Somewhat terrifying to imagine what the military is going to do with coordinated drone swarms. Like a smart cluster bomb with no wasted explosions. I suppose, though, less unexploded munitions laying around afterwards.
Ahh, I don't see how that last thing follows. I expect many expired, disabled and crippled drones lying around.
Edit: I watched to video after reading your comment. That's fucking terrifying. It's hard to make scary "action" sci-fi nowadays, and China is working hard to make dystopian sci-fi look tame, too.
Make them smaller and give them a sharp pointy end. Ask the public to name it. Reject “Stabby McDroneface” as “it sounds like a joke”. Be surprised by the first amateur reproduction. Use the moral outrage to ban all drones — even those tiny little drones that can’t life a single jelly baby. Realise drones are still useful and make and expensive license scheme that doesn’t really scale up because it’s organised with industrial era thinking instead of information era thinking. And what is old is new again.
There's already been reports of the use of toy drones for scouting and bombing in the Syrian civil war. Then there was that strange maybe-fake bombing attempt on Maduro: https://www.bbc.co.uk/news/world-latin-america-45073385
This sort of tech will probably be used to make better cluster munitions (though I wouldn't be surprised if this technology is already used in them), but I don't think there's much tactical value in this kind of tight coordination for a drone swarm. In combat you wouldn't want your drones this close together, one missile could take out a bunch of them at once.
I always wonder about people who see some awful technology in a dystopian sci-fi flick and think to themselves "I want to be the guy that invents that." The ability of people to sprint headfirst toward a shittier world if it means they can make a few bucks is just depressing.
The most positive aspect of the drone aerial demo concept I've seen is that it's an alternative to firework displays that's silent, smokeless, generally safer, and re-usable.
Exploding chemicals setting things on fire if mishandled (anywhere from factory to actual show). As opposed to small drones falling? (Those are most likely micro-drones)
Panic and disorientation from fireworks noise can impact wildlife and domestic animals. Firework explosions do not last long enough for animals to become accustomed to the sound. I see the lack of explosion noise as a plus
This holds for humans too. I have a childhood friend who's dad was an Army Ranger in Vietnam. Literally the last guy on a hill once. He would always try to be a good dad and attend Fourth of July festivities. It wouldn't take long before he was shaking and sweating. As a former weapons officer myself, I give zero shits about fireworks and would love to see some drone acrobatics instead.
We condition our cuddly civilians not to freak out when they hear large explosions by having fireworks as a plausible cover. I don't know how you make all of society 'soldier safe'.
I saw these live, followed by real fireworks afterwards. Surprisingly, the fireworks paled in comparison (and this was a large professional fireworks show). Re the noise: In both cases there was music, but IMHO fireworks noise gets rather monotonous after a while.
They could be paired with wireless headphones distributed to attendees, similar to silent discos. You could even pipe music to the masses, along with explosion sfx synced with the drones.
It's surprising that GPS signals are that important to these displays. I figured they use GPS to provide a rough center of the display, for the "leader" drone, and the other drones would position themselves relative to their leader using some local peer-to-peer positioning signals.
GPS + accelerometers in fusion are accurate enough, and what you describe AFAIK doesn't exist in off-the-shelf variant, which is usually the core consideration.
Drones like this use DGPS. GPS is inherently inaccurate when you need centimeter level precision, especially with spinny bladed machines in formation. DGPS uses a ground station that helps the drones gain the accuracy they need, but it's much more complex than a simple GPS unit.
I suspect the specific implementation here uses "carrier-phase RTK" which means the recievers are locking into the pure sinusoid of the carrier frequency to get sub-centimeter positioning accuracy. It's mind blowing that this works at all, given that it really isn't how the GPS system was designed to be used.
The ones I saw worked just fine over a lake, and it was windy. I saw a few drift off and self-correct very quickly so it looks like they have this figured out. I assume they hold their position relative to one or more ground units.
As someone who knows a fair amount about the underlying mechanics behind how GPS signals work, it sounds like you are misinformed. That said, never wise to assume, so I’m happy to be proven wrong.
Especially this comment in regards to why Disney didn't take the 2016 trial at Disney Springs to completion at a larger park, like EPCOT:
"I was there when they did the Springs show. The techs told us the biggest issue the drones have is navigating over reflective surfaces (aka water). It messes with the GPS that designates their point in the formation. It would have been perfect if it wasn't for the lagoon."
I vaguely remember seeing videos shot from camera phones without the background music you find on youtube, and it wasn't silent. The high pitched sounds of the drone swarm was very clearly audible.
