Differential rotations about the x-, y-, and z-
axes are defined by  x,  y ,  z.

ECE-141 Introduction to Robotics 15

Differential trigonometric functions

 We use the following relationships:

sin  x  x (in radians)

cos  x 1

ECE-141 Introduction to Robotics 16

Differential Rotations about Reference Axes

1 0 0 0 1 0 0 0
0 1   x 0 0 1  n 0
Rot ( x,  x )    Rot ( n,  n)   
0 x 1 0 0 n 1 0
   
0 0 0 1 0 0 0 1

 1 0  y 0  1 0 o 0
 0 1 0 0  0 1 0 0
Rot ( y ,  y )    Rot (o,  o)   
y 0 1 0  o 0 1 0
   
 0 0 0 1  0 0 0 1

 1 z 0 0  1 a 0 0
 z 1 0 0  a 1 0 0
Rot ( z ,  z )    Rot ( a ,  a )   
0 0 1 0 0 0 1 0
   
0 0 0 1 0 0 0 1
ECE-141 Introduction to Robotics 17
Multiple Differential Rotations about Reference
Axes

 Multiple differential rotations:

 Results are order-dependent
 If higher-order differentials are eliminated, the results are the same.

1 0 0 0  1 0 y 0  1 0 y 0
 
0 1   x 0  0 1 0 0    x y 1   x 0 
Rot ( x,  x) Rot ( y,  y )   
0  x 1 0    y 0 1 0     y  x 1 0 
   
0 0 0 1  0 0 0 1  0 0 0 1 

 1 0  y 0  1 0 0 0  1  x y  y 0
 
 0 1 0 0  0 1   x 0   0 1 x 0 
Rot ( y,  y ) Rot ( x,  x)   
 y 0 1 0  0  x 1 0     y  x 1 0 
  
 0 0 0 1  0 0 0 1   0 0 0 1 
ECE-141 Introduction to Robotics 18
Differential Rotations about a General Axis q

 Based on the previous slide, the order does not matter:

(d )q ( x)i  ( y ) j  ( z )k
Rot (q, d ) Rot ( x,  x) Rot ( y,  y ) Rot ( z ,  z )
1 0 0 0  1 0  y 0  1   z 0 0
 0 1   x 0  0 1 0 0   z 1 0 0
    
0  x 1 0    y 0 1 0  0 0 1 0
    
 0 0 0 1   0 0 0 1  0 0 0 1
 1 z  y 0 z

  x y   z   x y z  1   x 0  z q

  d
   y   x z  x   y z 1 0
  y
 0 0 0 1 
x
y
ECE-141 Introduction to Robotics x 19
Differential Rotations about a General Axis q

 Eliminating the higher-order differentials we get:

Rot (q, d ) Rot ( x,  x) Rot ( y,  y ) Rot ( z ,  z ) 

 1 z y 0
 z 1 x 0
 
 y  x 1 0
 
 0 0 0 1

ECE-141 Introduction to Robotics 20

Differential Rotations of a Frame

 Assume we may have differential translation and rotations.

 dT is the change in a frame. Therefore:

T  dT  [Trans(dx, dy, dz ) Rot (q, d )]T 

or  dT  [Trans (dx, dy, dz ) Rot (q, d )  I ] T 

 dT    T 
where    [Trans (dx, dy, dz ) Rot (q, d )  I ]

ECE-141 Introduction to Robotics 21

Differential Operator

 Differential Operator can be derived.

 It is NOT a frame or transformation matrix and it does not
follow the same requirements.

 0   z  y dx 
 z 0   x dy 
  
 y x 0 dz 
 
 0 0 0 0

ECE-141 Introduction to Robotics 22

The Differential Change

 Represents the changes in each of the elements in the

matrix representation of a frame.
 dnx dox dax dpx 
 dn do y da y dp y 
dT  y
 dnz doz daz dpz 
 
 0 0 0 0 
 The new frame is:
Tnew dT  Told

ECE-141 Introduction to Robotics 23

Differential Changes between Frames

 A differential change can be relative to the reference

frame or the current frame:

         
dT   T  T  T

T    T  T      
1 1
 T  T

   T    T   
T 1
 0  T z T
y T
dx 
 T 
 z 0  x
T T
dy 
  T y T
x 0 T
dz 
 
 0 0 0 0 
ECE-141 Introduction to Robotics 24
Differential Changes between Frames

 Differential change can be can be calculated directly, or

from:
T
 x δ n
T
 y δ o
T
 z δ a
T
d x n [δ p  d]
T
d y o [δ p  d]
T
d z a [δ p  d]
ECE-141 Introduction to Robotics 25
Differential Motions of a Robot and Its Hand
Frame

