[OpenRelief Developer] [OpenCV] Requesting an assessment of BeagleBoard DSP from Daniel Firth

[Andrew Back]

Hi Shane,

On 19 March 2012 08:24, Shane Coughlan <shane at openrelief.org> wrote:
> Hi Daniel!
>
> I wanted to check in with a brief question. In a discussion with Karl and Dan recently it became apparent that things like having our UAV recognize roads may be relatively computationally intense. Even our existing smoke code will use a fair bit of power. That means if we use a light computer like the BeagleBoard, we will need to use the DSP processing ability. Now, you are a math ninja, so I was wondering if you could help assess whether the DSP on the BeagleBoard is powerful enough to do that, and whether you could assist with some prototyping down the line.
>
> Just for reference, here is an overview of the BeagleBoard xM:
> http://beagleboard.org/hardware-xM

The somewhat beefier PandaBoard, which costs around the same and uses
a faster dual core OMAP4, may also be worth a look:

http://pandaboard.org/

Although it's not clear to me whether the DSP resources available will
be as flexible as those present in the OMAP3 used on the BeagleBoard.
This is getting outside my area of comfort... Though this resource may
be of assistance in determining this.

http://omappedia.org/wiki/Ducati_For_Dummies

Cheers,

Andrew

-- 
Andrew Back
http://carrierdetect.com