They almost certainly are safer in principle, but accidents happen. Due to safeguards and small dimensions and mass, I think the drones are safer, but they likely take just as much due diligence as traditional fireworks displays in order to reach parity safety levels.
I think you can simulate this all in advance and have each drone executing a pre-arranged gcode-like program. I'm guessing the complexity of choreography is similar to something like autorouting a circuit board.
Other comments mention that using GPS for relative positioning is fine. The largest source of error in GPS is ionospheric delays, which will affect all these drones equally, so I can believe that. Meanwhile, the flight computer handles figuring out how much power each motor needs to move or maintain its position. What if there is high wind on the day that you want to do this? Either a single drone can remain stationary in that amount of wind, or it can't; if it can't, then you cancel the performance because there is no computer program you can write to suddenly give the drone more control authority.
I doubt there is any additional complexity here; I doubt the drones are looking out for other drones and coordinating with each other. They just go to a set position at a set time at a set speed. Any complexity lies in the positioning; whether GPS + barometric altimeter + IMS (gyro/accelerometer) is enough to get the desired accuracy. It probably is.
Differential GPS is needed to achieve this accuracy. At some point in the future, vision systems + coded lighting may suffice, but for now, more elaborate tech is needed to keep these machines in tight formation.
I don't think differential GPS does anything here. It corrects for large-scale ionospheric delay error, which is going to be exactly the same for every drone at this scale. And they have WAAS receivers anyway, which provides better accuracy than differential GPS. (Maybe you mean RTK, which could help achieve centimeter accuracy.)
This is for sure done programmatically, as you might've guessed. And I'd guess they've built a software that models these animations and treats each drone as a node in a plot and then drones move independent of other drones since their routes are already plotted for them by the software.
So yes engineering wise this is amazing, both software and hardware, but the concept is not that complex.
Think of it as SwarmTouch but without the human "puppeteering" the drones ..
Don't know about these drones specifically, but off the top of my head there are a variety of different navigation systems you could use to supplement GPS for extra precision.
1. Inertial Navigation: basically taking measurements of your speed and direction of travel and using that information to update your position. This is not very accurate, and the longer you go without some other guidance system the less accurate it will get, so I'm not sure how useful it would be here. But it's fairly easy to implement, you just need to give each drone the ability to measure its speed and direction, and you would probably want that to correct for wind anyway.
2. Terrestrial guidance, where you used some fixed landmark or set of landmarks and look at your position relative to it, which can be done with cameras or radar or lidar. Probably too expensive to use radar or lidar, and I'd worry about interference between drones.
3. Command-based guidance. Rather than having each drone try to figure out where it is, you have a separate system on the ground that looks at the drones and tells them where they are. Since you only need one of these command units, it wouldn't be too expensive to use radar or lidar, and no worries about interference.
4. Real-time kinematic GPS: kind of a mix of 3 and 4. You basically have some radio stations broadcasting, and the drones triangulate their position based on that.
Not sure what kind of drones these are, but if they are the Intel Falcon 8+, they have GPS + intertial built in, and they are meant for survey work, so they might have the Real-time kinematic thing also.
Here's an older video about the run-up to Intel's Drone 100 simultaneous attempt. It's the most comprehensive overview of the coordination that I have found. It shows some screenshots of their custom drone control software. In this generation they used 4 pilots for 100 drones; Intel currently holds the world record for simultaneous drones controlled by a single operator at 500 drones, so they have taken a couple steps since this video was made in January 2016.
Edit2: Found a screengrab of the so-called ControlCenter v0.1.6-3-gf79c79. It's pretty hard to read but it's there at 1:06. Also in that same Intel 500 drone video is a screengrab for some software by Ascending Technologies GmbH at 0:40, which Intel bought in 2016. They appear to have created the Falcon 8 drone, which Intel also acquired in the deal.
Assume each drone is $500. That’s $1M or so in total. If I were renting out these things for fireworks displays, I would probably want to recover a healthy return over my financing cost.
Imagine I could obtain money for 10% - $100K per year against the $1M inventory. Then assume the drones depreciate to zero in three years - that’s another $333K. Operating cost is per-event.
Assuming no cost for my time to operate the drones, the rental charge would have to be enough to generate $433K a year. Assume you can run 10 events per year. You thus have to charge $43K per event just to break even on the financing of the drones. Add in a few thousand for operating costs of each event and maybe its $50K?
I don’t know what fireworks cost, but it seems likely to be within the same range to me.
>Then assume the drones depreciate to zero in three years
I'm not entirely sure you can make this assumption. Other than the battery, which would have to be replaced in 3 years, the rest of the electronics should hold up as it's just consumer grade smartphone parts.
