I recently came across this Raspberry Pi alternative - Libre Computer Le Potato on Amazon for $45 and notably in stock. It seems to be a Raspberry Pi 3B+ alternative that’s been around since 2017 or so.
For my own curiosity, has anyone here used this board either in FRC (e.g: vision coprocessor) or for personal projects?
A lot of effort was put into photonvision to get high frame rates on the rpi3
Specifically in frame acquisition and HSV conversion I believe have been tailored to the rpi3
I don’t know if the mali-450 supports this (it’s pretty barebones). I am pretty sure it won’t right out of the box.
You might have to make a custom pipeline from scratch.
It looks like it runs Raspbian, so you should at least be able to boot Pi images on it (e.g. WPILibPi) for camera streaming and OpenCV camera processing.
Of course, the pressing question in my mind is whether it’s pronounced po-tay-to, or po-tah-to.
The wplibpi image is probably going to be missing the GPU drivers and anything customized for that chip.
I’d go for an image from the manufacturer.
Best bet might be a raw python based pipeline with a flask http server for the MJPEG and control via pynetworktables.
I think I’ll pick one up and try it out.
Any news on this? I’m currently looking for alternate hardware options for our team, as several of our RasPi units have… um… “experienced significant degradation due to handling outside of their design parameters.”
We’ve bought a few of them with the intent of using them for AprilTags this year after some initial experiments on one board (with a global shutter USB camera) were promising. They had the distinct advantages of being both cheap and readily available, where most of the other alternatives missed on one or the other of those. On the hardware side, we were pleasantly surprised that it does not seem especially demanding in terms of power draw. Worth noting is that unlike the Pi, the Le Potato does not support a MIPI/CSI camera connection, if that matters to you.
The students doing the investigation haven’t gotten very far into it yet, so I do not have much in the way of specifics to share on performance. It’s very much in the “trying stuff” stage. Note that we were expecting to set up our own software stack for the image processing. I believe the students installed the standard Armbian image for it to start with. Without any particular optimization, decompression of the image stream from the camera looks like it would consume most of 1 core (it’s a quad core processor). That still leaves 3 cores - plus as much GPU power as we can take advantage of, which is the big unknown - for doing the rest of the image processing. We were also looking at the option of taking uncompressed image data from the camera, though this is USB bandwidth limited to about 5fps which is obviously not ideal.
Once we have anything of interest, we will share information about what we’re doing in our Open Alliance posts.
@signalfade and @bbonner thanks for the updates. With 3B+ systems non-existent any other hardware is helpful. Thanks!
@billbo911 may be able to help here. They recently got PhotonVision running on an Orange Pi, which is avaliable on amazon.
@stmayhem Here is the post where I posted the info about setting up the Orange Pi 4 LTS with PhotonVision.
This topic was automatically closed 365 days after the last reply. New replies are no longer allowed.