Spatial Descriptions

• Task Description

Kinematics • Transformations
• Representations

Configuration Parameters
Manipulator Prismatic Joint
A set of position parameters that describes
the full configuration of the system.
Revolute 9 parameters/link
k i
Joint Lin

End-Effector
n moving link Generalized coordinates
Base Links:
1 fixed link A set of independent configuration parameters
Revolute (1 DOF) Degrees of Freedom
Joints: Number of generalized coordinates
Prismatic (1 DOF)

Generalized Coordinates Generalized Coordinates

6 parameters { 3 positions
3 orientations

n moving links: 6n parameters

1
 Generalized Coordinates End-Efector Configuration
Parameters

{0} 0n+1
6 parameters { 3 positions
3 orientations
5 constraints

A set of m parameters:
( x1 , x2 , x3 ,… , xm )
n moving links: 6n parameters that completely specifies the end-effector
n 1 d.o.f. joints: 5n constraints
position and orientation with respect to{0}
d.o.f. (system): 6n - 5n = n

Operational Coordinates Joint Coordinates Joint Space

θ3
On+1 : Operational point θ3
θ2 θ

{0} 0n+1 θ2
α θ1 α
( xy )
θ1
A set x1 , x2 ,… , xm 0 y

of m0 independent configuration parameters

x
m 0 : number of degrees of freedom
Operational Space
of the end-effector. Operational Coordinates

Redundancy Position of a Point

With respect to a fixed

A robot is said to origin O, the position
be redundant if P of a point P is described
by the vector OP or
n > m0 p simply by p.

O
Degrees of redundancy: n − m0

2
 Rigid Body Configuration Coordinate Frames

Q P p = OP
E q q = OQ
Euclidian p
Space
0
z
p P
Cartesian Frame 0 y

Rigid Body Configuration Rotation Matrix ^ {A}

ZA ^
Y
^
Y ⎡r11 r12 r13 ⎤ ^ {B}
B

^ {B}
ZB
⎢ ⎥
B
^ {A}
B R = ⎢r21 r22 r23 ⎥
ZA ZB A
^A
YBP
^ {B}
⎣⎢r31 r32 r33 ⎥⎦
A
P ZB ^
XB
^
XB ^
Y A
A
XˆB = BAR BXˆB ^
XB
^
Y ^
A X A

Position: AP
^
X A ⎡1⎤ ⎡0⎤ ⎡0⎤
Orientation: {A^ ^ ,A^Z }
XB,AY ⎢ ⎥ ⎢ ⎥ A ⎢ ⎥
XˆB = R ⎢0⎥ YˆB = BR ⎢1⎥ ZB = BR ⎢0⎥
B B A A A A Aˆ
B
R=⎡⎣AXˆB AYˆB AZˆB⎤⎦
Describe rotations of {B} with respect to {A}
⎣⎢1⎦⎥
A
⎢⎣0⎥⎦ ⎢⎣0⎥⎦ B
Rotation Matrix

Rotation Matrix ^ {A}

Rotation Matrix ⎡ B Xˆ AT ⎤
ZA ⎢ ⎥ T
A
R= ⎡⎣AXˆB AYˆB AZˆB⎤⎦
A
BR = ⎡⎣ AXˆB YˆB
A
ZB ⎤⎦= ⎢ BYˆAT ⎥ = ⎡⎣ B Xˆ A BYˆA B Zˆ A ⎤⎦ = AB RT
Aˆ
B
^ ⎢ B ˆT ⎥
⎣⎢ Z A ⎦⎥
Y
^ {B}
B
ZB
Dot Product
⎡ Xˆ B . Xˆ A ⎤ ^
Y
A
B R = AB RT
A
⎢ ⎥ ^
X B = ⎢ Xˆ B .YˆA ⎥
A ˆ
^
XB
⎢ ˆ ˆ ⎥
X A Inverse of Rotation Matrices
⎣⎢ X B .Z A ⎦⎥
⎡XˆB.Xˆ A YˆB.Xˆ A ZˆB.Xˆ A⎤
B
X T
A
A
B R−1 = AB R = BART
⎢ ⎥
B R = ⎢ XB.YA
A ˆ ˆ YˆB.YˆA ZˆB.YˆA ⎥
⎢ˆ ˆ ⎥
YˆB.ZˆA ZˆB.ZˆA ⎦⎥
A
B R −1 = BA R T Orthonormal Matrix
⎣⎢ XB.ZA

