Some context: ARGUS was critical in winning the war in Iraq. It was the critical centerpiece of Task Force ODIN, and they set it up in hotspots to monitor 24/7.
Once a roadside IED bomb was detected, they rewound the video, tracked the person who setup the bomb, then tracked that person back to the supplier for the IED/Financials. ODIN killed about ~3000 in Iraq and at least ~1000 in Afghanistan See: http://www.npr.org/templates/story/story.php?storyId=1142211...
From what I remember what was critical in 'winning the war' in Iraq was the Sunni awakening when the military agreed to let local tribes take care of business in exchange for political power because they were tired of being murdered by foreign militants and sheltering them out of fear. In a few weeks Anbar province was totally cleared out using on the ground regular intel no flashy space camera.
Yeah, that guy who dealt marihuana on rollerblades in NYC could get tracked in a sting operation which would also catch most of his customers. All they'd have to do is to "tail" him with such a device over a month.
Right now, it would be way too expensive to devote such a system to catch one dealer, but the scary thing is that systems like this could surveil and process entire regions/neighborhoods as a batch, and the expense of doing so is driven by the cost of communications and processing power. However, we know those costs are going to go down. From this, we should be able to estimate the time when large metropolitan police departments will be able to afford such capabilities.
That gives a concrete deadline for interested parties who wish to act.
EDIT: To clarify -- that would be a deadline for when it will be too late not a deadline for when to safely act by!
> Argus is now obsolete, they had a replacement RFP back in 2009.
Scary. I wonder if it would be possible to do the same thing in infrared? What would it take to surveil the entirety of the US coastline and borders, for example?
Of course! There are multiple birds/sensors with SAR, IR, and multispectral capabilities always surveying various parts of the world. Organizations like NGA (formerly National Imagery and Mapping Agency to give some context) are very keen to using such geospatial data and provide resources in acquiring certain types of unclassified data sets:
> Of course! There are multiple birds/sensors with SAR, IR, and multispectral capabilities
I don't mean just taking pictures of them all. I mean to do the equivalent of Argus -- real-time 12 frames/second surveillance of the entirety of the borders and coasts of the continental United States of America. Just having the birds with the cameras is only a fraction of that task.
Entire startups could be built around the gathering and analysis of visual data from drones. A lot of businesses can a benefit from a little bit of being in the right place at the right time, and some business are only possible with reliable real time information. Imagine the applications to all kinds of businesses if they could put a 24 hour, 10,000 foot tail on their rivals. It won't be long before these drones have the ability to read a piece of paper from thousands of feet away.
From a machine learning perspective, I see drones as flying data factories. So much can be learned and predicted based on the movements of people, animals and low pressure systems. The level of "eyes on" analysis provided by drones can really improve a lot of key technologies. Privacy will be an issue of course, along with the logistics of many flying machines in the same air space. Smart cities of the future will operate their own drones and charge a subscription to the data feed for businesses.
I think it's definitely work watching the PBS Nova episode about drones [1]. I walked away thinking about the implications of 10 years of advancement in this field.
Drones should never be privatized, combining that much surveillance data with your Facebook data, it can basically predict your next move. That would be too much power in the hands of big corporations or malicious hackers.
The next logical step is for a private drone company with the ability to collect, and analyse this data in near real time and sell it to Wall Street. Maybe even sell the data to WalMart so they can monitor staffing levels, and have the right amount of staff on the tills at all times for maximum throughput.
You could also monitor shipping at ports, rail yard delivery volume, trucks at warehouses and so on.
I couldn't sleep at night knowing I used my engineering degree to enable global spying and summary executions of 'militant safehouses' full of kids, or enabling western friendly dictatorships like Yemen and Bahrain by helping them crush dissidents.
It's like when I hear about Physicists who no doubt got into the field to explore the universe and everything in it then ended up making audio weapons for the military for the sole reason of dispersing protests.
It's great that you wouldn't. But it's really very arrogant to assume that there exists no reasonable moral system that would allow an engineer to participate in drone development or audio weapons.
According to this article[0] it means 'all military-age males in a strike zone', unless there is evidence posthumously that indicates a combatant is innocent.
Seems to be wider than you are claiming: "unless there is explicit intelligence posthumously proving them innocent" is the quote. What burden of proof are they using?
So it can resolve 6 inches at 20k feet. About 1 pixel of resolution for the "terrorist's" face. At least they have a roughly 4x12 pixel image to make a sound determination as to which person to target.
FWIW I think the editorial modification of the HN title here from terrorist->person confuses the existing debate a bit. Additionally, it no longer matches the article headline.
