Re: [Flightgear-devel] Positioning Aircraft: accuracy
If it helps, when building Flightgear, I get compile warnings saying
that there's a narrowing conversion from double to float in some YASim
methods:
|flightgear/src/FDM/YASim/YASim.cxx:261:30: warning: narrowing
conversion of
‘((YASim*)this)->YASim::.FGInterface::get_uBody()’ from
‘double’ to ‘float’ inside { } is ill-formed in C++11 [-Wnarrowing]||
||flightgear/src/FDM/YASim/YASim.cxx:261:43: warning: narrowing
conversion of
‘((YASim*)this)->YASim::.FGInterface::get_vBody()’ from
‘double’ to ‘float’ inside { } is ill-formed in C++11 [-Wnarrowing]||
||flightgear/src/FDM/YASim/YASim.cxx:261:56: warning: narrowing
conversion of
‘((YASim*)this)->YASim::.FGInterface::get_wBody()’ from
‘double’ to ‘float’ inside { } is ill-formed in C++11 [-Wnarrowing]|
On Sun 23 Jun 2013 06:09:32 PM CDT, Curtis Olson wrote:
>
> The hang glider is using the UIUC fdm, right? Perhaps that is using a
> float (6-7 digits of precision) to represent it's initial conditions
> ... or somewhere in the startup pipeline or the flightgear interface
> to the uiuc model, the initial lon/lat is stored as a float?
>
>
> On Sun, Jun 23, 2013 at 5:27 PM, D-NXKT > wrote:
>
> Hmm, everything is fine if I use a different plane. Seems to be a hang
> glider specific problem.
> Monitoring the latitude and longitude variabel from the very beginning
> shows, that both variables have at first indeed the correct
> values. Then
> the latitude value suddenly jumps from 47.306263 to 47.306206.
>
> Sorry for the noise.
>
> Best regards
> D-NXKT
>
>
>
> --
> This SF.net email is sponsored by Windows:
>
> Build for Windows Store.
>
> http://p.sf.net/sfu/windows-dev2dev
> ___
> Flightgear-devel mailing list
> Flightgear-devel@lists.sourceforge.net
>
> https://lists.sourceforge.net/lists/listinfo/flightgear-devel
>
>
>
>
> --
> Curtis Olson:
> http://www.atiak.com
Re: [Flightgear-devel] Positioning Aircraft: accuracy
A few years ago, I helped squash a bug that occurred when reading in data with generic IO (I think that my patch was included in the code). At the time, it was reading in just a float. So, when using it to read in position, the doubles for lat/long were being cut (rounded or truncated?) and it meant that during playback of flight the aircraft would jump around a little bit (on the order of meters), even when there was actually smooth position data in the input file. The look was like a bad frame rate, because it would jump from position to position. So, I concur that it sounds a lot like just a float is being used for initial position, when a double is necessary. -- Adam From: Curtis Olson mailto:curtol...@flightgear.org>> Reply-To: FlightGear developers discussions mailto:flightgear-devel@lists.sourceforge.net>> Date: Tuesday, June 25, 2013 12:05 PM To: FlightGear developers discussions mailto:flightgear-devel@lists.sourceforge.net>> Subject: Re: [Flightgear-devel] Positioning Aircraft: accuracy It really smells like a double -> float conversion somewhere in the pipeline (more than a geocentric / geodetic conversion or something like that). Curt. On Tue, Jun 25, 2013 at 2:01 PM, D-NXKT mailto:d_n...@yahoo.de>> wrote: > How far off is the aircraft placement - how much is the "jump"? Is it > many hundreds of meters, or on the order a meter or two, or just > centimeters? > > Jon Several meters! Or more precise: first: lat= 47.306263 lon= 11.379070 "jump" to: lat= 47.3062060567 lon= 11.3790703145 ^ Difference = 6.34 meters (20.80 ft) Best regards D-NXKT -- This SF.net email is sponsored by Windows: Build for Windows Store. http://p.sf.net/sfu/windows-dev2dev ___ Flightgear-devel mailing list Flightgear-devel@lists.sourceforge.net<mailto:Flightgear-devel@lists.sourceforge.net> https://lists.sourceforge.net/lists/listinfo/flightgear-devel -- Curtis Olson: http://www.atiak.com<http://www.atiak.com/> - http://aem.umn.edu/~uav/ http://www.flightgear.org<http://www.flightgear.org/> - http://gallinazo.flightgear.org<http://gallinazo.flightgear.org/> -- This SF.net email is sponsored by Windows: Build for Windows Store. http://p.sf.net/sfu/windows-dev2dev___ Flightgear-devel mailing list Flightgear-devel@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/flightgear-devel
Re: [Flightgear-devel] Positioning Aircraft: accuracy
