4 ms·
Have you looked into the beaglebone products? I assume you're talking about the business listed in your about page? If you're doing any kind of stepper motor co
by ahepp 4y ago
Have you looked into the beaglebone products? I assume you're talking about the business listed in your about page? If you're doing any kind of stepper motor control ,the AM335X and AM57XX SoC families seem to have some pretty cool realtime units and cortex M mcus onboard.
- grepLeigh 4y agoThe Raspberry Pi doesn't drive 3D printer motors - that's typically an ATMega, Duet, or another 32-bit ARM SoC like the boards you mentioned. 3D printer manufacturers typically distribute a 32-bit SoC with either Marlin or Klipper firmware. https://www.klipper3d.org/ https://www.klipper3d.org/ I distribute an aarch64 Linux distro for the Pi (PrintNanny OS), which handles monitoring and automation tasks. The Pi also hosts a web GUI.
- ahepp 4y agoMakes sense, looks like a cool project! Are you saying you don't think your software could run on a 32 bit SoC? That would surprise me, but you would certainly no better than I would. Edit: I didn't realize there was CV involved, that probably uses a lot more resources. Really awesome project, I hope the supply issues get better.
- grepLeigh 4y agoThank you! Yeah, I run a WebRTC stack and gstreamer+nnstreamer application to stream video from the Pi. I'm running a TensorFlow Lite model on-device for object detection. Getting that stack to run on a 32-bit SoC without an operating system would be tough. That said, there are some incredibly talented people in the world advancing the field of "TinyML" https://www.tinyml.org/ https://www.tinyml.org/ - it's not like you're suggesting something ridiculous! There are already specialized camera + NPU (Neural Processing Units) boards on the market for specialized edge CV tasks. PrintNanny's workflow/automation requires general computing capability though, which rules out a lot of specialized CV boards.