While I'm generally in favor of title rewrites so people don't karma whore, editing the HN title away from the original headline seems a step too far.
Are you saying the (albeit original) word "terrorist" wasn't editorializing? All terrorists are persons, but not all persons are terrorists and, realistically, a drone can't tell the difference. Heck, people can't tell... (Remember those old "how to spot a communist" videos? https://www.youtube.com/watch?v=SkYl_AH-qyk ) I'm saying I think the word people is an improvement in objectivity.
When I think "spotting a terrorist", I think seeing a guy put a bomb on the ground, not necessary picking out facial features, which need more resolution. When I hear spot a person, I think "pick out a particular person" [facial features]. This may not be what other people infer, but it's my general gist.
It confuses the existing debate because there were already 60 minutes of comments, like my own, which reference this editorialization. Now that it has been obfuscated away this is slightly confused.
Then that's an interesting statistic in its own right: propaganda is the central debate in such cases. (It's not confusion, by definition.) If it were originally submitted to HN verbatim, with the word "terrorist," I would actually expect more of the focus of the debate to be on that reporting detail.
What does the face have to do with it? Do terrorists have different faces?
I imagine context is important. Camping in a tree with something that looks like a RPG, overlooking a supply route? Certainly suspect, no matter what your face looks like.
All that said, the article has very little to say about terrorists. At least as far as the article itself goes, it looks like the terrorist bit in the headline was just an attention grabber.
USA redefined "terrorist" to mean any person physically close to person that is known to be terrorist. It simplifies targeting and civilian casualties counting.
Don't assume what's actually happening is all that close to what's being described in an article like this. The goal is mostly just passive surveillance, a crappy video of something interesting is far better than a really high resolution video of an empty street.
PS: Don't forget your not limited to having one drone at a time.
In the other comment achille links to a Frontline description of how a similar system was used in Iraq. The 20,000 foot system with poor resolution was used to continually track an area, when something suspect occurred they would send in a higher resolution Predator and then if it checked out further they would follow and target the people on the ground.
I don't think they are identifying the people by their faces. Secondly, I don't think the drone strikes only hit one person. I think it hits a small area.
This must be a really valuable information asset. Think about when a IED goes off, they can reply the events leading up to this attack, trace the vehicles/people involved back to houses/buildings. What about when a murder or bank robbery happens. You can just watch when/where/who is doing what! This is some jack bauer stuff!
My thoughts exactly. It would be almost trivial to find out where someone has driven their car after committing a crime. I'm curious to see what might happen if they deploy these in US cities.
A little more context from that episode - this 1.8 GP camera thing was in part because operators would get tunnel vision when running the mega-zoom camera on the Predator drone for too long. They'd be tracking a van and miss the larger context that the van is full of school children. Nova painted this camera as a way to never lose the bigger picture. Of course now you've got a massive bandwidth problem.
No mention was made of the obvious solution: add a second low-zoom camera. A basic usability feature would be to click on a spot in the low-zoom area and have the high-zoom camera immediately track over to that spot, making it very easy to flip between several locations with the high-zoom lens.
edit: Other bit of bad design mentioned in that show. Apparently if you bank too sharply the satellite connection will break. I do not understand how this is possible. Gimbel the antenna and slave the servo to the artificial horizon gyroscope. This was a solved problem decades ago.
I find the technology and engineering behind this extremely interesting, but on the other hand the potential end use of this very scary (if used inappropriately by the wrong people).
The fact that today we can already buy a UAV for under $1,000 and a 4k camera for $500 (http://www.lehmannaviation.com/la/la100.php) just gives you a sense of how easy and fast this type of surveillance is evolving.
I thought I would chime in on my own experience with related imagery. It's not uncommon to use a drone similiar to ARGUS that is capable of NIIRS 5-7 (see http://www.fas.org/irp/imint/niirs.htm) resolution for surveying. The guys at the ground station will find some region of interest and then use another bird/sensor--whether that is SAR, IR, multispectral, etc--with NIIRS 8-9 resolution and have it give a better look at the finer details of the area.
In terms of detecting objects: anybody with any sort of computer vision background will probably be scoffing at drawing rectangles around objects that are moving within a stable image. However, at a broader scope, change detection and broad area detection are heavily research fields and are always of interest to whatever 3-letter agency that deals with the imagery.
So, not only does ARGUS solve problems, it also provides data for creating new problems. Cool little piece of hardware.
Bottom-up lighting on the face. Ominous, Hans-Zimmer-ish, military-industrial complex musical score. Dark lighting in the lab. Statements like: "It's important for people to know the truth." Even the title of the program: Rise of the Drones.
