Ganz nah dran.

Prof. Dr.-Ing. Gavin Kane ■ 01.10.2024 ■ 1

Robotic Systems

Lecture 8. - Structures
and Transformations

Prof. Dr.-Ing. Gavin Kane ■ 01.10.2024 ■ 2

 Lecture Content

Lecture Content:
Definitions
• Kinematics / Dynamics
• Manipulators
• Joints and Links
• Degrees of Freedom / Degrees of
Actuation
• Coordinate Frames
• Transformations

Prof. Dr.-Ing. Gavin Kane ■ 01.10.2024 ■ 3

 Kinematics and Dynamics

Kinematics Dynamics
Kinematics is the science of In robot Dynamics, the body is
motion, which treats motion augmented with masses and
without regard for the forces moments of inertia, and its
that cause it. movement analyse augmented
Position, velocity and with the necessary forces and
acceleration of the body are torques that are required to
regarded in this field cause the motion
The kinematics of manipulators
involves the study of the
geometric and time based
properties of the motion, and in
particular how the various links
move with respect to one
another and with time

Prof. Dr.-Ing. Gavin Kane ■ 01.10.2024 ■ 4

 Robot Structures

Manipulator
A set of connecting bodies
forming a chain (open or
closed). Note these bodies
are called links, and the
connection points the
joints.

Prof. Dr.-Ing. Gavin Kane ■ 01.10.2024 ■ 5

 Robot Structures

Basic Topologies of Kinematic Chains

Prof. Dr.-Ing. Gavin Kane ■ 01.10.2024 ■ 6

 Robot Structures

Joint Types

Prof. Dr.-Ing. Gavin Kane ■ 01.10.2024 ■ 7

Basic Joints

Two Basic Types of

Kinematic Pairs – R & P
All other joint pairing can be
produced with a combination
of revolute (R) and Prismatic (P)
joints

Prof. Dr.-Ing. Gavin Kane ■ 01.10.2024 ■ 8

Basic Joints Joint Notation Scheme
Translational Motion: Uses the joint Symbols (L,O,R,T,V)
to designate joint types used to
Linear Joint (type L) construct a particular robot
Orthogonal Joint (type O) manipulator
Rotational Motion: Separates body-and-arm assembly
Rotational Joint (type R) from wrist assembly using a colon
(:)
Twisting Joint (type T)
Revolving Joint (type V)
Example: TLR : TR

Common body-and-arm
configurations …
Prof. Dr.-Ing. Gavin Kane ■ 01.10.2024 ■ 9
Polar Coordinate
Body and Arm Assembly
Notation TRL:

Consists of a sliding arm (L joint) actuated relative Source: NUI Galway

to the body, which can rotate about both a

vertical axis (T joint) and horizontal axis (R joint)

Prof. Dr.-Ing. Gavin Kane ■ 01.10.2024 ■ 10

Cylindrical Body-and-Arm
Assembly
Notation TLO:

Consists of a vertical column, relative to

which an arm assembly is moved up or down Source: NUI Galway

The arm can be moved in or out relative to

the column

Prof. Dr.-Ing. Gavin Kane ■ 01.10.2024 ■ 11

SCARA Robot
Notation VRO:

SCARA stands for Selectively Compliant Assembly

Robot Arm
Source: NUI Galway
Similar to jointed-arm robot except that vertical axes
are used for shoulder and elbow joints to be compliant
in horizontal direction for vertical insertion tasks

Prof. Dr.-Ing. Gavin Kane ■ 01.10.2024 ■ 12

Wrist Configurations Robot
Wrist assembly is attached to end-of-arm
End effector is attached to wrist assembly
Function of wrist assembly is to orient end
effector Source: NUI Galway
Body-and-arm determines global position of
end effector
Anthropic wrist concepts:
Two or three degrees of freedom:
Roll
Pitch
Yaw
Notation :RRT

Prof. Dr.-Ing. Gavin Kane ■ 01.10.2024 ■ 13

Example
Sketch the following manipulator
configurations:
(a) TVT:R
(b) TVR:TR
(c) RR:T

Prof. Dr.-Ing. Gavin Kane ■ 01.10.2024 ■ 14

Example
Sketch the following manipulator
configurations:
R
(a) TFT:R R
(b) TVR:TR T T R
R T
R
(c) RR:T R V

T T

(a) TRT:R (b) TVR:TR (c) RR:T

Prof. Dr.-Ing. Gavin Kane ■ 01.10.2024 ■ 15

Open or Closed Chain Robots
Most Robots studied in this
course are open chain robots.
Serial Kinematics, means each
joint is connected to a link
leading from the previous joint.
Closed Chain Robots, or parallel
kinematics offer different
advantages from high speed or
high stability with disadvantages
in other areas, such as working
space.

