Navigation Systems and Sensors

Assignment - 5
Jimil Bhatt
SC21B083
September 15, 2024

Q.1 Explore the other methods of co-ordinate

transformations other than studied in the class.
The co-ordinate systems transformations other than Direction Cosine Matrix,
Euler angles and axis, Quaternions are:
• Rodrigues’ rotation formula,
• Homogeneous transformation matrix

(i) Rodrigues’ Rotation Formula

• It provides a way to rotate a vector in 3-D space given an axis and angle
of rotation.
• If V
⃗ is a vector in R3 and k̂ is a unit vector describing an axis of rota-
tion about which V ⃗ rotates by an angle θ, then by the right-hand rule,
Rodrigues’ formula is,

⃗rot = V
V ⃗ cos θ + (k̂ × V
⃗ ) sin θ + k̂(k̂ · V
⃗ )(1 − cos θ)

Proof: We can write,

⃗ =V
V ⃗∥ + V
⃗⊥
where,
⃗∥ = Parallel component of V
V ⃗,
⃗⊥ = Perpendicular component of V
V ⃗.

Then,
⃗∥ = (V
V ⃗ · k̂)k̂

1
and
⃗
V⃗⊥ = V⃗ −V ⃗∥ = V
⃗ − (V⃗ · k̂)k̂ = −k̂ × (k̂ × V )
1
(from the vector triple product relation)

⃗a × (⃗b × ⃗c) = (⃗a · ⃗c)⃗b − (⃗a · ⃗b)⃗c

In rotation, component V⃗∥ will be unchanged, i.e.,

⃗rot∥ = V
V ⃗∥
The perpendicular component will be,
⃗rot⊥ = cos θ V
V ⃗⊥ + sin θ(k̂ × V
⃗⊥ ) = cos θ V
⃗⊥ + sin θ(k̂ × V
⃗)

Then the full vector is,

⃗rot = V
V ⃗∥ + V
⃗rot⊥ = V
⃗∥ + cos θ V
⃗⊥ + sin θ(k̂ × V
⃗)

⃗⊥ = V
Putting V ⃗ −V
⃗∥ ,

⃗rot = cos θ V
V ⃗ + (1 − cos θ)(k̂ · V
⃗ )k̂ + sin θ(k̂ × V
⃗)

Thus,
⃗rot = V
V ⃗ + (1 − cos θ)k̂(k̂ · V
⃗ ) + sin θ(k̂ × V
⃗) (1)

In matrix notation,
The linear transformation on R3 given as V ⃗ → k̂ × V
⃗ is written in coordinates
by the column matrix:

(k̂ × V⃗ )x   ky Vz − kz Vy   0 −kz ky
 
Vx
(k̂ × V⃗ )y  = kz Vx − kx Vz  =  kz 0 −kx  Vy 
⃗ )z
(k̂ × V kx Vy − ky Vx −ky kx 0 Vz

Then,  
0 −kz ky
K =  kz 0 −kx 
−ky kx 0
⃗ = KV
Also, k̂ × V ⃗ , and k̂ × (k̂ × V
⃗ ) = K(KV
⃗ ) = K2 V
⃗.
Then, Eq. (1) becomes:
⃗rot = V
V ⃗ + sin θKV
⃗ + (1 − cos θ)K2 V
⃗

⃗rot = I + sin θK + (1 − cos θ)K2 V⃗

V
Thus, the rotation matrix is given by:

R = I + sin θK + (1 − cos θ)K2

2
(ii) Homogeneous Transformation Matrix
• It is used in 3D space to represent both rotation and translation of the
object in a single matrix.
• It allows easy combination of multiple transformations such as rotating
an object.
• A homogeneous matrix is given as,
⃗t
 
R
T =
0 1
where,
– R → 3 × 3 rotation matrix,
– ⃗t → 3 × 1 translational matrix,
– 0 → Zero vector,
– 1 → Scaling factor in homogeneous coordinates.
More explicitly, we can write,
 
r11 r12 r13 tx
r21 r22 r23 ty 
T = r31

r32 r33 tz 
0 0 0 1

Q.2 If ⃗u the unit vector is a rotation axis and ϕ

is the angle of rotation, then prove that the basis
vectors are related to the original basis by:
V12 + (V22 + V32 ) cos ϕ
 
