Spatial Descriptions and

Transformations

December 14, 2006

Representing positions and
orientations
„ Define coordinate systems and develop
conventions for representation
„ A universe coordinate system
Y2
X2
P
Y1
P2
P1

X1
Coordinate System

„ A system for specifying points using coordinates

measured in some specified way. Depending on
yp of problem
the type p under consideration,,
coordinate systems possessing special properties
may allow particularly simple solution.
solution
Cartesian (or rectangular)
Coordinates
„ The simplest coordinate system consisting of
coordinate
di axes oriented
i d perpendicularly
di l l to eachh other.
h
Cylindrical Coordinates

r = x 2 + y 2 , r ∈ [0, ∞)
⎛ y⎞
θ = tan −1 ⎜ ⎟, θ ∈ [0,2π )
⎝x⎠
z = z , z ∈ (−∞, ∞)

x = r cosθ
y = r sin θ
z=z
Spherical Coordinates

r = x 2 + y 2 + z 2 , r ∈ [0, ∞)
⎛ y⎞
θ = tan ⎜ ⎟, θ ∈ [0,2π )
−1

⎝x⎠
⎛z⎞
φ = cos −1 ⎜ ⎟, φ ∈ [0, π )
⎝r⎠

x = r sin φ cosθ
y = r sin φ sin θ
z = r cos φ
Description of a position

V
Vector relative
l i to frame
f
Location
„ A 3 x 1 position vector
„ A leading superscript
indicates the coordinate
system ⎡ px ⎤
A ⎢
P = ⎢ py ⎥ ⎥
⎢⎣ pz ⎥⎦
Description of an orientation

Positions of points are

described with vectors and
orientations of bodies are
described with an attached
coordinate system.
Description of an orientation

„ In order to describe the orientation of a body,

we will attach a coordinate system to the body
and then ggive a description
p of this coordinate
system relative to the reference system.
„ One way to describe
des ribe the body-attached
bod tt hed
coordinate system, {B}, is to write the unit
vectors of its three principal axes in terms of the
coordinate system
y {{A}.
} A ˆ A ˆ A ˆ
X B , YB , Z B
Rotation matrix

⎡ r11 r12 r13 ⎤

A
B
B
[
A ˆ
R =R
A
B [
X = Yˆ X
A ˆˆ
A
BB
A
⎡r r
Z ] = ⎢⎢r Y
B

⎢⎣ r r
11
A ˆ
r
21
B
r ⎤
12
A ˆ
r ⎥⎥. Z
22

r ⎥⎦
23 ]
⎢
B = ⎢ r21
13

r22 ⎥
r23 ⎥.
⎢⎣r31 r32 r33 ⎥⎦
31 32 33

The orientation of a body is represented with a matrix.

Note that the components
p of anyy vector are simply
p y the
projections of that vector onto the unit directions of its
reference frame.

⎡ Xˆ B ⋅ Xˆ A YˆB ⋅ Xˆ A Zˆ B ⋅ Xˆ A ⎤
⎢ ˆ ˆ ⎥
A
B R = ⎢ B A
X ⋅ Y Yˆ ⋅ Yˆ
B A Zˆ B ⋅ YˆA ⎥.
⎢ Xˆ ⋅ Zˆ Yˆ ⋅ Zˆ Zˆ B ⋅ Zˆ A ⎥⎦
⎣ B A B A

Direction cosines
 ⎡ Xˆ B ⋅ Xˆ A YˆB ⋅ Xˆ A Zˆ B ⋅ Xˆ A ⎤
⎢ ˆ ˆ ⎥
A
B R = ⎢ B A
X ⋅ Y Yˆ ⋅ Yˆ
B A Z B ⋅ YA ⎥.
ˆ ˆ
⎢ Xˆ ⋅ Zˆ Yˆ ⋅ Zˆ Zˆ B ⋅ Zˆ A ⎥⎦
⎣ B A B A

