Board index FlightGear Development Aircraft Autopilot and route manager

INS / navigation systems

Designing a stable autopilot is one of the hardest things. Need help?

INS / navigation systems

Postby zakalawe » Fri Oct 23, 2009 9:27 am

Now my navigation system changes are in CVS, I would like to see about making the position computation be selectable. The current code uses the 'perfect' position values, and I plan to add 'realistic' GPS values, i.e with some error introduced. If someone can propose a sensible INS reference platform model, I can add that as an option too.

If that's done, I believe creating a 'GPS' instance with name=ins should provide quite a reasonable foundation for building a 1960s or 1970s era INS device. Mostly it will be about hiding the generic features from the user interface! I've skimmed pilot's manual for some real-world INS devices, and generally the features seem very standard - add waypoints, compute time to waypoints, and so on - which the current code is doing quite happily. If someone wants to work on modelling an INS device, I'm happy to do any C++ work required to make my code sufficiently generic and configurable.
zakalawe
 
Posts: 1152
Joined: Sat Jul 19, 2008 4:48 pm
Location: Edinburgh, Scotland
Callsign: G-ZKLW
Version: next
OS: Mac

Re: INS / navigation systems

Postby jonbourg » Fri Oct 23, 2009 11:48 am

Whoa, that sounds complicated and I used to work on these things! A lot of airliners used the carousel IV INS system from Delco in the 80s-90s. I don't know what they use now, mostly the ring laser gyro type. Most planes would use two or three units together just in case one failed. One system I worked on had an electronic sextant that shot stars for triangulation with an INS unit on the bottom, it also interfaced with a doppler radar system and a GPS system. I was told you could fly around the world and return to within a 3 ft circle of where you left. Anyway keep up the good work!
jonbourg
 
Posts: 329
Joined: Tue Mar 11, 2008 12:43 am
Location: Omaha, NE
Callsign: BU2B
OS: Windows8

Re: INS / navigation systems

Postby zakalawe » Fri Oct 23, 2009 1:21 pm

I've found a site selling copies of the Carousel IV manual, will probably pick up a copy, it's always good reading to know about more of these things. It would be great to get one (or several!) of these into the Concorde, 727 and other aircraft of the era.
zakalawe
 
Posts: 1152
Joined: Sat Jul 19, 2008 4:48 pm
Location: Edinburgh, Scotland
Callsign: G-ZKLW
Version: next
OS: Mac

Re: INS / navigation systems

Postby flameout » Sat Oct 24, 2009 12:45 am

I have been working on the FlightGear INS model, but I have limited experience when it comes to mathematics (and programming.)

If you would like to work on a replacement model, go ahead. For now, I will at least fix the bugs in my current INS implementation.

As far as I am currently concerned, however, I can continue with it myself, as I do believe I have the knowledge to implement a fairly reasonable model. Additionally, I am thinking of (sometime) porting it from Nasal to C++. As for the model error, however, I do not have the experience necessary to come up with a real good model. So far, at least, I have been planning just to model a near-constant location drift, with a slowly-changing direction. This is an extreme simplification, but should work fine. Unlike your GPS code (at least, as far as I know about it,) mine effectively integrates velocity for position (and acceleration for velocity) instead of just adding an error offset to the position.

As I have said, I do not have the necessary experience to develop a real accurate INS model, and I would be welcome to an implementation from a developer such as you.

Thank you for the GPS work, by the way.
Also known as Johnathan Van Why.
flameout
 
Posts: 443
Joined: Thu Dec 25, 2008 5:43 am
Location: Oregon, USA
Callsign: MSJF
Version: GIT
OS: Gentoo

Re: INS / navigation systems

Postby MAKG » Sat Oct 24, 2009 1:34 am

I'll suggest that a random walk may be the best model here.

That MAY be what you said, provided the error is cumulative. Constant velocity, random 3D direction, added as a vector to the aircraft velocity.

FYI, getting uniformly distributed random directions in 3D can be difficult. One relatively simple way is to pick uniform values in [-1..1] for X, Y, and Z separately, throw away those values with magnitude >1 and then set the others to have magnitude=1.
MAKG
 
Posts: 1156
Joined: Sun Oct 19, 2008 6:11 pm
Location: California Central Coast

Re: INS / navigation systems

Postby flameout » Sat Oct 24, 2009 1:46 am

I would like to add (in case it wasn't clear from my earlier post) that I would be glad for someone else to work on an INS model. Frankly, I've recently been very busy (it has been weeks since I have even read the forums.)

Just wanted to clear this up.
Also known as Johnathan Van Why.
flameout
 
Posts: 443
Joined: Thu Dec 25, 2008 5:43 am
Location: Oregon, USA
Callsign: MSJF
Version: GIT
OS: Gentoo


Return to Autopilot and route manager

Who is online

Users browsing this forum: No registered users and 0 guests