If it's remotely possible I can almost guarantee the three letter agencies around the world are using it already (or have tried).
MI5 were lifting encryption keys via acoustic analysis of typing something like 65 years ago [Wright, Spycatcher]. And their budget was fairly limited, as opposed to the US agencies with their basically unlimited budgets (i.e. NRO)
I'm less concerned about the bug than the tiny remote camera with 6 hours of battery-powered live streaming capability from up to 120 meters distance. I'm concerned about more surveillance devices becoming smaller and pervasive. I hear this dismissal of "be more concerned about smartphones" frequently though. We can be equally worried about both can't we?
I don't think camera system size has been limiting factor in surveillance for a while. For robotics (either walking or flying) the locomotion and power are more restricting aspects, and using live insects for surveillance seems too slow and unpredictable to be useful.
If they cheap enough, imagine the fun of trying to secure government sites against everything from random conspiracy nuts to journalists and state actors when you need to secure it against something bug sized entering...
Moore's Law doesn't really work anymore. It hasn't held for the last 4 years, IIRC. That's why a lot of focus has been on multicore systems to try and get new speedups based on parallel computing as opposed to getting purely single-chip improvements.
They are, but performance in teraflops is not the same as performance in general computing activities. We are now scaling horizontally within a single cpu package.
Often they can survive quite a while without food or water, as long as the temperature and humidity are within some fairly generous extremes, although very dry air poses the risk of dehydration since their respiratory systems and body volume offer little ability to retain a reserve. But they rarely have to go that long without a meal, because if a spider can get in, so can suitably sized prey - and will. Next time you find yourself wondering how a spider thrives in an empty room, take a close look at and under her web. Most likely you'll find cocooned prey in both places - fresher prey items in the web, spent ones underneath it.
Spinning a web is a metabolically very expensive process, requiring as it does a great deal of protein - protein which can only be recovered by either eating the spun silk again, or eating prey trapped by it. Spiders consequently have to be pretty smart about where and how they do it, because the ones who aren't tend not to reproduce. So it's usually a very safe bet that, wherever you see a spider web, it's there because that's where the prey tends to be. It can be interesting to see what sort of prey that is - for example, right now in the basement of my ~130-year-old house there are a couple of very happy cellar spiders enjoying a steady diet of pill bugs, and another by the kitchen radiator who all on her own solved most of an ant problem for me before I ever realized I had it.
They're actually very desirable animals to have around, in my thoroughly considered opinion, despite getting about as bad a name as wasps - and with just as little reason.
wow had no idea they eat the silk. They are so small that I only notice them because of the webs and long strands flying around. I'll keep an eye out for their prey :) Thanks for that interesting info filled comment!
I've seen another study, which in my opinion is way more useful, where they are able to control a cockroaches using electrical pulses on their antennas. With a camera attached to its back, these cockroachs will be able to find people underneath rubbles.
Could a cockroach carry both the cockroach control device and the camera? Does the micro camera have enough low light capabilities to be useful in the dark? Are there any animal ethics to be taken into consideration for attempting this route?
I can’t tell if that idea is 2 years or 20 years in the future, but it sounds neat.
You'd probably be better off using a microphone if we are taking about search in low light conditions. Or maybe a CO2 sensor, if you can miniaturize it enough.
But I'm not so sure it's that much better than dogs for search and rescue. Much more promising for surveillance.
R2R, or Roach 2 Roach would be more apt. Jokes aside, this tech later applied to other animals, say that pesky 2 legged kind could make for a rather interesting future.
That's the only image stuck in my head since reading the headline. Glad to see they are in fact reality. Such a great movie, except the ten minutes of ruby rod screaming.
Just ignore it whenever researchers say their thing could be used for search and resuce. Nearly every fun robot invention is promoted as having that purpose, probably because the inventors can't really think of an actual use but they still realize it's cool.
The title made it sound like a fly or an ant, but in reality it’s a rather large beetle. It’s neat work though.
For a redteam project, I once made a spy camera that fit in a belt buckle. I’m not sure you’ll need to worry about surveillance implications of insect-mounted cameras; surveillance is already pervasive.
Imagine releasing the bug near a restricted area. The attacker can hope it gets past security and gain images from inside the restricted area. If it gets caught then the attacker if up to 120M away, which is a good head start.
Good point. The maximum distance the attacker could be would be limited by the act of releasing an insect and the biological limitations of insects (lifespan, travel speed), not by the range of the wireless communication.