⎡ Xˆ A ⋅ Xˆ B YˆA ⋅ Xˆ B Zˆ A ⋅ Xˆ B ⎤
⎢ ˆ ˆ ⎥
B
A R = ⎢ A B
X ⋅ Y Yˆ ⋅ Yˆ
A B Z A ⋅ YB ⎥.
ˆ ˆ
⎢ Xˆ ⋅ Zˆ Yˆ ⋅ Zˆ Zˆ A ⋅ Zˆ B ⎥⎦
⎣ A B A B
 B
A R= RA
B
T

⎡ A Xˆ BT ⎤
A T A ⎢ A ˆT ⎥ A ˆ
B R B R = ⎢ YB ⎥ X B [ YˆB
A A
]
Zˆ B = I 3
⎢ A Zˆ T ⎥
⎣ B⎦

A
B R = ABR −1 = AB R T .

The inverse of a matrix with orthonormal columns

is equal to its transpose.
transpose
Description of a frame: graphical
representation

For convenience
convenience, the point whose position we will describe
is chosen as the origin of the body-attached frame.
A frame: a position and an orientation pair

„ A set of four vectors giving position and

orientation information.
„ A position vector and a rotation matrix
„ Frame {B} is described by

{B} = { A
B R, A
PBORG .}
Mapping
„ Changing the description
f
from one frame
f to A
another
T maps
B
„ The quantity itself is not
changed; only its B
P→ P.
A
description is changed.
Mappings involving translated
frames

Translational mapping
 A
P = P + PBORG
B A

Both vectors are defined relative to frames of

the same orientation.
Mappings involving rotated frames

Rotating the description of a vector

 A
P= R P
A
B
B

The origins of the two frames are coincident.

Example 2.1
⎡0.866 − 0.500 0.000⎤
A
R = ⎢0.500 0.866 0.000⎥
B ⎢ ⎥
⎢⎣0.000 0.000 1.000 ⎥⎦
⎡ 0 .0 ⎤
B
P = ⎢⎢2.0⎥⎥
⎢⎣0.0⎥⎦

o
⎡− 1.000⎤
P = BAR BP = ⎢⎢ 1.732 ⎥⎥
30 A

⎢⎣ 0.000 ⎥⎦
Mappings involving general frames

General transform of a vector

A
P = R P + PBORG
A
B
B A

A
P= T P
A
B
B

an operator
t ini matrix
t i form
f
A Homogeneous Transform
Combines
C bi the
h operations
i off rotation
i and
d translation
l i
into a single matrix multiplication

⎡ P⎤ ⎡
A
R M A
B
A
PBORG ⎤ ⎡ P ⎤
B

⎢ ⎥ ⎢ ⎥⎢ ⎥
L
⎢ ⎥ ⎢= L L L M L ⎥⎢ L ⎥
⎢ 1 ⎥ ⎢0 0 0 M 1 ⎥⎦ ⎢⎣ 1 ⎥⎦
⎣ ⎦ ⎣

A
BT
A Homogeneous Transformation

„ Simultaneously represent the position and

orientation of one coordinate frame relative to
another.
„ Can be used to perform coordinate
tr nsform tions
transformations.
Example 2.2

⎡0.866 − 0.500 0.000 10.0⎤

⎢0.500 0.866 0.000 5.0 ⎥
BT =
A ⎢ ⎥
⎢0.000 0.000 1.000 0.0 ⎥
⎢ ⎥
⎣ 0 0 0 1 ⎦
⎡3.0⎤
B
P = ⎢⎢7.0⎥⎥
⎢⎣0.0⎥⎦
⎡ 9.098 ⎤
A
P = ABT BP = ⎢⎢12.562⎥⎥
⎢⎣ 0.000 ⎥⎦
Translating
„ A translation moves a
point
i iin space a fi
finite
i T operates on P2 A
distance along a given
to create A P1.
vector direction.
di i
„ Only one coordinate
system need be involved.
 Translational operators
A
P2 = AP1 + AQ
A
P2 = DQ (q ) AP1
⎡1 0 0 qx ⎤
⎢0 1 0 q y ⎥⎥
DQ (q) = ⎢ ,
⎢0 0 1 qz ⎥
⎢ ⎥
⎣0 0 0 1⎦
q = qx2 + q y2 + qz2

Translating a point in space is accomplished with the same mathematics

as mapping the point to a second frame.
Rotational Operators: Example 2.3

A
P2 = R AP1
A
P2 = RK (θ ) AP1

⎡cosθ − sin θ 0 0⎤
⎢ sin θ cosθ 0 0⎥⎥
RZ (Θ) = ⎢
⎢ 0 0 1 0⎥
⎢ ⎥
⎣ 0 0 0 1⎦
A homogeneous transform
Transformation operators: Example
2.4
The transform that rotates by R
andd translates by Q is the same
as the transform that describes a
frame rotated by R and
translated by Q relative to the
reference frame.