I guess the electric motors would be a strong replacement contender, but even then if you are buying in bulk it shouldn't hurt the budget too much.
Maybe I'm applying my hobby perspective to the industrial world, but I don't really see a drone as "broken" until it suffers a gnarly crash that breaks the frame. Everything until then can be swapped out in a few minutes, and even if the frame breaks you can transplant the rest of it to another frame. It's not like the smartphone world where everything is soldered shut to save space.
> I'm not entirely sure you can make this assumption. Other than the battery, which would have to be replaced in 3 years, the rest of the electronics should hold up as it's just consumer grade smartphone parts.
The drones not blowing up in 3 years just means you make extra money. If the cost is competitive when assuming the worst case, then that means you stand to make money if things go well.
> I'm not entirely sure you can make this assumption.
I agree with you, but not for the same reason. They could easily get more than 30 performances (aka 3 years using numbers in the example) in my experience. Probably in the hundreds per drone. But over the course of three years, they’ll have vastly improved HW designs so they’ll want to replace their primary fleet. That doesn’t mean the older models are now trash (aka worth $0), just that they won’t likely pull top dollar performances anymore, which is fine as there’s plenty of room for them to be used at less viewed events.
The scary thing is that this is using a not-very-amazing flight control system and if the better flight controller systems did this they wouldn't have to worry about stray drones or speeding up the video in post-production.
I'd imagine the software flying the drones would have to compensate for the downdraft of the nearby drones and the turbulence. It would be interesting to see the software side of things. Can't wait for the open source project.
I suspect it doesn't explicitly - provided it's got good enough positional information, it can just use all the traditional techniques of closed-loop control to get there and stay there.
How it gets the positional information isn't clear. Some sources say GPS, I'm guessing that relies on the relative error being very small and not caring about the absolute error.
see RTK GPS. uses a fixed ground station with a known position. the phase shift of the moving object’s gps signal is used to achieve centemeter-level precision.
ODESZA used what appears to be a simpler version this technology during their live performance at Coachella in 2018 [1]. It was definitely sponsored by HP/Intel.
why don't they start putting fireworks ON drones to let them explode in special directions while they are already flying in formation. you can take a lot of the excess propellant out, and pack much more of the shiny colored flaming stuff to make for one hell of a precision directed fireworks display. this is so damn obvious to me , but no ones thought to do it yet.
75 comments
[ 2.6 ms ] story [ 155 ms ] threadEdit: I watched to video after reading your comment. That's fucking terrifying. It's hard to make scary "action" sci-fi nowadays, and China is working hard to make dystopian sci-fi look tame, too.
I really don't look forward to any of that.
I think much safer is more precise.
Like with racing, the noise is a big part of the experience. And this comes from someone who gets really annoyed by loud motor vehicles.
The Black Hole Bomb and Black Hole Civilizations - Kurzgesagt: https://www.youtube.com/watch?v=ulCdoCfw-bY
Fireworks also work better in rainy/windy conditions.
Then again, I know nothing about this.
For swarms, drones need precise location relative to each other, for which there are other techniques than GPS.
I think you are speculating - or do you have information showing that these drones use DGPS for this display?
As someone who knows a fair amount about the underlying mechanics behind how GPS signals work, it sounds like you are misinformed. That said, never wise to assume, so I’m happy to be proven wrong.
https://twitter.com/wdwbusdriver/status/1184643020894953473
Especially this comment in regards to why Disney didn't take the 2016 trial at Disney Springs to completion at a larger park, like EPCOT:
"I was there when they did the Springs show. The techs told us the biggest issue the drones have is navigating over reflective surfaces (aka water). It messes with the GPS that designates their point in the formation. It would have been perfect if it wasn't for the lagoon."
https://techcrunch.com/2018/09/18/an-intel-drone-fell-on-my-...
Does anybody know of a blog post where this is explained? In particular, choreographing the movements of the swarm looks terrifyingly hard.
Other comments mention that using GPS for relative positioning is fine. The largest source of error in GPS is ionospheric delays, which will affect all these drones equally, so I can believe that. Meanwhile, the flight computer handles figuring out how much power each motor needs to move or maintain its position. What if there is high wind on the day that you want to do this? Either a single drone can remain stationary in that amount of wind, or it can't; if it can't, then you cancel the performance because there is no computer program you can write to suddenly give the drone more control authority.