Once they get it down to flying insects, would it be possible to deploy a swarm of these (thousands) and use positioning data to generate high resolution 3D imagery?
Hopefully the smaller it gets the less power it needs, the easier it is to power using environmental factors like wind, earths magnetic fields and the movement of the insect in such a scenario.
It would be a very hard challenging to build a high resolution image from several low level images. I haven’t heard of any similar work. This is an interesting problem space.
Could you use an array of these mounted at precise, known, differing angles and an FPGA to calculate high resolution localised depth maps? Applications you say? Entomological gesture studies!
Last time I spoke to Omnivision they wouldn't even consider a purchase order (from an R&D company) unless we were buying 2k units. So don't bother unless you're building consumer electronics.
Other companies like e2v do have sampling packs available for most of their standard sensors, and they're pretty friendly. Not quite so small though!
Interesting. That OV06948 camera is also 40k (200x200) pixels, 30fps, and RGB. Critically, the page you link lists its weight at 0.87g = 870mg, which would make it 157x denser than lead.[1] The actual spec sheet says 0.87 mg, i.e. less than one milligram, which is incredible.[2] It has up to a 4 meter range, which I understand to be wired, so it would still need some sort of transmitter.
The camera used by these researchers, including its lens and panning head, is 200 mg (77 for the sensor and lens and mount, 7 mg bluetooth chip, and 96 mg boost converter). It's only 20k (160x120) resolution, monochrome. It has a wireless 120 m range.
The killer, however, seems to be power. The 10 mAh battery alone weighs 500 mg, which means the 200 mg vs <1 mg comparison becomes more like 700 vs 501.
The OV06948's power usage is 25 mW, which means a 10 mAh / (25 mW / 3.3 V) = 1.32 hour runtime, capturing 142k frames of video. The bug rig gets 6 hours, capturing only 108,000 frames of video. Perhaps you could get longer battery life out of the OV06948 if you slowed the framerate.
Unfortunately this seems outside the bounds for ambient RF energy harvesting, even if you don’t mind dropping a few frames to charge back up if the capacitor runs low.
The rectenna would have to be much too large to produce anywhere near 25mW even at a 10% duty cycle.
I suppose if you didn’t mind ramping the RF output power way up you could shrink the antenna size, but not much smaller than 1cm^2 I think.
Co-author here. One of our primary constraints was power usage. I’m not sure exactly why, but the camera we used, by Himax, uses much less power than any omnivision camera we were aware of.
OK kids, I need help here. There was a sci-fi novel, probably the sixties, called something like "little bugs have little bugs on them". The premise was basically that.. A cold war where one side puts bugs on the bugs of the other side. Rings any bells? Potentially there were also some aspects of bug driven assassinations... But I am unsure. This thing been bugging me.
I think I may have found it on the wikipedia page for nanotechnology in fiction:
> The 1984 novel Peace on Earth by Stanislaw Lem tells about small bacteria-sized nanorobots looking as normal dust (developed by artificial intelligence placed by humans on the Moon in the era of cold warfare) that has later came to Earth and are replicating, destroying all weapons, modern technology and software, leaving living organisms (as there were no living organisms on the Moon) intact.
It's not a perfect match for your description, nor is it from the 60s, but it was the closest match I could find. Is that the book you were thinking of?
OK. The plot thickens. https://www.phrases.org.uk/meanings/big-fleas-have-little-fl... so .. this is the phrase I remember. Probably the "motto" used by the SF author. Now I remember Peace on earth... and BTW à propos the article cited .. the spy bugs there have very partial vision .. and if I remember correctly they recompose scenes from the multitude of images... I think this is as far as I am going to get. Thank you!
Am I the only one here at least considering the ethical aspect of attaching some device to an insect?
Maybe I'm sentimental and contradictory because when I drive a car over the Autobahn I'm killing hundreds of insects in an hour. At the same time I cannot harm a bug intentionally.
I can relate, but considering how much we exploit animals, including agricultural animals, breeding, raising and killing animals for food, using rats and monkeys for testing etc., this would not surface in the priority list for fighting animal welfare for me for a very, very long time. Even in thought.
I agree - there's a big difference between the ability of various species to experience suffering or to experience happiness, and I think that has an impact on the ethics of how we treat them. Humans, monkeys, pigs, dogs, etc. are definitely social/emotional intelligences that deserve better treatment than we give them. Rats? Chickens? Insects? Jellyfish? Plants? Bacteria? Somewhere on that spectrum I cease to worry about the suffering of the animal and worry more about those affected by it.