3
 ^
Example ^
Z A Description of a Frame
Y B {A}
Z^B
{A} ^
XB {B}

{B} A
PBORG ^
YB
^
Z O ^
B Y A

^
X ^
B X
F1 I
A
0 0 ← X AT B

R = G0 JJ A A A A
0 −1 ← B YAT Frame {B}: X B , YB , Z B , PBorg
GH 0
A

lBq = n
B

0K ← Z s
B T
1 A A
A
B R PBorg
↑ ↑ ↑
Aˆ
XB AYˆB AZˆB

Mapping Translations {B} P

A
changing descriptions from frame to frame P B
{A} P
Rotations PBORG
OB
{A}
{B} If P is given in {B}: BP OA
P
F X . PI F X
B B B T
I
P = G Y .P J = G Y JJ . P
A A
changing the position description of a point P
A
GG Z . P JJ GG Z
B
A
B B T
A
B

JK
H K H
B
A
B B T
A
OBP OAP (Two different vectors)

PBORG : POB POA B

=
A
P= R P A
B
B A
PO = A
POB + A PBORG
A

General Transform {B} Homogeneous Transform {B}

A A
P P
{A} B {A} B
P P
A A
PBORG PBORG

A
P = ABRBP + APBORG A
P = ABRBP + APBORG
⎡ AP⎤ ⎡ BAR A
PBorg ⎤⎡ BP⎤
⎢ ⎥ =⎢ ⎥⎢ ⎥
⎣ 1 ⎦ ⎣0 0 0 1 ⎦⎣ 1 ⎦
A A B
P = B T P
(4x1) (4x4) (4x1)

4
 Example LM0OP
{B} YB Operators
MM11PP
{A} ZA ? ZN QB
Mapping: changing descriptions from frame to frame
Operators: moving points (within the same frame)
LM0OP XB
MM3PP {A}
N1Q {B} A A
YA
Mapping
P
P = B R BP
XA
LM1 0 0 0 OP LM0OP
T=M PP P=M P R: P1 P2
Homogeneous 0 0 −1 3 1 R
MM0 MM1PP
A B
Transform B P2
N0
1 0 1
0 0 1
PQ N1Q Rotational Operator
LM0OP P1

P=M P
2
A
P = ABT . B P MM2PP
A

P2 = R P1
N 1Q

{B} P2
Rotational Operators Translations
RK(θ): P1 P2 {A}
POA POB
P2 = RK(θ) P1
Z P2
Example
L1 0 0 O θ P1 PBORG

RX (θ ) = MM0 cosθ − sin θ PP

MN0 cosθ PQ
Y
sin θ Mapping: PBORG : POB POA (same point)
X
2 diff. vectors
POA = POB + PBORG
⎡1 0 0 ⎤ ⎡ 0⎤ ⎡ 0 ⎤
P2 = RX (θ ) P1 = ⎢⎢0 0.8 −0.6⎥⎥ ⎢⎢2⎥⎥ = ⎢⎢1⎥⎥
⎢⎣0 0.6 0.8 ⎥⎦ ⎢⎣1⎥⎦ ⎢⎣2⎥⎦

{B} P2 {B} P2
Q e
Translations P1 Translations P1 mov
{A} POB {A} POB
POA POA

PBORG PBORG

Mapping: PBORG : POB POA (same point) Mapping: PBORG : POB POA (same point)
2 diff. vectors 2 diff. vectors
POA = POB + PBORG POA = POB + PBORG
Translational Operator: Translational Operator:
Q: P1 P2 (2 points, 2 diff vectors)
P2 = P1 + Q

5
 {B} P2
Translations P1 Q e
mov
Translation Operator P2
{A}
{A} P1
POA
Q

Operator: AP2 = AP1 + AQ

Homogeneous Transform:

Translational Operator:
LM1 0 0 qx OP
=M PP
0 1 0 qy A A
P2 = DQAP1
Q: P1 P2 (2 points, 2 diff vectors)
DQ
MM0 0 1 qz
PQ
P2 = P1 + Q N0 0 0 1

General Operators Inverse Transform

{B}
{A}
P2 = RK(θ) Q P1 PAORG
PBORG
0 0 0 1
A
T=
LM A
B R A
PBorg OP
P2 = T P1
B
N0 0 0 1 Q
R-1 = RT (T-1 = TT) B
PAORG
A
T −1
L
= T=M
B
A
B R T
− R . PBorg
A
B
T A
OP
B A
N0 0 0 1 Q

Homogeneous Transform Interpretations

Transform Equation
{B}
Description of a frame {A}
A
BT: lBq = n A
B R A
PBorg s PBORG
A {B}
BT
{B} P
A {A}
Transform mapping {A} P B
P
A B A PBORG O
BT : P P B
B
OA CT
{C}
P2
Transform operator {A}
P1 Q
T: P1 P2

6
 Transform Equation Compound Transformations
{C}
{A} C
P
A
P
{B}
{B}
A
BT
{A}
B
A CT
B
BT
CT B B C
{C}
A A B
P = CT P
P = BT P
A A B
A
P = A B C
BT CT P CT = B T CT

