2022 Iitb Ae 410 641 3

Navigation and Guidance
Dr. Shashi Ranjan Kumar
Assistant Professor
Department of Aerospace Engineering
Indian Institute of Technology Bombay
Powai, Mumbai, 400076 India
Dr. Shashi Ranjan Kumar IITB-AE 410/641 Quaternions 1 / 31

Coordinate Transformation
Computations of Euler Angles
When θ = π/2, we have some issues.

Equivalent rotation matrix
 
cos θ cos ψ cos θ sin ψ − sin θ
C =  cos ψ sin θ sin ϕ − sin ψ cos ϕ sin ψ sin θ sin ϕ + cos ψ cos ϕ cos θ sin ϕ 
cos ψ sin θ cos ϕ + sin ψ sin ϕ sin ψ sin θ cos ϕ − cos ψ sin ϕ cos θ cos ϕ
 
0 0 −1
=  cos ψ sin ϕ − sin ψ cos ϕ sin ψ sin ϕ + cos ψ cos ϕ 0 
cos ψ cos ϕ + sin ψ sin ϕ sin ψ cos ϕ − cos ψ sin ϕ 0
 
0 0 −1
=  sin(ϕ − ψ) cos(ϕ − ψ) 0 
cos(ϕ − ψ) − sin(ϕ − ψ) 0
First and third rotations are not distinguishable.

Transformation of Angular Velocities
Similar to DCM orientation, Euler angles also vary

with time when an input angular velocity vector is
applied between the two reference frames.
Angular velocity vector ω, in body-fixed coordinate
system, has components p, q, and r in the x, y, and
z directions, respectively.
Consider each derivative of an Euler angle as the
magnitude of the angular velocity vector in the
coordinate system in which the angle is defined.
For example, ψ̇ is the magnitude of ψ̇ that lies along
the Z axis of the Earth-fixed coordinate system.
     
ψ̇x 0 −ψ̇ sin θ
ψ̇ =  ψ̇y  =C  0  =  ψ̇ cos θ sin ϕ 
ψ̇z ψ̇ ψ̇ cos θ cos ϕ

Similarly, the components of θ̇ in X ′ Y ′ Z ′ are given

by (0, θ̇, 0)T .
In body frame, it can be obtained as
   
θ̇x 0
θ̇ =  θ̇y  = DB  θ̇ 
θ̇z 0
   
1 0 0 cos θ 0 − sin θ 0
=  0 cos ϕ sin ϕ   0 1 0   θ̇ 
0 − sin ϕ cos ϕ sin θ 0 cos θ 0
 
0
=  θ̇ cos ϕ 
−θ̇ sin ϕ

′′ ′′ ′′
Similarly, the components of ϕ̇ in X Y Z are
given by (ψ̇, 0, 0)T .
In body frame, it can be obtained as
   
ϕ̇x ϕ̇
ϕ̇ =  ϕ̇y  = D  0 
ϕ̇z 0
  
1 0 0 ϕ̇
=  0 cos ϕ sin ϕ   0 
0 − sin ϕ cos ϕ 0
 
ϕ̇
= 0 
0

Components of ω in body-fixed coordinate system is given by

ω = ψ̇ + θ̇ + ϕ̇
Now, we have
       
ωx p ψ̇x + θ̇x + ϕ̇x ϕ̇ − ψ̇ sin θ
 ωy  =  q  =  ψ̇y + θ̇y + ϕ̇y  =  ψ̇ cos θ sin ϕ + θ̇ cos ϕ 
ωz r ψ̇z + θ̇z + ϕ̇z ψ̇ cos θ cos ϕ − θ̇ sin ϕ
Euler angle rates

  q sin ϕ + r cos ϕ
 
ψ̇
 θ̇  =  cos θ 
 q cos ϕ − r sin ϕ 
ϕ̇ p + tan θ(q sin ϕ + r cos ϕ)
What happen when θ = ±90◦ ? Gimbal lock problem

How to avoid such difficulties? Nonsingular representation, e.g.,
quaternions
Singularity of Euler Angle Rates
For θ = π/2, p = ϕ̇ − ψ̇, q = θ̇ cos ϕ, r = −θ̇ sin ϕ