 The differential motions of the robot can be expressed by:

 dx     d1 
 dy     d 
     2
 dz   Robot   d3 
     or  D   J  D 
  x   Jacobian   d 4 
 y    d 5 
    
 z    d 6 

ECE-141 Introduction to Robotics 26

Differential Motions of a Robot and Its Hand
Frame

 Use the previous equation to relate the robot joint

differential motions to the end-effector motions,
 Use the data in D to form  .
 Calculate dT.
 Calculate the new position and orientation of the end-
effector
 Repeat.

ECE-141 Introduction to Robotics 27

Calculation of the Jacobian

 Direct derivation is possible for position equations, but not

practical for orientation.
 Indirect derivation, especially relative to the current frame
is easier and more practical.
 For:  T6 d x   T6 J11 T6 J12 . . . T6 J16   d1 
 T6   T6  
 d y   J 21
T6
J 22 . . . T6
J 26   d 2 
 T6 d z   T6 J 31 . . . . T6
J 36   . 
 T6   T6 T6  
  x   J 41 . . . . J 46   . 
 T6   T6 J . . . . T6
J 56   . 
 T y   T 51  
 6 z   6 J 61 . . . . T6
J 66   d 6 
ECE-141 Introduction to Robotics 28
Calculation of the Jacobian: Cont.

 If joint i under consideration is a revolute

joint:
T6
J1i ( nx p y  n y px )
T6
J 2i ( ox p y  o y px )
T6
J 3i ( ax p y  a y px )
T6
J 4i nz
T6
J 5i oz
T6
J 6i az
ECE-141 Introduction to Robotics 29
Calculation of the Jacobian: Cont.

 If joint i under consideration is a prismatic

joint: T6
J1i nz
T6
J 2i oz
T6
J 3i az
T6
J 4i 0
T6
J 5i 0
T6
J 6i 0
ECE-141 Introduction to Robotics 30
Calculation of the Jacobian: Cont.

 Where the information comes from:

For column 1, use oT6  A1 A2 A3 A4 A5 A6

For column 2, use 1T6  A2 A3 A4 A5 A6
For column 3, use 2T6  A3 A4 A5 A6
For column 4, use 3T6  A4 A5 A6
For column 5, use 4T6  A5 A6
For column 6, use 5T6  A6

ECE-141 Introduction to Robotics 31

Inverse Jacobian

 Inverse Jacobian is the real objective

 D   J  D 
 J  1   D   J  1   J  D 

 D   J  1   D 
 And similarly

 T6 J  1   T6 D   T6 J  1   T6 J   D 
T6  1
D  J   D 
   T6

ECE-141 Introduction to Robotics 32

Calculation of Inverse Jacobian

 Can be calculated from the Jacobian, but impractical for a

6 by 6 large matrix.
 Can be derived directly from the forward kinematic
equations instead.

ECE-141 Introduction to Robotics 33

Derivation of the Inverse Jacobian

 Example: 6-axis robot first joint:

px S1  p y C1 0
 py 
1
1 tan   and 1 1  180
 px 

px S1  p y C1
dpx S1  px C1d1 dp y C1  p y S1d1
d1 ( px C1  p y S1 )  dpx S1  dp y C1
 dpx S1  dp y C1
d1 
( px C1  p y S1 )
ECE-141 Introduction to Robotics 34
Derivation of the Inverse Jacobian: Cont.

 Similarly:
S 234 (C1ax  S1a y ) C234 az
C234 (d 2  d3  d 4 )(C1ax  S1a y )  S 234 [ a x S1d1  C1da x  a yC1d1  S1da y ]
 S234 (d 2  d3  d 4 )az  C234 daz

S 234 [ax S1d1  C1dax  a y C1d1  S1da y ]  C234da z

( d 2  d 3  d 4 ) 
C234 (C1ax  S1a y )  S 234 a z

ECE-141 Introduction to Robotics 35

Derivation of the Inverse Jacobian: Cont.

 and:
2a2 a3C3 ( pxC1  p y S1  C234 a4 ) 2  ( p z  S 234a4 ) 2  a22  a32

 2a2 a3 S3d3 2( pxC1  p y S1  C234 a4 )