A
P2 =T P1 A
Compound transformations
B
P = T P,
B
C
C

A
P = T P.
A
B
B

A
P = T T P,
A
B
B
C
C

T = T T.
A
C
A
B
B
C
 ⎡ A
B
B
R R
C M A
B R PCORG + PBORG ⎤
B A

⎢ ⎥
CT = ⎢L
A
L L M L ⎥
⎢0 0 0 M 1 ⎥
⎣ ⎦
Inverting a transform

⎡ B
A R M B
PAORG ⎤
⎢ ⎥
AT = ⎢L
B
L L M L ⎥
⎢0 0 0 M 1 ⎥⎦
⎣
−1
B
AT= T
A
B
B A T
A R= R
B
A
P = R P + PBORG
A
B
B A

0B
B A
( PBORG )= A R PBORG + PAORG
A B

B
PAORG =− R PBORG =− R
B
A
A A
B
T A
PBORG
 ⎡ A
B R T
M − R
A
B PBORG ⎤
T A

⎢ ⎥
AT = ⎢L
B
L L M L ⎥
⎢0 0 0 M 1 ⎥
⎣ ⎦
Example 2.5
⎡0.866 − 0.500 0.000 4.0⎤
⎢0.500 0.866 0.000 3.0 ⎥
BT =
A ⎢ ⎥
⎢0.000 0.000 1.000 0.0⎥
⎢ ⎥
⎣ 0 0 0 1 ⎦

⎡ 0.866 0.500 0.000 − 4.964⎤

⎢− 0.500 0.866 0.000 − 0.598⎥
AT =
B ⎢ ⎥
⎢ 0.000 0.000 1.000 0.0 ⎥
⎢ ⎥
⎣ 0 0 0 1 ⎦
Check byy MATLAB.
Transform equations

U
D T= T T,
U
A
A
D
U
D T = T T T.
U
B
B
C
C
D
U
A T T = T T T.
A
D
U
B
B
C
C
D
−1 U A C −1
T= T
B
C
U
B AT T T .
D D
Example of a transform equation

U
CT =UAT DAT −1 DCT ,
U
CT = T T.
U
B
B
C
−1 D
U
AT= T T T
U
B
B
C
D
C A T.
Example 2.6: Manipulator reaching for
a bolt GT
T

−1 B
T
GT= T
B
T T T.
S
S
G
Proper orthonormal matrices
„ Rotation matrices are special in that all columns are
mutually
ll orthogonal
h l and
dhhave uniti magnitude.
i d
„ The determinanat of a rotation matrix is always
equal to +1.
 Special Orthogonal group of order n

For any R ∈ SO (n) the following properties hold.

R T = R −1 ∈ SO(n)

The columns (and therefore the rows) of R are mutually

orthogonal.

Each column (and therefore each row) of R is a unit vector.

det R = 1.
For any proper orthonormal matrix R, there exists a
skew-symmetric matrix such that

R = ( I 3 − S ) −1 ( I 3 + S ),
)
⎡ 0 − sx sy ⎤
⎢ ⎥
S = ⎢ sx 0 − sz ⎥
⎢− s y sz 0 ⎥⎦
⎣

Any 3 x 3 rotation matrix can be specified by just three

parameters.
Proper orthonormal matrices

Six constraints on the nine matrix elements:

Xˆ = 1, Yˆ = 1, Zˆ = 1,

Xˆ ⋅ Yˆ = 0, Xˆ ⋅ Zˆ = 0, Yˆ ⋅ Zˆ = 0.