Azimuth and elevation rates
q sin ϕ + r cos ϕ θ̇ cos ϕ sin ϕ − θ̇ sin ϕ cos ϕ 0
ψ̇ = = =
cos θ cos θ 0
sin θ(q sin ϕ + r cos ϕ) 0
ϕ̇ = p + =p+
cos θ 0
Indeterminate forms!!!
Using L’Hospital rule, and the fact that d() d() dt
dθ = dt dθ , we have
d
dθ (q sin ϕ + r cos ϕ)
ψ̇|θ=π/2 = lim d(cos θ)
θ→π/2
dθ
q̇ sin ϕ + q cos ϕϕ̇ − r sin ϕϕ̇ + ṙ cos ϕ
= lim
θ→π/2 −θ̇ sin θ
q̇ sin ϕ + ṙ cos ϕ + ϕ̇θ̇
=−
θ̇
Singularity of Euler Angle Rates
Also, for θ = π/2, p = ϕ̇ − ψ̇
ϕ̇|θ=π/2 = p + ψ̇|θ=π/2
q̇ sin ϕ + ṙ cos ϕ + ϕ̇θ̇
= p−
θ̇
On solving this equation,

p q̇ sin ϕ + ṙ cos ϕ
ϕ̇|θ=π/2 = −
2 2θ̇
Also, θ̇ = q cos ϕ − r sin ϕ.

For θ ≈ π/2, use these limiting values, else use usual update equations.

Quaternions
Quaternion: Complex Numbers
Complex numbers C = {a + bi| a, b ∈ R, i2 = −1} form a plane.

Their operations are related to two-dimensional geometry.
Any complex number has a length, given by the Pythagorean formula
p
|a + bi| = a2 + b2
We can add and subtract in C.

a + bi + c + di = (a + c) + (b + d)i
We can also multiply in C.

(a + bi)(c + di) = (ac − bd) + (ad + bc)i
What does this last formula mean? difficult to interpret

Fortunately, there is a better way to multiply complex numbers.
Quaternions
Quaternion: Complex Numbers
Geometrically, this formula says eiϕ lies

on the unit circle in C.
If we multiply eiϕ by a positive number
r, we get a complex number of length r,
reiϕ .
If we denote a + bi = r1 eiθ1 and
c + di = r2 eiθ2 then
(a + bi)(c + di) = r1 r2 ei(θ1 +θ2 )
To multiply two complex numbers,

multiply their lengths and add their
We can use complex arithmetic angles.
(multiplication) to perform a In particular, if we multiply a given
geometric operation (rotation). complex number z by eiϕ then it is
rotated by ϕ degrees.
Quaternions
Quaternions: History
The 19th century Irish mathematician and physicist William Rowan Hamilton
was fascinated by the role of C in two-dimensional geometry.
For years, he tried to invent an algebra of “triplets” to play the same role in
three dimensions.
On October 16th, 1843, while walking with his wife to a meeting of the Royal
Society of Dublin, Hamilton discovered a 4-D division algebra called the
quaternions.
Although, similar concept was developed by Gauss in 1819 (but unfortunately

not published).
Quaternions
Quaternions: History

Quaternions
Quaternions Definitions
Hamilton noticed that

i2 = j 2 = k2 = ijk = −1

The quaternions are denoted as
H = {a + bi + cj + dk : a, b, c, d ∈ R}.
Cyclic symmetry:
ij =k = −ji
jk =i = −kj
ki =j = −ik
Quaternions don’t commute.

i, j and k are recognized as unit vectors.

Quaternions
Quaternions Operations
The quaternion product is the same as the cross product of vectors.
i × j =k
j × k =i
k × i =j
However, unlike the unit vectors i × i = j × j = k × k = 0, we have
i×i=−1
j×j =−1
k×k =−1
i×j×k =−1

Quaternions
Quaternions Operations
A Hamilton quaternion can be considered as scalar part and a vector part

[Q] = ⟨q0 , q⟩ = q0 + q1 i + q2 j + q3 k
| {z }
q

1, i, j, k serves as basis for quaternion vector space.

Quaternions span the space of real and imaginary numbers.
Quaternion algebra includes scalar and vector algebra.
Addition, subtraction and multiplication: Similar way as in vector algebra.
Addition:
[Q] + [S] =(q0 + q1 i + q2 j + q3 k) + (s0 + s1 i + s2 j + s3 k)

=(q0 + s0 ) + (q1 + s1 )i + (q2 + s2 )j + (q3 + s3 )k

Quaternions
Quaternion Operations
Subtraction: Addition of negative quaternion −[Q] = (−1)[Q]

[Q] − [S] =(q0 + q1 i + q2 j + q3 k) − (s0 + s1 i + s2 j + s3 k)
=(q0 − s0 ) + (q1 − s1 )i + (q2 − s2 )j + (q3 − s3 )k
Multiplication:
[Q][S] =(q0 + q1 i + q2 j + q3 k)(s0 + s1 i + s2 j + s3 k)
=(q0 s0 − q1 s1 − q2 s2 − q3 s3 ) + (q0 s1 + q1 s0 + q2 s3 − q3 s2 )i
+ (q0 s2 − q1 s3 + q2 s0 + q3 s1 )j + (q0 s3 + q1 s2 − q2 s1 + q3 s0 )k
Using dot product of vectors q.s = q1 s1 + q2 s2 + q3 s3 , we have