Transform Equation
A {B}
A A B BT B
CT = B T CT CT
{A}
{C}

⎡ R R A B A
R PCorg + PBorg ⎤
B A
A
CT =⎢ B C B
⎥ {D}

⎣0 0 0 1 ⎦ D
AT
C
DT
A B C D
BT CT DTAT = I
B
AT = BCT CDTDAT

D
{D} Spatial Descriptions
{A} A T
D
T
• Task Description
C
U
A T {C}

• Transformations
{U}
{B}
B
T
U
B T
C
• Representations

D
A T −1 . DC T . CB T −1 .UB T −1 .UA T ≡ I
U
A T = UBT . CB T . DC T −1 . DA T

7
 Position Representations
End-Effector Configuration
{E} z

r
y
{B} φ
θ
ρ
x Cartesian: (x, y, z)
B
T : position + orientation
Cylindrical: (ρ, θ, z)
E

End-Effector Configuration Parameters

X =
LM X OP P
position Spherical: (r, θ, φ)
NX Q R
orientation

Rotation Representations Three Angle Representations

Rotation Matrix {A}
Lr r12 r13 OP {B}

R = MMr
11

MNr
21 r22 PP
r23 = r1 r2 r3
31 r32 r33 Q
Direction Cosines LMr OP1

x r = r2
MM PP
Nr Q3 ( 9 x1)

Constraints
r1 = r2 = r3 = 1
r1 .r 2 = r1 .r3 = r 2 .r3 = 0

Three Angle Representations Euler Angles (Z-Y-X)

ZA
{A}
{B’} ZB’ {A} ZB
{A} ZB”
{B’} α B ′′
A YB’
{B}
YB
B R
B′ R YA YB”
Euler Angles
Fixed Angles γ
(12 sets)
(12 sets) XA
XB’ XB
XA
{A} {A} ZB’ XB”
ZB”
{B”} {B”} YB’
B′ YB”
A
B R = BA′ R. BB′′′ R. BB′′ R
YA
YB”
B ′′ R β
A
B R = R Z (α ). RY ( β ). R X (γ )
XB” XB’

8
X-Y-Z Fixed Angles
Z-Y-X Euler Angles
ZA

R = R Z ′ (α ). RY ′ ( β ). R X ′ (γ )
{A} A
{B} α (yaw) B

(roll) γ
YA
β (pitch)
LMcα − sα 0 OP LM cβ 0 sβ OP LM1 0 0 OP
MMsα cα 0. 0 PP MM PP MM
1 0 . 0 cγ − sγ PP
XA N0 0 1 − sβQN QN
0 cβ 0 s γ cγ Q
R X (γ ): v → R X (γ ). v
RY (β ): ( R X (γ ). v ) → RY ( β ).( R X (γ ). v )
LMcα . cβ X X OP
A
MM
R = BARZ ′Y ′X ′ (α , β , γ ) = sα . cβ X X P
cβ . cγ QP
B
RZ (α ): ( RY ( β ). R X (γ ). v ) → R Z (α ).( RY ( β ). R X (γ ). v ) N − sβ cβ . sγ
A
B R = R XYZ (γ , β , α ) = R Z (α ). RY ( β ). R X (γ )
A
B

Z-Y-Z Euler Angles Example

ZA
A
B R = R Z ′ (α ). RY ′ ( β ). R Z ′ (γ ) YB {A}

LM X X c α . sβ OP ZB
{B}
YA
A
B R = BARZ ′Y ′Z ′ (α , β , γ ) = MM− sβX. cγ X s α . sβ PP
N s β . sγ cβ Q XB X
A

R Z ′Y ′X ′ (α , β , γ ): α =0
β=0
γ = 90 o

Fixed & Euler Angles Inverse Problem

Given BAR find (α , β , γ )
X-Y-Z Fixed Angles R Z ′Y ′X ′

R XYZ (γ , β , α ) = R Z (α ). RY ( β ). R X (γ ) Lr r12 r13 OP LM

cα . cβ cα . sβ . sγ − sα . cγ cα . sβ . cγ + sα . sγ OP
R = MMr
11
A
B

MNr
21 r22 PP MM
r23 = sα . cβ sα . sβ . sγ + cα . cγ
N
sα . sβ . cγ − cα . sγ PP
Q
31 r32 r33 Q
− sβ cβ . s γ cβ . cγ
Z-Y-X Euler Angles
cos β = cβ = r112 + r212 U|
V|
→ β = A tan 2( − r31 , r112 + r212 )
R Z ′Y ′X ′ (α , β , γ ) = R Z (α ). RY ( β ). R X (γ ) sin β = sβ = − r31 W
if cβ = 0 (β = + 90ο) Singularity of the representation
R Z ′Y ′X ′ (α , β , γ ) = R X YZ (γ , β , α )
Only (α + γ) or (α − γ) is defined

