Lesson 9: Euler Parameters

Euler Theorem
According to the Euler Theorem, there exist a unique axis that if the x-y-z frame (or the
frame) is rotated about it by an angle it becomes parallel to the frame (or the
x-y-z frame). This axis is denoted by

u and it is called the orientational axis of rotation.

usin
o
2

where e contains the x-y-z or components of

e
Vector

e is along the orientational axis of rotation having a magnitude of

2

The Euler parameters are denoted in any of the following forms:

p =
e
0
e

=
e
0
e
1
e
2
e
3

The four Euler parameters are not independent--they are related through one constraint
equation that can be expressed in any of the following forms:

e
0
2
+ e
1
2
+ e
2
2
+ e
3
2
=1
e
0
2
+ e
T
e =1
p
T
p =1

This process does not tell us how to locate the orientational axis of rotation!
The transformation matrix is expressed as:

A = 2
e
0
2
+ e
1
2

1
2
e
1
e
2
e
0
e
3
e
1
e
3
+ e
0
e
2
e
1
e
2
+ e
0
e
3
e
0
2
+ e
2
2

1
2
e
2
e
3
e
0
e
1
e
1
e
3
e
0
e
2
e
2
e
3
+ e
0
e
1
e
0
2
+ e
3
2

1
2

In compact form this matrix is written as

A = (2e
0
2
1)I + 2(ee
T
+ e
0
e)
Note that the elements of this matrix are quadratic in terms of the Euler parameters


x
z

e

Inverse Problem
Assume that the values of the nine direction cosines; i.e., all the nine elements of the
transformation matrix, are known. How do we determine the four Euler parameters?

2
e
0
2
+ e
1
2

1
2
e
1
e
2
e
0
e
3
e
1
e
3
+ e
0
e
2
e
1
e
2
+ e
0
e
3
e
0
2
+ e
2
2

1
2
e
2
e
3
e
0
e
1
e
1
e
3
e
0
e
2
e
2
e
3
+ e
0
e
1
e
0
2
+ e
3
2

1
2

=
a
11
a
12
a
13
a
21
a
22
a
23
a
31
a
32
a
33

We first equate the traces of the two matrices:

trA = a
11
+ a
22
+ a
33
= 2 (e
0
2
+ e
1
2

1
2
)+ (e
0
2
+ e
2
2

1
2
)+ (e
0
2
+ e
3
2

1
2
)
( )

This yields the value of

e
0
2
:

e
0
2
=
trA+1
4
(a)
The other three parameters are found by equating the diagonal terms:

2
e
0
2
+ e
1
2

1
2
e
1
e
2
e
0
e
3
e
1
e
3
+ e
0
e
2
e
1
e
2
+ e
0
e
3
e
0
2
+ e
2
2

1
2
e
2
e
3
e
0
e
1
e
1
e
3
e
0
e
2
e
2
e
3
+ e
0
e
1
e
0
2
+ e
3
2

1
2

=
a
11
a
12
a
13
a
21
a
22
a
23
a
31
a
32
a
33

We find:

e
1
2
=
1+ 2a
11
trA
4
, e
2
2
=
1+ 2a
22
trA
4
, e
3
2
=
1+ 2a
33
trA
4
(b)
We note that these formulas give the values of the squared Euler parametershow do
we determine the correct signs for the parameters?
In order to determine the correct signs for the Euler parameters, we first find the value of

e
0
2
from (a). Assume that

e
0
2
0. We assign either a positive or a negative sign to

e
0
.
We then subtract the off-diagonal terms to compute the other three parameters

e
1
=
a
32
a
23
4e
0
, e
2
=
a
13
a
31
4e
0
, e
3
=
a
21
a
12
4e
0
(c)
Since

e
0
0, we can find the other three parameters without any difficulties.
Note: It makes no difference what sign we give to

e
0
(why?)
What if

e
0
= 0? We add the off-diagonal terms of the transformation matrix to obtain the
following equations:

a
21
+ a
12
= 4e
1
e
2
(d)

a
13
+ a
31
= 4e
1
e
3
(e)

a
32
+ a
23
= 4e
2
e
3
(f)
Compute

e
1
2
,

e
2
2
, and

e
3
2
from (b). At least one of these parameters must be
nonzero!
If

e
1
0, use (d) and (e) to compute

e
2
and

e
3
. It makes no difference what sign we
give to

e
1
(why?)
If

e
2
0, use (d) and (f) to compute

e
1
and

e
3
; etc.

Singularity?
The inverse process in determining the Euler parameters shows that there is no singularity
associated with these parametersas long as the direction cosines are known, we can find
the corresponding Euler parameters! This is a major advantage in using these parameters
Another advantage of these parameters over the Euler angles is the transformation matrix is
simpler and more efficient to compute
The disadvantage is that we need to make sure that the constraint on Euler parameters is
satisfied; i.e., the sum of square of the four parameters must be exactly equal to one!