Prof. Dr.-Ing. Gavin Kane ■ 01.10.2024 ■ 16

The Problem of Kinematics
Using Kinematics to desribe
the spatial configuration of
the robot gives us two Forwards Kinematics
scenarios.
Forward Kinematics
(Direct) Joint Space Cartesian Space
[J1, J2, J3, J4, J5, J6] [X, Y, Z, R, P, W]
Given the joint angles of the robot,
what is the orientation and
position of the end-effector?
Inverse Kinematics Inverse Kinematics
Given a desired end-effector
position and orientation, what are
the required joint angles required
to achieve this?

Prof. Dr.-Ing. Gavin Kane ■ 01.10.2024 ■ 17

 Joint(s) Reference Frame(s)
World Reference Frame

Coordinates / Frames and

Transformations

End-Effector Reference Frame

Basis Reference Frame

Prof. Dr.-Ing. Gavin Kane ■ 01.10.2024 ■ 18

 Points in the plane can be denoted
Cartesian coordinate with tuples denoting their coordinates
Plane with respect to the frame of reference.
y These Tuples can also be used to talk
about directions or displacements
(2,1)
between points
P
x 𝑣=
2
1
y Here in the two dimensional plane, we
P1 (-1,1)
say 𝑣 ∈ ℝ2 , denoting the fact that 𝑣 is a
vector composed of two real numbers,
P0 as opposed to two complex numbers
x

Prof. Dr.-Ing. Gavin Kane ■ 01.10.2024 ■ 19

Operations on Vectors Geometric Interpretation
y a∙𝑣
Scalar Multiplication

2 10 𝑣
5∙ =
1 5 x
Addition
2
+
3
=
5 y
1 5 6 𝑣+𝑤 𝑤

𝑣 x
Prof. Dr.-Ing. Gavin Kane ■ 01.10.2024 ■ 20
Inner Product Inner Product
Properties
Given two vectors 𝑣, 𝑤 ∈ ℝ2 we Inner Product is commutative:
can define their inner product 𝑣 ∙ 𝑤 = 𝑤 ∙ 𝑣 = 𝑣1 𝑤1 + 𝑣2 𝑤2
as follows:
𝑣 ∙ 𝑤 = 𝑣1 𝑤1 + 𝑣2 𝑤2
Inner Product is distributive
where:
𝑣1 𝑤1 with respect to vector addition
𝑣 = 𝑣 ,𝑤 = 𝑤 𝑢∙ 𝑣+𝑤 =𝑢∙𝑤+𝑢∙𝑣
2 2
Note that the inner product
returns a single scalar value.
Example:
2 3
∙ = 2 ∙ 3 + 1 ∙ 5 = 11
1 5

Prof. Dr.-Ing. Gavin Kane ■ 01.10.2024 ■ 21

Euclidean Norm
The inner product of a vector with y
itself, results in a sum of the square
of the vectors values.
𝑣1 P
𝑣 = 𝑣 ⟹ 𝑣 ∙ 𝑣 = 𝑣1 𝑣1 + 𝑣2 𝑣2
2
We can interpret this quantity as
𝑥
the square of the length of the
vector. This value is referred to as
the Euclidean norm of the vector x
and is denoted by 𝑣 𝑦
𝑣 2 = 𝑣∙𝑣
𝑣 = 𝑣∙𝑣

Prof. Dr.-Ing. Gavin Kane ■ 01.10.2024 ■ 22

 Cosine Rule
Consider the figure with two vectors in the
plane separated by an angle θ y
Using the properties of inner Products and
Euclidean Norms we can prove that: V
(v-w)
𝑣−𝑤 2
= 𝑣 2
−2 𝑣 ∙ 𝑤 + 𝑤 2

From the well known Cosine Rule, we also

know that all three sides of the triangle are θ
w
related as follows:
𝑣−𝑤 2 = 𝑣 2 + 𝑤 2 −2 𝑣 𝑤 𝑐𝑜𝑠𝜃
x
So, we can conclude that:
𝑣∙𝑤 = 𝑣 𝑤 𝑐𝑜𝑠𝜃

