Re: [Paparazzi-devel] Navigation schema ?

[Famille Chateauvieux-Hellwig]

        Ok, I think I've got it now.
        That can really help me. The Navigation algorithm that I used until now 
is 
only based on the angle between heading and bearing (btw. Heading is the 
Course Over Ground µ-blox message). The Problem is that the angles are small 
when I'm far from the target, and become larger when I come closer. The 
result is that the plane is doing sharp turns when reaching the Target. This 
is a big problem as I am only taking high resolution fixed pictures on the 
way point (Surveying).  

Thanks a lot guys. I'll report the results if I succeed.
Martial


Le lundi 16 juin 2008 12:32, address@hidden a écrit :
> Famille,
>
> It is important to note that we do not have any "heading" function
> whatsoever.  The navigation control operates only on bearing and uses one
> of two schemes:
>
> In a simple "goto wp" the carrot jumps to the wp and the plane just tries
> to turn toward that point.  The result is some random path that eventually
> reaches the wp (of course it's theoretically possible to end up circling
> the wp many times before reaching it).  This is the navigation scheme used
> between non-adjacent blocks, when Auto2 is first initiated, or when an
> active waypoint is moved.
>
> For a "route" or circle, the carrot exactly follows the line, some fixed
> distance ahead of the plane, and the plane always tries to turn toward the
> carrot.  This gives a very accurate path as the course is tightly
> controlled.
>
> -- Famille Chateauvieux-Hellwig <address@hidden> wrote:
> >  What do you call a real route ?
>
>       Assume the plane flying north (heading 0°). The waypoint you target is
> currently 200m east from the current position (bearing 90°). My PID loop
> has 90 - 0 as input and acts on the rudder. the plane will start flying
> more to the right. In the next cycle, I'll have may be heading = 5, 
> bearing=92, but that's not something I can really compute in advance. I
> just let the PID loop regulate it.
>
> >  It does not make a difference only if you are currently on your desired
> > track. Usually you don't want only to go somewhere, you want to follow a
> > track all the way until the waypoint is reached.
> >
> >  You can compute a carrot position as soon as you are able to project the
> > current position of your vehicle on the route you want to follow. Easy
> > with straight lines and circles ...
> >
> > --Pascal
>
>       Ach so !
>       Hmmm. That's not always easy.
>       Assuming you're close enough to the desired route, and the difference
> between heading and bearing is small, you can place the carrot at the
> intersection between the desired route and the "Carrot circle", carrot
> seconds around the plane. Correct ?
> Figure 1
>
>       Now If you're far from the route or your current heading is way off the
> bearing to the carrot, you have to change the strategy, and compute a more
> complex desired route (ex: figure 2) ? This will typically be the case just
> when you switch the autopilot on. Can you do this on the AVR ?
>
> Is this the idea ?
>
> Thanks Pascal
> Martial
>
> ____________________________________________________________
> Beauty Product Reviews
> Read Unbiased Beauty Product Reviews and Join Our Product Review Team!
> http://thirdpartyoffers.juno.com/TGL2141/fc/JKFkuJi7UvsgxRXY2T2Fh6fXhKk3E7s
>PXPdd8QXJFu48ZahC8kOZgw/