It really smells like a double -> float conversion somewhere in the pipeline (more than a geocentric / geodetic conversion or something like that). Curt. On Tue, Jun 25, 2013 at 2:01 PM, D-NXKT wrote: > > How far off is the aircraft placement - how much is the "jump"? Is it > > many hundreds of meters, or on the order a meter or two, or just > > centimeters? > > > > Jon > > > Several meters! > > Or more precise: > > first: lat= 47.306263 lon= 11.379070 > "jump" to: lat= 47.3062060567 lon= 11.3790703145 >^ > Difference = 6.34 meters (20.80 ft) > > Best regards > D-NXKT > > > > -- > This SF.net email is sponsored by Windows: > > Build for Windows Store. > > http://p.sf.net/sfu/windows-dev2dev > ___ > Flightgear-devel mailing list > Flightgear-devel@lists.sourceforge.net > https://lists.sourceforge.net/lists/listinfo/flightgear-devel > -- Curtis Olson: http://www.atiak.com - http://aem.umn.edu/~uav/ http://www.flightgear.org - http://gallinazo.flightgear.org -- This SF.net email is sponsored by Windows: Build for Windows Store. http://p.sf.net/sfu/windows-dev2dev___ Flightgear-devel mailing list Flightgear-devel@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/flightgear-devel
Re: [Flightgear-devel] Positioning Aircraft: accuracy
> How far off is the aircraft placement - how much is the "jump"? Is it > many hundreds of meters, or on the order a meter or two, or just > centimeters? > > Jon Several meters! Or more precise: first: lat= 47.306263 lon= 11.379070 "jump" to: lat= 47.3062060567 lon= 11.3790703145 ^ Difference = 6.34 meters (20.80 ft) Best regards D-NXKT -- This SF.net email is sponsored by Windows: Build for Windows Store. http://p.sf.net/sfu/windows-dev2dev ___ Flightgear-devel mailing list Flightgear-devel@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/flightgear-devel
Re: [Flightgear-devel] Positioning Aircraft: accuracy
> It seems that we have a problem with the FG/JSBSim-interface or it's a > rounding error during JSBSim-initialisation?! > > Summarizing my observations: > - immediately after FG is started the property browser shows the > correct values > - this values hold true as long as the splash-screen appears > - as soon as the splash-screen disappears and the plane pops up, the -- > lon/lat values jump to the wrong (rounded?) values > - I also noticed that the "correct" values in the property browser are > displayed with the same numbers of digits as typed. Simultaneous with > the jump to the wrong values the number of digits increases to their > count for "double" values > > Maybe this observations help to debug this issue. > > Best regards > D-NXKT How far off is the aircraft placement - how much is the "jump"? Is it many hundreds of meters, or on the order a meter or two, or just centimeters? Jon -- This SF.net email is sponsored by Windows: Build for Windows Store. http://p.sf.net/sfu/windows-dev2dev ___ Flightgear-devel mailing list Flightgear-devel@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/flightgear-devel
Re: [Flightgear-devel] Positioning Aircraft: accuracy
> The hang glider is using the UIUC fdm, right? Perhaps that is using a > float (6-7 digits of precision) to represent it's initial > conditions ... or somewhere in the startup pipeline or the flightgear > interface to the uiuc > model, the initial lon/lat is stored as a float? Curt, you pointed me in the right direction! But it's the other way around. The UIUC-Version of the hang glider works perfectly. The JSBSim-Version is wrong. Further investigations with several planes show that only (but all) JSBSim-aircrafts are affected! I've tested the ASW20-JSBSim, C172p and Rascal110-JSBSim. The YASim- and UIUC-aircrafts work fine (ASK13, bo105, Rascal110-YASim)! It seems that we have a problem with the FG/JSBSim-interface or it's a rounding error during JSBSim-initialisation?! Summarizing my observations: - immediately after FG is started the property browser shows the correct values - this values hold true as long as the splash-screen appears - as soon as the splash-screen disappears and the plane pops up, the --lon/lat values jump to the wrong (rounded?) values - I also noticed that the "correct" values in the property browser are displayed with the same numbers of digits as typed. Simultaneous with the jump to the wrong values the number of digits increases to their count for "double" values Maybe this observations help to debug this issue. Best regards D-NXKT -- This SF.net email is sponsored by Windows: Build for Windows Store. http://p.sf.net/sfu/windows-dev2dev ___ Flightgear-devel mailing list Flightgear-devel@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/flightgear-devel
Re: [Flightgear-devel] Positioning Aircraft: accuracy
