OpenCV--Is NAVIO+ the right board?

Navio / Edge Flight stack
1

Hello All,

I am starting a rover project that will require the use of OpenCV libraries to do object recognition. At first glance, the NAVIO+ seems to be an easy way to handle the navigational aspects of the project, i.e. telemetry, servo control, and other guidance related issues…(“easier” than piecing the components together from scratch).

But the forum replies on the topic seem to suggest that there may not be enough computational power left on my RPi2 after navigational issues are handled.

I imagined that the Navio+ would do the work of the nav controls, and my RPi would go along for the ride, occasionally interjecting directional commands…and that the GUI aspect of my RPi would remain intact.

Can someone help me understand if the NAVIO+ is the way to go, or am I better off looking for a more self-sufficient nav control solution…Arduino based…etc…

Again, I do not have in depth knowledge, about NAVIO+ at this point so I may very well be missing some of the higher level concepts and details involved.

Thanks in advance for your time.

2

I’m working on a project that requires OpenCV and object recognition for navigation. I’m installing and configuring openCV right now and will test several Python codes I’ve written soon.

I’ve decided to go ahead and use RPI II and NAVIO+ with OpenCV and optimize my code if I get performance degradation. Another option is to to setup a server instead of native processing.
There’s no doubt OpenCV and image processing tasks are demanding.

I recommend to see the comparison made by initial-state on the difference between RPI II and Beagle Bone Black to get some statistics on the performance of the 900Mhz quad core the RPi has.

3

What you want for CV is high memory throughput. The sensors on the navio should be fine.Arduino has a factor of around 62.5 less CPU power (or ram throughput) vs PI 2, If you want something extendable then either works, depends on how you intend to extend. I have a DE0 nano fpga available for example not sure if I can create a fast enough comms between them for real time video.