http://www.gpsworld.com/Wireless/exp...location-12540

> A-GNSS + Wi-Fi Positioning + Cellular Positioning + Sensors. You might have guessed it, but we think accurate everywhere location will be enabled by a combination of all the above hybrid techniques plus one more important technology: sensors. Integrated sensors like accelerometers, magnetometers, and barometers enable devices to sense changes in direction, orientation, and elevation. Given an accurate starting location (for example, GNSS position fix), sensors can track location accurately for several minutes (and this will continue to get better). Location error will accumulate over time, but this can be minimized when Wi-Fi, cellular, and GNSS positioning are used in conjunction to constrain the error. Furthermore, barometers can be used to track elevation changes, thereby allowing devices to know exactly what floor of a building a user is on. Other technologies, or signals of opportunity, may be used in the future to further improve performance, but we think this mix of A-GNSS, W
i-Fi, cellular, and sensor positioning is the key to accurate everywhere location in mobile devices.

On 2012-02-04 11:18 , Sam Wormley wrote:
> http://www.gpsworld.com/Wireless/exp...location-12540
>
>
>> A-GNSS + Wi-Fi Positioning + Cellular Positioning + Sensors. You might
>> have guessed it, but we think accurate everywhere location will be
>> enabled by a combination of all the above hybrid techniques plus one
>> more important technology: sensors. Integrated sensors like
>> accelerometers, magnetometers, and barometers enable devices to sense
>> changes in direction, orientation, and elevation. Given an accurate
>> starting location (for example, GNSS position fix), sensors can track
>> location accurately for several minutes (and this will continue to get
>> better). Location error will accumulate over time, but this can be
>> minimized when Wi-Fi, cellular, and GNSS positioning are used in
>> conjunction to constrain the error. Furthermore, barometers can be
>> used to track elevation changes, thereby allowing devices to know
>> exactly what floor of a building a user is on. Other technologies, or
>> signals of opportunity, may be used in the future to further improve
>> performance, but we think this mix of A-GNSS, W
> i-Fi, cellular, and sensor positioning is the key to accurate everywhere
> location in mobile devices.

The same was true when automobile GPS's first came out - odometer
integration and magnetic heading were to aid the GPS in urban canyons
and in the presence of SA. That all went away as the constellation
became full and SA was removed.

Certainly the new combinations of sensors will do better.

--
"We demand rigidly defined areas of doubt and uncertainty."
Douglas Adams - (Could have been a GPS engineer).

Alan Browne <(E-Mail Removed)> writes:
> The same was true when automobile GPS's first came out - odometer
> integration and magnetic heading were to aid the GPS in urban canyons
> and in the presence of SA. That all went away as the constellation
> became full and SA was removed.
>
> Certainly the new combinations of sensors will do better.

I really wish GPS firmware engineers tried a bit harder to calibrate the
magnetic compass. I have 1 Garmin GPS and 2 android phones with
magnetic sensors. Both are mostly worthless in the car due to the car's
own magnetic fields superimposing its own field on the earth's
background field. I've never understood why they couldn't calibrate out
that field by simply taking the observed heading from the GPS whenever
the GPS indicated a reasonable velocity and slowly building up a lookup
table for the times when the GPS didn't indicate a high enough velocity
to generate a good heading.

-wolfgang
--
g+: https://plus.google.com/114566345864337108516/about

On 2/4/12 3:36 PM, Wolfgang S. Rupprecht wrote:
>
> Alan Browne<(E-Mail Removed)> writes:
>> The same was true when automobile GPS's first came out - odometer
>> integration and magnetic heading were to aid the GPS in urban canyons
>> and in the presence of SA. That all went away as the constellation
>> became full and SA was removed.
>>
>> Certainly the new combinations of sensors will do better.
>
> I really wish GPS firmware engineers tried a bit harder to calibrate the
> magnetic compass. I have 1 Garmin GPS and 2 android phones with
> magnetic sensors. Both are mostly worthless in the car due to the car's
> own magnetic fields superimposing its own field on the earth's
> background field. I've never understood why they couldn't calibrate out
> that field by simply taking the observed heading from the GPS whenever
> the GPS indicated a reasonable velocity and slowly building up a lookup
> table for the times when the GPS didn't indicate a high enough velocity
> to generate a good heading.
>
> -wolfgang

I've had similar thoughts. If motion were detected, then the
COG could be used to correct (or override) any magnetic
compass.