V1 V2 (1 − cos ϕ) − V3 sin ϕ V1 V3 (1 − cos ϕ) + V2 sin ϕ
T
Q = V1 V2 (1 − cos ϕ) + V3 sin ϕ V22 + (V12 + V32 ) cos ϕ V2 V3 (1 − cos ϕ) − V1 sin ϕ
V1 V3 (1 − cos ϕ) − V2 sin ϕ V2 V3 (1 − cos ϕ) + V1 sin ϕ V32 + (V12 + V22 ) cos ϕ

ANS -
⃗ is rotating about ⃗u axis with
Let ⃗u be the axis of rotation unit vector, and V
angle of rotation θ. After rotation, we get:
⃗ =V
V ⃗∥ + V
⃗⊥

⃗∥ is ∥ to ⃗u and V
where V ⃗⊥ is perpendicular to ⃗u and the plane. If ω
⃗ is any vector
⃗
containing ⃗u and V⊥ , we have:
⃗∥ = (V
V ⃗ · ⃗u)⃗u · · · (1)

3
 ⃗′ =V
V ⃗∥ + V
⃗⊥ · · · (2)
Therefore,
⃗ =V
V ⃗∥ + V
⃗⊥ ⇒ ⃗⊥ = V
V ⃗ −V
⃗∥ = V
⃗ − (V
⃗ · ⃗u)⃗u · · · (3)

After rotation by angle θ:

⃗⊥
⃗u × V
ω
⃗ = · · · (4)
⃗⊥ |
|⃗u × V
⃗⊥ becomes V
Also, V ⃗ ′ , but the magnitude remains the same:
⊥

⃗ ⊥ | = |V
|V ⃗⊥′ | · · · (5)

Then,

⃗ ⃗
⃗⊥′ | cos θ V⊥ + |V
⃗⊥′ = |V
V ⃗⊥′ | sin θ ⃗u × V⊥
⃗⊥ |
|V |V⃗⊥ |
 
⃗⊥′ = cos θV
V ⃗⊥ + sin θ ⃗u × V ⃗⊥ · · · (6)
 
⃗ ′ = cos θ(V
V ⃗ −V ⃗∥ ) + sin θ ⃗u × (V ⃗ −V ⃗∥ )
⊥
 
⃗⊥′ = cos θ(V
V ⃗ − (V ⃗ · ⃗u)⃗u) + sin θ ⃗u × V ⃗ · · · (7)
Now we put,
⃗′ =V
V ⃗∥ + V
⃗⊥′
⃗ ′ = (V
V ⃗ · ⃗u)⃗u + cos θ(V
⃗ − (V
⃗ · ⃗u)⃗u) + sin θ(⃗u × V
⃗) · · · (8)

If ⃗u = ux Iˆ + uy Jˆ + uz K̂ along the axis of rotation, and the basis is rotated

by an angle θ, then:

î = cos θIˆ + (1 − cos θ)ux (ux Iˆ + uy Jˆ + uz K̂) + sin θ(−uy K̂ + uz J)

ˆ

î = u2x (1 − cos θ) + cos θ Iˆ + [ux uy (1 − cos θ) + uz sin θ] Jˆ

 

+ [ux uz (1 − cos θ) − uy sin θ] K̂ · · · (9)

Similarly,

ĵ = [ux uy (1 − cos θ) − uz sin θ] Iˆ + u2y (1 − cos θ) + cos θ Jˆ

 

4
 + [uy uz (1 − cos θ) + ux sin θ] K̂ · · · (10)

k̂ = [ux uz (1 − cos θ) + uy sin θ] Iˆ + [uy uz (1 − cos θ) − ux sin θ] Jˆ

+ u2z (1 − cos θ) + cos θ K̂

 
· · · (11)
From (9), (10), and (11), we get:
   ˆ  ˆ  
î I I î
 ĵ  = Q  Jˆ  and  Jˆ  = QT  ĵ 
k̂ K̂ K̂ k̂