That's so ignorant. How can we possibly know how other animals experience pain or suffering? You just assume that the further you diverge from us in the animal kingdom the less animals are able to suffer. That certainly makes things easier.
I think a lot of it has to do with the means of reproduction of the animal. One might make an argument that [r-type species](https://en.m.wikipedia.org/wiki/R/K_selection_theory#r-selec...) have decimating loss factored into their everyday experience and survival strategy. As a different commenter pointed out, driving a car down a highway kills hundreds of insects. As a practical matter, insects are the most successful clade ever, with the largest biodiversity, largest number of individuals, and largest biomass of any land animals (by far). If the end is to reduce human impact on ecosystems, then farming insects seems like a reasonable trade off from a utilitarian perspective.
There has been a ton of research into how different animals experience pain. You are correct that we don't know how they experience it, but based on nervous system structure and other behavior, we can make some conclusions about the nature of the pain and whether it's a basic avoidance mechanism or something more complex that would involve suffering.[1]
That said, I think a good rule is to not cause unnecessary pain/suffering in any living thing (e.g. if trapping rodents, use something that leads to a humane death).
I think the assumption that anything with a brain can experience pain, suffering and happiness is a pretty safe one. Synthesising information to decide what is suffering and happiness, and then trying to maximise happiness, seems to be one of the basic functions of any brain.
It's like killing a spider in the house. Killing it instead of taking it outside doesn't bring me any benefit the way eating a chicken does, so why kill it? Just like putting the camera rig on an actual beetle for the demo doesn't bring a benefit, so why do it?
I think this is just a reflection of your personality. Respecting nature, small or big and taking the time to give it space, let’s you value and organise your life in a more balanced fashion.
That’s only true if your time has no value. Killing is faster. Also, as someone who has gotten multiple infections from non-venomous spider bites, I can tell you catch and release is riskier than killing.
this is much better than previous things i've seen that implanted electrodes in the brains of cockroaches and used electrical impulses to steer them around.
Attaching a camera to an insect is probably not even a blip on the radar of a university ethics board compared to all the mammals that are used and killed annually in various experiments.
You also kill billions of bacteria without even knowing it... and of course every time you wash your hands, which I hope is a frequent occurrence in these times.
Excellent work. I'm old enough to remember when putting wireless vision systems on a robot required a kilogram at least :-). Of course this would be made even simpler if you integrated the vision sensor, the BLE radio, and the control system into a single die. Something that would be quite accessible to nation states and sufficiently bored billionaires. Ambient energy harvesting has also made great strides so the days of persistent and difficult detect surveillance are getting closer.
With high power towers for cell network and plenty of devices that emit radio, you can have a wireless power harvester that can be powered forever if you lower you energy consumption below a given threshold.
Also was an experiment a while ago where photons were both feeding the camera sensor and used to get the image itself. Granted resolution and clarity was bad but the images still could be used to recognize stuff. Combine that with a smart AI at receiver site and you can have full surveillance, at least for counting bodies (humans + cars) for your desired place (think gathering places like Tienanmen).
> The power system is the primary limitation here, but it might be possible to use a solar cell to cut down on battery requirements ... with a long-range wireless link and a vision system, it’s possible to add sophisticated vision-based autonomy to tiny robots by doing the computation remotely
Since WiFi can be used to see through walls and ceilings of buildings and homes, using low-cost passive sensors, such devices can be autonomously navigated. E.g. for indoor use, knowing light source locations for charging and camouflage.
> our dataset provides a collection of Wi-Fi signals that are recorded for 40 different pairs of subjects while performing twelve two-person interactions. The presented dataset can be exploited to advance Wi-Fi-based human activity recognition in different aspects, such as the use of various machine learning algorithms to recognize different human-to-human interactions.
No, considering how large that thing on the bug is, and how easily it would be seen, it would need to be much closer to something in order to see anything. It would be cheaper/more effective to simply use normal means.
I'be been banging on about MOOP - Massive open online psychology- where we basically monitor our daily interactions such as conversations with kids partners etc, and build a society wide set of best practises and can then be guided in real time by these best practises (if you have ever seen those shows where a Nanny lives full time for a week with a family that the kind of thing)
But it only works if the data is treated medically. we won't ever stop the three letter agencies from abusing it (entirely) but we need to make PII more than GDPR and make it as sacrosanct in law as lawyer client / doctor patient confidentiality.