Result: Drones, engineers, and government projects are scary.
Not saying I agree or disagree with the sentiment, just that the bias is a bit over the top.
All I can think about here is some kind of big brother eye in the sky watching every subject go about their daily lives as soon as they leave their house.
/just going to be over here with my tin-foil hat...
They claimed they can stream 1.8 gigapixel video live or near real-time in the PBS video. Does anyone have an idea of how they can get that much bandwidth wirelessly ?
It's not absurd though for the military to have the capacity to bounce 600 gigabits/s around the globe. There are commercial sattelites with capacities beyond 110 gigabits/s
You'll notice that all the windows the guy brings up are simply video players with progress bars that show where in the video it displays. So that's how the real-time is done for PBS. :) All that wonderful dragging around of windows is standard touchscreen mouse cursor functionality. I wouldn't be surprised if the top-secret stuff is just a bunch of empty boxes and a beach ball covered by a tarp.
In a real-life scenario, it likely only streams enough pixels to show you the area(s) you're interested in. So you can take advantage of 1.8 gigapixels by selecting which 1M pixel area you wish to focus in on and that feed gets streamed to you. If you focus in on multiple areas simultaneously, it is likely to support as many 1M pixel streams as the bandwidth allows with degrading quality (lossy compression) beyond that.
I personally can't wait for them to use ARGUS and persistence stare in North St. Louis, East St. Louis, Chicago, Detroit, and all the other cities with extreme violent crime problems. That's one of the biggest reasons for why so many have moved out to the suburbs far far away from the cities. Look at your cities crime-map data and notice how crime completely destroys entire neighborhoods and surrounding areas. This new technology, if put to use to combat crime could revitalize downtowns and cities for hundreds of millions of people. Reversing suburban sprawl, "white-flight", and forest destruction.
People DO want to move inward, especially America's younger generations that grew up isolated in the suburbs where everything important and all their friends were miles apart. Young people in America are leading the urban renewal philosophy, and it's sad to see their efforts, and their businesses affected by crime.
I don't like the Privacy violation, but... With everything comes good and bad. Look at cars, consumerism, religion, etc... Things that were set up with good intentions, had some bad outcomes, and in the end were accepted when people saw both the good and bad as general progress.
«Panopticism is taken literally by at least one Air Force commander in Kandahar. ‘He
knows we’re there,’ Colonel Theodore Osowski told a reporter, referring to the Taliban,
‘and when we’re not there, he thinks we might be there’: Brian Mockenhaupt, ‘We’ve
seen the future and it’s unmanned’, Esquire, 14 October 2009.»
in: Dis/Ordering the Orient: scopic regimes and modern war by Derek Gregory
Can spot a "target" from 20k feet would be a more accurate headline. Also, if it can be that accurate for this, why can't they turn rocket drones into sniper drones by now? Or do they actually prefer the capability of killing the target and everyone around him at once?
> why can't they turn rocket drones into sniper drones
Because it's really really hard to hit someone with a ballistic projectile from a distance of 20k feet in 100 mph winds (or however fast a drone flies).
I don't like "collateral damage," but if this was a war, and if they needed to take out enemy without losing friendlies, they'd usually call in an artillery strike. This is way more accurate than that.
I understand your sentiment, but the historical context is that this is much cleaner than previous methods.
Notice that I said in war. I'm not sure how we piece together a moral system to deal with this kind of asymmetric enemy like terrorists/rebels. War is horrible. Is this lower grade conflict any better?
> why can't they turn rocket drones into sniper drones by now?
If I had to take a guess, it's because 20,000 feet is too much air for a ballistic object to go through without being nudged out of its intended trajectory.
Rockets can more-or-less correct their flight path, plus it doesn't matter if they're off by an inch, a foot, or a meter.
You might only get one shot out of one, but they can be quite powerful. (Considering when Real Genius was made, it's kinda prophetic. We even have a military mini shuttle around that size.)
While handy for use in civil areas, or low intensity conflicts, this will be useless in what we consider a "traditional" high intensity war. The airspace in such a war is non-permissive. Any loitering drones will be easily shot down, or jammed so that they have to return to base. GPS will be jammed, and any AD system worth its salt will be able to down these with ease. And if the enemy has fighter aircraft, it'll be a turkey shoot.
I'm running for office, and I have a database of the political registrations of houses. So I can build up and track the red or blue houses, and see which places people are going to vote.
I now have the data I need to perform a DOS attack on the most threatening poll places based upon raw, real time data.