We have:
⃗ = V1 Iˆ + V2 Jˆ + V3 K̂
V
Also, we have:
⃗|=1
|V ⇒ V12 + V22 + V32 = 1
Thus,
1 − V12 = V22 + V32
1 − V22 = V12 + V32
1 − V32 = V12 + V22
⃗ and rotation ϕ similar to ⃗u and rotation θ,
⇒ For unit vector V
From (9), (10), (11):

V12 (1 − cos θ) + cos θ

 
V1 V2 (1 − cos θ) − V3 sin θ V1 V3 (1 − cos θ) + V2 sin θ
T
Q = V1 V2 (1 − cos θ) + V3 sin θ
 V22 (1 − cos θ) + cos θ V2 V3 (1 − cos θ) − V1 sin θ
V1 V3 (1 − cos θ) − V2 sin θ V2 V3 (1 − cos θ) + V1 sin θ V32 (1 − cos θ) + cos θ

From,

1 − V12 = V22 + V32 ⇒ V12 − V12 cos θ + cos θ = V12 + (V22 + V32 ) cos θ

Hence, QT becomes:

V12 + (V22 + V32 ) cos ϕ

 
V1 V2 (1 − cos ϕ) − V3 sin ϕ V1 V3 (1 − cos ϕ) + V2 sin ϕ
T
Q = V1 V2 (1 − cos ϕ) + V3 sin ϕ V22 + (V12 + V32 ) cos ϕ V2 V3 (1 − cos ϕ) − V1 sin ϕ
V1 V3 (1 − cos ϕ) − V2 sin ϕ V2 V3 (1 − cos ϕ) + V1 sin ϕ V32 + (V12 + V22 ) cos ϕ

Hence Proved.

5
Q.3 (A)Find Euler angle sequence from DCM ma-
trix for all 12 rotation schemes, and find how an-
gles are determined. Also show that angle values
used in different sequence or about different axis
result in different attitude.
ANS -
• Based on sequence of rotation we have,
– Symmetric euler angles
∗ ZXZ, XYX, ZYZ, YXY, YZY, XZX
– Asymmetric Tait-Bryan angles
∗ XYZ, YXZ, ZXY, XZY, YZX, ZYX
then, for

(i) XYX

R = Rx (ϕ)Ry (θ)Rx (ψ)

   
1 0 0 cθ 0 sθ 1 0 0
= 0 cϕ −sϕ   0 1 0  0 cψ −sψ 
0 sϕ cϕ −sθ 0 cθ 0 sψ cψ
 
cθ sψ sθ cψ sθ
Rxyx = sϕ sθ cϕ cψ − sϕ sψ cθ −cϕ sψ − sϕ cψ cθ 
cϕ sθ cϕ sψ + sϕ cψ cθ −sϕ cψ + cϕ sψ cθ

(ii) ZXZ

R = Rz (ϕ)Rx (θ)Rz (ψ)

 
cϕ cψ − sϕ cθ sψ sϕ cψ + cϕ cθ sψ sθ sψ
Rzxz = −cϕ sψ − sϕ cθ cψ −sϕ sψ + cϕ cθ cψ sθ cψ 
sϕ sθ −cϕ sθ cθ

(iii) YZY
 
−sϕ sθ + cϕ cθ cψ sθ sψ −cψ sθ − sϕ cθ cϕ
R = Ry (ϕ)Rz (θ)Ry (ψ)R = −cϕ sθ cθ sϕ sθ 
sϕ sθ + cϕ cθ cψ sθ sψ cϕ cθ − sθ sϕ sψ

(iv) XZX
 
cθ sψ cθ sψ sϕ
R = −sϕ cϕ cϕ cψ − sϕ sψ cϕ cψ + sϕ cθ 
sψ sθ −sϕ cψ − cϕ sθ cθ cψ − sϕ cψ

6
 (v) YXY
 
cϕ cψ − sψ sθ sϕ sθ −sϕ cψ − cϕ sψ
R= sψ sθ cθ sϕ cψ + sθ cθ 
cϕ sθ + sϕ cψ −sϕ cθ cψ cθ − sϕ sθ

(vi) ZYZ
 
cϕ cψ − sϕ sθ cϕ cψ + sϕ cθ −sϕ cψ
R = −sϕ cθ − cϕ sθ cψ cθ − sϕ sψ sϕ sθ 
sθ sψ sϕ sψ + cϕ cθ cθ

