Hi Thomas
First off, I very much appreciate what you are doing with your channel.
I set out to build an open source CubeSat a few years ago and earlier this year, I have started a YT channel to document the progress
It's definitely early days, but progress has been steady and folks seem to like the project :) Here is the repo on Codeberg.
One of the big ticket items down the road in terms of complexity will be ADCS. For the determination part, using a star tracker is a typical approach (here are a few commercial ones to peruse).
As you can imagine, these are supremely expensive. A major goal of mine is to keep cost low and accessibility high. So I was wondering if you would be interested in partnering up on this. I'd feel right at home with the hardware aspects, but when it comes to software, I'm a noob.
Here is how i think these work: At the core of the thing, a plate solver would need to run sporadically and ingest one or multiple images (that may or may not have 'bad' data in them that needs to be accounted for, like motion blur, optical flares, the sun or the earth), filter out start position patterns, run these patterns against a catalog of known patterns, and then, given the precise time, calculate the spacecraft's rotation in 3 axes, probably relative to earth).
Ideally, this would be written in Python or MicroPython, same as most of the CubeSats's subsystems.
I have a vague feeling that a chip like the RP2350, maybe together with a TPU accelerator, could get the job done, but that's pretty much at the edge of my knowledge.
I do think this is a really nice challenge. It's a bit of a longer-term thing as I won't be getting into ADCS until late next year, but I have been thinking about getting in touch with you about this for a while and I though I'd seize this opportunity and post here :)
Anyways, I'd love to collaborate on this. Also, we could meet up irl to discuss this further at some point as I am also a member of the Germanic accent club.
Best,
Manuel
Hi Thomas
First off, I very much appreciate what you are doing with your channel.
I set out to build an open source CubeSat a few years ago and earlier this year, I have started a YT channel to document the progress
It's definitely early days, but progress has been steady and folks seem to like the project :) Here is the repo on Codeberg.
One of the big ticket items down the road in terms of complexity will be ADCS. For the determination part, using a star tracker is a typical approach (here are a few commercial ones to peruse).
As you can imagine, these are supremely expensive. A major goal of mine is to keep cost low and accessibility high. So I was wondering if you would be interested in partnering up on this. I'd feel right at home with the hardware aspects, but when it comes to software, I'm a noob.
Here is how i think these work: At the core of the thing, a plate solver would need to run sporadically and ingest one or multiple images (that may or may not have 'bad' data in them that needs to be accounted for, like motion blur, optical flares, the sun or the earth), filter out start position patterns, run these patterns against a catalog of known patterns, and then, given the precise time, calculate the spacecraft's rotation in 3 axes, probably relative to earth).
Ideally, this would be written in Python or MicroPython, same as most of the CubeSats's subsystems.
I have a vague feeling that a chip like the RP2350, maybe together with a TPU accelerator, could get the job done, but that's pretty much at the edge of my knowledge.
I do think this is a really nice challenge. It's a bit of a longer-term thing as I won't be getting into ADCS until late next year, but I have been thinking about getting in touch with you about this for a while and I though I'd seize this opportunity and post here :)
Anyways, I'd love to collaborate on this. Also, we could meet up irl to discuss this further at some point as I am also a member of the Germanic accent club.
Best,
Manuel