Prof. Dr.-Ing. Gavin Kane ■ 01.10.2024 ■ 23

 Geometric Interpretation of the Cosine Rule
𝜋 𝜋 𝜋
𝜃< 𝜃= 𝜃>
2 2 2

y y y

V
V

V
θ θ θ
w w w
x x x
𝑣∙𝑤 >0 𝑣∙𝑤 =0 𝑣∙𝑤 <0

Prof. Dr.-Ing. Gavin Kane ■ 01.10.2024 ■ 24

 Geometric Interpretation of the Cosine Rule
𝜋 𝜋 𝜋
𝜃< 𝜃= 𝜃>
2 2 2

y y y

V
V

V
θ θ θ
w w w
x x x
𝑣∙𝑤 >0 𝑣∙𝑤 =0 𝑣∙𝑤 <0

Vectors are Orthogonal, when Inner Product = 0

Prof. Dr.-Ing. Gavin Kane ■ 01.10.2024 ■ 25
 Representation from Poses
in ℝ2
• Cartesian Coordinates on a plane
• Axes are parallel to unit vectors 𝑥ො and 𝑦ො {A}
• A point P = [x,y] is equal to the vector 𝒑 = yA
𝑥ෝ
𝒙 + 𝑦ෝ
𝒚
• A relative coordinate system has two PA
components
• Translation with Vector 𝑡 = [𝑥𝑡 , 𝑦𝑡 ] xA
• Rotation by the Angle θ
• A pose can be given as AξB~(𝑥𝑡 , 𝑦𝑡 , 𝜃)

Prof. Dr.-Ing. Gavin Kane ■ 01.10.2024 ■ 26

 Representation from Poses
in ℝ𝑛 Representation from Poses
• Coordinate representation in n in ℝ3
dimensional space
𝑣1 𝑤1
𝑣2
𝑣 = ⋮ ,𝑤 = ⋮
𝑤2 {A}
zA
𝑣𝑛 𝑤𝑛
PA
• Definitions of Inner Products, Euclidean
Norms and their properties remain the
same.
yA
• Orthogonality remains in n dimensional
space. xA
𝑣∙𝑤 =0

Prof. Dr.-Ing. Gavin Kane ■ 01.10.2024 ■ 27

 Basis Vectors
• Note that every element in ℝ𝑛 can be written as
a linear combination of n vectors. For example:
𝑥 2
𝑦 ∈ ℝ
𝑥 1 0
𝑦 = 𝑥 + 𝑦
0 1 {A}
yA
1 0
• In this example the vectors 𝑒1 = and 𝑒2 =
0 1
are referred to as basis vectors. PA

• This choice of basis vectors is convenient but xA

not unique. It is only important that basis
vectors are orthogonal.

Prof. Dr.-Ing. Gavin Kane ■ 01.10.2024 ■ 28

Linear Transformations
• Let T be a function that takes elements in one vector
space 𝑉1 and maps them onto elements in another • Some examples:
vector space 𝑉2 . We can say that the function T is a
Linear Transformation iff it satisfies the following • Scaling:
property: 𝑥 𝑠𝑥 𝑥
𝑇 𝑦 = 𝑠 𝑦
𝑦
𝑇 𝛼𝑣 + 𝛽𝑤 = 𝛼𝑇 𝑣 + 𝛽𝑇 𝑤 • Reflection:
𝑥 𝑦
𝑇 𝑦 =
where 𝑥
• Projection:
𝑥
𝛼, 𝛽 ∈ ℝ, 𝑣, 𝑤 ∈ 𝑉1 and T 𝑣 , 𝑇 𝑤 ∈ 𝑉2 𝑇 𝑦 = 𝑥+𝑦
• or
𝑇 𝛼𝑣 = 𝛼𝑇 𝑣
• and
𝑇 𝑣+𝑤 =𝑇 𝑣 +𝑇 𝑤

