Would_This_run_on_original_Beaglebone?

62 views
Skip to first unread message

Monroe King

unread,
Dec 26, 2018, 7:03:25 PM12/26/18
to OpenStartracker
Would this run on the beaglebone white? I have one from another project and I'd like to try this tracker out.

How would this tracker compair with the Sinclair? @$120k

Great work here very glad to see this project.

Monroe

Umair Khan

unread,
Dec 27, 2018, 12:02:56 AM12/27/18
to OpenStartracker
Hey Monroe,

Full disclosure: I'm not the original developer of OpenStarTracker, that would be Andrew Tennenbaum. But I am working with it as part of the Portland State Aerospace Society -- we're building a CubeSat (called OreSat) and we've decided to use OpenStarTracker as the basis for our star tracking solution. As a member, I feel obligated to advertise our little crew wherever I go, so here are some links if you want to learn more: Portland State Aerospace Society, OreSat, OreSat general GitHub, OreSat Star Tracker repo (star tracker software is in "openstartracker" folder -- be warned, the rest of the repo is a bit of a mess at the moment).

I'm hesitant to flat-out say it won't work on an original BeagleBone, but I'm pretty confident that it won't be too useful even if it does work. For reference, I've used OpenStarTracker with the BeagleBone Enhanced and the PocketBeagle. You'll note that both of these run at 1GHz, in comparison to the original BeagleBone's 720MHz, so everything's gonna run slower. The biggest sticking point is RAM -- I'm a little apprehensive about using it with the PocketBeagle's 512MB after coming from the Enhanced's 1GB, much less the 256MB of the original BeagleBone. It may work, but you will run a serious risk of running out of memory at any point during the calibration and solution stages. But it could be interesting to try regardless just to see what happens :)

With regards to comparing it to the Sinclar tracker -- I'll just use this marketing datasheet and go down the categories based on my experiences:
  • absolute accuracy -- it's probably possible to achieve 5 arcsecond accuracy, but normally you'd be closer to tens of arcseconds, and I'd say the upper bound would be an arcminute of accuracy. Of course, this hasn't been demonstrated in orbit. I'm basing these thoughts off of comparing what OpenStarTracker gives to what astrometry.net gives, which seems to be a pretty good gold-standard here on Earth. But we have tested OpenStarTracker with images we've taken out in the Oregon desert and it does work very well!
  • maximum slew rate -- this depends on the camera system you're using with OpenStarTracker, not the software itself. OpenStarTracker is moderately robust, but obviously you put garbage in and you'll get garbage out.
  • lens / detector -- again, you pick whatever camera system you want with OpenStarTracker 
  • output solution -- OpenStarTracker gives you the right-ascension and declination of the center of the input image, and an orientation angle about that center. You could feasibly do something to get different outputs, though.
  • processing -- OpenStarTracker also does lost-in-space solutions, and it'll be running on whatever hardware you have it on. In my experience, you can solve a 640x480 image in ~0.3 seconds (on the BeagleBone Enhanced and PocketBeagle), so assuming you have a good camera and data transfer system you could probably hit 2Hz.
  • command / telemetry / supply voltage / power consumption / environment -- again, you pick the hardware it runs on
  • heritage -- open-source!!!
In summary, from a pure software perspective, OpenStarTracker actually compares very well. But there's a lot of hardware decisions to be made that can affect the quality of your overall star tracking system, and this is presumably what Sinclair excels at.

Hope that was helpful!

- Umair Khan.

Monroe King

unread,
Dec 27, 2018, 3:07:33 AM12/27/18
to OpenStartracker


Extreamly helpful thank you ! I'll pick up the BBE and take a shot at it. I will look at your links! I think I've seen a lot of stuff from you guys in the past? Maybe that's another group I'll have to take a look.

Yeah you need a pretty high end CCD camera and a fast lens to really make it work hard I'm sure. I've got some astrophotography cameras that will proablywork for testing. I'll have to come up with an aproprate lens.

Pretty exciting stuff!
Reply all
Reply to author
Forward
0 new messages