[C1dpx  px S1d1  S1dp y  p y C1d1  a4 S 234 (d 2  d3  d 4 )]
 2( pz  S 234 a4 )[dpz  a4C234 (d 2  d3  d 4 )]

ECE-141 Introduction to Robotics 36

Derivation of the Inverse Jacobian: Cont.

 and similarly, equations for the remaining can be derived,

including:

S6  S 234 (C1nx  S1n y )  C234 nz

C6 d 6  C234 (C1nx  S1n y )(d 2  d3  d 4 )

 S 234 ( S1nx d1  C1dnx  C1n y d1  S1dn y )
 S 234 nz (d 2  d3  d 4 )  C234 dnz

ECE-141 Introduction to Robotics 37

Derivation of the Inverse Jacobian: Cont.

 These equations may be used to control the differential

motions of the robot.

ECE-141 Introduction to Robotics 38

Calculation of Jacobian with screw-based
mechanics

 Same principles apply. The meaning and the use of

Jacobians is similar in screw-based mechanics. The
derivation is different. ŝ 3

ŝ4
ŝ2

S3 S4 z
ŝ1

S2
x
Fixed
base

S1 y

ECE-141 Introduction to Robotics 39

Cont.

 Denoting the Jacobian of the robot as J, we write:

  
 v  J   D
 
 

  is the vector of angular velocities of the end effector

with respect to the reference frame, v is the vector of
linear velocities of the end effector with respect to the
reference frame, and  D   is the angular velocities of the
joints

ECE-141 Introduction to Robotics 40

Cont.

ECE-141 Introduction to Robotics 41

Cont.

ECE-141 Introduction to Robotics 42

Process of calculating the Jacobian

ECE-141 Introduction to Robotics 43

Cont.

ECE-141 Introduction to Robotics 44

Cont.

ECE-141 Introduction to Robotics 45

Example

 Intermediate reference frame at z4 (s5)

zU zn

ŝ1 ŝ5
a4
a2 ŝ3 a3 ŝ6
ŝ2 xU xn
ŝ4

ECE-141 Introduction to Robotics 46

Joint 5:

ECE-141 Introduction to Robotics 47

Joint 6:

ECE-141 Introduction to Robotics 48

Joint 4:

ECE-141 Introduction to Robotics 49

Joint 3:

ECE-141 Introduction to Robotics 50

Joint 2:

ECE-141 Introduction to Robotics 51

Joint 1:

ECE-141 Introduction to Robotics 52

Jacobian:
 S234 0 0 0 0 S5 
 0 1 1 1 0  C5 
 
5
 C234 0 0 0 1 0 
j  
 0 a2 S34  a3S4 a3 S 4 0 0 0 
 a2C2  a3C23  a4C234 0 0 0 0 0 
 
 0 a2C34  a3C4  a4 a3C4  a4 a4 0 0 
5
x S2341  S56
5
 y  
2   3   4  C5 6
5
z C2341  5
5
v x a2 S34
2  a3 S 4 ( 2   3 )
5
v y ( a2C2  a3C23  a4C234 )1
5
v z ( a2C34  a3C4  a4 )
2  ( a3C4  a4 ) 3  a45

ECE-141 Introduction to Robotics 53

Inverse Jacobian

 Inverting the matrix is too involved.

 Direct differentiation is easier and quicker.

ECE-141 Introduction to Robotics 54

Jacobian of parallel robots

ECE-141 Introduction to Robotics 55

Cont.

ECE-141 Introduction to Robotics 56

Example: Jacobian of 6-6 Stewart platform

 We write 6 loop equations and differentiate

them to form the Jacobian as follows
P  bi ai  d i for i 1 6 B3
B4 d̂1
B2
P b1
P  b bi d i +di d i B5

B6
B1

A3
P
A4 A2
zo d1
yo
O a1
xo
A5 A1

A6
ECE-141 Introduction to Robotics 57
Notes:

ECE-141 Introduction to Robotics 58

Derivation:

ECE-141 Introduction to Robotics 59

 CHAPTER FIVE
DIFFERENTIAL MOTIONS AND
VELOCITIES
Dr. Sohail Khan

ECE-141 Introduction to Robotics 60

 Introduction to Robotics: Analysis, Control, Applications, Third Edition.
Saeed B. Niku.
© 2020 John Wiley & Sons Ltd. Publishe

