Finding absolute position of an object: which kind of sensor?

carlorossi

Dear all,

I'm looking for a sensors kit able to give me its X-Y space coordinates in 2D.

I would like to put two or three of these objects in a table and get their positions in that table in realtime.

I already tried with a gyroscope and with an accelerometer, trying to calculate their position starting from their moving speed and their rotation speed.
But there is always a notable error... when I move the object and then I put the object in the same position again, the coordinates don't return on the zero position.
There is always a little error that after some movement will become huge.

I think it is not an error in the algorithm, I think a gyroscope/accelerometer is not a good sensor to find an absolute position.

I also tried with a simple 2D mouse, but the same error occurs:
if you try to put a mouse over a paper, you see that after few movement its position in the paper does not correspond to the pointer position in the computer screen anymore.

I would like to try with the Nintendo Wii, but I'm pretty sure it will not give me its absolute position.

And I don't like videotracking (like Reactable), it is too much delicate and unstable.

So, what kind of sensor do you think will be good for getting its absolute position over a table?

Thank you very much.
Carlo

sfogar

Hi,

what about a graphic tablet ?

All the best

--
Alessandro Fogar

http://www.fogar.it

2007/9/28, carlo rossi :
>
> Dear all,
>
> I'm looking for a sensors kit able to give me its X-Y space coordinates in 2D.
>
> I would like to put two or three of these objects in a table and get their positions in that table in realtime.
>
> I already tried with a gyroscope and with an accelerometer, trying to calculate their position starting from their moving speed and their rotation speed.
> But there is always a notable error... when I move the object and then I put the object in the same position again, the coordinates don't return on the zero position.
> There is always a little error that after some movement will become huge.
>
> I think it is not an error in the algorithm, I think a gyroscope/accelerometer is not a good sensor to find an absolute position.
>
> I also tried with a simple 2D mouse, but the same error occurs:
> if you try to put a mouse over a paper, you see that after few movement its position in the paper does not correspond to the pointer position in the computer screen anymore.
>
> I would like to try with the Nintendo Wii, but I'm pretty sure it will not give me its absolute position.
>
> And I don't like videotracking (like Reactable), it is too much delicate and unstable.
>
> So, what kind of sensor do you think will be good for getting its absolute position over a table?
>
> Thank you very much.
> Carlo
>

Mark-David Hosale

Hi Carlo-
Have you thought about something that works like a graphics tablet?
There are a bunch of DIY sources for graphic tablets if you google.
The general principle can be found here:
http://en.wikipedia.org/wiki/Graphics_tablet

Another solution might be to use an interactive whiteboard interface
like the mimio or the ebeam:
http://www.e-beam.com/

I was running into a similar problem in the development of a project
recently and found the reactable to be complicated for what I needed.
I finally decided that the best thing for my needs was to buy an OEM
touch screen:
http://www.magictouch.com
The extra money I spent was made-up for by the time I saved.

Hope this helps-
MarkDavid

On Sep 28, 2007, at 3:31 AM, carlo rossi wrote:

>
> Dear all,
>
> I'm looking for a sensors kit able to give me its X-Y space
> coordinates in 2D.
>
> I would like to put two or three of these objects in a table and
> get their positions in that table in realtime.
>
> I already tried with a gyroscope and with an accelerometer, trying
> to calculate their position starting from their moving speed and
> their rotation speed.
> But there is always a notable error... when I move the object and
> then I put the object in the same position again, the coordinates
> don't return on the zero position.
> There is always a little error that after some movement will become
> huge.
>
> I think it is not an error in the algorithm, I think a gyroscope/
> accelerometer is not a good sensor to find an absolute position.
>
> I also tried with a simple 2D mouse, but the same error occurs:
> if you try to put a mouse over a paper, you see that after few
> movement its position in the paper does not correspond to the
> pointer position in the computer screen anymore.
>
> I would like to try with the Nintendo Wii, but I'm pretty sure it
> will not give me its absolute position.
>
> And I don't like videotracking (like Reactable), it is too much
> delicate and unstable.
>
> So, what kind of sensor do you think will be good for getting its
> absolute position over a table?
>
> Thank you very much.
> Carlo
>
>

carlorossi

Thank you for answers.
However, I'd like to have more than one object to move simultaneously.
Also, I really like the idea to "touch" a real object.
So, putting some kind of sensor inside of every objects will be really perfect for what I would like to do.