Prof. Dr.-Ing. Gavin Kane ■ 01.10.2024 ■ 29

 Matrices and Linear
Transformations • With this concept, we can push to represent the Linear
transformation in a Matrix form, where the columns of
• We can characterise the action of a the matrix represent the action of the function on the
Linear Transformation on vectors in basis vectors.
ℝ𝑛 by reporting it‘s action on the 𝑥 5𝑥 + 6𝑦
canonical basis vectors: 𝑓 𝑦 =
2𝑥 − 3𝑦
This function f, is a linear transformation so:
• For example if 𝑣 ∈ ℝ2 1 1 6
0 0 5
𝑓 𝑥 +𝑦 = 𝑥𝑓 + 𝑦𝑓 =𝑥 +𝑦
0 1 0 1 2 −3
𝑣 = 𝑥𝒆𝟏 + 𝑦𝒆𝟐
Let
5 6
𝑇(𝒗) = 𝑥𝑇(𝒆𝟏 ) + 𝑦𝑇(𝒆𝟐 ) 𝐴=
2 −3
We can say that:
𝑥 5 6 𝑥 5 6
𝑓 𝑦 = = 𝑥 + 𝑦
2 −3 𝑦 2 −3

Prof. Dr.-Ing. Gavin Kane ■ 01.10.2024 ■ 30

 Matrix Vector
Multiplication
• In general if 𝐴 ∈ ℝ𝑚×𝑛 is a matrix with m rows and n columns
• Examples:
and 𝑣 ∈ ℝ𝑛 we say that the matrix vector product: Produces
1 −3 1 1 −3 −5
𝑤 = 𝐴𝑣 =1 +2 =
0 4 2 0 4 8
• Produces a vector 𝑤 ∈ ℝ𝑚 whose elements are defined as 2
1 2 3 1 2 3 9
follows: −1 = 2 −1 +3 =
4 5 6 4 5 6 21
3
𝑛

𝑤𝑖 = ෍ 𝐴𝑖𝑗 𝑣𝑗 • This second example shows a property of Matrix

𝑗=1 Multiplication mapping a Vector from ℝ3 into ℝ2
Space
• We can think of this product as a weigthed sum of the
columns of A or as the column vector formed by by taking the
inner product of v with each row of the matrix A.

Prof. Dr.-Ing. Gavin Kane ■ 01.10.2024 ■ 31

 Composing Linear
Transformations
• Linear Compositions are also linear. • The same can be seen for linear transformations
• That is, if 𝑉1 , 𝑉2 and 𝑉3 are all vector spaces, f is a linear in Matrix form.
transformation from 𝑉2 to 𝑉3 and g is a linear transformation 𝑣𝑥
• Assuming ℝ2 , with 𝑣 = 𝑣
from 𝑉1 and 𝑉2 , then their composition 𝑓 ∘ 𝑔 will also be a 𝑦
linear transformation. 𝑣𝑥
𝑇2 𝑇1 𝑣 = 𝑇2 𝑣𝑥 𝑇1 𝑒1 + 𝑣𝑦 𝑇1 𝑒2
𝑦
𝑓 ∘ 𝑔 𝛼𝑣 + 𝛽𝑤 = 𝑓 𝑔 𝛼𝑣 + 𝛽𝑤 = 𝑣𝑥 𝑇2 𝑇1 𝑒1 + 𝑣𝑦 𝑇2 𝑇1 𝑒2
= 𝑓 𝛼𝑔 𝑣 + 𝛽𝑔 𝑤 𝑣𝑥
= 𝑇2 𝑇1 𝑒1 𝑇2 𝑇1 𝑒2 𝑣𝑦
= 𝛼𝑓 𝑔 𝑣 + 𝛽𝑓 𝑔 𝑤
• If the transformations 𝑇1 and 𝑇2 are represented
with matrices 𝑀1 and 𝑀2 we can see that the net
• Thus, the composition 𝑓 ∘ 𝑔 is a linear transformation from
result of applying the two transformations back
𝑉1 to 𝑉3
to back can be seen in a third matrix 𝑀3
𝑀3 = 𝑀2 𝑀1 : , 1 𝑀2 𝑀1 : , 2

Prof. Dr.-Ing. Gavin Kane ■ 01.10.2024 ■ 32

Composing Linear
Transformations
• Example:
𝑥 𝑥
𝑓 𝑦 = 𝑥 + 2𝑦 = 1 2 𝑦
𝑥 5𝑥 + 3𝑦 5 3 𝑥
𝑔 𝑦 = =
2𝑥 − 𝑦 2 −1 𝑦
• Consider the composition 𝑓 ∘ 𝑔 :
𝑥 1 0
𝑓∘𝑔 𝑦 =𝑥 𝑓∘𝑔 +𝑦 𝑓∘𝑔
0 1
5 3
=𝑥 1 2 +𝑦 1 2
2 −1
= 9𝑥 + 1𝑦
• So the matrix associated with 𝑓 ∘ 𝑔 is:
𝑓∘𝑔 = 9 1

