idea for Neo 2nd generation: Accelerometer

[Jose Ildefonso Camargo Tolosa]

On 1/24/07, Gabriel Ambuehl <gabriel_ambuehl at buz.ch> wrote:
> On Wednesday 24 January 2007 12:33, Sean Moss-Pultz wrote:
> > On 1/24/07 11:41 AM, "kkr" <kkr22.mpt56 at bluewin.ch> wrote:
> > > Does anybody know more about price and precision of the accelerometers's
> > > chips on the market now?
> >
> > About US$3.
>
> Are they good enough to twice integrate the acceleration data (which obviously
> puts quite some demand onto accuracy) to get a better estimate of where a car
> would be after entering a tunnel? If so, there's your killer app.

Here:

http://www.analog.com/en/subCat/0,2879,764%255F800%255F0%255F%255F0%255F,00.html

and here:

http://www.freescale.com/webapp/sps/site/taxonomy.jsp?nodeId=01126911184209

Yes, they are.  You just need a "fast enough" sampling rate.... the
problem is: how fast, and the other problem: filter out gravity in
order to avoid integrating it.  Thus, you need to "follow gravity", or
high-pass filter the thing, and you will lose ALL constant
accelerations.

Another issue is the "fast acceleration scenarios", where your
accelerometer would be "saturated", and you could need a less-precise,
higher-acceleration accelerometer in order
to "still know" where are you moving.

I have been working on a device like this (blind navigation system,
please don't steal the idea, if you wanna work in something like that,
we could work together, I just want it to be free)... just for fun.
If, someday, I finish it, I'll post it somewhere for everybody to
build it and use it!.

c-ya!

Ildefonso Camargo