Electronics > Projects, Designs, and Technical Stuff
Radar to detect mosquitoes?
PartialDischarge:
--- Quote from: elektrolitr on July 31, 2020, 07:21:30 am ---Hawk-Eye system used in tennis etc is said to be accurate to <4mm and is using only high-speed cameras and image processing. I think that tracking a tennis ball in a court is the same order of difficulty as tracking a mosquito in a regular size room.
--- End quote ---
Besides the yellow color of a tennis ball and its 20-30 bigger size, the main problem I see is that tennis matches are played in sunlight and mosquitoes do their work at night.
magic:
--- Quote from: phil from seattle on July 31, 2020, 04:50:00 am ---No, they really thought they would be conquering malaria. I know several people involved. And the patents, maybe.
But on your last point, I've edited it to make it more realistic: Never take anything coming from America at facevalue
--- End quote ---
Eh, fair enough :D
Maybe it was the plan before the budget blew up or maybe the people you knew (engineers?) had different concept than the for profit company that hired them.
It's Myhrvold by the way and some Microsoft coworkers described him as an eternal megalomaniac, so who knows, perhaps these guys were serious. It seems like every other company wants to change the world these days, simply making money doesn't cut it anymore.
Besides, from the description, I think IV's modus operandi is roughly "patent a PoC of just about anything conceivable" and license it when / if ever the technology becomes economically viable. So who cares that nobody can afford it, meh.
Siwastaja:
--- Quote from: MosherIV on July 31, 2020, 07:39:59 am ---Triangulating sound method works for a single target.
I doubt it would work if there are a number of mosquitoes in the room.
--- End quote ---
FFT it before triangulating, I'm sure mosquittoes are buzzing at fairly slowly drifting frequency (which you can lock into and track despite the drift), and each have a different frequency. This is the feeling I have trying to sleep despite mosquittoes: different pitch everywhere, can track every moment quite accurately with the human auditory system. Obviously this only works with fairly limited number of mosquittoes and requires a bit of luck so that they don't happen to fall within too close frequencies to begin with, but I would guesstimate this could work quite well up to maybe 5-10 simultaneous mosquittoes.
In any case, the range of such system would be limited (square law of amplitude), so you would use a distributed approach and use multiple such systems independently covering different areas. They could share information whenever a mosquitto is getting near to the end of the range of one system.
As a side note, before starting to automagically kill insects, make sure they are not the pollinating kind. For example, UV light attracts insects that are genetically programmed to go for flowers (with their UV signals), these insects are extremely important for the ecosystem and fairly beningn. The bad guys you want to get rid off are those who come for your blood, and these detect CO2, body temperature, and so on.
If all this sounds too easy, here's a true engineering challenge: before zapping the mosquitto with laser, figure out how to remotely test the flying mosquitto for malaria, and only zap it if positive.
nfmax:
--- Quote from: janoc on July 30, 2020, 04:04:43 pm ---Given the size of a mosquito and the speed with which it moves, that would have to be one heck of a camera to be able to pick them up, especially at a distance of a few meters.
--- End quote ---
Swallows manage it though (Yay swallows!)
janoc:
--- Quote from: elektrolitr on July 31, 2020, 07:21:30 am ---Hawk-Eye system used in tennis etc is said to be accurate to <4mm and is using only high-speed cameras and image processing. I think that tracking a tennis ball in a court is the same order of difficulty as tracking a mosquito in a regular size room.
The key difference would be that a ball has already a contrasting color. But, if one can't paint mosquitos, there is always option to paint the room!
--- End quote ---
Ummm, not at all. First, the Hawk-Eye uses 6 to 10 cameras for the tracking. High-speed, high resolution expensive cameras tracking a ball of a contrasting color against the uniform color of the court in an extremely well lit environment. Also the fact that the ball is round and uniformly colored simplifies its detection immensely. Conditions about as good as it gets.
Compare that with tracking an essentially black mosquito against non-uniform and uncontrolled background, in a poorly lit room (so you have a lot of image noise). From my experience with optical tracking systems (both using and building them), I can guarantee you, that you wouldn't be able to reliably track a fly (10x larger target than a mosquito) optically even against white walls in a well lit room from a distance of few meters using common machine vision hardware. The bug would be the size of one or two pixels and lost in the noise.
If you don't believe me, take your photographic camera or smartphone (which likely has a much better lens and resolution than machine vision cameras do) and try to photograph a fly or a mosquito on a wall from several meters (without zoom, obviously). And see how big the critter will be in the resulting image, if it will be visible at all. For any sort of reliable detection you need the resulting "blob" of color to be at least (ballpark figure) 10-20 pixels across, likely more because the image will be averaged/filtered during the processing and very fine details get lost.
And now imagine you are doing this at perhaps 1/4 resolution because machine vision cameras rarely exceed 2000px per line. Both because of costs and also because the gear required to process such enormous images in real time (unlike taking a static photo) would be extremely non-trivial. The high resolution (e.g. 15 megapixel and more) cameras also tend to have a slow framerate, making them unsuitable for motion tracking.
Navigation
[0] Message Index
[#] Next page
[*] Previous page
Go to full version