Excellent idea if you get caught say you "misinterpreted the rules" that's what they claim when busted robo calling voters with the wrong time or polling location
I couldn't stop thinking about armed drones during the recent gun-control furor. I was wondering how it would change the debate if people were really thinking about it - about the near inevitability that they will be used domestically by government and non-governmental actors. The issue doesn't win clean points on either side, but it does seem to make a lot of today's fears about guns seem trivial.
In the past a dictator always had to depend on large numbers of other people to enforce his power. That means that if enough people are fed up with him, he will lose his power. It is really scary to imagine a future where a handful of people or even a single person can physically control the entire population. This is of course not a binary change, but a gradual one. Before guns, a dictator had to keep perhaps 30% of the population happy. With guns a much smaller number of people could control a much larger number. Autonomous weapon systems like drones are just the logical conclusion.
All the guns you can hold aren't going to help when the only potable water is rationed. Historically, poisoning the public water supply was suicide. Now that we can efficiently distribute bottled water for weeks, it's a feasible strategy to break a resistance.
Second amendment remedies will be rather... lacking when it comes to 24/7 surveillance by armed drones flying above the sky.
It's scary but increasingly inevitable that panopticon surveillance of all open physical spaces will happen. Having a rifle isn't going to do much to escape or counteract that.
Was about to post what a perversion that interpretation is of the 2nd amendment, but holy cow can't actually argue with that.
What a shame they wrote "free state", but then again, they were forming a new government, so what should we have expected?
And if actions speak louder than words, most of the founders who served as presidents didn't exactly "walk the walk" constitutionally when they held that position.
> It is really scary to imagine a future where a handful of people or even a single person can physically control the entire population.
What weaponry do you think would be necessary for a civilian population to successfully repel a US Marine battalion? Is it even reasonable to expect that a civilian population would possess anywhere near the appropriate level of training for such a situation?
Without major advances in artificial intelligence technology, drones have a key weakness: the wireless link. Miniaturization may win the drones arms race, which makes drones accessible to non-governmental actors, but when it comes to controlling the spectrum, bigger is better. So governmental actors are always going to have an advantage because they can build the biggest antennas attached to the biggest power sources.
Not a problem if you know where the target's going to be.
Imagine a swarm of explosive-carrying drones hitting the President's podium at inaguration from a dozen different directions. Known time, known position, open-air approaches. Must be something that keeps the Secret Service up at night.
This kind of drones are relatively slow. The perimeter around in the president in any exposed situation is pretty large. Any appearance of unidentified flying objects on the scene is likely to trigger an evacuation of any key targets. For the kind of payloads small drones can carry, and with no possibility of changing the target, you don't even need to move the target very far for the strike to be non-lethal.
On the other hand, you could build a cruise missile (essentially a drone) that might be big and fast enough to hit your target. But those date back to WWII, so if they keep the Secret Service up at night, it's a pretty exhausted security detail by now.
Designing a drone that operates purely based on inertial navigation is a good bit harder and more expensive than something that uses an off the shelf GPS device.
You don't necessarily need a wireless link. You can always use preprogrammed routes. Scouts can simply log data and return with it later, much like the spy planes of old using film cameras. Attacks can similarly be preprogrammed; bombers in war, for example, tend to have fixed targets.
It's the interceptors that would prove the most difficult to operate autonomously; they are the only one whose primary purpose is never fixed.
Rolling with the idea of a drone dystopia, as drones get smart enough you could even rely on the drone to make its own decisions and respond without an external operator.
One of the things that makes cheap drones possible is easily accessible, off-the-shelf, cheap and lightweight computer/GPS/wireless hardware. Losing GPS and having to use inertial guidance makes the problem a lot more expensive and complicated, and advanced AI is like manned missions to mars--always further off than it looks.
Ok, so you are specifically referring to GPS. I agree, GPS is important for cheap and easy navigation. I was presuming you were talking about a link to a base-station.
Of course, aren't all the GPS satellites already government satellites in the first place? Seems like they wouldn't need to jam the signal, they could just shut them down, or maybe GPS can be set to require authentication...
Once a roadside IED bomb was detected, they rewound the video, tracked the person who setup the bomb, then tracked that person back to the supplier for the IED/Financials. ODIN killed about ~3000 in Iraq and at least ~1000 in Afghanistan See: http://www.npr.org/templates/story/story.php?storyId=1142211...
Argus is now obsolete, they had a replacement RFP back in 2009. http://www.wired.com/dangerroom/2009/08/new-army-camera-prom...
Also a quick note, the aircraft had a 274 mbit/s uplink for streaming the live feeds.
Edit: Fixed 27/7