GPS - GPS hysteresis

Author

GPS hysteresis

mortenbisgaard@gmail.com

2007-07-05, 3:33 pm

Hi

I an using a Novatel OEM4 stand alone GPS receiver for control of a
flying system. For this application I am ok with fairly bad accuracy,
but I cannot tolerate sudden changes in the measured postion. My
problem is that when the satellite number change, the solution jumps
around perhaps up to 1m in one timestep.

What I would like is to reduce the number of times that the receiver
changes number of satellites by applying some kind of hysteresis on
the satellites:
Ie. when a satellite comes above the horizon it needs to be above 30
degrees to be used. When i goes down again it needs to be below eg. 15
degrees to be discarted. Thereby I would get the position jumps when a
satellite goes in and out of the solution many times in a row.

However, I don't know if this is possible (with the Novatel OEM4) can
anyone help with this? Or perhaps some other clever idea. . .

Thanks, hope this made sense
Morten Bisgaard

Dominic Sexton

2007-07-05, 10:33 pm

In message <1183665808.078416.125370@n2g2000hse.googlegroups.com>,
mortenbisgaard@gmail
.com writes
>Hi
>
>I an using a Novatel OEM4 stand alone GPS receiver for control of a
>flying system. For this application I am ok with fairly bad accuracy,
>but I cannot tolerate sudden changes in the measured postion. My
>problem is that when the satellite number change, the solution jumps
>around perhaps up to 1m in one timestep.
>
>What I would like is to reduce the number of times that the receiver
>changes number of satellites by applying some kind of hysteresis on
>the satellites:
>Ie. when a satellite comes above the horizon it needs to be above 30
>degrees to be used. When i goes down again it needs to be below eg. 15
>degrees to be discarted. Thereby I would get the position jumps when a
>satellite goes in and out of the solution many times in a row.
>
>However, I don't know if this is possible (with the Novatel OEM4) can
>anyone help with this? Or perhaps some other clever idea. . .

You can set the mask angle or in Novatel speak the elevation cut-off
angle with the ECUTOFF command. The default is five degrees.

You cannot have a different cut off for adding a satellite to the
solution and removing it from the solution.

However you may well be better off applying some filtering to the output
of the receiver rather than trying to constrain the receiver to smooth
the position fixes.

For example you might feed the current position / velocity solution from
the receiver into a comparison with the predicted position / velocity
(calculated from the recent position and velocity data). If they agree
well (as they should when there are no jumps in the solution) accept the
new current position / velocity.

If they do not agree well you could average the predicted solution with
that currently reported by the receiver. The average solution will not
be correct but it will change much more smoothly than the raw solution
from the receiver.

When the system gets past the cause of the glitch in a few solutions the
prediction will once again agree with the current solution from the
receiver.

You can further develop this to take account of the dynamics of the
previous solutions and when there significant changes (quick changes
heading or ascent / descent rate) allow a larger margin of error between
the prediction and the current solution. Where the craft is in a steady
state you can add extra significance to the prediction.
--
Dominic Sexton

utilsea@aol.com

2007-07-05, 10:33 pm

Hi,
it's a typical application for hybrided system (GPS + attitude
sensor).

GPS give the position and the attitude sensor is used for filtering
GPS data.

A little bit expensive but the most efficient.
If you are using NOVATEL board have a look on this side they provide
some GPS board ready for hybridation.

Http://utilsea.free.fr

LinkBot