My take on this is that old saw - the great ideas seem like bad ideas to most other people.
We are already in a surveillance society, and it is not going away, so it needs to have benefits for us or it is just dystopia all the way.
At the moment the average school leaving age in the UK just became 21 as now 59% of leavers attend university. We are educated / trained for the first 1/4 of our lives - but the remaining 3/4 ? We need life long learning but that does not happen in a classroom - it's most effective right at the moment we need it - and what delivery mechanism exists - there is one in every pocket.
And it's not good to anyone just to straight up insult other posters. Honestly, I am happy for folks to disagree but try and figure out why you disagree with me - express your thoughts.
It’s possible to remote control a cockroach sized bot, so now you can be a sort of “fly on the wall” in any conversation. Drop the roach off by drone which can double up as the access point.
Unless they've managed to somehow hook the output of the cam to the cortex of the bug thereby allowing it to do things it previously couldn't, I find the article a tad disappointing: this is one huge bug. Does it even fly?
I saw "steerable" and was hoping they had wired something up the pleasure centers of the insect's brains so they they could steer them around like little RC vehicles. (nope... they can only point the camera)
I believe I saw somewhere they did that with rats. Fascinating, and yet so f'd up.
Generally, there' a tradeoff between quality vs size (mostly through energy: processing and sending more data needs more energy, which needs a bigger power source).
Presumably they also picked some reasonably available and suitable parts and tried what they can get, so there might be room for improvements now there is a baseline established.
189 comments
[ 4.5 ms ] story [ 247 ms ] thread...a joke that hits way different after hearing the news out of Portland...
MI5 were lifting encryption keys via acoustic analysis of typing something like 65 years ago [Wright, Spycatcher]. And their budget was fairly limited, as opposed to the US agencies with their basically unlimited budgets (i.e. NRO)
I read a book suggesting that programme was discontinued as the cats kept getting involved in road traffic accidents.
Check out Black Hornet Nano for what was state of art almost a decade ago https://en.wikipedia.org/wiki/Black_Hornet_Nano
Spinning a web is a metabolically very expensive process, requiring as it does a great deal of protein - protein which can only be recovered by either eating the spun silk again, or eating prey trapped by it. Spiders consequently have to be pretty smart about where and how they do it, because the ones who aren't tend not to reproduce. So it's usually a very safe bet that, wherever you see a spider web, it's there because that's where the prey tends to be. It can be interesting to see what sort of prey that is - for example, right now in the basement of my ~130-year-old house there are a couple of very happy cellar spiders enjoying a steady diet of pill bugs, and another by the kitchen radiator who all on her own solved most of an ant problem for me before I ever realized I had it.
They're actually very desirable animals to have around, in my thoroughly considered opinion, despite getting about as bad a name as wasps - and with just as little reason.
Who will own the copyright on the video?
[1]: https://www.youtube.com/watch?v=DrHMBletjXg
I love using sci fi as inspiration for new products.
[1] https://en.wikipedia.org/wiki/Danny_Dunn,_Invisible_Boy
[2] It's not invisibility ala The Invisible Man, but instead a dragonfly sized drone.
I can’t tell if that idea is 2 years or 20 years in the future, but it sounds neat.
But I'm not so sure it's that much better than dogs for search and rescue. Much more promising for surveillance.
...and those Things are bugs.
It might not be the only research lab doing this, but its the one I'm most aware of.
For a redteam project, I once made a spy camera that fit in a belt buckle. I’m not sure you’ll need to worry about surveillance implications of insect-mounted cameras; surveillance is already pervasive.
https://en.wikipedia.org/wiki/Super-resolution_imaging#Multi...
https://arxiv.org/abs/1905.03277
https://en.wikipedia.org/wiki/Photosynth
https://en.wikipedia.org/wiki/Spotted_lanternfly
https://en.wikipedia.org/wiki/Prey_(novel)
The smallest known camera module is the OmniVision OV6948, has the dimensions: 0.65 x 0.65 x 1.158mm, including 120 degree wide lens.
Sadly it cannot be purchased in single quantities :(
[1] https://www.digitalcameraworld.com/news/worlds-smallest-came...
Edit: It may help to describe your sample order request as being part of an evaluation.
Other companies like e2v do have sampling packs available for most of their standard sensors, and they're pretty friendly. Not quite so small though!