Prof. Dr.-Ing. Gavin Kane ■ 01.10.2024 ■ 33

Matrix Multiplication
• In general, we can compute the product of • Examples:
two matrices: 𝐴 ∈ ℝ𝑙×𝑚 and B ∈ ℝ𝑚×𝑛 as
follows: 1 −3 1 3 1 −3 1 1 −3 3
=
𝐶 = 𝐴𝐵 0 4 2 0 0 4 2 0 4 0
𝑚 −5 3
=
𝐶𝑖𝑗 = ෍ 𝐴𝑖𝑘 𝐵𝑘𝑗 8 0
𝑘=1
• The resulting matrix, C, represents the 2 1 0
1 2 3 9 3 10
compostion of the two linear −1 4 −1 =
4 5 6 21 12 19
3 −2 4
transformations, A and B.
• One way to think of C is as the matrix
formed by applying the matrix A, to each of
the columns of B in turn.

Prof. Dr.-Ing. Gavin Kane ■ 01.10.2024 ■ 34

Matrix Operations
• Matrix Multiplication by scalar represents a • Matrix Addition scalar multiplication are distributive and
scaling. commutative:

• Matrix addition, of two matrices the same 𝐴+𝐵 =𝐵+𝐴

size, occurs by adding element by 𝛼 𝐴 + 𝐵 = 𝛼𝐴 + 𝛼𝐵
corresponding element.
• Matrix Multiplication is Distributive
• If A, B, and C are matrices and 𝛼, 𝛽 ∈ ℝ are 𝐴 𝐵 + 𝐶 = 𝐴𝐵 + 𝐴𝐶
scalars, then: 𝐴 + 𝐵 𝐶 = 𝐴𝐶 + 𝐵𝐶
• Matrix multiplication is associative: 𝐴(𝛼𝐵 + 𝛽𝐶) = 𝛼𝐴𝐵 + 𝛽𝐴𝐶

𝐴 𝐵𝐶 = 𝐴𝐵 𝐶

Prof. Dr.-Ing. Gavin Kane ■ 01.10.2024 ■ 35

Matrix Operations
• Matrix Multiplication by scalar represents a • Matrix Addition scalar multiplication are distributive and
scaling. commutative:

• Matrix addition, of two matrices the same 𝐴+𝐵 =𝐵+𝐴

size, occurs by adding element by 𝛼 𝐴 + 𝐵 = 𝛼𝐴 + 𝛼𝐵
corresponding element.
• Matrix Multiplication is Distributive
• If A, B, and C are matrices and 𝛼, 𝛽 ∈ ℝ are 𝐴 𝐵 + 𝐶 = 𝐴𝐵 + 𝐴𝐶
scalars, then: 𝐴 + 𝐵 𝐶 = 𝐴𝐶 + 𝐵𝐶
• Matrix multiplication is associative: 𝐴(𝛼𝐵 + 𝛽𝐶) = 𝛼𝐴𝐵 + 𝛽𝐴𝐶

𝐴 𝐵𝐶 = 𝐴𝐵 𝐶 • Matrix Multiplication is NOT COMMUTATIVE

𝐴𝐵 ≠ 𝐵𝐴

Prof. Dr.-Ing. Gavin Kane ■ 01.10.2024 ■ 36

 Transposition and Matrix
Matrix Transposition Multiplication
• Transposition is an operation that can be • A related relationship that can later be
applied to a matrix to swap the rows and useful is:
columns
𝐶 = 𝐴𝐵 ⟹ 𝐶 𝑇 = 𝐵𝑇 𝐴𝑇
• The transpose of a matrix 𝐴 is denoted by
𝐴𝑇 • Which also implies:

• In MATLAB you can produce the transpose (𝐴𝐵)𝑇 = 𝐵𝑇 𝐴𝑇

of a matrix 𝐴 by typing A‘
• Example:

1 4
1 2 3 𝑇
𝐴= 𝐴 = 2 5
4 5 6
3 6

Prof. Dr.-Ing. Gavin Kane ■ 01.10.2024 ■ 37

Coordinate Transformation
• Let’s begin by imaging 2 distinct
coordinate frames in the plane, {A} and {B} p
yA