Determining Euler Parameters

From Direction Cosines
In the previous lesson we discussed the necessary formulas for determining the Euler
parameters if the direction cosines (the transformation matrix) are known
The direction cosines can be determined if we have a set of three angles describing the
rotation (Euler angles), or if we have the coordinates of several points that are attached to
the body

From Euler Angles
Assume that the three Euler angles (z-x-z convention) are known. What are the
corresponding Euler parameters?
We equate the two transformation matrices

2
e
0
2
+ e
1
2

1
2
e
1
e
2
e
0
e
3
e
1
e
3
+ e
0
e
2
e
1
e
2
+ e
0
e
3
e
0
2
+ e
2
2

1
2
e
2
e
3
e
0
e
1
e
1
e
3
e
0
e
2
e
2
e
3
+ e
0
e
1
e
0
2
+ e
3
2

1
2

1
]
1
1
1
=
c co s c so c so s c co s s
s co + c c so s so + c c co c s
s so s co c

1
]
1
1
1

We compute the trace of both matrices to obtain

e
0
, then we equate the diagonal
terms to get

e
0
= cos

2
cos
+
2
e
1
= sin

2
cos

2
e
2
= sin

2
sin

2
e
3
= cos

2
sin
+
2

It is obvious that if we have the three angles we can always compute the four
parameters

Inverse Conversion
The four Euler parameters are known, what are the corresponding Euler angles?
The conversion formulas are


cos = 2(e
0
2
+ e
3
2
)1
cos =
2(e
2
e
3
+ e
0
e
1
)
sin
sin =
2(e
1
e
3
e
0
e
2
)
sin
cos =
2(e
0
e
1
e
2
e
3
)
sin
sin =
2(e
1
e
3
+ e
0
e
2
)
sin


We note that the singularity problem still exists when

sin = 0
The singularity problem is not caused by the Euler parametersit is associated with
the Euler angles!

From Coordinates of Points
Assume that the x-y-z coordinates of four points, O, A, B and C are known
We compute the components of three vectors

a = r
A
r
O
b = r
B
r
O
c = r
C
r
O

The vectors are normalized to obtain three unit vectors

u
()
=
1
a
a u
()
=
1
b
b u
( )
=
1
c
c
Now we have the nine direction cosines:

A = u
()
u
()
u
( )

This process can also be achieved with only three points, O, A, and B as shown
We compute the components of two vector

a = r
A
r
O
b = r
B
r
O

The vectors are normalized to obtain two unit vectors

u
()
=
1
a
a u
()
=
1
b
b
The third unit vector is obtained as

u
( )
= u
()
u
()

The transformation matrix is

A = u
()
u
()
u
( )

Special cases
The Euler parameters can be found easily in he following special cases
The x-y-z frame and the frame are parallel

p = 1 0 0 0
{ }
T

Any axis can be the orientational axis of
rotation; the angle of rotation is zero

z


C
B
A

z


B
A

b
z


The x-axis is parallel to the -axis

p = cos

2
sin

2
0 0

T

The x- axis (or -axis) is the orientational
axis of rotation


The y-axis is parallel to the -axis

p = cos

2
0 sin

2
0

T

The y- axis (or -axis) is the orientational
axis of rotation


The z-axis is parallel to the axis

p = cos

2
0 0 sin

T

The z- axis (or axis) is the orientational axis of rotation

Matrices and Identities with Euler Parameters

G and L Matrices
Two

3 4 matrices are defined as

G=
e
1
e
0
e
3
e
2
e
2
e
3
e
0
e
1
e
3
e
2
e
1
e
0

L =
e
1
e
0
e
3
e
2
e
2
e
3
e
0
e
1
e
3
e
2
e
1
e
0

In compact form:

G = e e + e
0
I

L = e e + e
0
I

The transformation matrix A can be expressed as

A = GL
T

Since G and L matrices are linear in Euler parameters, it is obvious now why A is quadratic
in Euler parameters!
These two matrices have useful characteristics where some identities are listed here. Refer
to your textbook for the proof of these identities

Gp = 0 GG
T
= I G
T
G = pp
T
+ I
*
G p =

Gp

G p = 0
Lp = 0 LL
T
= I L
T
L = pp
T
+ I
*
L p =

Lp

L p = 0

where

I
*
is a

4 4 identity matrix
z

z

z

Note that most of these identities are interchangeable between G and L
These and many other identities can become useful in the derivation of the kinematic
constraints and the equations of motion

Constraints on Euler Parameters
We have already seen that the four Euler parameters must satisfy the following constraint

p
T
p 1= 0
The time derivative of this constraint provides the constraint on the first time derivative of
Euler parameters

p
T
p = 0
The time derivative of the velocity constraint provides the constraint on the second time
derivative of Euler parameters

p
T
p + p
T
p = 0

Additional Matrices and Identities
Two

4 4 matrices in terms of the elements of a 3-vector a are defined as

a
+

0 a
T
a a

0 a
T
a a

These matrices can be used in deriving many identities to be used in the derivation of
constraint equations, etc.

G
T
a = a
+
p Ga
+
= aG+ ap
T

G
T
a = a
+
p
L
T
a = a

p La

= aL+ ap
T

L
T
a = a

p

Other useful identities:

Aa = 2Ga

A
T
a = 2La
+
p

p
(Aa) = 2Ga

+ 2ap
T

p
(A
T
a) = 2La
+
+ 2ap
T

There is no need for us to memorize any of these identities. However, it will be useful to
understand what they are and how to use them. Check the textbook and learn how some of
these identities are derived. Some of these identities can be used to derive the partial
derivative of a constraint equation with respect to Euler parameters. Other identities can be
used to determine the first and second time derivatives of a constraint equation.