Each is a unit vector, and all three must be mutually

perpendicular.
perpendicular
Example 2.7
⎡0.866 − 0.500 0.000⎤
Rz (30) = ⎢⎢0.500 0.866 0.000⎥⎥
⎣⎢0.000 0.000 1.000 ⎥⎦
⎡1.000 0.000 0.000 ⎤
Rx (30) = ⎢⎢0.000 0.866 − 0.500⎥⎥
⎢⎣0.000 0.500 0.866 ⎥⎦

Rz (30) Rx (30) ≠ Rx (30) Rz (30)

R i
Rotations don’t
d ’ generally
ll commute.
X-Y-Z fixed angles (roll
(roll--pitch
pitch--yaw)

RX (γ ) RY ( β ) RZ (α )
A
B RXYZ (γ , β ,α ) = RZ (α ) RY ( β ) RX (γ )

⎡cα − sα 0⎤ ⎡ cβ 0 sβ ⎤ ⎡1 0 0 ⎤
⎢ ⎥
= ⎢ sα cα 0⎥ ⎢ 0 ⎢ 1 ⎥ ⎢
0 ⎥ ⎢0 cγ − sγ ⎥ ⎥
⎢⎣ 0 0 1⎥⎦ ⎢⎣− sβ 0 cβ ⎥⎦ ⎢⎣0 sγ cγ ⎥⎦
⎡cαcβ cαsβsγ − sαcγ cαsβ cγ + sαsγ ⎤
⎢
= ⎢ sαcβ sαsβsγ + cαcγ ⎥
sαsβcγ − cαsγ ⎥.
⎢⎣ − sβ cβ sγ cβcγ ⎥⎦
Inverse problem:

⎡ r11 r12 r13 ⎤

A
R (γ , β , α ) = ⎢r r r23 ⎥⎥.
B XYZ ⎢ 21 22
⎢⎣ r31 r32 r33 ⎥⎦
β = A tan 2(−r31 , r + r ),
2
)
11
2
21

α = A tan 2(r21 / cβ , r11 / cβ )),

γ = A tan 2(r32 / cβ , r33 / cβ ).
A two argument arc tangent function
Trigonometric Functions

We need to express the

full range of angles,
angles but
the usual inverse tangent
function returns an angle
g
in the range (−π / 2, π / 2).
π 3π
A tan 2(−1,1) = − , A tan 2(1,−1) = +
4 4
Euler’s Rotation Theorem

„ An arbitrary rotation may be described by only

three parameters (angles).
Euler angles

If the rotations are written in terms of rotation matrices B, C, and

D, then a general rotation A can be written as A = BCD.
Th three
The h angles l giving
i i theh three
h rotation
i matrices
i are called
ll d Euler
E l
angles. There are several conventions for Euler angles, depending
on the axes about which the rotations are carried out.
out [Appendix
B]
Z-Y-X Euler angles
Each
E h rotation
i iis performed
f d about
b an axis
i off the
h moving
i
system {B} rather than one of the fixed reference {A}.

RZ (α ) RY ( β ) RX (γ )
 B′ B′′
A
B R= R R R
A
B′ B′′ B

A
B RZ ′Y ′X ′ = RZ (α ) RY ( β ) RX (γ )
⎡cα − sα 0⎤ ⎡ cβ 0 sβ ⎤ ⎡1 0 0 ⎤
⎢
= ⎢ sα cα 0⎥ ⎢ 0⎥ ⎢ 1 ⎥ ⎢
0 ⎥ ⎢0 cγ − sγ ⎥ ⎥
⎢⎣ 0 0 1⎥⎦ ⎢⎣− sβ 0 cβ ⎥⎦ ⎢⎣0 sγ cγ ⎥⎦
⎡cαcβ cαsβsγ − sαcγ cαsβ cγ + sαsγ ⎤
= ⎢⎢ sαcβ sαsβ sγ + cαcγ sαsβ cγ − cαsγ ⎥⎥.
⎢⎣ − sβ cβsγ cβ cγ ⎥⎦
Three rotations taken about fixed axes yield the same final orientation as the same
three rotations taken in opposite order about the axes of moving frame.
Homework #5 : Z Z--Y-Z Euler Angles
(1 pt.) – Due Dec. 21
„ Derive the rotational matrix
„ Extract Z-Y-Z Euler angles
Rules for Composition of Rotational Transformations:
Relative to the current frame: postmultiply
Relative to fixed frame: premultiply