• Let 𝑝𝐴 denote the coordinates of a point pA

wrt frame {A} and 𝑝𝐵 the coordinates of the
same point wrt frame {B}
{A} xA

Prof. Dr.-Ing. Gavin Kane ■ 01.10.2024 ■ 38

Homogenous Coordinates
• In order to model the coordinate
transformations that we will be interested in
with matrix operations we will adopt
homogenous coordinates. p
• We will associate points in the plane with yA
tuples consisting of 3 numbers where the last
number is a 1.
𝑥𝐴 𝑥𝐵 pA
𝑝𝐴 = 𝑦𝐴 , 𝑝𝐵 = 𝑦𝐵
1 1
• Vectors in the plane are associated with {A} xA
tuples where the last number is a 0.
𝑣𝐴𝑥 𝑣𝐵𝑥
𝑉𝐴 = 𝑣𝐴𝑦 , 𝑉𝐵 = 𝑣𝐵𝑦
0 0

Prof. Dr.-Ing. Gavin Kane ■ 01.10.2024 ■ 39

Coordinates Transformations
• By adopting homogenous
coordinates we will be able to
associate the coordinate
transformations that we are
p
interested in with matrix operations yA
as follows:
𝑝𝐴 = 𝑔𝐴𝐵 𝑝𝐵
pA
𝑥𝐴 𝑥𝐵
• Where 𝑝𝐴 = 𝑦𝐴 , 𝑝𝐵 = 𝑦𝐵 and 𝑔𝐴𝐵 {A} xA
1 1
is a 3 by 3 matrix.

Prof. Dr.-Ing. Gavin Kane ■ 01.10.2024 ■ 40

Elementary
Transformations - Scaling

yA p
𝑥𝐴 𝑠𝑥 0 0 𝑥𝐵
𝑦𝐴 = 0 𝑠𝑦 0 𝑦𝐵 yB
1 0 0 1 1

xB xA{A}
{B}

Prof. Dr.-Ing. Gavin Kane ■ 01.10.2024 ■ 41

Elementary
Transformations -
Rotation
yA p
𝑥𝐴 cos 𝜃 −sin 𝜃 0 𝑥𝐵
𝑦𝐴 = sin 𝜃 cos 𝜃 0 𝑦𝐵
1 0 0 1 1
𝜃
xA{A}

Prof. Dr.-Ing. Gavin Kane ■ 01.10.2024 ■ 42

Elementary
Transformations - yB
Translation
yA p
𝑥𝐴 1 0 𝑡𝑥 𝑥𝐵
𝑦𝐴 = 0 1 𝑡𝑦 𝑦𝐵
xB {B}
1 0 0 1 1

xA{A}

Prof. Dr.-Ing. Gavin Kane ■ 01.10.2024 ■ 43

 Representation from
Poses in ℝ3
• Position Coordinates are now vectors, 1x3 {A}
𝑥𝐴 zA
𝑝𝐴 = 𝑦𝐴
𝑧𝐴
• Rotations in ℝ3 represent the rotation of
PA
the base vectors
𝐴𝑅 = 𝐴 𝑖Ƹ 𝐴𝑗Ƹ 𝐴𝑘
෠𝐵
yA
𝐵 𝐵 𝐵
• These can be specified in a number of
ways as vectors, but most commonly
xA
Rotational Matrices 3x3

Prof. Dr.-Ing. Gavin Kane ■ 01.10.2024 ■ 44

Elementary
1 0 0
Transformations - 𝑅𝑜𝑡𝑥(𝜃) = 0 cos(𝜃) −sin(𝜃)
Rotation in ℝ3 0 sin(𝜃) cos(𝜃)
• Can be formed by rotations around
the individual axes. cos(𝜙) 0 sin(𝜙)
• The rotations are then indiviudally 𝑅𝑜𝑡𝑦(𝜙) = 0 1 0
seen as planar rotations −sin(𝜙) 0 cos(𝜙)
• Should the complete rotation
include multiple rotations, the
cos(𝜓) −sin(𝜓) 0
order of rotations are important!! 𝑅𝑜𝑡𝑧(𝜓) = sin(𝜓) cos(𝜓) 0
0 0 1

