Slashdot Mirror
Too Many Asteroids To Keep Track Of?
BLAMM! writes: "Space.com has an article explaining how the increase in asteroid detection has outstripped our ability to examine the data. And the problem is only going to get worse. A distributed workload is mentioned as a solution, but only among a few other "astereroid hunter" organizations. It's clear that an abundance of data isn't the only problem, but perhaps an @home type distributed solution would help? Anyone have a contact inside Seti@home? Looking for aliens is cool, but checking asteroids for NEOs is a real-world, in-your-face kind of problem."
In order to find an astroid, you have to compare two images taken of the same piece of the sky some time apart and then look for 'moving stars' (This was the method used to find Uranus, Neptune and Pluto as well). Computers are good for this type of search, but each client must be able to know the orbits of the already known objects so it do not report duplicates. This can be overcome by allowing the client to report those, and then use a central computer to compare new results and discard already known ones.
The problem comes when a lot of people have to download images and compare. These images must be of the highest resolution for the search to work.
Analysing an image would not take as long as FFT-analysis of radio-telescope data, as it is basically a pixel for pixel comparison (although this is a bit of an oversimplification). Therefore a LOT of images have to be transmited across the internet.
Lets make a rough estimate. a high resolution image would be something like 1200x1200 pixels (8 bit greyscale) this is 1.4 MB This can perhaps be compressed to somthing like 300KB. If each user analyses one set of images each day then each user has to download 600KB a day. This corresponds to the server having to transmit 55 megabit/s on average. This might not sound of much, but this is only the raw data, results have to be transmitted back and negotiations between server and client also have to be transmitted.
I think that this calculation shows that unless some realy smart could be done (or the analysis of the images takes much longer than 1 day on average), a seti@home style distributed analysis would be impractical.
Yours Yazeran
Plan: to go to Mars one day with a hammer
Yours Yazeran
Plan: To go to Mars one day with a hammer.
I doubt the problem of cataloging new asteroids lies with the actual finding of them. A simple image blinking discovery method would consume a trivial amount of computing resources. I'd suspect the tedious portions of the cataloging are due to the measurement of the angle/direction the asteroid moves in relation to the imaged field, it's correlation to the orientation the telescope was in at the instant it took the image, the time of year it was taken, the location of the observer, etc. and the calculations required to derive the asteroids actual orbit from these data. However I don't see why all that cant be computerized and automated with sensors (gps, atomic clock time, telescope orientation sensors, orbit calculation algorythms etc.) so that people are only needed to maintain the equipment. All that really needs to be invented is the software.
- "Hear that?! The percolations are imminent! Cease your ingress!"
The asteroids are already detected, but their orbits are not well known. There's no need to send around bitmaps to compare for moving stars. That step is already completed. The article said that the number of known asteroids will double in the near future, and there are two problems.
1) There are not enough scopes to do followup observations. OK, you geeks. You want hackable hardware? How about you pick up a good Meade telescope with a laptop, CCD, appropriate cables and software, and set up your own asteroid discovery observatory. This is how these asteroids are currently found, regardless of the size of the telescope. A modest 11" scope can do excellent quality science without breaking the bank. You can use this equipment to do followup observations that are needed to track the object, calculate the orbit, and keep it from getting "lost".
2) A small distributed.net style project could possibly be used to do orbit calculations. Orbital mechanics, especially the type involving multiple objects, such as Jupiter, the Earth, and the Sun (maybe mars too) require a lot of horsepower. Running 60,000 asteroids through 100 years of orbit prediction might not require the same effort required to crack a 64-bit RSA key, but it's more than my Celery 300 can handle by itself.
If tits were wings it'd be flying around.
Now there is a distributed system that would help out this project. Granted, you need some hardware, but it shouldn't be that expensive. Those computerized Meade telescope like this one, aren't that expensive. Now if they could be centrally controlled and linked, very high resolution images could be rendered? I bet there are already some people set up with this. Any projects started? What are they tracking these days?
-- these are only opinions and they might not be mine.
Consider the number of times some idiot has tried to modify the seti@home client to falsify its work in order to improve his or her account statistics. That's not really a huge risk to human life when we're searching for aliens (Unless, of course, their first message to us is something along the lines of "We are the Borg. You will be assimilated. Resistance is futile. Your biological and technological distinctiveness will be added to our own.") However, if some script kiddie spoofs his work units, and so we miss a near-earth orbiting asteroid - especially one traveling so close it could concievably hit the Earth - well, that would be bad. I know the folks working on seti@home have implemented error-checking, but it just seems to me that the risk of depending on thousands of insecure home PCs to assess a threat that could destroy civilization might be too great. Whatever happened to "slow and steady wins the race"?
I know, there's always the possibility that an asteroid or comet is on a collision course RIGHT NOW, but just considering the time spans involved (65 million years since the last -ahem - dinosaur killer),I would thing that the more likely result of our search would be to find an asteroid that might hit us in 10 years. Or 20. Or 100, I dunno. The point is, I would feel much safer knowing that the asteroid data were being analyzed by secure, trusted machines that might take longer, that if it were being handeled by a very insecure internet distributed computing system.
Is that sinking feeling my karma falling through the floor?
I'm the stranger...posting to