I'm looking in the Electrotap and Makingthings webpage, without success...

Sukandar Kartadinata

have a look at this:
http://tangible.media.mit.edu/content/papers/pdf/SenseTable_CHI01.pdf
http://www.media.mit.edu/physics/publications/papers/02.PatRec.NIME.pdf

they're using LC tags inside objects on top of a modified Wacom
tablet if I remember correctly

On 28.09.2007, at 13:41, carlo rossi wrote:

>
> Thank you for answers.
> However, I'd like to have more than one object to move simultaneously.
> Also, I really like the idea to "touch" a real object.
> So, putting some kind of sensor inside of every objects will be
> really perfect for what I would like to do.
>
> I'm looking in the Electrotap and Makingthings webpage, without
> success...

sfogar

Hi,

I remember that on the old Mattel Powerglove they detected the
position with ultrasonic sensors.

There was a little buzzer emitting the 'tone' and the sensor (mics)
then measured the distance calculating the delay.

For example:

http://www.parallax.com/detail.asp?product_id=28015

Just an idea...

--
Alessandro Fogar

http://www.fogar.it

yair reshef

"Motion Tracking:. No Silver Bullet,. but a Respectable. Arsenal"
( www.cs.unc.edu/~tracker/media/pdf/cga02_welch_tracking.pdf )
this pdf sums the state of the art in the field. intresting reading.
i dont have a silver bullet to offer.
the "dead reckoning" methods you tried are troublesome, mybe give some
thought to the video tracking. its cheaper then most. and if you build the
table with care can work outside of the lab.

On 9/28/07, Alessandro Fogar wrote:
>
> Hi,
>
> I remember that on the old Mattel Powerglove they detected the
> position with ultrasonic sensors.
>
> There was a little buzzer emitting the 'tone' and the sensor (mics)
> then measured the distance calculating the delay.
>
> For example:
>
> http://www.parallax.com/detail.asp?product_id=28015
>
> Just an idea...
>
> --
> Alessandro Fogar
>
> http://www.fogar.it
>
>
> 2007/9/28, carlo rossi :
> >
> > Dear all,
> >
> > I'm looking for a sensors kit able to give me its X-Y space coordinates
> in 2D.
> >
> > I would like to put two or three of these objects in a table and get
> their positions in that table in realtime.
> >
> > I already tried with a gyroscope and with an accelerometer, trying to
> calculate their position starting from their moving speed and their rotation
> speed.
> > But there is always a notable error... when I move the object and then I
> put the object in the same position again, the coordinates don't return on
> the zero position.
> > There is always a little error that after some movement will become
> huge.
> >
> > I think it is not an error in the algorithm, I think a
> gyroscope/accelerometer is not a good sensor to find an absolute position.
> >
> > I also tried with a simple 2D mouse, but the same error occurs:
> > if you try to put a mouse over a paper, you see that after few movement
> its position in the paper does not correspond to the pointer position in the
> computer screen anymore.
> >
> > I would like to try with the Nintendo Wii, but I'm pretty sure it will
> not give me its absolute position.
> >
> > And I don't like videotracking (like Reactable), it is too much delicate
> and unstable.
> >
> > So, what kind of sensor do you think will be good for getting its
> absolute position over a table?
> >
> > Thank you very much.
> > Carlo
> >
>

Mark-David Hosale

Interactive whiteboards like the ebeam or mimio will work for
multiple objects. They need to track several objects at once in order
to tell the colors apart on the pens. There are several hacks on the
web if you poke around, including:

Graphonic Interface (uses ebeam)
http://www.create.ucsb.edu/~dano/graphonic/

Drafting Poems: Inverted Potentialities (uses mimio)
http://www.aiaesthetic.com/index.php?work=DraftingPoems

Of course you are limited to the number of color of pens, or type of
objects that are being tracked (mimio has an eraser for example). But
my guess is that these systems are fairly complex to rebuild from
scratch, therefore hacking either the ebeam or the mimio is probably
you're best approach.

hope this helps-
good luck in your quest-
md

On Sep 28, 2007, at 4:41 AM, carlo rossi wrote: