[Opengc-devel] nav display status

[Damion S. <be...@cs...>]

Hi,

A quick update as to where things are with the nav display (which has 
been a thorn in my side for quite some time)...

It works! (sort of) I made a simple test gauge that only displays VORs, 
but it is in principle fairly simple to go ahead and write readers for 
the other types of objects displayed in ND. It will likely be some time 
before a feature complete ND is available, but full "moving map" 
functionality should be approximately 2-3 weeks away. I have designed 
the framework to hopefully be relatively generic, so making a 
type-specific ND will not involve rewriting any of the fundamental nav 
code.

A few specifics:

1) I decided to go with Robin Peel's nav database, which is used by 
X-Plane and Flightgear (specifically, the X-Plane version of this 
database). It includes airports, radio nav beacons, airways, and 
intersections. He follows the DAFIF cycle, making it easy to get 
up-to-date information with minimal effort.

2) Please handle the CVS repository with care - I have been making 
fairly substantial changes, and not everything is stable just yet. As 
of last night it builds on both Mac and Windows, but you will have to 
manually edit ogcNavDatabase.cpp to point it at the right nav path - it 
will be updated to respect the nav path specified in opengc.ini in the 
near future, but there is a technical problem related to file parsing 
that needs to be sorted out. There is a bug (probably pointer related) 
that causes a segfault upon exiting OpenGC, which I still have to sort 
out.

3) The problem of mapping a spherical/ellipsoidal earth onto a flat 
plane is not trivial. The current implementation does a Mercator 
projection of the lat/lon coordinates of each geographic object and 
uses these cached values along with a real-time projection of the 
aircraft lat/lon to produce a rectangular map. This works surprisingly 
well, and is quite fast since it becomes easy to apply a visibility 
constraint during rendering and only render the objects actually 
visible on the map.

I tried flying around my home airport (KPIT) at approximately 40 
degrees north, and the Mercator distortion was minimal. The main 
advantage of a Mercator projection is of course the preservation of 
absolute heading relative to true north, but the coordinate conversion 
process renders it useless as you approach the poles. I've toyed with 
the idea of doing a local real-time mercator projection (i.e. centered 
around the current aircraft location rather than the intersection of 
the equator and prime meridian) but it's not clear to me what the 
performance hit would be versus the cached implementation currently in 
use.

In any case, do any of you know how this is handled in the real world? 
There are of course other projections available, some of which I have 
the math for and some which I don't. On the scale that navigation 
occurs (as distinct from flight planning) it's mostly safe to assume 
that the world is flat, but I'd like to have a reasonably correct 
implementation.

Cheers,
-Damion-

---------
Damion Shelton
Carnegie Mellon University, Robotics Institute
A408-o Newell Simon Hall
412.268.3866 (office)
412.818.8829 (cell)
http://www.cs.cmu.edu/~beowulf
---------
If you ever crawl inside an old hollow log and go to sleep, and while 
you're in there some guys come and seal up both ends and then put it on 
a truck and take it to another city, boy, I don't know what to tell 
you.