The hang glider is using the UIUC fdm, right? Perhaps that is using a float (6-7 digits of precision) to represent it's initial conditions ... or somewhere in the startup pipeline or the flightgear interface to the uiuc model, the initial lon/lat is stored as a float? On Sun, Jun 23, 2013 at 5:27 PM, D-NXKT wrote: > Hmm, everything is fine if I use a different plane. Seems to be a hang > glider specific problem. > Monitoring the latitude and longitude variabel from the very beginning > shows, that both variables have at first indeed the correct values. Then > the latitude value suddenly jumps from 47.306263 to 47.306206. > > Sorry for the noise. > > Best regards > D-NXKT > > > > > -- > This SF.net email is sponsored by Windows: > > Build for Windows Store. > > http://p.sf.net/sfu/windows-dev2dev > ___ > Flightgear-devel mailing list > Flightgear-devel@lists.sourceforge.net > https://lists.sourceforge.net/lists/listinfo/flightgear-devel > -- Curtis Olson: http://www.atiak.com - http://aem.umn.edu/~uav/ http://www.flightgear.org - http://gallinazo.flightgear.org -- This SF.net email is sponsored by Windows: Build for Windows Store. http://p.sf.net/sfu/windows-dev2dev___ Flightgear-devel mailing list Flightgear-devel@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/flightgear-devel
Re: [Flightgear-devel] Positioning Aircraft: accuracy
Hmm, everything is fine if I use a different plane. Seems to be a hang glider specific problem. Monitoring the latitude and longitude variabel from the very beginning shows, that both variables have at first indeed the correct values. Then the latitude value suddenly jumps from 47.306263 to 47.306206. Sorry for the noise. Best regards D-NXKT -- This SF.net email is sponsored by Windows: Build for Windows Store. http://p.sf.net/sfu/windows-dev2dev ___ Flightgear-devel mailing list Flightgear-devel@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/flightgear-devel
Re: [Flightgear-devel] Positioning Aircraft: accuracy
Hi Tom, > How have you checked these values? I've just tried it myself and in > the property browser exactly the same values appear. I started FG with --lon=11.379070 --lat=47.306263 --heading=155 --units-meters --on-ground and found me at --lon=11.3791 --lat=47.3063. Manual input in the property browser moved me to the correct position. > Maybe you are using an operating system with a low precision atof? In > the property browser strtod is used to convert the number whereas for > the dialog and arguments atof is used. I'm using Suse 12.3 64bit. How can I check the atof precision? Thanks for the quick response D-NXKT -- This SF.net email is sponsored by Windows: Build for Windows Store. http://p.sf.net/sfu/windows-dev2dev ___ Flightgear-devel mailing list Flightgear-devel@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/flightgear-devel
Re: [Flightgear-devel] Positioning Aircraft: accuracy
Hi, Am 2013-06-23 22:27, schrieb D-NXKT: > I noticed that positioning an aircraft at FG startup with --lon= and > lat= has only a limited accuracy. It seems that only the first 7 > digits are read. This leads to an uncertainty of several > meters/feet! For example lon=11.379070 --lat=47.306263 is needed to > place our hang glider directly in front of the ramp (LOWI-Scenery). > Unfortunately the hang glider pops up several meters right of the > ramp. I found out that FG internally uses this (rounded) values: > --lon=11.3791 --lat=47.3063. How have you checked these values? I've just tried it myself and in the property browser exactly the same values appear. > Same issue with input via the Menu "Location/Position Aircraft In > Air". Here by default values are shown with four decimal places, but also values with higher precision are accepted. > Only positioning via the property browser leads to exact results. Maybe you are using an operating system with a low precision atof? In the property browser strtod is used to convert the number whereas for the dialog and arguments atof is used. Tom -- Thomas Geymayer www.tomprogs.at / C-Forum und Tutorial: www.proggen.org Student of Computer Science @ Graz University of Technology --- Austria -- This SF.net email is sponsored by Windows: Build for Windows Store. http://p.sf.net/sfu/windows-dev2dev ___ Flightgear-devel mailing list Flightgear-devel@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/flightgear-devel
[Flightgear-devel] Positioning Aircraft: accuracy
Hello, I noticed that positioning an aircraft at FG startup with --lon= and lat= has only a limited accuracy. It seems that only the first 7 digits are read. This leads to an uncertainty of several meters/feet! For example lon=11.379070 --lat=47.306263 is needed to place our hang glider directly in front of the ramp (LOWI-Scenery). Unfortunately the hang glider pops up several meters right of the ramp. I found out that FG internally uses this (rounded) values: --lon=11.3791 --lat=47.3063. Same issue with input via the Menu "Location/Position Aircraft In Air". Only positioning via the property browser leads to exact results. Maybe someone could fix this at the next opportunity?! Thanks in advance D-NXKT -- This SF.net email is sponsored by Windows: Build for Windows Store. http://p.sf.net/sfu/windows-dev2dev ___ Flightgear-devel mailing list Flightgear-devel@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/flightgear-devel
