Positioning delay?

Help
2012-08-01
2013-05-29
  • john3voltas
    john3voltas
    2012-08-01

    Hi guys.
    Yesterday I went out on a read trip by car using one of my phones+bt antenna (Nokia E51).
    I used Gpsmid 0.7.7+Portuguese map that's available on the site.
    It worked fine (as usual) but I found something strange: looking at the map, the cursor would be ~2 seconds late regarding my actual position.
    Suppose I was approaching a 90º corner.
    On the map I would be still approaching the corner while in reality I would have already gone past the corner.
    This happened everywhere that I tested.
    Plus: if I stopped for a moment (for instance traffic lights) the map/cursor would hit the exact spot where I was stopped.
    Which means that the bt antenna seems to be quite accurate.
    Is this ~2 seconds delay normal? I don't know if it's normal or not because I don't have other positioning software to compare.
    Cheers

     
  • Libor Striz
    Libor Striz
    2012-08-01

    My experience wit GPS ( OVI maps, TrekBuddy, GPSMid on Nokia E52 with internal GPS ) is
    there is always some short positional delay during navigation, at least 1s, maybe more.

    Major part of it seem due time "quantization", I guess GPS is read or provides data every seconds.

    Some extra delay can be caused by processing, and / or if there is some error with actual GPS reading ( street signal shadow, low signal, whatever )

    My point of view is the should be implemented some forward position estimation,
    for displayed position error to oscilate around real position.

    Developers could say more.

     
  • john3voltas
    john3voltas
    2012-08-01

    Some extra delay can be caused by processing, and / or if there is some error with actual GPS reading ( street signal shadow, low signal, whatever )

    That was not my case. I had a quite strong reading from 9 satellites with no tall buildings/trees around.
    GPS conditions were pretty close to optimal.
    Besides, as said, when I stopped moving map/cursor would hit my exact location thus proving that the GPS signal was pretty accurate.

    My point of view is the should be implemented some forward position estimation, for displayed position error to oscilate around real position.

    But not in every situation.
    While routing, I am inclined to agree with you.
    But while recording/tracing then let me disagree because I need recording/tracing to be as much accurate as possible in order to improve the OSM maps.
    If recording/tracing starts guessing the position…
    Cheers

     
  • Libor Striz
    Libor Striz
    2012-08-01

    All depends on systematic error, that occurs with respect to movement of receiver.
    My suspicion is such delay is not limited to routing, but I have not verified it.

    E.g. iff you drive a car 72 km/h and there is 1s GPS reading interval,
    in time the new position is about to get, your last known position is already 20 meters wrong.
    (not counting other reasons )

    My idea was to shift 0..20 meters error to interval -10..+10 meters.

    For 14-15 km/h on bicycle is is about 0..4 meters. versus -2..+2.

    Hm, I see now what you mean.  I have not meant the correction of position determination.

    The correction was meant to display optical position on map and related eventual navigation info.

    By other words, to display position being in the middle of the current last known position
    and predicted position at next reading.

     
  • Libor Striz
    Libor Striz
    2012-08-02

    What if GPS is implemented ( in GPSMid, but more probably in firmware ) in such a way
    that makes (weighted) averaging of last few positions ?

    It would make for filtering and stabilization of static position.
    But can implement such a noticed delay for faster movement, like in car.