On 2012-02-04 16:36 , Wolfgang S. Rupprecht wrote:
>
> Alan Browne<(E-Mail Removed)> writes:
>> The same was true when automobile GPS's first came out - odometer
>> integration and magnetic heading were to aid the GPS in urban canyons
>> and in the presence of SA. That all went away as the constellation
>> became full and SA was removed.
>>
>> Certainly the new combinations of sensors will do better.
>
> I really wish GPS firmware engineers tried a bit harder to calibrate the
> magnetic compass. I have 1 Garmin GPS and 2 android phones with
> magnetic sensors. Both are mostly worthless in the car due to the car's
> own magnetic fields superimposing its own field on the earth's
> background field. I've never understood why they couldn't calibrate out
> that field by simply taking the observed heading from the GPS whenever
> the GPS indicated a reasonable velocity and slowly building up a lookup
> table for the times when the GPS didn't indicate a high enough velocity
> to generate a good heading.

You remind of a Kalman-Filter a friend implemented at work back in the
90's that calibrated out the TARSYN magnetic hdg error curve using other
inputs including GPS track.

My etrex 30 has a 3 axis magnetic heading sensor. There is a
calibration routine where you rotate the compass in sequence around the
z, y and x axis' and then calibration is declared complete (or start
over). The mag heading on that thing is worse than 10°. The iPhone 4
mag heading I can often get to better than 5°.

--
"We demand rigidly defined areas of doubt and uncertainty."
Douglas Adams - (Could have been a GPS engineer).

On 2012-02-04 17:43 , Sam Wormley wrote:
> On 2/4/12 3:36 PM, Wolfgang S. Rupprecht wrote:
>>
>> Alan Browne<(E-Mail Removed)> writes:
>>> The same was true when automobile GPS's first came out - odometer
>>> integration and magnetic heading were to aid the GPS in urban canyons
>>> and in the presence of SA. That all went away as the constellation
>>> became full and SA was removed.
>>>
>>> Certainly the new combinations of sensors will do better.
>>
>> I really wish GPS firmware engineers tried a bit harder to calibrate the
>> magnetic compass. I have 1 Garmin GPS and 2 android phones with
>> magnetic sensors. Both are mostly worthless in the car due to the car's
>> own magnetic fields superimposing its own field on the earth's
>> background field. I've never understood why they couldn't calibrate out
>> that field by simply taking the observed heading from the GPS whenever
>> the GPS indicated a reasonable velocity and slowly building up a lookup
>> table for the times when the GPS didn't indicate a high enough velocity
>> to generate a good heading.
>>
>> -wolfgang
>
> I've had similar thoughts. If motion were detected, then the
> COG could be used to correct (or override) any magnetic
> compass.

Heading is not track (COG).

--
"We demand rigidly defined areas of doubt and uncertainty."
Douglas Adams - (Could have been a GPS engineer).

On Sat, 04 Feb 2012 13:36:15 -0800, Wolfgang S. Rupprecht wrote:
>
> Alan Browne <(E-Mail Removed)> writes:
>> The same was true when automobile GPS's first came out - odometer
>> integration and magnetic heading were to aid the GPS in urban canyons
>> and in the presence of SA. That all went away as the constellation
>> became full and SA was removed.
>>
>> Certainly the new combinations of sensors will do better.
>
> I really wish GPS firmware engineers tried a bit harder to calibrate the
> magnetic compass. I have 1 Garmin GPS and 2 android phones with
> magnetic sensors. Both are mostly worthless in the car due to the car's
> own magnetic fields superimposing its own field on the earth's
> background field. I've never understood why they couldn't calibrate out
> that field by simply taking the observed heading from the GPS whenever
> the GPS indicated a reasonable velocity and slowly building up a lookup
> table for the times when the GPS didn't indicate a high enough velocity
> to generate a good heading.

Magnetic deviation is a tricky thing: it changes depending on the
heading of the device, which may not be the same as the heading of the
car (Alan's "heading is not track" comment elsewhere), it's going to
vary by heading in general (it's a different mix of forces when you're
facing 0° versus 10° versus 20°), and shifting it by as little as an
inch or two in a metal-heavy vehicle like a car will result in vastly
different readings, often by as much as 5 or 10°. Good map data and good
track with reasonable assumptions regarding road lock are going to give
you MUCH better information about direction than a magnetic compass
inside a box with energized antennae, someplace inside a much larger,
irregularly shaped highly ferric box. Not generally worth the trouble
when track will give so much better results for effort that's already
been undertaken ANYWAY, when results without track don't have much of a
real use for better than gross answers anyway.

--
94. When arresting prisoners, my guards will not allow them to stop and
grab a useless trinket of purely sentimental value.
--Peter Anspach's list of things to do as an Evil Overlord

Alan Browne <(E-Mail Removed)> writes:
> Heading is not track (COG).

