Navio2 on NVIDIA Jetson Nano

Hi,
I am not sure where to ask this question but I am trying to get the navio2 software to run on a Jetson Nano (which has a Pi compatible interface). I have managed to compile most of the navio2 / ardupilot code but I am struggling a bit with emlidtools as I am having import version clashes from modules like humanfriendly. I am wondering which version of python is required for emlidtools to work and whether anyone has tried doing what I am attempting or if there is a suggested roadmap on how to get it to run on new hardware.

At the moment I only know enough to be dangerous so pardon my ignorance.

  • Charles

Hi Charles,

We haven’t tried to accomplish this so I’m afraid I can’t be of much help.

At the moment, Navio2 can be used with Raspberry Pi 2, 3 and 4 only. You can check this guide for more information about Raspberry Pi configuration with Navio2.

Hello Polina,

No problem I was really just asking in case someone else had done
it already. I have a PI3 so I can compare installations if I get stuck.

So far I have installed ROS on the Nano and compiled ardupilot with
navio2 support. I have also built the emlidtool without errors but my
problem only occurs when I run it. I think I have an idea on how to fix that.

The Nano is only a stop gap as I am waiting for the Jetson Xavier NX
to be released in March it is the same DIMM form factor as the Nano but
is (as) or more powerful than the TX2 used in the skydio 2. It is what NVIDIA is
pushing for autonomous drones. Obviously I don’t have the experience
of Skydio so my project will be rather lame compared to them but it will keep
me busy during the dark winter nights.

  • Charles

I have come to the conclusion a better approach will be to use a PI4
with the Emlid software and a Google Coral USB device for the
inference. https://coral.ai/products/accelerator/
So that is the approach I intend to take.

  • Charles
2 Likes

Charles,
You may also want to check out googles vision bonnet. I no longer have the citation but recently read it’s faster than the coral. On amazon it’s also ~20$ less. It’s sold as part of “googles vision kit”. Also has the benefit of being a pi hat form factor.

EDIT: While looking for the citation. For anyone coming along later, I found that they were referring to the vision bonnet vs Intel NCS not the vision bonnet vs the Google Coral which is indeed faster than both of them (probably combined :slight_smile: ).

1 Like