DPRG List  

[DPRG] Expected accuracy of heading

Subject: [DPRG] Expected accuracy of heading
From: John Swindle swindle at compuserve.com
Date: Tue Apr 30 11:40:40 CDT 2013


One of the design criteria for the audio localizer was that it be 
robust in real environments. Folks complained that the sun coming 
in the windows interfered with IR, that the rebar in the floor 
interfered with compasses, that the room lighting messed up the 
camera's colors, etc. None of that seemed reasonable, not robust. 
So, I made the localizer work even if a kid is screaming at it. 
No bellyaching.

If I were doing triangulation, the headings would be more 
accurate. The problem with triangulation is that it is terrible 
for localization. To get an inch of location accuracy at 50 feet, 
you'd need better than 0.1 degree accuracy in the thing that was 
doing the pointing (actually, better than one third of that). 
LORAN and GPS don't do triangulation. They do multilaterization, 
which gives excellent positional results, but is poor for giving 
direction. I'm trying to improve the direction info without 
adding much, and without compromising the location info.

The localizer is running on a PC, but there's no fancy 
processing, so I think it could be done on $2 processors.

Thanks for the example. It sounds like fast craft need more 
stable direction info than I know how to provide in a small 
package. I still wonder if 5 degree accuracy would be useful for 
a slower, ground-based robot, knowing that there wouldn't be any 
drift or interference.

Good luck with your flying craft.

John Swindle

More information about the DPRG mailing list