[Q][S] = ⟨q0 , q⟩⟨s0 , s⟩ = ⟨q0 s0 − q.s, q0 s + s0 q + q × s⟩
Product of two quaternions is still a quaternion, with scalar part (q0 s0 − qs)
and vector part (q0 s + s0 q + q × s).
Quaternions
The set of quaternions is closed under multiplication and addition.
Example
Consider two quaternions below and find their quaternion product.
[Q] =3 + i − 2j + k
[S] =2 − i + 2j + 3k
⇒ q.s = −2, q × s = −8i − 4j.

⇒ We know that
⟨q0 , q⟩⟨s0 , s⟩ =⟨q0 s0 − q.s, q0 s + s0 q + q × s⟩

=6 − (−2) + 3(−i + 2j + 3k) + 2(i − 2j + k) + (−8i − 4j)
=8 − 9i − 2j + 11k

Quaternions
Scalar multiplication:
λ[Q] = λq0 + λq1 i + λq2 j + λq3 k
Conjugate:
[Q]⋆ = q0 − q = q0 − q1 i − q2 j − q3 k
We have the following relations

([Q]⋆ )⋆ =q0 − (−q) = [Q]
[Q] + [Q]⋆ =2q0
[Q]⋆ [Q] =(q0 + q)(q0 − q)
=q0 q0 − (−q).q + q0 q + (−q)q0 + (−q × q)
=q02 + q.q
=q02 + q12 + q22 + q32 = [Q][Q]⋆
Quaternions
Norm of Length of quaternion:

N ([Q]) =[Q][Q]⋆ = [Q]⋆ [Q] = (q0 + q)(q0 − q)
= q02 + q.q = q02 + q12 + q22 + q32
For two quaternions [Q] and [S], we have

([Q][S])⋆ = [S]⋆ [Q]⋆
⇒ Proof:
⋆
([Q][S])⋆ = [(q0 + q) (s0 + s)]
⋆
= (q0 s0 − q.s + q0 s + s0 q + q × s)
= (q0 s0 − q.s − q0 s − s0 q − q × s)
[S]⋆ [Q]⋆ = (s0 − s) (q0 − q)

=q0 s0 − (−s).(−q) + q0 (−s) + s0 (−q) + (−s) × (−q)
= (q0 s0 − q.s − q0 s − s0 q − q × s) = ([Q][S])⋆
Quaternions
Norm of product of two quaternions is equal to product of their norms.
N ([Q][S]) = N ([Q])N ([S])
⇒ Proof:
N ([Q][S]) = ([Q][S])([Q][S])⋆
= [Q][S][S]⋆ [Q]⋆
= [Q]N ([S])[Q]⋆
= N ([Q])N ([S])
Also, by using mathematical induction, one may write
N ([Q1 ][Q2 ] . . . [Qn ]) = N ([Q1 ])N ([Q2 ]) . . . N ([Qn ])

Quaternions
Inverse of quaternion: If [Q] ̸= 0 then its inverse is defined by
[Q][Q]−1 = [Q]−1 [Q] = 1
[Q]⋆
Using norm concept, [Q]−1 = , N (Q) ̸= 0. Does it make sense?
N (Q)
[Q][Q]⋆
[Q][Q]−1 = [Q]−1 [Q] = =1
N (Q)
If [Q] is unit quaternion then
[Q]⋆
[Q]−1 = = [Q]⋆
N (Q)
Inverse and conjugate for the unit quaternions are the same.

Quaternions
Identities:
How to define zero and unit quaternions?
A zero quaternion is quaternion with zero scalar and zero vector.
A unit quaternion is defined as any quaternion whose norm is 1.
[0] = 0 + 0i + 0j + 0k, [1] = 1 + 0i + 0j + 0k
Unlike direction cosine matrix, where six redundancies are present, the
quaternion has only one.
For unit quaternion,
q02 + q12 + q22 + q32 = 1

p
q02 + q12 + q22 + q32 can be used for normalizing factor for each parameter.

Quaternions
How to identify if two quaternions are equal?

Equality of quaternions: Two quaternions are equal if both their scalars as
well as their vectors are equal.
[Q] =[S]
⇒ q0 =s0 , q1 = s1 , q2 = s2 , q3 = s3
⇒ [Q]i =[S]i ∀ i = 0, 1, 2, 3
Can we express 3D vector as quaternion?
Any three dimensional vector can be expressed as quaternion with zero scalar.