I doubt there is any additional complexity here; I doubt the drones are looking out for other drones and coordinating with each other. They just go to a set position at a set time at a set speed. Any complexity lies in the positioning; whether GPS + barometric altimeter + IMS (gyro/accelerometer) is enough to get the desired accuracy. It probably is.
So yes engineering wise this is amazing, both software and hardware, but the concept is not that complex.
Think of it as SwarmTouch but without the human "puppeteering" the drones ..
https://www.hackster.io/news/swarmtouch-lets-one-person-cont...
1. Inertial Navigation: basically taking measurements of your speed and direction of travel and using that information to update your position. This is not very accurate, and the longer you go without some other guidance system the less accurate it will get, so I'm not sure how useful it would be here. But it's fairly easy to implement, you just need to give each drone the ability to measure its speed and direction, and you would probably want that to correct for wind anyway.
2. Terrestrial guidance, where you used some fixed landmark or set of landmarks and look at your position relative to it, which can be done with cameras or radar or lidar. Probably too expensive to use radar or lidar, and I'd worry about interference between drones.
3. Command-based guidance. Rather than having each drone try to figure out where it is, you have a separate system on the ground that looks at the drones and tells them where they are. Since you only need one of these command units, it wouldn't be too expensive to use radar or lidar, and no worries about interference.
4. Real-time kinematic GPS: kind of a mix of 3 and 4. You basically have some radio stations broadcasting, and the drones triangulate their position based on that.
Not sure what kind of drones these are, but if they are the Intel Falcon 8+, they have GPS + intertial built in, and they are meant for survey work, so they might have the Real-time kinematic thing also.
https://www.youtube.com/watch?v=eZ-js5zn-I0
Edit: found a link for Intel Mission Control. Download it and try it out yourself?
https://downloadcenter.intel.com/download/28715/Intel-Missio...
Edit2: Found a screengrab of the so-called ControlCenter v0.1.6-3-gf79c79. It's pretty hard to read but it's there at 1:06. Also in that same Intel 500 drone video is a screengrab for some software by Ascending Technologies GmbH at 0:40, which Intel bought in 2016. They appear to have created the Falcon 8 drone, which Intel also acquired in the deal.
https://www.youtube.com/watch?v=aOd4-T_p5fA
http://www.asctec.de/en/intel-acquires-ascending-technologie...
Imagine I could obtain money for 10% - $100K per year against the $1M inventory. Then assume the drones depreciate to zero in three years - that’s another $333K. Operating cost is per-event.
Assuming no cost for my time to operate the drones, the rental charge would have to be enough to generate $433K a year. Assume you can run 10 events per year. You thus have to charge $43K per event just to break even on the financing of the drones. Add in a few thousand for operating costs of each event and maybe its $50K?
I don’t know what fireworks cost, but it seems likely to be within the same range to me.
I'm not entirely sure you can make this assumption. Other than the battery, which would have to be replaced in 3 years, the rest of the electronics should hold up as it's just consumer grade smartphone parts.
I guess the electric motors would be a strong replacement contender, but even then if you are buying in bulk it shouldn't hurt the budget too much.
Maybe I'm applying my hobby perspective to the industrial world, but I don't really see a drone as "broken" until it suffers a gnarly crash that breaks the frame. Everything until then can be swapped out in a few minutes, and even if the frame breaks you can transplant the rest of it to another frame. It's not like the smartphone world where everything is soldered shut to save space.
The drones not blowing up in 3 years just means you make extra money. If the cost is competitive when assuming the worst case, then that means you stand to make money if things go well.
https://www.accountingtools.com/articles/2017/5/11/useful-li...
Consumer grade smartphones, of course, last as long as...uh, three years?
I mean, I'm pretty sure the lifespan of whichever Google phone I had that started rebooting and wouldn't stop was less than that.
I agree with you, but not for the same reason. They could easily get more than 30 performances (aka 3 years using numbers in the example) in my experience. Probably in the hundreds per drone. But over the course of three years, they’ll have vastly improved HW designs so they’ll want to replace their primary fleet. That doesn’t mean the older models are now trash (aka worth $0), just that they won’t likely pull top dollar performances anymore, which is fine as there’s plenty of room for them to be used at less viewed events.
How it gets the positional information isn't clear. Some sources say GPS, I'm guessing that relies on the relative error being very small and not caring about the absolute error.
Similar to how WAAS works for aircraft, it's GPS plus a ground "enhancer".
Shooting with what?
Lucky HN isn't to technical since it's kinda obvious.
1,500 drones
And more importantly speed up.
Source - https://youtu.be/KhDEEN4gcpI?t=110
[1] https://www.youtube.com/watch?v=OZi2cN7LEfM