HOMOGENEOUS TRANSFORMATION MATRICES

CARTESIAN COORDINATES

Corkscrew rule
z (right-hand rule) y
to define positive
directions of the
axes

y x

x z Cartesian coordinates
Cartesian coordinates used in computer
used in robotics graphics

Prof. Emil. M. Petriu

 RECOVERING THE LOCATION OF MOVING OBJECTS: TRANSLATION
Using a reference frame N affixed to the object allows the calculation of the variable
location of the point Q on the moving object in a given reference frame R as the sum
z0 of two vectors, a constant one representing the location of Q relative to N and another
variable representing the location of N relative to R:
Rq = Rp + Nq

z1
Recovering the 3D location of a point on a moving object
Q in a given reference frame R. when the object suffers only
a translation movement relative to R..
R p* = R p + Rr
Nq
z1 R q* R Nq
= p* +
Rq y1 Q*
Rp R q* R q* R Nq
N = p+ + Rr
x1
Rr Nq
R
y0
R P*
x0 N* y1
x1

Prof. Emil. M. Petriu

 RECOVERING THE LOCATION OF MOVING OBJECTS: ROTATION

Recovering the 3D location of a point on an object Vectors can not

R q* = ?
rotating in a given reference frame R. describe the rotation !

z0
z1 z0
Q z1

R q* = ? Q*
Nq

y1 Nq

Rp
x1
ROT N N*
Rp
x1
y0 y1
R
y0
x0
x0

Prof. Emil. M. Petriu

RECOVERING THE LOCATION OF MOVING OBJECT: ROTATION > continued

z0

z1
Q z
1
Rot( z, φ )
Q*
Nq
Rot( y, θ)
Nq y1y1
x N
R p1
ROT N
N*
xR1 p
x1 Rot( x, ψ ) y1
R
y0
x0

Prof. Emil. M. Petriu

 RECOVERING THE LOCATION OF MOVING OBJECT: ROTATION > continued

The general rotation of an object in 3D

z0 can be conveniently decomposed into a
sequence of three elementary rotations
about the axes of the Cartesian reference
frame attached to that object, namely:
z1 Rot( z, φ ) a rotation about z-axis, “roll,”
by an angle φ ;
Q Rot( y, θ) a rotation about y-axis, “pitch,”
Rot( z, φ ) by an angle θ ;
Rot( x, ψ ) a rotation about x-axis, “yaw,”
by an angle ψ .
Nq
Rot( y, θ)
N y1y1
Rp
ROT N
x1
x1 Rot( x, ψ )
R
y0
x0

Prof. Emil. M. Petriu

 RECOVERING THE LOCATION OF MOVING OBJECT: ROTATION > continued

z0
The general rotation in 3D is often known as the
RPY rotation , a name reminding of the Roll,
Pitch, and Yaw components.
z1
Q z
1
Rot( z, φ )
Q*
Nq
Rot( y, θ)
Nq y1y1
x N
R p1
RPY
ROT N
N*
xR1 p
x1 Rot( x, ψ ) y1
R
y0
x0

Prof. Emil. M. Petriu

RECOVERING THE LOCATION OF MOVING OBJECT: ROTATION >> continued

z z

Rot( z, φ )
Roll z*
Q
Q* y*

Nq

y y

N N
Rot( y, θ)
Pitch x*
Rot( x, ψ )
x x
Yaw

Prof. Emil. M. Petriu

 RECOVERING THE LOCATION OF MOVING OBJECTS: Roll, Rot( z, φ )

z z

Rot( z, φ )
Roll
z Q (x,y,z) z’ = z
Q* (x’,y’,z’)

y*
y φ
N y y’ y
x x’

x x x*

Prof. Emil. M. Petriu

 RECOVERING THE LOCATION OF MOVING OBJECTS: Rot( z, φ ) > continued

y* y’ = d . sin(α+ φ)
y’ = d . sin(α) . cos(φ) + d . sin(φ) . cos(α)
φ
y y’ x = d . cos(α)
y y = d . sin(α)

y’ = y . cos(φ) + x . sin(φ)
d
d
α
Q* (x’,y’) In a similar way we get:
α
x
Q (x,y) x’ = x . cos(φ) − y . sin(φ)

φ
x*
x x’ cos(φ) − sin(φ) x
=
y’ sin(φ) cos(φ) y

Prof. Emil. M. Petriu

3D HOMOGENEOUS TRANSFORMATION MATRICES FOR ROBOTICS

Homogeneous transformation 3-by-3 3-by-1

matrices are general 4-by-4 rotation translation
matrices accounting for both T = sub-matrix sub-matrix
object translation and object
rotation in 3D. 0 0 0 1

The homogeneous
transformations matrix 1 0 0 px
for a general Trans (px, py, pz) = 0 1 0 py
3D translation 0 0 1 pz
by a vector 0 0 0 1
px . i + py . j + pz . k is:

Prof. Emil. M. Petriu

 The homogeneous transformations matrices for 3D rotations about the x, y, and z axes:

z 1 0 0 0
0 cos (ψ) -sin (ψ) 0
Rot( x, ψ ) = 0 sin (ψ) cos (ψ) 0
Rot( z, φ )
0 0 0 1
Roll
Q (x,y,z)
cos (θ) 0 sin (θ) 0
0 1 0 0
Rot( y, θ ) = -sin (θ) 0 cos (θ) 0
0 0 0 1
y

N
Rot( y, θ)
cos (φ) -sin (φ) 0 0
Pitch sin (φ) cos (φ) 0 0
Rot( z, φ ) = 0 0 1 0
Rot( x, ψ )
x 0 0 0 1
Yaw

Prof. Emil. M. Petriu

Homogeneous transformation matrices allow to calculate the final
effect of a sequence of object transforms (translations and/or rotations) by
multiplying the homogeneous matrices corresponding to these transforms.

l2 l3

θ3
θ2
l1 { x 3 , y3 , z 3 }
θ1

Prof. Emil. M. Petriu