Yocto Project

Like that, that thing that you have in your car where like on your dashboard when you plug your phone, it starts showing Google Maps and whatnot.

Uh, typically this would be a fancy MPU and big Like Yocto-based distros probably that take a lot of time to build and that br- uh, bring, uh, with them, uh, like lots of dependencies and like effectively a, a large software bill of material and whatnot.

It turns out that if you have a reasonably powerful, uh, MCU, um, just enough RAM, uh, and a display and a USB, because this happens over USB, then you can do, uh, Android Auto head unit on Zephyr, and I have it running on SD boards, Renaissance board, Espressif boards, Native SIM.

Um, so again, this one is also, I'm in the process of polishing it up.

You also have the ability for BSP skills if you need to have coding to actually tap into code level serial console, the ability to access the BSP, make changes that you need.

And then third that is upcoming is the ability for Yocto support.

Yocto and the embedded community is becoming more and more important because you can actually cherry pick from the Jetpack suite, specifically things that will enable Yocto support for embedded customization.

Another exciting news for the show related to NVIDIA Jetson is we won the Embedded Computing Design Award for our JCO 1000 Orin package with NVIDIA's NemoClaw to really showcase agentech AI workflows, but most importantly, the memory optimization within NVIDIA Jetson's SKU stack.

Why that's important is you can actually run the command within the BSP and the skill to really enable what is memories being used and essentially run AI workloads with the best memory capacity that is available in the module.

We'll also show you a short demo on our JCO6000-ORIN running a live VLM with AJAX-ORIN 32GB Super Mode enabled.

Last but not least, another important update is the official Yocto project support for NVIDIA Jetson.

In short, Yocto helps developers create customized Linux environments optimized for embedded deployments, which can be especially useful when moving robotic systems towards production.

For the JCO series, our Jetpack 7.2 BSP does not yet include the Yocto image package, and we expect this to be available around mid-Q4 this year, so stay tuned for the update.

Alright, so let's put these capabilities into practice.

We'll start with memory optimization using Jetson agent skills on the JCO1000RN series, followed by a live VLM demonstration on the JCO6000RN with AGXRN32GB Super Mode.

Um-

Uh, is it Yocto?

No, I don't think so.

Yo- Yocto Linux, that's, uh, that's kind of its own thing for embedded if I remember correctly.

Hmm.

Um-

So, like, it's, it's similar.

So if, if you're, if you're familiar with building Yocto, just think that you're building Yocto.

So build a system that you need to build Yocto, and then use that as your AI system instead.

Got it.

[laughs]

We have customers that use GCC, for example, to compile code.

We work in various build environments, so Yocto and Buildroot.

We also support many different operating systems on which those applications may be targeting.

We work with real-time operating systems like Lynx and VxWorks and QNX and ThreadX and others.