Page 1 of 1

Re: Searching for Meteorites with the Raspberry Pi

Posted: Sat Dec 24, 2011 11:47 am
by Brian Crosse
Hi All,

I work as an RF / digital processing engineer in the astronomy department of a University. Mostly I do work on a radio telescope. One of our "Summer Projects" (Australia - Southern Hemisphere!) is building a tool for high schools to hunt for fresh meteorites. It's a system where a bunch of solar powered webcams & cheap laptops video the skies at night and record to the disk when a meteor trail is seen.

Later; perhaps months later, the "grabs" are retrieved and compared and our software will (when complete) be able to project a likely impact point if two or more units capture the same meteor trail.

I'd really like to see us be able to bring the price per unit down and the Raspberry Pi seems like the perfect tool. It would need the Raspberry, a camera with decent resolution (via the CSI-2?), a clock calender chip, a 12v SLA battery, a small solar panel, a cheap switch-mode 5v regulator and a rain-proof case per node.

The school would need a gps to record the location of the units and some PCs to do the data analysis but very little else.

This seems to me to be well within the spirit of the Raspberry Pi as an educational tools but for science and mathematics as well as comp-sci.

This post is not really asking for help. We have plenty of students looking for projects to work on, but more as a way to canvas ideas and let you know why we're looking forward to the release of the camera module!

For those still interested:
We won't be looking for good video compression. We'll want enough processing power to search for and recognize a meteor trail in real time. We won't mind a few false positives. We'd ideally grab in highish res to memory and go off-line to dump to SD-Card on a positive detection. The unit itself wouldn't be able to convert camera X and Y to azimuth & elevation; that would be done by post processing. A "Learning mode" would grab images on regular intervals so that the post-processing system could locate bright stars and planets to train the camera lens deconvolution stage.

Please keep us in mind when you're working on the camera system! We'll publish and provide software and training to schools free!

If we can keep the price down, every school in rural / outback Australia might just want a few! Who knows, maybe we can search for Extensive Air Showers, rare animals and other great educational ideas.

Re: Searching for Meteorites with the Raspberry Pi

Posted: Sun Dec 25, 2011 2:35 pm
by Warringer
Why not use a cheap webcam? The Pi got an USB port.

Use two cheap webcams and stick them about a mater apart from each other and the Pi should have enough processsing power to run a simple pattern registering software and you are able to get the position of the object in question with some simple trigonometry by using the differences between the two grabbed images from both cameras.

Okay maybe stick them even further apart...

Re: Searching for Meteorites with the Raspberry Pi

Posted: Tue Dec 27, 2011 3:43 am
by Brian Crosse
Cheap webcams is the current plan, but I'm not sure they'll have the resolution to let us plot the trajectory accurately.  We need to calculate speed as well as path.

We think that the cameras should be about 10Km apart so that's a pretty long USB cable.

Once we need multiple units to capture the video, there's no compelling reason to get the RaspberryPi to do the math.  Chances are that the school would have two or three camera units out in the field and every few weeks they'd swap SD cards and bring the data back to school for the data reduction.  They may not even have access to a spare RaspberryPi there.  The first step would be to compare the time stamps a look for events captured by multiple cameras.  That's why false positives aren't a problem; we'd just ignore 'meteors' that didn't trip two cameras.

It's cool that the RaspberryPi has enough grunt to do the data reduction but in practice it's a school PC that has the needed information.

Re: Searching for Meteorites with the Raspberry Pi

Posted: Tue Dec 27, 2011 7:31 pm
by gimp
How's your budget?

I've seen setups where a decent webcam is connected (though an intermediary device) to a wifi setup, so that the camera's picture is shown elsewhere in real-time.

What if you did something like this:

Each camera -> one R-Pi:

R-pi will process data, crunch numbers, continuously save the last x seconds; if it sees something that your algorithms decide is a meteor, it will save (in addition to last x seconds) the next y seconds until after the meteor is no longer visible, and then save the packet of video with timestamp.

You could then use the data however you wanted; you could analyze it further, or you could even set up a way to connect your nodes to do analysis on the spot. You could use a set of wireless access points with decent antennas, though that might be expensive to reach ten miles (or maybe not, if it's a population-free area with line-of-sight, in which case cantennas would do the work for you just fine). You could use one of those satellite internet usb plugins to send the other nodes, or yourself, the video files when they appear, so you could analyze the data further without waiting for months. Linux tools such as scp would make such a process pretty damn seamless.

My point is that the extremely low cost of the R-Pi might make it possible to buy one for each camera, and that gives you a pretty sweet ability to process things on the spot and collect data without physically visiting the setups.

Re: Searching for Meteorites with the Raspberry Pi

Posted: Wed Dec 28, 2011 4:04 am
by Brian Crosse
I like the way you're thinking.

The "last x seconds" plus the "next y seconds" is exactly the sort of thing I was thinking but you put it far more succinctly than I could have.

The WiFi idea sends shivers down my spine though.  I spend most of my life trying to stay 80dB below MIL-STD 461 so that we don't contaminate the radio observatory site and you want to DELIBERATELY emit RF!!?!

THATS...THATS...THATS... probably not a problem for schools now that I think about it.

Even if the price was high(ish) for long range WiFi, we could maybe ask schools to co-operate and piggyback off the schools internet connection for some/most of the links.  I suppose my vision (and that of the other people here) was unattended boxes placed in the middle of nowhere that were visited every month or so.  In retrospect, that may not be the best plan.  I shall muse on this some more.

Re: Searching for Meteorites with the Raspberry Pi

Posted: Wed Dec 28, 2011 4:03 pm
by gimp might be useful.

Again, nothing wrong with unattended boxes. Just saying that even in that case, a bit of processing would save you a lot of time further down the road. At the very least, I think we can agree that you would be using 10x less disk space by only recording things selectively. (Where I live, it would probably be 100x less.) This could let you take higher resolution videos, or use a higher framerate, with fewer worries about size constraints.

Re: Searching for Meteorites with the Raspberry Pi

Posted: Wed Dec 28, 2011 7:09 pm
by Warringer
Maybe a tad relevant...

Re: Searching for Meteorites with the Raspberry Pi

Posted: Wed Oct 24, 2012 10:50 am
by tnabil
Hi Brain,

Although a bit irrelevant to the forum, I'm really interested in knowing which schools are working on this project with you. I might want to put them on the short list for my kids.

Re: Searching for Meteorites with the Raspberry Pi

Posted: Tue Oct 30, 2012 6:28 am
by Thalass
Has there been anything further done with this? It would be interesting to see what kind of results you could get with a cheap array of skywatching webcams, with syncronised time and Motion to automatically save videos and images of targets. Even better if the Pis were driving telescopes, aimed at a specific target.

Obviously it wouldn't come close to rivalling the Square Kilometre Array or anything like that, but still it'd be pretty neat, especially as it gets schools involved.

Re: Searching for Meteorites with the Raspberry Pi

Posted: Wed Oct 25, 2017 7:50 pm
by kevinse
Hi there, this is an old post but I'd be interested in knowing the results of your project :-). I'm diving into the Pi and this seems awesome to learn various things (imaging, connexion, etc.). Looking forward to hear more!