[Top][All Lists]
[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
Re: [Paparazzi-devel] Final year project and uni of Adelaide
From: |
Gareth Roberts |
Subject: |
Re: [Paparazzi-devel] Final year project and uni of Adelaide |
Date: |
Fri, 30 Oct 2009 16:53:46 +0000 |
In terms of Gareth, quick 'n' dirty (wrote most of it drunk the night
before EMAV) for the balloon detection, just did colour detection and
then a centroid, then moved the balloon waypoint based on that (it was a
bit slow due to the uplink speed). I used python and opencv - the trick
is not to iterate through cvImages in python (I took apart the source
code, and every time you call an iteration, it generates a lamba),
instead use IPLImages and pyscho on the main loop, it was as fast as the
C implementation for my particular application.
For my final year project, which isn't directly related to paparazzi,
I'm using a pair of webcams, a BeagleBoard (which runs full linux) and
currently am using a wifi draft-n link to get the two video streams back
to a groundstation laptop. I only started a couple of weeks ago, and it
takes a worrying amount of time to generate a depthmap on a core 2 duo
with 4 gig of ram, so god knows how it's going to perform on a ~600MHz
arm.
Todd: were you using the processing opencv library? I looked at it, but
it was painfully slow.
If you want any code, let me know.
--G
On Fri, 2009-10-30 at 19:35 +1100, Deane Chadwick wrote:
> Hi Todd,
>
> Good job with the aircraft looks like you guys did well, pitty about
> the wind. I did the challenge last year with QUT and was d their image
> processing. Your image processing looks heaps good and am very
> interested in what camera you used and whether you tested it with the
> IR lamps? Also how did you sync the images with telemetry data?
>
> Garath,
>
> How are you working your stereo imaging? Are you using two cameras or
> subsequent images (not sure I follow the exactly what you are doing)?
> Either way how did you find it worked? I was going to implement stereo
> imaging but didnt have the time to do it.
>
> Deane
>
> On Fri, Oct 30, 2009 at 10:27 AM, Todd Sandercock
> <address@hidden> wrote:
> I think its possible to run on board of the aircraft but the
> frame rate might have to be reduced a bit. It currently runs
> on a mac and chews the 2.4 GHz core 2 duo pretty bad. It is
> written using a hacked java open cv library too so that would
> surely make it a heap slower too.
>
> Todd
>
>
>
> ______________________________________________________________
> From: Marcus Wolschon <address@hidden>
> To: address@hidden
> Sent: Fri, 30 October, 2009 2:36:08 AM
> Subject: Re: [Paparazzi-devel] Final year project and uni of
> Adelaide
>
>
> On 2009-10-29, Todd Sandercock <address@hidden>
> wrote:
> > http://www.vimeo.com/7307880
>
> > Once everything was up and running our image
> > processing software was able to pick out the target, send
> the position to
> > paparazzi over the ivy bus then instruct the aircraft to
> loiter over the
> > target for confirmation by the ground operator. The ground
> operator from
> > there has to
> > confirm or reject the target with paparazzi being told the
> correct action
> > depending on the answer.
>
> Sounds really cool!
> How much does the image-processing need?
> Any chance of running that in the drone?
>
> Marcus
>
>
> _______________________________________________
> Paparazzi-devel mailing list
> address@hidden
> http://lists.nongnu.org/mailman/listinfo/paparazzi-devel
>
>
>
> ______________________________________________________________
> Get more done like never before with Yahoo!7 Mail. Learn more.
>
> _______________________________________________
> Paparazzi-devel mailing list
> address@hidden
> http://lists.nongnu.org/mailman/listinfo/paparazzi-devel
>
>
> _______________________________________________
> Paparazzi-devel mailing list
> address@hidden
> http://lists.nongnu.org/mailman/listinfo/paparazzi-devel