Prev: LPCTSTR: whats going wrong here?
Next: Good MFC Book
From: Andreas Hoffmann on 22 Apr 2008 15:18
Hello,

how can I calculate the regression wir MFC, C++?


Regards Andreas
x y
-----------------------------------------
1 10 10
2 10 20
3 10 30
4 10 40
5 20 40
6 20 30
7 20 20
8 20 10
9 30 10
10 30 20
11 30 30
12 30 40
13 40 40
14 40 30
15 40 20
16 40 10
17 50 10
18 50 20
19 50 30
20 50 40

~~~~~~~~~~~~~~~~~~~~~

x-measure y-measure
1 10,05 10,10
2 10,05 20,11
3 10,05 30,14
4 10,05 40,15
5 20,08 40,15
6 20,08 30,20
7 20,08 20,12
8 20,08 10,09
9 30,15 10,10
10 30,15 20,16
11 30,15 30,21
12 30,15 40,16
13 40,22 40,17
14 40,22 30,22
15 40,22 20,18
16 40,22 10,09
17 50,31 10,10
18 50,31 20,12
19 50,31 30,19
20 50,31 40,18


Transformation:
-------------------------------------------------------------
X = -0.023 + 0.99*x + -0.000001*y + -0.000000*x*y +
-0.000084*x^2 + 0.000000*y^2

Y = 0.04 + -0.0031*x + 0.989*y + -0.00002*x*y +
0.00005*x^2 + 0.00019*y^2


X(x,y) = a0 + a1*x + a2*y + a3*x^2 + a4*y^2 + a5*x*y + ...
Y(x,y) = b0 + b1*x + b2*y + b3*x^2 + b4*y^2 + b5*x*y + ...
From: Joseph M. Newcomer on 22 Apr 2008 15:35
Probably by writing the code that performs the computation?

Seriously, I have no idea how any of these values are computed or derived.

MFC is probably irrelevant.
joe

On Tue, 22 Apr 2008 12:18:07 -0700 (PDT), Andreas Hoffmann <andreas_hoffmann2(a)arcor.de>
wrote:

>Hello,
>
>how can I calculate the regression wir MFC, C++?
>
>
>Regards Andreas
> x y
>-----------------------------------------
>1 10 10
>2 10 20
>3 10 30
>4 10 40
>5 20 40
>6 20 30
>7 20 20
>8 20 10
>9 30 10
>10 30 20
>11 30 30
>12 30 40
>13 40 40
>14 40 30
>15 40 20
>16 40 10
>17 50 10
>18 50 20
>19 50 30
>20 50 40
>
>~~~~~~~~~~~~~~~~~~~~~
>
> x-measure y-measure
>1 10,05 10,10
>2 10,05 20,11
>3 10,05 30,14
>4 10,05 40,15
>5 20,08 40,15
>6 20,08 30,20
>7 20,08 20,12
>8 20,08 10,09
>9 30,15 10,10
>10 30,15 20,16
>11 30,15 30,21
>12 30,15 40,16
>13 40,22 40,17
>14 40,22 30,22
>15 40,22 20,18
>16 40,22 10,09
>17 50,31 10,10
>18 50,31 20,12
>19 50,31 30,19
>20 50,31 40,18
>
>
>Transformation:
>-------------------------------------------------------------
>X = -0.023 + 0.99*x + -0.000001*y + -0.000000*x*y +
>-0.000084*x^2 + 0.000000*y^2
>
>Y = 0.04 + -0.0031*x + 0.989*y + -0.00002*x*y +
>0.00005*x^2 + 0.00019*y^2
>
>
>X(x,y) = a0 + a1*x + a2*y + a3*x^2 + a4*y^2 + a5*x*y + ...
>Y(x,y) = b0 + b1*x + b2*y + b3*x^2 + b4*y^2 + b5*x*y + ...
Joseph M. Newcomer [MVP]
email: newcomer(a)flounder.com
Web: http://www.flounder.com
MVP Tips: http://www.flounder.com/mvp_tips.htm
From: Andreas Hoffmann on 27 Apr 2008 11:47
Hello,
> Seriously, I have no idea how any of these values are computed or derived.  
>
> MFC is probably irrelevant.
>                                 joe
>
ok.
I need a similar function like
http://www.mn-electronic.de/v3/index.php?val=en&l1=10&l2=20&l3=0
http://de.search.yahoo.com/search;_ylt=A1f4cfjxnhRIExEAyk8zCQx.?p=camera+fiducial+pick+place&y=Suche&fr=yfp-t-501&ei=UTF-8&rd=r1

How can I calibrate a machine, like a pick & place machine.
I need the algorithm in C++, MFC.
Or a good math library.

Regards Andreas
From: Joseph M. Newcomer on 27 Apr 2008 21:37
What's this got to do with the regression computation? Are you doing robotics?

The answer is, it depends on what data you have available. How accurate it is. What your
robotic controls are capable of. What kind of backlash the physical activators have, what
resolution they provide, and about twenty or thirty other parameters.

At this point, you are into control theory, and the algorithms depend on the nature of the
feedback loop you have. It would depend on, for example, the camera which is being used,
its resolution, perhaps its color depth. Very little seems to have anything to do with
regression analysis.

You are asking for a very complicated piece of software, perhaps several hundred thousand
lines of code, with interactions with robotic devices, probably with real-time feedback
from cameras. Such software is typically proprietary.

I've never done robotics, but I have friends who do this, and the problems of doing
real-time component placement were issues they were addressing 20-30 years ago. I have no
idea how the technology has changed in the intervening years, but it certainly isn't a
simple mathematical algorithm.
joe


On Sun, 27 Apr 2008 08:47:51 -0700 (PDT), Andreas Hoffmann <andreas_hoffmann2(a)arcor.de>
wrote:

>Hello,
>> Seriously, I have no idea how any of these values are computed or derived. �
>>
>> MFC is probably irrelevant.
>> � � � � � � � � � � � � � � � � joe
>>
>ok.
>I need a similar function like
>http://www.mn-electronic.de/v3/index.php?val=en&l1=10&l2=20&l3=0
>http://de.search.yahoo.com/search;_ylt=A1f4cfjxnhRIExEAyk8zCQx.?p=camera+fiducial+pick+place&y=Suche&fr=yfp-t-501&ei=UTF-8&rd=r1
>
>How can I calibrate a machine, like a pick & place machine.
>I need the algorithm in C++, MFC.
>Or a good math library.
>
>Regards Andreas
Joseph M. Newcomer [MVP]
email: newcomer(a)flounder.com
Web: http://www.flounder.com
MVP Tips: http://www.flounder.com/mvp_tips.htm
 | 
Pages: 1
Prev: LPCTSTR: whats going wrong here?
Next: Good MFC Book