This solution manual may not be copied, posted, made available to students,
placed on any outlet, any other electronic media system, and the Internet
without prior expressed consent of the copyright owner.
This material may not be used except in conjunction with the “Introduction to
Robotics: Analysis, Control, Applications”

ECE-141 Introduction to Robotics 61

Introduction

 Differential motions, divided by differential time intervals,

constitute velocity relationships.
 We cover :
 Differential motions of frames
 Differential motions of robots
 Relationship between the above

ECE-141 Introduction to Robotics 62

Differential Motions of a 2-DOF Mechanism

 A mechanism and its velocity diagram

v B v A  v B / A
. . .
l1 1 [ to l1 ]  l2 (1   2 )[ to l2 ]
. . . . . .
 l1 1 sin 1i  l1 1 cos 1 j  l2 (1   2 ) sin(1   2 )i  l2 (1   2 ) cos(1   2 ) j
y

B
y

l2
VB/A

2
l1 VA
A VB

o 1 x x

(a) (b)
ECE-141 Introduction to Robotics 63
Differential Motions

 In matrix form we get:

 vBx   .
  l1 sin 1  l2 sin(1   2 )  l2 sin(1   2 )   1 
    .
 vBy  l
 1 cos   l cos(   ) l cos(   )
2 
2 
1 2 1 2 2 1

 Differential motions are the product of the Jacobian and

the joint differential motions

ECE-141 Introduction to Robotics 64

Differential Motions

 The same relationship can be derived from direct

differentiation:
 xB l1 cos 1  l2 cos(1   2 )

 yB l1 sin 1  l2 sin(1   2 )
dxB  l1 sin 1d1  l2 sin(1   2 )(d1  d 2 )

dyB l1 cos 1d1  l2 cos(1   2 )(d1  d 2 )
 dxB    l1 sin 1  l2 sin(1   2 )  l2 sin(1   2 )   d1 
 dy   l cos   l cos(   ) l cos(   )   d 
 B  1 1 2 1 2 2 1 2   2

Differential Jacobian Differential

motion of B ECE-141 Introduction to Robotics motion of joints65
The meaning of a Jacobian

 A representation of the configuration of the mechanism

Resulting
motion
B

Re
m sulti
ot ng
io
A n
B
1

Resultin
motion
g
o o o
A B A
1 1

ECE-141 Introduction to Robotics 66

What is the Jacobian
 For an equation in the form of
yi  fi ( x1 , x2 , x3 ,..., x j )
 The derivatives are:
 f1 f1 f1
 y 
 1 x 1 x  x   x 2  L   xj
x j
 1 2

 f 2 f 2 f 2
 y2   x1   x2  L   xj
 x1 x2 x j
M

 f i f i f i
 y 
 i x 1 x  x   x2  L   xj
 1 2 x j
ECE-141 Introduction to Robotics 67
Jacobian
 This can be rearranged into:

 f1 f1 f1 

  y1   x    x1 
  x2 x j  
 1 
   f 2  
  y2   K K K    x2   fi 
   x1   or  yi      x j 
   M    x j 
 M    M
   fi fi   
  yi   x
 x1
 x j   j 

ECE-141 Introduction to Robotics 68

Robot’s Jacobian
 For a robot, we get:

 dx     d1 
 dy     d 
     2
 dz   Robot   d3 
     or  D   J  D 
  x   Jacobian   d 4 
 y    d5 
    
 z    d 6 

ECE-141 Introduction to Robotics 69

Differential versus Large-Scale Motions

 Large scale motions happen once; differential motions

happen continuously.
 The final consequences are different.

l

B C

l
l A 

ECE-141 Introduction to Robotics 70

Differential Motions of a Frame versus a Robot

 A frame may move by any means, including as a result of

the motions of a robot arm when the frame is attached to
the end effector.
 We will study the differential motions of a frame and of a
robot separately, then combine them together.

z z
z
o' Hand differential
o n' o'
a' motions a' o n'
a n a n

Joint differential bot

Ro
motions

x y x y x
y
ECE-141 Introduction to Robotics 71
Differential Motions of a Frame

 Differential motions of a frame can be divided into the

following:
 Differential translations
 Differential rotations
 Differential transformations (combinations of translations
and rotations)

ECE-141 Introduction to Robotics 72

Differential Translations

 The translation of a frame at differential values or

Trans(dx,dy,dz). This means the frame has moved a
differential amount along the x-, y-, and z-axes.