9
Singularities - Example Equivalent angle-axis representation, R K (θ )
c β = 0, s β = + 1 LMθ . k OP
{A}
X

F0 − s(α − γ ) c(α − γ ) I
{B}
θ
K X r = θ . K = θ . kY
MM PP
GG 0 JJ
ZA {A}
Nθ . k Z Q
A
B R= c(α − γ ) s(α − γ ) YB

H −1 0 0 K (α − γ )
YA
⎡ k x .k x .vθ + cθ
⎢
RK (θ ) = ⎢ k x .k y .vθ + k z .sθ
⎢
⎣ k x .k z .vθ − k y .sθ
k x .k y .vθ − k z .sθ
k y .k y .vθ + cθ
k y .k z .vθ + k x .sθ
k x .k z .vθ + k y .sθ ⎤
⎥
k y .k z .vθ − k x .sθ ⎥
k z .k z .vθ + cθ ⎦⎥

(α − γ ) X LM OP
c β = 0, s β = − 1 XA B
with vθ = 1 − cθ
r11 r12 r13

{B} ZB
R K (θ ) = r21
MM r22 r23
PP
F0 − s(α + γ ) − c(α + γ ) I Nr31 r32 r33 Q
R = G0 JJ r + r + r −1
c(α + γ ) − s(α + γ ) θ = Ar cos( 11 22 33 )
GH 1
A
B 2
LM OP
0 0 K A
K=
1
2.sin θ
r32 − r23
MM
r13 − r31 , PP singularity for sin θ = 0
N
r21 − r12 Q

Euler Parameters Inverse Problem Given A

B R find ε
θ
ε 1 = W x . sin
2 W LMr11 r12 r13 OP LM 1 − 2ε − 2ε 2
2
2
3 2 ( ε 1 ε 2 − ε 3 ε 4 ) 2 (ε 1 ε 3 + ε 2 ε 4 ) OP
θ
ε 2 = W y . sin θ
MMr
21 r22
PP MM
r23 = 2(ε 1ε 2 + ε 3ε 4 ) 1 − 2ε 12 − 2ε 23 2 ( ε 2 ε 3 − ε 1ε 4 ) PP
ε 3 = W z . sin
2
θ Nr
31 r32 r33 Q N2 ( ε ε − ε ε )
1 3 2 4 2 ( ε 2 ε 3 + ε 1ε 4 ) 1 − 2ε 12 − 2ε 22 Q
2
θ r11 + r22 + r33 = 3 − 4 ( ε 12 + ε 22 + ε 23 )
ε 4 = cos
2 (1 − ε 24 )
1
Normality Condition ε4 =
1 + r11 + r22 + r33
2
W = 1, ε + ε + ε + ε = 1
2
1
2
2
2
3
2
4 r −r r −r r −r
ε 1 = 32 23 , ε 2 = 13 31 , ε 3 = 21 12
ε : point on a unit hypershpere 4ε 4 4ε 4 4ε 4
in four-dimensional space ε 4 = 0?

Lemma For all rotations one of the • ε 1 = max

i
εi l q
Euler Parameters is greater than 1
ε1 = r11 − r22 − r33 + 1
or equal to 1/2 2
4
(∑ ε = 1)
l q
2

• ε 2 = max εi
i
1
i

Algorithm Solve with respect to m a x ε

i
l q i ε2 =
1
− r11 + r22 − r33 + 1
2
• ε 1 = max
i
εi l q • ε 3 = max εi l q
i
1
ε1 = r11 − r22 − r33 + 1 1
2 ε3 = − r11 − r22 + r33 + 1
2
(r21 + r12 ) (r31 + r13 ) (r − r23 )
ε2 = ε3 = , ε 4 = 32
4ε 1
,
4ε 1 4ε 1 • ε 4 = max
i
{ε i }
1
ε4 = 1 + r1 1 + r2 2 + r3 3
2

10
 ZA
Euler Parameters / Euler Angles Quiz YB
ZB
β α −γ 60o
ε 1 = sin cos YA Direction Cosines
2 2
β α −γ
XB
LM 11 OP
ε 2 = sin sin XA
MM 000 PP r 1
2 2 Euler Parameters
β α +γ MM 00 PP
ε 3 = cos sin LM11/ 22PO = M 11 // 22 P r
2 2 xr
MM 3 // 22 PP
x =M
MM 0 PPP
2
00
β α +γ
ε 4 = cos cos r
MM 00 PP
2 2 MN 33/ 22PQ MM−− 3 / 22PP r 3
N 11//22 Q

11