[FOSS-GPS] Short moving baseline implementation

[Danny Miller]

I'm confused.  You've got LSA two meters apart??  Was that what you 
meant to say?

This is relevant.  If you're tens of meters apart, you would likely get 
acceptable- if not superior- performance with plain GPS with 
high-performance units.

http://www.nvs-gnss.com/products/receivers/item/2-nv08c-csm.html

1M accuracy.  Almost all the time, better than that.  But it's also 
capable of RAW outputs.

RTKLib takes time to resolve carrier phase and can sometimes hop around 
when a signal problem confuses the carrier phase.  The simple onboard 
non-RTK solution is quicker and, while less accurate, may be more reliable.

If this is a question of "I'd like to autopilot two manned LSAs in very 
close proximity to one another for 2 hours", no, there's just no way 
that's acceptable.  It'll be too dangerous.  RTKLib is an alpha/beta-ish 
product and not suitable for life-critical function.

If you get two units and antenna and play with 'em, you'll see that the 
reading can jump around at times on its own.  If you put your hand over 
the antenna, it may jump 10M to the left.  Maybe it'll drop out of FIX 
and maybe not.  Fearing your unmanned rover will drive into a swimming 
pool due to error is one thing.  LSA is another.

I also wondered if STM32F4 with DSP would do this.  The thing is, that's 
not float-point DSP.  AFAIK it cannot be used to calculate RTKLib's 
matrices as they are done.

It MIGHT be possible to rework RTKLib with 32-bit integer math for the 
high-throughput matrix calcs, but you'd have to understand the math's 
possible ranges really well.  I decided it was both beyond my skillset, 
AND too much work even if I had that skillset down. Now that the RPI and 
BBB are so prevalent and a lot of other high-hp stuff is waiting to 
replace them, going to such lengths to support a much lower-power core 
does not seem that important.

Danny

On 11/4/2014 2:32 PM, MarekM wrote:
> Hi Danny,
> Thanks for reply.
> This project is not for UAV's, its for small ultralight aircrafts (LSA).
> In first stage i've used Trimble BD982 with two antennas, but cost of it is
> too high.
> I have IMU with magnetometr/compass - only problem is long time stability -
> mission time is more than 4 your, and there are lot of drift.
> BD982 is also based on ARM procesor and give 50 Hz refresh rate so... (Few
> days ago there was adds of one US company with small board with two uBlox
> modules, RTK and heading).
> I will try to use Raspberry Pi ( have one), but STM32F4 will be interesting
> (maybye DSP functions can be usefull).
> Expexted accuracy with 2 meter baseline is 0.5 - 0.2 deg ( with speed approx
> 80 -120 mph).
>
>
>
>
> --
> View this message in context: http://open-source-gps-related-discussion-and-support.1099874.n2.nabble.com/Short-moving-baseline-implementation-tp7573069p7573071.html
> Sent from the Open Source GPS-related discussion and support mailing list archive at Nabble.com.
> _______________________________________________
> This message is sent to you from FOSS-GPS at lists.osgeo.org mailing list.
> Visit http://lists.osgeo.org/mailman/listinfo/foss-gps to manage your subscription
> For more information, check http://wiki.osgeo.org/wiki/FOSS-GPS
>