The camera used by these researchers, including its lens and panning head, is 200 mg (77 for the sensor and lens and mount, 7 mg bluetooth chip, and 96 mg boost converter). It's only 20k (160x120) resolution, monochrome. It has a wireless 120 m range.
The killer, however, seems to be power. The 10 mAh battery alone weighs 500 mg, which means the 200 mg vs <1 mg comparison becomes more like 700 vs 501.
The OV06948's power usage is 25 mW, which means a 10 mAh / (25 mW / 3.3 V) = 1.32 hour runtime, capturing 142k frames of video. The bug rig gets 6 hours, capturing only 108,000 frames of video. Perhaps you could get longer battery life out of the OV06948 if you slowed the framerate.
[1] https://www.wolframalpha.com/input/?i=%28870+mg+%2F+%280.65+...
[2] https://www.ovt.com/download/sensorpdf/553/OmniVision_OVM694...
The rectenna would have to be much too large to produce anywhere near 25mW even at a 10% duty cycle.
I suppose if you didn’t mind ramping the RF output power way up you could shrink the antenna size, but not much smaller than 1cm^2 I think.
They are not cheap: https://www.digikey.com/product-detail/en/ams/NE2D-CHIP-B-W-...
> The 1984 novel Peace on Earth by Stanislaw Lem tells about small bacteria-sized nanorobots looking as normal dust (developed by artificial intelligence placed by humans on the Moon in the era of cold warfare) that has later came to Earth and are replicating, destroying all weapons, modern technology and software, leaving living organisms (as there were no living organisms on the Moon) intact.
It's not a perfect match for your description, nor is it from the 60s, but it was the closest match I could find. Is that the book you were thinking of?
Maybe I'm sentimental and contradictory because when I drive a car over the Autobahn I'm killing hundreds of insects in an hour. At the same time I cannot harm a bug intentionally.
That said, I think a good rule is to not cause unnecessary pain/suffering in any living thing (e.g. if trapping rodents, use something that leads to a humane death).
[1]https://en.wikipedia.org/wiki/Pain_in_invertebrates
Also was an experiment a while ago where photons were both feeding the camera sensor and used to get the image itself. Granted resolution and clarity was bad but the images still could be used to recognize stuff. Combine that with a smart AI at receiver site and you can have full surveillance, at least for counting bodies (humans + cars) for your desired place (think gathering places like Tienanmen).
> The power system is the primary limitation here, but it might be possible to use a solar cell to cut down on battery requirements ... with a long-range wireless link and a vision system, it’s possible to add sophisticated vision-based autonomy to tiny robots by doing the computation remotely
Since WiFi can be used to see through walls and ceilings of buildings and homes, using low-cost passive sensors, such devices can be autonomously navigated. E.g. for indoor use, knowing light source locations for charging and camouflage.
From https://www.sciencedirect.com/science/article/pii/S235234092... (2020):
> our dataset provides a collection of Wi-Fi signals that are recorded for 40 different pairs of subjects while performing twelve two-person interactions. The presented dataset can be exploited to advance Wi-Fi-based human activity recognition in different aspects, such as the use of various machine learning algorithms to recognize different human-to-human interactions.
Related project: https://dhalperi.github.io/linux-80211n-csitool/
But it only works if the data is treated medically. we won't ever stop the three letter agencies from abusing it (entirely) but we need to make PII more than GDPR and make it as sacrosanct in law as lawyer client / doctor patient confidentiality.
We are already in a surveillance society, and it is not going away, so it needs to have benefits for us or it is just dystopia all the way.
At the moment the average school leaving age in the UK just became 21 as now 59% of leavers attend university. We are educated / trained for the first 1/4 of our lives - but the remaining 3/4 ? We need life long learning but that does not happen in a classroom - it's most effective right at the moment we need it - and what delivery mechanism exists - there is one in every pocket.
And it's not good to anyone just to straight up insult other posters. Honestly, I am happy for folks to disagree but try and figure out why you disagree with me - express your thoughts.
- Someone somewhere on a DARPA project
But I'll be stocking up on insect-killing sprays.
To coin a phrase, this is the "progress of d!ckheads".
Make cameras smaller, and super glue them to a bug. Is there really a point, besides the trendy headline?
You might be able to do some interesting wildlife documentary movies, though.
I believe I saw somewhere they did that with rats. Fascinating, and yet so f'd up.
Presumably they also picked some reasonably available and suitable parts and tried what they can get, so there might be room for improvements now there is a baseline established.
[007]: https://www.youtube.com/watch?v=ye8KvYKn9-0