Suppose R is defined by the following sequence of basic

rotations in order specified:

11. A rotation of θ about the current x-axis

2. A rotation of φ about the current z-axis
3 A rotation of
3. α about the fixed z-axis
4. A rotation of β about the current y-axis
5 A rotation of
5. δ about the fixed x-axis
x axis

R = Rx (δ ) Rz (α ) Rx (θ ) Rz (φ ) Ry ( β )
Equivalent angle
angle--axis representation

If the
th axis
i is
i a generall direction,
di ti
any orientation may be obtained
through proper axis and
orientation selection.
 ⎡ k x k x vθ + cθ k x k y vθ − k z sθ k x k z vθ + k y sθ ⎤
⎢ ⎥
RK (θ ) = ⎢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 ⎥⎦
⎣
vθ = 1− cosθ

⎡ r11 r12 r13 ⎤ Homework #6 (1 pt.)

A
R (θ ) = ⎢r r r ⎥, – Due Jan.
Jan 13
B K ⎢ 21 22 23 ⎥
⎣⎢ r31 r32 r33 ⎥⎦ Verify this equation.
⎛ r11 + r22 + r33 − 1 ⎞
θ = A cos⎜ ⎟
⎝ 2 ⎠
⎡r32 − r23 ⎤
1 ⎢ ⎥.
K=
ˆ r − r
i θ⎢
13 31 ⎥
2 sin
⎢⎣ r21 − r12 ⎥⎦
Change of Coordinates

An n x n matrix A represents a linear transformation from

n n
R to R in the sense that it takes a vector x to a new
vector y according to
y = Ax
The vector y is called the image of x under the
transformation A. If the vectors x and y are represented in
terms of the standard unit vectors
e1 = [1,0,K,0] ,K, en = [0,0,K,1]
T T

then the column vectors of A represent the images of the

basis vectors e1 ,K, en .
Change of Coordinates (cont
(cont’d)
d)

Often it is desired to represent vectors with respect to a

second coordinate frame with different basis vectors
f1 ,K, f n . In this case the matrix representing the same
linear transformation as A, but relative to this new basis, is
given by
A′ = T −1 AT
where T is a nonsingular matrix with column vectors
f1 ,K, f n . The transformation T −1 AT is called a
similarity transformation of the matrix A.
The use of similarity transformations allows us to
express the same rotation easily with respect to
different frames.
frames

Relative to frame {A} Frames {A} and {B} are related by

⎡cos θ − sin θ 0⎤ ⎡ 0 0 1⎤
Rz (θ ) = ⎢⎢ sin θ cos θ 0⎥⎥, T = BAR = ⎢⎢ 0 1 0⎥⎥
⎢⎣ 0 0 1⎦⎥ ⎢⎣− 1 0 0⎥⎦

Relative to frames {B}, we have

⎡1 0 0 ⎤ A rotation about Ẑ A
Rz′ (θ ) = T −1 Rz (θ )T = ⎢⎢0 cos θ sin θ ⎥⎥ but expressed relative to {B}
⎢⎣0 − sin θ cos θ ⎥⎦
Rotation about an arbitrary axis
The rotational transformation R = Rz (α ) Ry ( βwill
) bringg
the world z-axis into alignment with the vector K.
Therefore,, a rotation about the axis K can be computed
p
using a similarity transformation as
RK (θ ) = RRz (θ ) R −1

= Rz (α ) Ry ( β ) Rz (θ ) Ry (− β ) Rz (−α )

ky kx
sin α = cos α =
k +k
2
x
2
y k +k
2
x
2
y

sin β = k x2 + k y2 cos β = k z
S ppos R iss generated
Suppose g d by a rotation
o o of o 90o about
bo z0
followed by a rotation of 30o about y0 followed by a
rotation of 60o about x0 . Then

