Blogs

Moderator

An Idea for ArduPilot-based Differential GPS

Posted by Mark Harrison on May 11, 2011 at 12:35pm
Here's an idle thought I had that might make help ArduCopter 2.0 loitering precision:
  • Hook a GPS to GCS.
  • GCS calculates GPS difference.
  • GCS transmits GPS correction over xbee MAVLINK connection
  • ArduCopter Stabilize mode gets very fine resolution.

 

Details here: http://eastbay-rc.blogspot.com/2011/05/idea-ardupilot-differential-gps.html

 

DifferentialGPSPos.jpg

DGPS background: http://www.ngs.noaa.gov/CORS

E-mail me when people leave their comments –
Follow

You need to be a member of diydrones to add comments!

Join diydrones

Comments

  • Pieter van Woerkom August 5, 2011 at 6:08am

    What about WAAS ? Mediatek is supposed to support WAAS so no need for DGPS it think.

     

  • Moderator
    Mark Harrison May 15, 2011 at 12:47am
    Thanks all, I've learned a lot.  Note: A few people left comments at the original blog, they were unhappily eaten when blogger went down this week.
  • Developer
    John Arne Birkeland May 12, 2011 at 5:57am

    If the goal is to get a steady position hold/loiter on the cheap (not using commercial GCS equipment), this could work if we do some assumptions. But it would not be GCS since the corrected GPS coordinates would not necessary be "correct" ones. They would only be corrected in reference to a fixed stationary position with an unknown GPS error.

    For this to work we would have to assume that GPS coordinates reported by a unit are correct, and that the only limiting factor is the random signal error generated by the GPS satellites.

    We also have to assume that two identical GPS units (A and B) at the same location, will produce identical GPS coordinates.

    If this is the case, then by placing GPS A in a fixed stationary position. You would know that any variation in the GPS position are the result of a GPS signal random error. So if you transmit the GPS A (stationary unit) error variation to GPS B (mobile unit) and subtract it, GPS B should reflect a true position without random error.

    Will it work? Frankly I have no clue, since the are a lot of assumptions that has to be true.

    If not, the alternative (true GCS) will require specialized/expensive hardware.

  • Clark Hymas May 12, 2011 at 4:15am

    This is what is used in our AutoFarm tractor guidance systems.  Another unique feature is they use 2 GPS receiver antenas on the tractor mounted about 3 feet apart.  By using 2 antenas the system knows how much of a roll there is left to right and what dirrection the tractor is facing without having any movement.  This provides us with RTK sub inch accuracy.

    The board inside the tractor has 3 "GPS clocks" one for each tractor antenna and one for the base station packets.  I am guessing, but to do something similar would require several more componants added to this system.

     

  • simonl May 12, 2011 at 12:38am
    My experience of a EM406 that's stationary is that the position moves around a LOT. I guess if you average the lat/lon over time you can better estimate the actual position , but I'm not seeing how that improves much? Sounds an interesting idea though.
  • me May 11, 2011 at 7:39pm
    Thales AC12 does carrier phase. I've post processed to within a foot of an NGS benchmark. Lots of hurdles but possible.
  • DaveyWaveyBunsenBurner May 11, 2011 at 3:45pm
    And then you plot a waypoint in GCS with a 30m radius :)
  • Russell B. Sutton May 11, 2011 at 3:44pm
    This is a very good concept, been thinking about it myself.  Also through in a Baro sensor with that GPS and CSG and get way better ALT. readings as well.
  • Agustin Monteverdi May 11, 2011 at 3:35pm

    why... for the luzz...

    i once read about it  seems a good idea complex buts its a good idea..

     

  • hamish May 11, 2011 at 2:14pm

    sooo from what i gathered off poor man GPS , If you could get both recievers to lock onto the same satelites then you're game?

     

    Neglecting ionosphere interference and assuming identical recievers

This reply was deleted.