Quaternions
Quaternions obey the associative and commutative laws of addition, and the
associative and distributive laws of multiplication.
For three quaternions, Q1 , Q2 , Q3
□ Associative addition
(Q1 + Q2 ) + Q3 =Q1 + (Q2 + Q3 )
□ Commutative addition
Q1 + Q2 = Q2 + Q1
□ Associative multiplication
(Q1 Q2 )Q3 = Q1 (Q2 Q3 )
□ Distributive multiplication
Q1 (Q2 + Q3 ) = Q1 Q2 + Q1 Q3
Is the multiplication of quaternions commutative?

Quaternions
Pure quaternions: Quaternion with zero real or scalar part

Any vector in R3 is a pure quaternion.

Quaternions
Quaternion Rotation Operator
How quaternion [Q] ∈ R4 operate on a vector in R3 ?

Define quaternion operator with unit quaternion [Q] as
LQ (v) =[Q]v[Q]⋆ = (q02 − ∥q∥2 )v + 2(q.v)q + 2q0 (q × v)
⇒ Proof:
LQ (v) =[Q]v[Q]⋆ = (q0 + q)(0 + v)(q0 − q)
=(q0 + q)(v.q + {q0 v − v × q})
=q0 (v.q) − q.{q0 v − v × q} + (v.q)q + q0 {q0 v − v × q}
+ q × {q0 v − v × q}
=q.{v × q} + (v.q)q + q0 {q0 v − v × q} + q × {q0 v − v × q}
=(v.q)q + q02 v + 2q0 {q × v} + q × {q × v}
=(v.q)q + q02 v + 2q0 {q × v} + q(q.v) − v(q.q)
=(q02 − ∥q∥2 )v + 2(q.v)q + 2q0 {q × v}

Quaternions
Operation of Unit Quaternion on Vector

This operator LQ does not change the length of the vector v.
∥LQ (v)∥ =∥[Q]v[Q]⋆ ∥ = |[Q]|∥v∥|[Q]⋆ | = ∥v∥
The direction of v, if along q (say v = kq), is left unchanged by the operator LQ .
[Q]v[Q]⋆ =[Q](kq)[Q]⋆ = (q02 − |q|2 )(kq) + 2(q.(kq))q + 2q0 (q × (kq))

=k(q02 + ∥q∥2 )q = kq
Any vector along q is thus not changed under operator LQ . This makes us
guess that the operator LQ acts like a rotation about q.
The operator LQ is linear over R3 . For any two vectors v 1 , v 2 ∈ R3 and any
a1 , a2 ∈ R
LQ (a1 v 1 + a2 v 2 ) = a1 LQ (v 1 ) + a2 LQ (v 2 ).

Quaternions
Rotation of Vector using Quaternion

For any unit quaternion
θ θ
[Q] = q0 + q = cos + q̂ sin
2 2
and for any vector v ∈ R3 , the action of the operator LQ (v) = [Q]v[Q]⋆ on v is
equivalent to a rotation of the vector through an angle θ, about q̂ as the axis of
rotation.
A vector v ∈ R3 , we decompose it as v = a + n, where a is the component

along the vector q and n is the component normal to q.
Under the operator LQ , a is invariant, while n is rotated about q through an
angle θ.
Since the operator is linear, the image [Q]v[Q]⋆ is indeed interpreted as a
rotation of v about q through an angle θ.

Quaternions
The operator Lq on vectors n
LQ (n) =(q02 − ∥q∥2 )n + 2(q.n)q + 2q0 (q × n)

=(q02 − ∥q∥2 )n + 2q0 (q × n)
=(q02 − ∥q∥2 )n + 2q0 ∥q∥(q̂ × n)
Denote n⊥ = q̂ × n. Now,
LQ (n) =(q02 − ∥q∥2 )n + 2q0 ∥q∥n⊥

2 θ 2 θ θ θ
= cos − sin n + 2 sin cos n⊥
2 2 2 2
= cos θn + sin θn⊥
Resulting vector is a rotation of n through an angle θ in the plane defined by

n and n⊥ .

Quaternions
L−[Q] = [−Q]v[−Q]⋆ = [Q]v[Q]⋆ How?

Negative quaternion −[Q]
2π + θ 2π + θ
−[Q] = cos + q̂ sin
2 2
It represents the rotation about the same axis through the angle 2π + θ,
essentially the same rotation.
Quaternions
Reference
1 George M. Siouris, Aerospace Avionics Systems: A Modern Synthesis,
Academic Press, Inc. 1993.
2 Bandhu N. Pamadi, Performance, Stability, and Control of Airplanes, AIAA
Education Series, 1998.
Thank you for your attention !!!