Similarly, for Asymmetric sequence,

(vii) XYZ

R = Rx (ϕ)Ry (θ)Rz (ψ)

   
1 0 0 cθ 0 sθ cψ −sψ 0
= 0 cϕ −sϕ   0 1 0  sψ cψ 0
0 sϕ cϕ −sθ 0 cθ 0 0 1
 
cθ cψ sϕ cθ sθ
= cϕ sθ − sϕ cψ sψ cθ + cϕ sθ −sθ 
cϕ sθ + sϕ sψ sϕ sψ − cθ cϕ sθ cϕ cψ

Similarly to (vii), we get other,

(viii) YZX
 
cθ cϕ cψ cθ + sψ sϕ sψ sθ
R= −sθ cθ cϕ sθ sψ + cθ cψ cϕ 
cψ sθ + cϕ sψ cϕ sψ + cθ cψ sϕ cϕ + sψ sθ

(ix) ZXY
 
cϕ cψ + sψ sθ sϕ cϕ sψ − sθ cψ sϕ cθ sϕ
R =  sθ cψ − sϕ sθ sϕ cϕ cψ − sψ sθ sϕ −sθ 
sψ sθ − cψ cθ sϕ cψ cθ sϕ + sψ sθ cϕ cθ

(x) XZY
 
cϕ cθ sθ −cψ sθ
R = −sψ cϕ + sθ sϕ sψ cϕ cθ sϕ sθ − cθ cϕ sψ 
sϕ sψ − cϕ cθ sϕ cθ + cϕ sθ sψ cθ

(xi) YXZ
 
cϕ cθ − sψ sϕ sθ cψ sθ + sϕ sθ cϕ −sϕ cθ
R= −cϕ sψ cϕ cψ sψ sθ 
sθ sψ + cϕ cψ sθ sϕ sθ + cθ cϕ sθ cθ

7
(xii) ZYX
 
cϕ cθ sθ sϕ sϕ sθ − sθ cϕ sψ
R =  −sθ cϕ cθ cϕ sθ − cψ sϕ 
sθ sψ −cψ sθ cθ cϕ − sϕ sθ

Looking at the rotation sequence matrix we see that,

Sequence Φ  Θ Ψ 
−1 Q21 −1 −1 Q12
XYX tan cos (Q11 ) tan
 Q31   Q13 
YZY tan−1 Q 32
cos−1 (Q22 ) tan−1 Q 23

 Q12   Q21 
ZXZ tan−1 Q 13
cos−1 (Q33 ) tan−1 Q 31

 Q23   Q32 
XZX tan−1 Q 31
cos−1 (Q11 ) tan−1 Q 32

 Q21   Q12 
XYX tan−1 Q 21
cos−1 (Q22 ) tan−1 Q 23

 Q31   Q12 
XYZ tan−1 Q 23
cos−1 (Q22 ) tan−1 Q 21

 Q33   Q11 
XYZ tan−1 −Q Q32
sin−1 (−Q23 ) tan−1 Q 21

 13  Q22 
YZX tan−1 Q 13
sin−1 (−Q12 ) tan−1 Q 31

 Q33   Q32 
ZXY tan−1 −Q Q12
sin−1 (−Q13 ) tan−1 Q 21

 32   Q22 
ZXY tan−1 −Q 32
sin−1 (−Q31 ) tan−1 −Q 12

 Q22   Q13 
YXZ tan−1 −Q 13
sin−1 (Q33 ) tan−1 −Q 12

 Q23   Q11 
ZYX tan−1 −Q 21
Q11 sin−1 (Q31 ) tan−1 −Q 32
Q33

Looking at sequence (7) and (12), we see XYZ and ZYX sequence give
different values of Φ, Θ, Ψ angles. Hence it will be the case for other
sequences also. Therefore, it will result in different altitude.

8
Q3 (B). Determine appropriate Euler angles
from DCM matrix for already solved.

Q3 (C) Obtain classical Euler angles for trans-

formation from Geocentric equatorial to per-
ifocal system.