⎛ 1⎞
R = Rx (60) Ry (30) Rz (90) θ = cos −1 ⎜ − ⎟ = 120o
⎝ 2⎠
⎡ 3 1 ⎤
⎢ 0 − ⎥ ⎡ 1 ⎤
⎢ 2 2 ⎥ ⎢ 3 ⎥
⎢ ⎥
=⎢ − ⎥
1 3 3
− K̂ = ⎢
1 1
− ⎥
⎢ 4⎥ K
⎢
2 4 ⎢2 3 2⎥
3 1 3⎥ ⎢ 1 1⎥
⎢ ⎥ + ⎥
⎢⎣ 2 4 4 ⎥⎦ ⎢
⎣2 3 2⎦
Example 2.9
⎡0.707⎤
K = ⎢⎢0.707⎥⎥, θ = 30o
A ˆ

⎢⎣ 0.0 ⎥⎦
⎡1.0 ⎤
A
P = ⎢⎢2.0⎥⎥
⎣⎢3.0⎥⎦
 ⎡1.0 0.0 0.0 1.0 ⎤ ⎡1.0 0.0 0.0 − 1.0 ⎤
⎢0.0 1.0 0.0 2.0⎥⎥ B′ ⎢0.0 1.0 0.0 − 2.0⎥
A′T =
A ⎢ , BT = ⎢ ⎥
⎢0.0 0.0 1.0 3.0 ⎥ ⎢0.0 0.0 1.0 − 3.0 ⎥
⎢ ⎥ ⎢ ⎥
⎣0.0 0.0 0.0 1.0 ⎦ ⎣ 0 .0 0. 0 0. 0 1. 0 ⎦
⎡ 0.933 0.067 0.354 0.0⎤
⎢ 0.067 0.933 − 0.354 0.0⎥⎥
A′
B′T =
⎢
⎢− 0.354 0.354 0.866 0.0⎥
⎢ ⎥
⎣ 0.0 0.0 0.0 1.0 ⎦
⎡ 0.933 0.067 0.354 − 1.13⎤
⎢ 0.067 0.933 − 0.354 1.13 ⎥
A
T = A A′ B′
′ BT =
A′T BT
⎢ ⎥.
B
⎢− 0.354 0.354 0.866 0.05 ⎥
⎢ ⎥
⎣ 0. 000 0 .000 0 .000 1.00 ⎦
A rotation about an axis that does not pass through the origin causes
a change in position,
position plus the same final orientation as if the axis had
passed through the origin.
Euler parameters
„ The four parameters ε1 , ε 2 , ε 3 , and ε 4 describing a
i θ about
fi i rotation
finite b an arbitrary
bi i Kˆ = [k x k y k z ]T
axis
„ Defined by

θ θ θ
ε1 = k x sin , ε 2 = k y sin , ε 3 = k z sin
2 2 2
θ
ε 4 = cos
2
Euler parameters (cont’d)
„ A quaternion in scalar-vector representation

(ε , ε 4 ) = ε1i + ε 2 j + ε 3k + ε 4 .
ε ⋅ ε + ε = ε + ε + ε + ε = 1.
2
4
2
1
2
2
2
3
2
4

An arbitrary rotation may be described by only three parameters

parameters,
a relationship must exist between these four quantities.
Line vector

„ Refers to a vector that is dependent on its line of

action, along with direction and magnitude, for
causingg its effects.
„ A force vector
Free vector

„ Refers to a vector that may be positioned

anywhere in space without loss or change of
meaning,g, provided
p that magnitude
g and direction
are preserved.
„ Appure
re moment vector,
e tor the velocity
elo it of a point
Equal velocity vectors
Transforming velocities

B
V = 5 Xˆ , V = 5Yˆ
A
Rigid motions

A rigid
g motion is an ordered pair
p of ((d,, R).
)

d ∈R 3
R ∈ SO(3)
a pure translation
a pure rotation

SE (3) = R × SO(3)
3

Special Euclidean Group

Quiz #2 – Due Today

„ All the problems are from the course text.

Unless otherwise indicated, each problem is
worth 0.5 points.
p
„ 2.27, 2.28, 2.29, 2.30, 2.31, 2.33
Homework #7 – Dec. 21

„ 2.4, 2.12, 2.13, 2.37