UZB438E - Robotic Control Systems

Homework

Instructor: Murad Abu Khalaf

Uğur Baş
110210004
ugurbas@[Link]

October 13, 2025

UZB438E - Robotic Control Systems Uğur Baş – 110210004

1. Problem 2–35
Problem Statement
Find 𝐻 −1 explicitly for 

𝑅 𝑑
𝐻= , 𝑅 ∈ 𝑆𝑂 (3), 𝑑 ∈ R3 .
0 1

Solution
The homogeneous transformation matrix can be written as a composition of translation and rotation:
Using ( 𝐴𝐵) −1 = 𝐵−1 𝐴−1 ,
  𝐼 𝑑   𝑅 0  −1  𝑅 0 −1  𝐼 𝑑  −1
−1
𝐻 = = .
0 1 0 1 0 1 0 1

Since 𝑅 −1 = 𝑅 ⊤ and the inverse of a translation is a shift in the opposite direction,

 −1  −1
𝑅⊤ 0
     
𝑅 0 𝐼 𝑑 𝐼 −𝑑
= , = .
0 1 0 1 0 1 0 1

Multiplying them gives

𝑅 ⊤ −𝑅 ⊤ 𝑑
 
−1
𝐻 = ,
0 1
which matches the expected result. The inverse corresponds to undoing translation by 𝑑 and rotation
by 𝑅.

2. Problem 2–27
Given a unit quaternion
𝜃 𝜃
𝑞 = cos + (𝑛𝑥 i + 𝑛 𝑦 j + 𝑛 𝑧 k) sin ,
2 2
its scalar–vector form is
 𝜃 𝜃 𝜃 𝜃 
(𝑞 0 , 𝑞 1 , 𝑞 2 , 𝑞 3 ) = cos , sin 𝑛𝑥 , sin 𝑛 𝑦 , sin 𝑛 𝑧 .
2 2 2 2
Since 𝑛2𝑥 + 𝑛2𝑦 + 𝑛2𝑧 = 1, substituting gives

𝜃 𝜃
𝑞 20 + 𝑞 21 + 𝑞 22 + 𝑞 23 = cos2 + sin2 (𝑛2𝑥 + 𝑛2𝑦 + 𝑛2𝑧 ) = 1.
2 2
Hence, the quaternion has unit magnitude as required.

1
UZB438E - Robotic Control Systems Uğur Baş – 110210004

3. Problem 2–28
The quaternion–to–rotation mapping is given by
sin 𝜃 𝑛𝑥 = 2𝑞 0 𝑞 1 , sin 𝜃 𝑛 𝑦 = 2𝑞 0 𝑞 2 , sin 𝜃 𝑛 𝑧 = 2𝑞 0 𝑞 3 , cos 𝜃 = 2𝑞 20 − 1.
From these, the rotation matrix corresponding to 𝑞 becomes:
 2𝑞 2 + 2𝑞 2 − 1 2𝑞 1 𝑞 2 − 2𝑞 0 𝑞 3 2𝑞 1 𝑞 3 + 2𝑞 0 𝑞 2 
 1 0 
𝑅(𝑞) = 2𝑞 1 𝑞 2 + 2𝑞 0 𝑞 3 2𝑞 22 + 2𝑞 20 − 1 2𝑞 2 𝑞 3 − 2𝑞 0 𝑞 1  .
2𝑞 1 𝑞 3 − 2𝑞 0 𝑞 2 2𝑞 2 𝑞 3 + 2𝑞 0 𝑞 1 2𝑞 2 + 2𝑞 2 − 1 
 3 0 
Note that 𝑞 and −𝑞 represent the same spatial rotation.

4. Problem 2–29
For a given rotation matrix 𝑅 = [𝑟𝑖 𝑗 ], we can extract (𝑞 0 , 𝑞 1 , 𝑞 2 , 𝑞 3 ) using:
𝑟 32 − 𝑟 23 𝑟 13 − 𝑟 31 𝑟 21 − 𝑟 12
𝑞0 𝑞1 = , 𝑞0 𝑞2 = , 𝑞0 𝑞3 = ,
4 4 4
𝑟 32 + 𝑟 23 𝑟 13 + 𝑟 31 𝑟 21 + 𝑟 12
𝑞2 𝑞3 = , 𝑞1 𝑞3 = , 𝑞1 𝑞2 = . (1)
4 4 4
Also,
1 + (𝑟 11 + 𝑟 22 + 𝑟 33 ) 1 + (−𝑟 11 + 𝑟 22 + 𝑟 33 )
𝑞 20 = , 𝑞 21 = ,
4 4
1 + (𝑟 11 − 𝑟 22 + 𝑟 33 ) 1 + (𝑟 11 + 𝑟 22 − 𝑟 33 )
𝑞 22 = , 𝑞 23 = . (2)
4 4
To ensure numerical accuracy, pick the term with the highest value and compute others accordingly.
If 𝑞 0 is dominant:
√︂
1 + (𝑟 11 + 𝑟 22 + 𝑟 33 ) 𝑟 32 − 𝑟 23 𝑟 13 − 𝑟 31 𝑟 21 − 𝑟 12
𝑞0 = , 𝑞1 = , 𝑞2 = , 𝑞3 = .
4 4𝑞 0 4𝑞 0 4𝑞 0

5. Problem 2–33
For quaternions 𝑎 = (𝑎 0 , a) and 𝑏 = (𝑏 0 , b), their product is defined as:
 
𝑎0 𝑏0 − a · b
𝑎◦𝑏= .
𝑎0b + 𝑏0a + a × b
Let 𝑞 = (𝑞 0 , q) be a unit quaternion and 𝑣 = (0, v) be a pure quaternion.