ANS -
In the geocentric equatorial system, the 3 axes are aligned as follows:
– X-axis: direction of vernal equinox
– Y-axis: 90° to the east along the equatorial plane
– Z-axis: Earth’s axis of rotation
In perifocal coordinate system:
– Satellite-based system
– Fundamental plane is that of satellite
– P-axis: points towards perigee of satellite’s orbit
– Q-axis: 90° from P-axis along orbital plane
– R-axis: normal to orbital plane passing through Earth
Thus, for transforming x-y-z to p-q-r coordinates, we need to follow the
given steps.

Transformation matrix is given by:

Qi = Rz (Ω)Rx (i)Rz (ω)

Euler angle rotation matrix (ZYX convention):

 
cos ψ cos θ cos ψ sin θ sin ϕ − sin ψ cos ϕ cos ψ sin θ cos ϕ + sin ψ sin ϕ
R =  sin ψ cos θ sin ψ sin θ sin ϕ + cos ψ cos ϕ sin ψ sin θ cos ϕ − cos ψ sin ϕ
− sin θ cos θ sin ϕ cos θ cos ϕ

where:
ϕ is the rotation around the X-axis
θ is the rotation around the Y-axis
ψ is the rotation around the Z-axis

9
 Figure 1: Orbit transformation diagram

Q3 (D) In 2D show that rotation of a frame

by angle α followed by rotation by angle β is
equivalent to single rotation through angle α
+ β.

In 3D show that successive rotations about

same axis is equivalent to one rotation by same
axis with sum of the angles of successive ro-
tations.

ANS -
 
cos γ − sin γ
Rγ =
sin γ cos γ

10
     
cos α sin α cos β − sin β cos(α + β) sin(α + β)
R = Rα Rβ = = = Rα+β
− sin α cos α − sin β cos β − sin(α + β) cos(α + β)

Similarly, in 3-D:
 
1 0 0
Rz (θ) = 0 cos θ sin θ 
0 − sin θ cos θ

    
1 0 0 cos β 0 sin β cos β sin α sin β cos α sin β
Rx (α)Ry (β) = 0 cos α sin α   0 1 0 = 0 cos α − sin α 
0 − sin α cos α − sin β 0 cos β − sin β sin α cos β cos α cos β

Thus,

Ry (α)Rz (β) = Rz (α + β)

Q4 - Prove or disprove that intrinsic and ex-

trinsic rotations for a given scheme results in
some DCM matrix.

ANS -

Intrinsic Rotation:
Rotations about the object fixed axes
first rotate outer gimbal, then middle gimbal, finally inner gimbal

Figure 2: intrinsic Rotation

11
Extrinsic Rotation:
Rotations performed w.r.t fixed frame
first rotate inner gimbal, then middle gimbal, finally outer gimbal

Figure 3: Caption for the image

Any extrinsic rotation is equivalent to intrinsic rotation by the same angles

but in reversed order of elemental rotations.

x · y ′ · z”(Ext) = z” · y ′ · x(int)

– First rotation same for both:

Iz = E z

– Second rotation: Intrinsic rotation x can be represented of an inverse

rotation about z then about y (extrinsic) and then z again.
Ez
⃗ygx = Ez · Ey =
Ez

– Third rotation on Iy :

Iy = Iy · Ez · Ex = Ez · Ey · fy

Ix · fy · fz = R (fy Ez · Ex · Ey · fy ) Iy fz
= (fy Ez · Ex ) (Ez · Ey · fy · Iy fz )
= fy · Ez · Ex
Ix · Iy · Iz = Ez · Ey · Ex

Hence proved xyz and zyx

Ez · Ey · Ex ̸= Ex Ey Ez

Ix · Iy · Iz ̸= Ex · Ey · Ez

12
Q4 (A) Show that v0 is fixed with respect to
AT

ANS -
4. In slide 42, show v0 is fixed with respect to T̂ .
If Av0 = λv0 , v0 is the eigenvector of λ = 1.
(i) If A is a rotation matrix, i.e., an orthogonal matrix:

Av0 = v0 and AT A = I

Then,

(AT A)v0 = v0

AT (Av0 ) = v0

AT v 0 = v 0

Q4 (B) How to compute the rotation angle

about fixed axis for an arbitrary rotation ma-
trix?