ECE-141 Introduction to Robotics 73

Differential Rotations about Reference Axes

 Small rotations of the frame, represented by

Rot (q, d )
 Differential rotations about the x-, y-, and z-
axes are defined by  x,  y ,  z.

ECE-141 Introduction to Robotics 74

Differential trigonometric functions

 We use the following relationships:

sin  x  x (in radians)

cos  x 1

ECE-141 Introduction to Robotics 75

Differential Rotations about Reference Axes

1 0 0 0 1 0 0 0
0 1   x 0 0 1  n 0
Rot ( x,  x )    Rot ( n,  n)   
0 x 1 0 0 n 1 0
   
0 0 0 1 0 0 0 1

 1 0  y 0  1 0 o 0
 0 1 0 0  0 1 0 0
Rot ( y ,  y )    Rot (o,  o)   
y 0 1 0  o 0 1 0
   
 0 0 0 1  0 0 0 1

 1 z 0 0  1 a 0 0
 z 1 0 0  a 1 0 0
Rot ( z ,  z )    Rot ( a ,  a )   
0 0 1 0 0 0 1 0
   
0 0 0 1 0 0 0 1
ECE-141 Introduction to Robotics 76
Multiple Differential Rotations about Reference
Axes

 Multiple differential rotations:

 Results are order-dependent
 If higher-order differentials are eliminated, the results are the same.

1 0 0 0  1 0 y 0  1 0 y 0
 
0 1   x 0  0 1 0 0    x y 1   x 0 
Rot ( x,  x) Rot ( y,  y )   
0  x 1 0    y 0 1 0     y  x 1 0 
   
0 0 0 1  0 0 0 1  0 0 0 1 

 1 0  y 0  1 0 0 0  1  x y  y 0
 
 0 1 0 0  0 1   x 0   0 1 x 0 
Rot ( y,  y ) Rot ( x,  x)   
 y 0 1 0  0  x 1 0     y  x 1 0 
  
 0 0 0 1  0 0 0 1   0 0 0 1 
ECE-141 Introduction to Robotics 77
Differential Rotations about a General Axis q

 Based on the previous slide, the order does not matter:

(d )q ( x)i  ( y ) j  ( z )k
Rot (q, d ) Rot ( x,  x) Rot ( y,  y ) Rot ( z ,  z )
1 0 0 0  1 0  y 0  1   z 0 0
 0 1   x 0  0 1 0 0   z 1 0 0
    
0  x 1 0    y 0 1 0  0 0 1 0
    
 0 0 0 1   0 0 0 1  0 0 0 1
 1 z  y 0 z

  x y   z   x y z  1   x 0  z q

  d
   y   x z  x   y z 1 0
  y
 0 0 0 1 
x
y
ECE-141 Introduction to Robotics x 78
Differential Rotations about a General Axis q

 Eliminating the higher-order differentials we get:

Rot (q, d ) Rot ( x,  x) Rot ( y,  y ) Rot ( z ,  z ) 

 1 z y 0
 z 1 x 0
 
 y  x 1 0
 
 0 0 0 1

ECE-141 Introduction to Robotics 79

Differential Rotations of a Frame

 Assume we may have differential translation and rotations.

 dT is the change in a frame. Therefore:

T  dT  [Trans(dx, dy, dz ) Rot (q, d )]T 

or  dT  [Trans (dx, dy, dz ) Rot (q, d )  I ] T 

 dT    T 
where    [Trans (dx, dy, dz ) Rot (q, d )  I ]

ECE-141 Introduction to Robotics 80

Differential Operator

 Differential Operator can be derived.

 It is NOT a frame or transformation matrix and it does not
follow the same requirements.

