Lecture 6: More MatLab; Functions, Transformations

Professor Erik Cheever

Course web page:

http://www.swarthmore.edu/NatSci/echeeve1/Class/e5/E5Index.html
Remember…
 Thursday 10/8: Bridge report is due.
 Thursday 10/8: First MatLab lab is due.

 Tuesday (8:30-10:00), Wednesday (8:00-10:00)

Wizards available in Hicks 211/213.

 Start thinking about your project (last 3 weeks of

class). Some ideas are at “Projects” link on left side
of course web page.
The big picture…
 Today:
 Matlab questions?
 MatLab functions.
 Graphical Transformations.
 This week in lab: functions and Graphical
Transformations.
 1st week after break: “Control of Flow” in MatLab.
 Following three weeks: design, build, and program a
physical robotic arm (using MatLab, functions,
transformations…).

 Last three weeks of class – individual projects. Some

ideas on “Projects” link on left of course web page.
Also…
 Skim (i.e., be familiar with what is there):
 Sections 3.2 and 3.6,
 Sections 4.2 and 4.4.1, and
 Section 6.2

 If you are anxious about programming, you may

want to try tutorials 2.5 and 2.6.
From Elektronik
(a Swedish Engineering Magazine)
 … MatLab demo…
&
… MatLab questions…
Functions
Sometimes we would like to repeat a set of
commands repeatedly.

An example:
Suppose we want to find the perimeter of a shape
defined by a “patch.”

Recall – distance from (x1,y1) to (x2,y2) is

2 2
d x1 x2 y1 y2
Finding the perimeter d x1 x2
2
y1 y2
2

To find the perimeter we:

 find the difference between consecutive points
in x and in y,
 square each difference in x and in y,
 add the square for corresponding x and y
differences.
 take the square root of each of these to find the
distance between consecutive points, and
 sum these distances.
Finding perimeter (1) 3
>> %Define shape (a parallelogram)
2
>> x = [0 1 3 2]; y=[0 1 1 0];
>> myShape=patch(x,y,[0 0 1]); 1

>> axis([-3 3 -3 3],'square'); grid on 0

-1
>> %Difference between x values
>> xd=diff(x) -2

xd = 1 2 -1 -3
-3 -2 -1 0 1 2 3

This is no good because it leaves off difference between last

point and first point.
>> x1=[x x(1)]
x1 = 0 1 3 2 0

>> xd=diff(x1)
xd = 1 2 -1 -2
Finding perimeter (2) d x1 x2
2
y1 y2
2

>>>> x1=[x x(1)]; %Make new arrays

>> y1=[y y(1)]
y1 = 0 1 1 0 0

>> xd=diff(x1) %Find difference between consecutive points

>> yd=diff(y1)
yd = 1 0 -1 0

>> sqrd = xd.^2 + yd.^2 %Square difference and add (x^2+y^2)

sqrd = 2 4 2 4

>> sqrtSqd=sqrt(sqrd) %Get length of each side

sqrtSqd = 1.4142 2.0000 1.4142 2.0000

>> perim=sum(sqrtSqd) %Get perimeter

perim = 6.8284
Problems with this approach
 If we want to calculate the perimeter of 10
different shapes, we need to enter the code 10
different times.
 Even if we use a script file, we need to make 10
copies of the code
 If we find a mistake in the code, we need to
change the code every place it appears.

 We can fix all of these issues by defining a

“function”
Functions
 For our purposes a function is a special MatLab
file that does a calculation(s) and returns a
result(s)
 It is different from a script file:
 it returns a result
 it can’t change other variables that you are using
Anatomy of a Function
Keyword “function” tells us this is a function file (called “perim.m”)
Output variable(s) name(s); in this case only one: the variable “p”
Function name; also name of “.m” file Input variables

function p = perim(xvals,yvals)

x1=[xvals xvals(1)];%Form augmented vector

y1=[yvals yvals(1)];

xd=diff(x1); %Calculate differences

yd=diff(y1); Calculations

sqrd = xd.^2+yd.^2; %Square them

sqrtSqd=sqrt(sqrd); %Find distances

p=sum(sqrtSqd); %Find perimeter

Output variable (must be the same variable name as is in
function declaration)
Using Functions
Now we can call function as if it was built in:

>> q=perim(x,y)
q = 6.8284

A trick: if you add comments to the very top of the

function file, they will appear if you ask for
documentation (i.e., “>>doc perim”).
Caveats
 Function name must not be the same as a
variable name
 Function name must have no spaces in it
 Function must be in MatLab directory so MatLab
can find it.
 If you edit a function, you must save the file
before the changes will take effect in subsequent
calls
 If you edit a function, you must save the file
before the changes will take effect in subsequent
calls
Coordinate transformations
 Robot arm: transformation between joint
angles and x, y, z position of laser light
 Next two weeks in lab: we will use coordinate
transformations to move around objects.
From this… … to this
 Rotation of a point (algebraic)
(rotation around the origin)

x  cos φ , y  sin φ
x'  cos θ φ , y'  sin θ φ

Trig identities :
cos θ φ cos φ cos θ sin φ sin θ
sin θ φ cos φ sin θ sin φ cos θ

x'  cos θ φ y'  sin θ φ

 cos φ cos θ  sin φ sin θ  cos φ sin θ  sin φ cos θ
x cos θ y sin θ x sin θ y cos θ
Image adapted from: http://www.cs.umu.se/kurser/TDBC07/HT04/handouts/HO-lecture7.pdf
Rotation of a point (2x2 matrix)
x' cos θ x sin θ y
y' sin θ x cos θ y

x' cos θ sin θ x

y' sin θ cos θ y

cos θ sin θ
rotation matrix=R
sin θ cos θ
 Translation of a point (3x3)
x' x tx , y' y ty

How can we express this with matrices?

x' 1 0 tx x
y' 0 1 ty y
1 0 0 1 1

1 0 tx
0 1 ty translation matrix=M
0 0 1

Image adapted from: http://www.cs.umu.se/kurser/TDBC07/HT04/handouts/HO-lecture7.pdf

 Rotation redefined (3x3)
We can combine a rotation (first)
with a translation (second).
This can be done in one operation if we
redefine the rotation matrix to be 3x3
cos θ sin θ 0
sin θ cos θ 0 rotation matrix=R
0 0 1

x' cos θ x sin θ y

y' sin θ x cos θ y

x' cos θ sin θ 0 x x

y' sin θ cos θ 0 y R y
1 0 0 1 1 1
Image adapted from: http://www.cs.umu.se/kurser/TDBC07/HT04/handouts/HO-lecture7.pdf
 Rotation and translation (1)
We can combine a rotation (first)
with a translation (second).
x' x x
y' M R y MR y
1 1 1

x' 1 0 tx cos θ sin θ 0 x

y' 0 1 ty sin θ cos θ 0 y
1 0 0 1 0 0 1 1

x' cos θ sin θ tx x

y' sin θ cos θ ty y
1 0 0 1 1

Image adapted from: http://www.cs.umu.se/kurser/TDBC07/HT04/handouts/HO-lecture7.pdf

 Rotation and translation (2)
cos θ sin θ tx
transformation
sin θ cos θ ty =T
matrix
0 0 1

x' cos θ sin θ tx x x

y' sin θ cos θ ty y T y
1 0 0 1 1 1

x' cos θ x sin θ y tx

y' sin θ x cos θ y ty

Image adapted from: http://www.cs.umu.se/kurser/TDBC07/HT04/handouts/HO-lecture7.pdf

Extension to many points
If we have several points (x1,y1), (x2,y2), (x3,y3), … (xn,yn), we can still
use the transformation matrix.

x1' x'2 x'3 xn' cos θ sin θ tx x1 x2 x3 xn

y1' y'2 y'3  yn' sin θ cos θ ty y1 y2 y3  yn
1 1 1 1 0 0 1 1 1 1 1

x1' cos θ x1 sin θ y1 tx

'
etc 
y 1 sin θ x1 cos θ y1 ty

If the points define a shape, we can rotate and translate that shape with a
single matrix multiplication.
… MatLab demo…
Other transformations…
(that we won’t be using)

Scaling

Shearing

Images from: http://www.cs.umu.se/kurser/TDBC07/HT04/handouts/HO-lecture7.pdf

Even more transformations…
(that we also won’t be using)

Reflection

Composite transformations (e.g., rotation about an arbitrary point (x1,y1)):

• Translate to origin
• Rotate
• Translate back

Images from: http://www.cs.umu.se/kurser/TDBC07/HT04/handouts/HO-lecture7.pdf