Approach 1: Using Rodrigues Formula

 
∗ 0
𝑞◦𝑣◦𝑞 = .
(q · v)q + 𝑞 20 v + 2𝑞 0 (q × v) + q × (q × v)
Substituting 𝑞 = cos(𝜃/2) + n sin(𝜃/2), this simplifies to:
 
0
.
v + sin 𝜃 (n × v) + (1 − cos 𝜃)n × (n × v)

2
UZB438E - Robotic Control Systems Uğur Baş – 110210004

Approach 2: Direct Computation

Expanding the multiplication gives:
 (2𝑞 2 + 2𝑞 2 − 1)𝑣 1 + (2𝑞 1 𝑞 2 − 2𝑞 0 𝑞 3 )𝑣 2 + (2𝑞 1 𝑞 3 + 2𝑞 0 𝑞 2 )𝑣 3 
 1 0
𝑞𝑣𝑞 ∗ =  (2𝑞 1 𝑞 2 + 2𝑞 0 𝑞 3 )𝑣 1 + (2𝑞 22 + 2𝑞 20 − 1)𝑣 2 + (2𝑞 2 𝑞 3 − 2𝑞 0 𝑞 1 )𝑣 3  ,

 (2𝑞 1 𝑞 3 − 2𝑞 0 𝑞 2 )𝑣 1 + (2𝑞 2 𝑞 3 + 2𝑞 0 𝑞 1 )𝑣 2 + (2𝑞 2 + 2𝑞 2 − 1)𝑣 3 
 3 0 
which equals 𝑅v with 𝑅 as defined previously.

6. Angular velocity for Z–Y–Z Euler angles

Given:
𝑅(𝑡) = 𝑅𝑧,𝜓(𝑡) 𝑅 𝑦,𝜃 (𝑡) 𝑅𝑧,𝜙(𝑡) .
The aim is to find 𝑤(𝑡) such that:
𝑅¤ = 𝑆(𝑤)𝑅.

Step 1 — Differentiate the Rotation Matrix

Using the product rule:

𝑅¤ = 𝑅¤ 𝑧 (𝜓)𝑅 𝑦 (𝜃)𝑅𝑧 (𝜙) + 𝑅𝑧 (𝜓) 𝑅¤ 𝑦 (𝜃)𝑅𝑧 (𝜙) + 𝑅𝑧 (𝜓)𝑅 𝑦 (𝜃) 𝑅¤ 𝑧 (𝜙)

¤
= 𝜓𝑆(k)𝑅 ¤ 𝑧 (𝜓)𝑆(j)𝑅 𝑦 (𝜃)𝑅𝑧 (𝜙) + 𝜙𝑅
+ 𝜃𝑅 ¤ 𝑧 (𝜓)𝑅 𝑦 (𝜃)𝑆(k)𝑅𝑧 (𝜙). (3)

Step 2 — Express as 𝑆(𝑤)𝑅

Using 𝑅 𝑆(𝑎) 𝑅 ⊤ = 𝑆(𝑅𝑎), we can rewrite:
h i
¤ ¤ ¤ ¤
𝑅 = 𝜓𝑆(k) + 𝜃𝑆(𝑅𝑧 (𝜓)j) + 𝜙𝑆(𝑅𝑧 (𝜓)𝑅 𝑦 (𝜃)k) 𝑅.

Hence,
¤
𝑆(𝑤) = 𝜓𝑆(k) ¤
+ 𝜃𝑆(𝑅 ¤
𝑧 (𝜓)j) + 𝜙𝑆(𝑅 𝑧 (𝜓)𝑅 𝑦 (𝜃)k),

or equivalently,
𝑤 = 𝜓¤ k + 𝜃¤ 𝑅𝑧 (𝜓)j + 𝜙¤ 𝑅𝑧 (𝜓)𝑅 𝑦 (𝜃)k.

Step 3 — Compute Basis Rotations

− sin 𝜓   sin 𝜃 
   
𝑅𝑧 (𝜓)j =  cos 𝜓  , 𝑅 𝑦 (𝜃)k =  0  .
 0  cos 𝜃 
   
Then
cos 𝜓 sin 𝜃 
 
𝑅𝑧 (𝜓)𝑅 𝑦 (𝜃)k =  sin 𝜓 sin 𝜃  .
 cos 𝜃 
 

3
UZB438E - Robotic Control Systems Uğur Baş – 110210004

Step 4 — Express 𝑤 in Components

0 − sin 𝜓  cos 𝜓 sin 𝜃 
𝑤 = 𝜓¤ 0 + 𝜃¤  cos 𝜓  + 𝜙¤  sin 𝜓 sin 𝜃  .
     
1  0   cos 𝜃 
     
Simplifying:
¤
𝑤 𝑥 = cos 𝜓 sin 𝜃 𝜙¤ − sin 𝜓 𝜃,
¤
𝑤 𝑦 = sin 𝜓 sin 𝜃 𝜙¤ + cos 𝜓 𝜃,
𝑤 𝑧 = 𝜓¤ + cos 𝜃 𝜙.
¤

Summary. Thus, the angular velocity components in the inertial frame are defined by the
expressions above, corresponding directly to the Z–Y–Z Euler sequence.