 0   z  y dx 
 z 0   x dy 
  
 y x 0 dz 
 
 0 0 0 0

ECE-141 Introduction to Robotics 81

The Differential Change

 Represents the changes in each of the elements in the

matrix representation of a frame.
 dnx dox dax dpx 
 dn do y da y dp y 
dT  y
 dnz doz daz dpz 
 
 0 0 0 0 
 The new frame is:
Tnew dT  Told

ECE-141 Introduction to Robotics 82

Differential Changes between Frames

 A differential change can be relative to the reference

frame or the current frame:

         
dT   T  T  T

T    T  T      
1 1
 T  T

   T    T   
T 1
 0  T z T
y T
dx 
 T 
 z 0  x
T T
dy 
  T y T
x 0 T
dz 
 
 0 0 0 0 
ECE-141 Introduction to Robotics 83
Differential Changes between Frames

 Differential change can be can be calculated directly, or

from:
T
 x δ n
T
 y δ o
T
 z δ a
T
d x n [δ p  d]
T
d y o [δ p  d]
T
d z a [δ p  d]
ECE-141 Introduction to Robotics 84
Differential Motions of a Robot and Its Hand
Frame

 The differential motions of the robot can be expressed by:

 dx     d1 
 dy     d 
     2
 dz   Robot   d3 
     or  D   J  D 
  x   Jacobian   d 4 
 y    d 5 
    
 z    d 6 

ECE-141 Introduction to Robotics 85

Differential Motions of a Robot and Its Hand
Frame

 Use the previous equation to relate the robot joint

differential motions to the end-effector motions,
 Use the data in D to form  .
 Calculate dT.
 Calculate the new position and orientation of the end-
effector
 Repeat.

ECE-141 Introduction to Robotics 86

Calculation of the Jacobian

 Direct derivation is possible for position equations, but not

practical for orientation.
 Indirect derivation, especially relative to the current frame
is easier and more practical.
 For:  T6 d x   T6 J11 T6 J12 . . . T6 J16   d1 
 T6   T6  
 d y   J 21
T6
J 22 . . . T6
J 26   d 2 
 T6 d z   T6 J 31 . . . . T6
J 36   . 
 T6   T6 T6  
  x   J 41 . . . . J 46   . 
 T6   T6 J . . . . T6
J 56   . 
 T y   T 51  
 6 z   6 J 61 . . . . T6
J 66   d 6 
ECE-141 Introduction to Robotics 87
Calculation of the Jacobian: Cont.

 If joint i under consideration is a revolute

joint:
T6
J1i ( nx p y  n y px )
T6
J 2i ( ox p y  o y px )
T6
J 3i ( ax p y  a y px )
T6
J 4i nz
T6
J 5i oz
T6
J 6i az
ECE-141 Introduction to Robotics 88
Calculation of the Jacobian: Cont.

 If joint i under consideration is a prismatic

joint: T6
J1i nz
T6
J 2i oz
T6
J 3i az
T6
J 4i 0
T6
J 5i 0
T6
J 6i 0
ECE-141 Introduction to Robotics 89
Calculation of the Jacobian: Cont.

 Where the information comes from:

For column 1, use oT6  A1 A2 A3 A4 A5 A6

For column 2, use 1T6  A2 A3 A4 A5 A6
For column 3, use 2T6  A3 A4 A5 A6
For column 4, use 3T6  A4 A5 A6
For column 5, use 4T6  A5 A6
For column 6, use 5T6  A6

ECE-141 Introduction to Robotics 90

Inverse Jacobian

 Inverse Jacobian is the real objective

 D   J  D 
 J  1   D   J  1   J  D 

 D   J  1   D 
 And similarly

 T6 J  1   T6 D   T6 J  1   T6 J   D 
T6  1
D  J   D 
   T6

ECE-141 Introduction to Robotics 91

Calculation of Inverse Jacobian

 Can be calculated from the Jacobian, but impractical for a

6 by 6 large matrix.
 Can be derived directly from the forward kinematic
equations instead.

ECE-141 Introduction to Robotics 92

Derivation of the Inverse Jacobian

 Example: 6-axis robot first joint:

px S1  p y C1 0
 py 
1
1 tan   and 1 1  180
 px 

px S1  p y C1
dpx S1  px C1d1 dp y C1  p y S1d1
d1 ( px C1  p y S1 )  dpx S1  dp y C1
 dpx S1  dp y C1
d1 
( px C1  p y S1 )
ECE-141 Introduction to Robotics 93
Derivation of the Inverse Jacobian: Cont.