ANS -
To compute the rotation angle about a fixed axis for an arbitrary rotation
matrix, follow these steps:
General Rotation Matrix:
A general 3D rotation matrix R is of the form:
 
r11 r12 r13
R = r21 r22 r23 
r31 r32 r33

Steps to Compute the Rotation Angle θ

13
1. **Trace of the Matrix**: The trace of a matrix is the sum of its diagonal
elements, i.e.,
Tr(R) = r11 + r22 + r33

2. **Rotation Angle Formula**: The angle of rotation θ can be computed

using the trace of the matrix as follows:

Tr(R) − 1
cos θ =
2
Therefore, the rotation angle θ is:
 
Tr(R) − 1
θ = cos−1
2

Additional Consideration:
- The matrix R must be orthogonal, meaning RT R = I, and its determi-
nant should be det(R) = 1.
- If the axis of rotation is also required, you can compute the unit vector
along the axis using the following formula:
 
r − r23
1  32
v= r13 − r31 
2 sin θ
r21 − r12

This gives the unit vector v along the axis of rotation.

Q4 (c) Verify trace can be used to compute

the angle of rotation for rotation about coor-
dinate axes.

ANS -
For rotation about any arbitrary axis of rotation, as proved in Rz (θ):

 2
v1 + (v22 + v32 ) cos ϕ

v1 v2 (1 − cos ϕ) v1 v3 (1 − cos ϕ)
Qϕ =  v1 v2 (1 − cos ϕ) v22 + (v12 + v32 ) cos ϕ v2 v3 (1 − cos ϕ) 
v1 v3 (1 − cos ϕ) v2 v3 (1 − cos ϕ) v32 + (v12 + v22 ) cos ϕ

tr(a) = v12 + v22 + v32 + cos ϕ · (2(v12 + v22 + v32 ))

14
Since ⃗v is a unit vector, i.e., v12 + v22 + v32 = 1:

tr(a) = 1 + 2 cos ϕ

Thus,
 
−1 tr(Qϕ ) − 1
ϕ = cos
2

Q4 (D) Prove that

tr(AB) = tr(BA)

ANS -
tr(AB) = tr(BA)

If C = AB = BA, then:

n
X n
X
cij = aik bkj , then cij = aik bkj
k=1 k=1

X n
X
tr(C) = tr(AB) = cii = tr(A) = bki aki
k k=1
Scalar products hold commutative property
n X
X n
tr(AB) = bki aki = tr(BA)
i=1 k=1

(Q4 (E) Show that the incremental rotation

matrices commute under multiplication

ANS -
Consider:  
1 0 0
R̃x (θ) = 0 cos θ sin θ 
0 − sin θ cos θ

15
 we are using here approximation as:
cos(θ) ≈ 1;

sin(θ) ≈θ;

θ ≈ 0;
Now consider the rotation matrix;
   
1 0 0 1 0 0
R̃θ = 0 1 −θ , R̃ϕ = 0 1 −ϕ
0 θ 1 0 ϕ 1

where θ, ϕ are small angles.

  
1 0 0 1 0 0
R̃ϕ R̃θ = 0 1 −ϕ 0 1 −θ
0 ϕ 1 0 θ 1

Multiplying these matrices gives:

 
1 0 0
= 0 1 − θϕ −(θ + ϕ)
0 (θ + ϕ) 1 − θϕ

R̃ϕ R̃θ = R̃θ R̃ϕ

Q4 (F) Verify that the Direction Cosine Matrix

(DCM) from the coordinate system (I, ⃗ J,
⃗ K)
⃗ to
the coordinate system (⃗i, ⃗j, ⃗k) is:
ℓ2 (1 − cos θ) + cos θ
 
ℓm(1 − cos θ) + n sin θ ℓn(1 − cos θ) − m sin θ
ℓm(1 − cos θ) − n sin θ m2 (1 − cos θ) + cos θ mn(1 − cos θ) + ℓ sin θ
ℓn(1 − cos θ) + m sin θ mn(1 − cos θ) − ℓ sin θ n2 (1 − cos θ) + cos θ

ANS -
Answer is the same as the answer for the question 2.

16
Q5 what are the other techniques of the rotation?

ANS -
Answer is the same as the answer for the question 1.

17