Having learned to drive in ice and snow in the Boston area, I can
appreciate that heading isn't track. To be honest, during the times
when heading != track, I was much too busy to look down at anything
inside the car and in any case I would have gladly forgiven the GPS for
displaying an erronious heading. ;-)

-wolfgang
--
g+: https://plus.google.com/114566345864337108516/about

"Peter H. Coffin" <(E-Mail Removed)> writes:
> Magnetic deviation is a tricky thing: it changes depending on the
> heading of the device, which may not be the same as the heading of the
> car (Alan's "heading is not track" comment elsewhere), it's going to
> vary by heading in general (it's a different mix of forces when you're
> facing 0° versus 10° versus 20°), and shifting it by as little as an
> inch or two in a metal-heavy vehicle like a car will result in vastly
> different readings, often by as much as 5 or 10°.

That's why I suggested a lookup table, just in case the system was
non-linear.

On the other hand, I'm hard pressed to see how it could be anything but
a linear time-invariant system that one can use superposition on.

> Good map data and good track with reasonable assumptions regarding
> road lock are going to give you MUCH better information about
> direction than a magnetic compass inside a box with energized
> antennae, someplace inside a much larger, irregularly shaped highly
> ferric box. Not generally worth the trouble when track will give so
> much better results for effort that's already been undertaken ANYWAY,
> when results without track don't have much of a real use for better
> than gross answers anyway.

Well, the grotty part now is that every time I come to a stop in the car
the map spins after the stop. A smarter system would record the track
and the magnetic heading, see that the track velocity dropped to zero
but the magnetic heading didn't change. It could then determine that
the heading must not have changed.

-wolfgang
--
g+: https://plus.google.com/114566345864337108516/about

On 2012-02-04 22:37 , Peter H. Coffin wrote:
> On Sat, 04 Feb 2012 13:36:15 -0800, Wolfgang S. Rupprecht wrote:
>>
>> Alan Browne<(E-Mail Removed)> writes:
>>> The same was true when automobile GPS's first came out - odometer
>>> integration and magnetic heading were to aid the GPS in urban canyons
>>> and in the presence of SA. That all went away as the constellation
>>> became full and SA was removed.
>>>
>>> Certainly the new combinations of sensors will do better.
>>
>> I really wish GPS firmware engineers tried a bit harder to calibrate the
>> magnetic compass. I have 1 Garmin GPS and 2 android phones with
>> magnetic sensors. Both are mostly worthless in the car due to the car's
>> own magnetic fields superimposing its own field on the earth's
>> background field. I've never understood why they couldn't calibrate out
>> that field by simply taking the observed heading from the GPS whenever
>> the GPS indicated a reasonable velocity and slowly building up a lookup
>> table for the times when the GPS didn't indicate a high enough velocity
>> to generate a good heading.
>
> Magnetic deviation is a tricky thing: it changes depending on the
> heading of the device, which may not be the same as the heading of the
> car (Alan's "heading is not track" comment elsewhere), it's going to
> vary by heading in general (it's a different mix of forces when you're
> facing 0° versus 10° versus 20°), and shifting it by as little as an
> inch or two in a metal-heavy vehicle like a car will result in vastly
> different readings, often by as much as 5 or 10°. Good map data and good
> track with reasonable assumptions regarding road lock are going to give
> you MUCH better information about direction than a magnetic compass
> inside a box with energized antennae, someplace inside a much larger,
> irregularly shaped highly ferric box. Not generally worth the trouble
> when track will give so much better results for effort that's already
> been undertaken ANYWAY, when results without track don't have much of a
> real use for better than gross answers anyway.

I was really on about the notion that unless the heading system frame of
reference is tied to where the nose of the vehicle is pointing, then you
can't use the track vector to correct the heading system. In a car,
with most tracks aligned with heading it may not be such an issue.

In a "good" (aircraft) heading system, the mag heading error is
calibrated to less than 2.5° at all headings. The error function looks
like a sine over 360° with an amplitude of 2.5° (eg: amplitude=error for
that heading reading). It's actually 2nd order but grossly a sine. On
one nav system program we instituted a Kalman Filter that uses Doppler
velocity sensor inputs and GPS to real time calibrate that error out.
Thence, if the GPS was not usable, the Doppler nav would be more
accurate when using the magnetic heading disciplined gyro heading. It
took a few flight tests to get working correctly...

--
"We demand rigidly defined areas of doubt and uncertainty."
Douglas Adams - (Could have been a GPS engineer).