Prof. Dr.-Ing. Gavin Kane ■ 01.10.2024 ■ 45

Homogenous
Transformations in ℝ3
• A compact representation of the Example for translation in x,y and z and
translation and rotation is known rotation around y:
as the homogenous
transformation. cos(𝜙) 0 sin(𝜙) 𝑡𝑥
• Point Representation: 0 1 0 𝑡𝑦
𝑔𝐴𝐵 =
𝑥𝐴 −sin(𝜙) 0 cos(𝜙) 𝑡𝑧
𝑦 0 0 0 1
• 𝑝𝐴 = 𝐴
𝑧𝐴
1
• Transformation Representation:
𝑅𝐴𝐵 𝑡𝐴𝐵
• 𝑔𝐴𝐵 =
0 0 0 1

Prof. Dr.-Ing. Gavin Kane ■ 01.10.2024 ■ 46

Examples
1. H=trans(4,-3,7)roty(90°)rotz(90°) 2. H=trans(90°)rotz(90°)trans(4,-3,7)

Prof. Dr.-Ing. Gavin Kane ■ 01.10.2024 ■ 47

Frames of Reference
• A Frame of Reference defines a Joint(s) Reference Frame(s)
coordinate system relative to World Reference Frame
some point in space
• It can be specified by a position
and orientation relative to
other frames
• The initial frame is taken to be
a point that is assumed to be End-Effector
fixed in space Reference Frame
• We would like to be able to Basis Reference Frame
transform between frames of
reference
Prof. Dr.-Ing. Gavin Kane ■ 01.10.2024 ■ 48
 {B}
Coordinate System
• A Rigid Body B can be shown
as a collection of points

• The relative Pose of B (the local {A}

yA
coordinate System {B}) to the
reference Coordinate Frame is
given as ξ𝐵 , Xi ξB
xA

Prof. Dr.-Ing. Gavin Kane ■ 01.10.2024 ■ 49

 {B}
Coordinate System
• A Rigid Body B can be shown
as a collection of points p 𝑝𝐵

• The relative Pose of B (the local {A} 𝑝𝐴

yA
coordinate System {B}) to the
reference Coordinate Frame is
given as 𝑔𝐵
• Transformation of Rigid Body B
in Frame A xA
𝑝𝐴 = 𝑔𝐴𝐵 𝑝𝐵

Prof. Dr.-Ing. Gavin Kane ■ 01.10.2024 ■ 50

Chained Reference P
𝑝𝑐
Systems
• A series of reference frames
can be chained together {C}
𝑝𝐴 𝑔𝐵𝐶
𝑝𝐴 = 𝑔𝐴𝐶 𝑝𝑐
yA {B}
𝑔𝐴𝐶 = 𝑔𝐴𝐵 𝑔𝐵𝐶
𝑔𝐴𝐵
𝑝𝐴 = 𝑔𝐴𝐵 𝑔𝐵𝐶 𝑝𝑐
xA
{A}

Prof. Dr.-Ing. Gavin Kane ■ 01.10.2024 ■ 51

Example Reference System Fixed Camera
Reference Frame {C}
Transformation
𝑔𝐶𝐵
• Aim is to find Object B in Robot Reference Frame, to
determine the required movement of End-Effector to Object Object
Reference Frame
• Closed Loop gives: 𝑔𝐶 {B}
𝑔0𝐶 𝑔𝐶𝐵 = 𝑔0𝑅 𝑔𝑅𝐸 𝑔𝐸𝐵
𝑔𝐸𝐵
Noting: 𝑔𝐵
𝑔𝐵 is unknown {E}
End Effector
𝑔𝐶 and 𝑔𝑅 determined by calibration of system
𝑔𝑅𝐸 Reference Frame
𝑔𝑅𝐸 determined by forward kinematics 𝑔𝑅
𝑔𝐶𝐵 determined through camera measurments {R}
{O} Robot Base
𝑔𝐸𝐵 is required information, where should the World Reference Frame
end-effector more to? Reference Frame

Prof. Dr.-Ing. Gavin Kane ■ 01.10.2024 ■ 52

Example System Calculation
• A object with 3 pegs is placed on a
table as shown here:
• It needs to be picked up & placed on
another inclined plane part as shown.
Problem is about matching the two
coordinate systems viz (x2,y2,z2) to
(x5,y5,z5).
• Find position and orientation
matrices (transformation) of end 1
−300
1
−400
4
40
effecter in initial and final position for 𝑃2 = 200 , 𝑃4 = 250 , 𝑃5 = 60
the system shown on the right. 20 0 0
25
5
𝑃6 = 30 = 2𝑃3
10

Prof. Dr.-Ing. Gavin Kane ■ 01.10.2024 ■ 53