 Similarly:
S 234 (C1ax  S1a y ) C234 az
C234 (d 2  d3  d 4 )(C1ax  S1a y )  S 234 [ a x S1d1  C1da x  a yC1d1  S1da y ]
 S234 (d 2  d3  d 4 )az  C234 daz

S 234 [ax S1d1  C1dax  a y C1d1  S1da y ]  C234da z

( d 2  d 3  d 4 ) 
C234 (C1ax  S1a y )  S 234 a z

ECE-141 Introduction to Robotics 94

Derivation of the Inverse Jacobian: Cont.

 and:
2a2 a3C3 ( pxC1  p y S1  C234 a4 ) 2  ( p z  S 234a4 ) 2  a22  a32

 2a2 a3 S3d3 2( pxC1  p y S1  C234 a4 )

[C1dpx  px S1d1  S1dp y  p y C1d1  a4 S 234 (d 2  d3  d 4 )]
 2( pz  S 234 a4 )[dpz  a4C234 (d 2  d3  d 4 )]

ECE-141 Introduction to Robotics 95

Derivation of the Inverse Jacobian: Cont.

 and similarly, equations for the remaining can be derived,

including:

S6  S 234 (C1nx  S1n y )  C234 nz

C6 d 6  C234 (C1nx  S1n y )(d 2  d3  d 4 )

 S 234 ( S1nx d1  C1dnx  C1n y d1  S1dn y )
 S 234 nz (d 2  d3  d 4 )  C234 dnz

ECE-141 Introduction to Robotics 96

Derivation of the Inverse Jacobian: Cont.

 These equations may be used to control the differential

motions of the robot.

ECE-141 Introduction to Robotics 97

Calculation of Jacobian with screw-based
mechanics

 Same principles apply. The meaning and the use of

Jacobians is similar in screw-based mechanics. The
derivation is different. ŝ 3

ŝ4
ŝ2

S3 S4 z
ŝ1

S2
x
Fixed
base

S1 y

ECE-141 Introduction to Robotics 98

Cont.

 Denoting the Jacobian of the robot as J, we write:

  
 v  J   D
 
 

  is the vector of angular velocities of the end effector

with respect to the reference frame, v is the vector of
linear velocities of the end effector with respect to the
reference frame, and  D   is the angular velocities of the
joints

ECE-141 Introduction to Robotics 99

Cont.

ECE-141 Introduction to Robotics 100

Cont.

ECE-141 Introduction to Robotics 101

Process of calculating the Jacobian

ECE-141 Introduction to Robotics 102

Cont.

ECE-141 Introduction to Robotics 103

Cont.

ECE-141 Introduction to Robotics 104

Example

 Intermediate reference frame at z4 (s5)

zU zn

ŝ1 ŝ5
a4
a2 ŝ3 a3 ŝ6
ŝ2 xU xn
ŝ4

ECE-141 Introduction to Robotics 105

Joint 5:

ECE-141 Introduction to Robotics 106

Joint 6:

ECE-141 Introduction to Robotics 107

Joint 4:

ECE-141 Introduction to Robotics 108

Joint 3:

ECE-141 Introduction to Robotics 109

Joint 2:

ECE-141 Introduction to Robotics 110

Joint 1:

ECE-141 Introduction to Robotics 111

Jacobian:
 S234 0 0 0 0 S5 
 0 1 1 1 0  C5 
 
5
 C234 0 0 0 1 0 
j  
 0 a2 S34  a3S4 a3 S 4 0 0 0 
 a2C2  a3C23  a4C234 0 0 0 0 0 
 
 0 a2C34  a3C4  a4 a3C4  a4 a4 0 0 
5
x S2341  S56
5
 y  
2   3   4  C5 6
5
z C2341  5
5
v x a2 S34
2  a3 S 4 ( 2   3 )
5
v y ( a2C2  a3C23  a4C234 )1
5
v z ( a2C34  a3C4  a4 )
2  ( a3C4  a4 ) 3  a45

ECE-141 Introduction to Robotics 112

Inverse Jacobian

 Inverting the matrix is too involved.

 Direct differentiation is easier and quicker.

ECE-141 Introduction to Robotics 113

Jacobian of parallel robots

ECE-141 Introduction to Robotics 114

Cont.

ECE-141 Introduction to Robotics 115

Example: Jacobian of 6-6 Stewart platform

 We write 6 loop equations and differentiate

them to form the Jacobian as follows
P  bi ai  d i for i 1 6 B3
B4 d̂1
B2
P b1
P  b bi d i +di d i B5

B6
B1

A3
P
A4 A2
zo d1
yo
O a1
xo
A5 A1

A6
ECE-141 Introduction to Robotics 116
Notes:

ECE-141 Introduction to Robotics 117

Derivation:

ECE-141 Introduction to Robotics 118