No,
one has to KNOW the field of gravity and subtract these
values from measured values of the three accelerometers.
No alternative. This has nothing to do with rotated coor-
dinates. Ideally, we have an inertially stabilized plat-
form with three accelerometers.
These cannot distinguish between gravity effects and motion
acceleration.
Best regards --Gernot Hoffmann
Reply by Matt Timmermans●July 23, 20052005-07-23
"mik" <pennese@tiscali.it> wrote in message
news:0radnTmIJsrgT33fRVn-uA@giganews.com...
> is it possible to decouple gravity and inertial acceleration, or better
> filter out gravity from measures taken ONLY with a triaxial accelerometer
With just one, if rotation is unconstrained, not really. There are many
ways you could move the accellerometer to produce any given set of readings.
The closest thing you could do would be to devise a statistical model of the
motion and choose the most likely way of producing your readings. The
result would be "plausible", but not accurate.
> (without using other sensors as gyroscopes ecc);
You can do it with two more triaxial accelerometers. Will that do?
--
Matt
Reply by Bob Cain●July 23, 20052005-07-23
mik wrote:
> I have a simple question (with a complex answer i suppose):
>
> is it possible to decouple gravity and inertial acceleration, or better
> filter out gravity from measures taken ONLY with a triaxial accelerometer
> (without using other sensors as gyroscopes ecc); if it is, can a Kalman
> filter minimize effect of gravity?
Seems it would be easy to calibrate it out by a measurement
taken when it is known to be in at rest relative to the
earth surface where you are. Thereafter just subtract it.
Or do you have reason to believe gravity's acceleration will
be signifigantly changing while in operation? In that case
it should still be calculable if you know the local
relationship of gravity's acceleration with position. This
doesn't violate GR because you are taking into account more
than can be known from within a sealed elevator taking into
account only local measurements.
In short, it should be instantaneously calculable and not
require an adaptive filter.
Bob
--
"Things should be described as simply as possible, but no
simpler."
A. Einstein
Reply by Jerry Avins●July 22, 20052005-07-22
mik wrote:
> I have a simple question (with a complex answer i suppose):
>
> is it possible to decouple gravity and inertial acceleration, ...
Not according to Special Relativity.
> or filter out gravity from measures taken ONLY with a triaxial accelerometer
> (without using other sensors as gyroscopes ecc); if it is, can a Kalman
> filter minimize effect of gravity?
Kalman, Shmalman. Even Newton requires six axes: three for translation,
three more for rotation.
Jerry
--
Engineering is the art of making what you want from things you can get.
�����������������������������������������������������������������������
Reply by Andre●July 22, 20052005-07-22
Ask someone who makes inertial navigation systems?
I guess if you cannot measure rotation, you will not be able to solve
your problem. With 3 degrees of freedom for moving, and 3 more for
rotation, you will need at least 6 sensors inputs to get your 6 outputs.
Best regards,
Andre
mik wrote:
> I have a simple question (with a complex answer i suppose):
>
> is it possible to decouple gravity and inertial acceleration, or better
> filter out gravity from measures taken ONLY with a triaxial accelerometer
> (without using other sensors as gyroscopes ecc); if it is, can a Kalman
> filter minimize effect of gravity?
>
>
>
> This message was sent using the Comp.DSP web interface on
> www.DSPRelated.com
--
Please change no_spam to a.lodwig when replying via email!
Reply by mik●July 22, 20052005-07-22
I have a simple question (with a complex answer i suppose):
is it possible to decouple gravity and inertial acceleration, or better
filter out gravity from measures taken ONLY with a triaxial accelerometer
(without using other sensors as gyroscopes ecc); if it is, can a Kalman
filter minimize effect of gravity?
This message was sent using the Comp.DSP web interface on
www.DSPRelated.com