matrix pressure sensor interface

I have read with interest the way to interface a keyboard to a 8051 microcontroller. My question is this: I have developed a fabric that is made like a matrix keyboard. For example there are 12 conductive rows and 12 conductive columns. Everything works fine when you press only in one point, like on a touch screen. But I want to go a little further and use this fabric for shape recognition. In other words I'm trying to find a way to recognize the shape of the object that I put on my fabric.

The problem is that when I put i.e. a round object on my fabric all rows and columns that are touched by my object will short in a n area that has the maximum sizes of the circle, so I will not see a circle, but a square. With an oval shape I will see a rectangle.

Do you have any experience?

Thank you in advance for your help

Riccardo

Hello Riccardo.

Interesting project.
I think you should be able to tell a circle from a square using the row/column method.

I suspect your program logic is letting you down.
Could you post a schematic and your code ?

I'm not sure I agree, but would enjoy the discussion. Assuming a resistive strip for either the row or column, how would you discern a touch in multiple areas?

If you reduce the touched area to a single vertical strip, let's say the width of 4 horizontal strips, you can scan the horizontal strips to find the length of the touch. If, on the other hand, the vertical strips on either side of the original strip are touching, say the width of only 2 horizontal strips, the first strip will "shadow" the smaller area, because there is no way to scan the smaller contact area. The larger, first vertical strip would "shadow" the smaller vertical strip contact area.

I think???

hahahah

A drawing might help explain what I'm trying to say:

What do you think?

-Dave

Thank you for your prompt replies. I think Dave got the point. If the strips have a low resistance my logic will see an area where all strips are shorted, so it will not be possible to discern the proper shape (see Dave's picture).
I was thinking to use resistive strips instead of conductive ones, in order to be able to recognize in some way the position where the strips are shorted. See image below.



Suppose each square has a resistance of R. I ground all but one column and one row. When I measure resistance between point 1 and 3 or point 2 and 3 I think I could be able to see some difference. My question is how difficult it is to convert this system into a mathematical law that then I can transfer into a program. It is going to be a big matrix. I think I should start with a generic element, find the numerical model and then apply the same rules to all other ones. What do you think?
Riccardo

I'd be curious how you're planning on making the resistive squares :-)

me too.. constantly watching this thread... very innovative

I was thinking to simply knit resistive wire, instead of a conductive one. Actually I wouldn't do squares, but strips. I need a couple of weeks. I'll keep you update.

Its really something new inntex thanks for sharing this with us. i think more dense the resistive strips are more accurate will be the results. i am still thinking of design. maybe it will be an x-y kind of mesh design?

Yes, x-y axis is what I'm trying to do.
Density, I did not take that into consideration. Good idea, thank you.
Riccardo