2022 Iitb Ae 410 641 3

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

2022 Iitb Ae 410 641 3

Uploaded by

Copyright:

Available Formats

Navigation and Guidance

Dr. Shashi Ranjan Kumar

Dr. Shashi Ranjan Kumar IITB-AE 410/641 Quaternions 1 / 31

When θ = π/2, we have some issues.

First and third rotations are not distinguishable.

Dr. Shashi Ranjan Kumar IITB-AE 410/641 Quaternions 2 / 31

Similar to DCM orientation, Euler angles also vary

Dr. Shashi Ranjan Kumar IITB-AE 410/641 Quaternions 3 / 31

Similarly, the components of θ̇ in X ′ Y ′ Z ′ are given

Dr. Shashi Ranjan Kumar IITB-AE 410/641 Quaternions 4 / 31

Dr. Shashi Ranjan Kumar IITB-AE 410/641 Quaternions 5 / 31

Components of ω in body-fixed coordinate system is given by

Euler angle rates

What happen when θ = ±90◦ ? Gimbal lock problem

For θ = π/2, p = ϕ̇ − ψ̇, q = θ̇ cos ϕ, r = −θ̇ sin ϕ

Also, for θ = π/2, p = ϕ̇ − ψ̇

On solving this equation,

Also, θ̇ = q cos ϕ − r sin ϕ.

Dr. Shashi Ranjan Kumar IITB-AE 410/641 Quaternions 8 / 31

Complex numbers C = {a + bi| a, b ∈ R, i2 = −1} form a plane.

We can add and subtract in C.

We can also multiply in C.

What does this last formula mean? difficult to interpret

Geometrically, this formula says eiϕ lies

(a + bi)(c + di) = r1 r2 ei(θ1 +θ2 )

To multiply two complex numbers,

Although, similar concept was developed by Gauss in 1819 (but unfortunately

Dr. Shashi Ranjan Kumar IITB-AE 410/641 Quaternions 12 / 31

Hamilton noticed that

The quaternions are denoted as

Quaternions don’t commute.

Dr. Shashi Ranjan Kumar IITB-AE 410/641 Quaternions 13 / 31

The quaternion product is the same as the cross product of vectors.

However, unlike the unit vectors i × i = j × j = k × k = 0, we have

Dr. Shashi Ranjan Kumar IITB-AE 410/641 Quaternions 14 / 31

A Hamilton quaternion can be considered as scalar part and a vector part

1, i, j, k serves as basis for quaternion vector space.

[Q] + [S] =(q0 + q1 i + q2 j + q3 k) + (s0 + s1 i + s2 j + s3 k)

Dr. Shashi Ranjan Kumar IITB-AE 410/641 Quaternions 15 / 31

Subtraction: Addition of negative quaternion −[Q] = (−1)[Q]

Using dot product of vectors q.s = q1 s1 + q2 s2 + q3 s3 , we have

The set of quaternions is closed under multiplication and addition.

⇒ q.s = −2, q × s = −8i − 4j.

⟨q0 , q⟩⟨s0 , s⟩ =⟨q0 s0 − q.s, q0 s + s0 q + q × s⟩

Dr. Shashi Ranjan Kumar IITB-AE 410/641 Quaternions 17 / 31

We have the following relations

Norm of Length of quaternion:

For two quaternions [Q] and [S], we have

[S]⋆ [Q]⋆ = (s0 − s) (q0 − q)

Norm of product of two quaternions is equal to product of their norms.

N ([Q][S]) = N ([Q])N ([S])

Also, by using mathematical induction, one may write

N ([Q1 ][Q2 ] . . . [Qn ]) = N ([Q1 ])N ([Q2 ]) . . . N ([Qn ])

Dr. Shashi Ranjan Kumar IITB-AE 410/641 Quaternions 20 / 31

Inverse of quaternion: If [Q] ̸= 0 then its inverse is defined by

[Q][Q]−1 = [Q]−1 [Q] = 1

If [Q] is unit quaternion then

Dr. Shashi Ranjan Kumar IITB-AE 410/641 Quaternions 21 / 31

[0] = 0 + 0i + 0j + 0k, [1] = 1 + 0i + 0j + 0k

q02 + q12 + q22 + q32 = 1

Dr. Shashi Ranjan Kumar IITB-AE 410/641 Quaternions 22 / 31

How to identify if two quaternions are equal?

Dr. Shashi Ranjan Kumar IITB-AE 410/641 Quaternions 23 / 31