Disposable Drones

How do you deliver medical supplies to a war zone cheaply? The answer, according to this project, might be to make a disposable drone. Created by friends of Hackaday [Star Simpson] and the Sky Machines group at Otherlab, this project is looking to make drones out of cheap biodegradable products like cardboard.

Rather than risk an expensive drone that might never return, the project imagines a drone that flies to the target, delivers its cargo with an accuracy of about 10 meters and then be easily disposed of. The prototype the team is working on is part of a DARPA project called Inbound, Controlled Air-Releasable Unrecoverable Systems (ICARUS) and is a glider designed to be released from a plane or helicopter. Using a cheap GPS receiver and controller, the drone then glides to the destination.

It’s an interesting take on the drone: making it so simple and cheap that you can use it once and throw it away. And if you want to get a feel for how [Star] and Otherlabs approach problems like this, check out the awesome talk that [Star] gave at our recent SuperConference on making beautiful circuit boards.

Thanks for the tip, [Adrian]!

81 thoughts on “Disposable Drones

  1. As opposed to dropping in cargo by even cheaper parachute? While I can see missions for this device, I doubt there would be enough of them to justify the sort of mass production that would bring the unit price down to something reasonable.

        1. Dropping supplies by parachute is totally feasible but it would lack the accuracy that a GPS guided, cardboard drone offers. Parachutes fall slowly and get caught in trees or power lines too.
          When someone’s life is in danger and they may be in a great deal of danger, dropping those supplies within 10 meters can matter an awful lot.

          1. For the third time in this thread: I did not say there were no missions for this device, only that there will not be enough to make these drones cheap by manufacturing them in volume.

        2. Why would they? GPS-modules don’t cost more than like 15$ in small quantities (and the GPS SOCs cost only around 3$, so if you’re comfortable in laying out simple RF PCBs), and the rest is just a microcontroller, MEMS-IMU, pressure sensor, some mosfets, some voltage regulators and a few passives. So the whole electronics of this thing shouldn’t cost more than 30-50$ in parts even as a one-off.

          1. Parachutes cost on the order of hundreds to thousands of dollars. Lets say you cut that cost in half to estimate actual production costs… and the cardboard drone as a one off is still coming in ahead.

          2. So a parachute is cheaper.
            Factor in the cost and operational cost of an aircraft to get there to drop a load by parachute.
            And in a war zone, the delivery aircraft is exposed. A drone can be launched from a safe stand-off distance.

            And it just needs to get there. It doesn’t need to be a stable camera platform, nor stabilized to allow human POV perceptions to fly it, etc.. No one would care if it yaws as it turns or doesn’t have hyper responses to turbulence.

          3. @Rando Calrisian – We are not comparing these with skydiving canopies. I strongly doubt that one required to drop a load in the same weight range as one of these drones would be far less expensive than what you are suggesting.

            @Canoe – I am not arguing that this platform is not without utility, what I am questioning is does the need come up enough to create a market (as it were) for the sort of mass production that would make this an economically viable option (thus widening their possible use). Please don’t confound this question with that of practicality – I am questioning the cost of disposability here only.

          4. Facts of life… two places that matter, inside the camp/compound and outside it. Whether camp belongs to UN, US, international observers, charity with private security or whatever. Anything outside you have a 10:1 shot of retrieving intact possibly at cost of personnel. UN mission or NATO or whoever maybe does have air superiority but nobody wants to screw around below a couple of thousand feet due to small arms fire or RPGs, that’s also why it won’t come in by helicopter, helicopter mission would be several ship with half a dozen gunships to soften up and suppress local area. Plus there’s an uncomfortable number of advanced ground to air missiles unaccounted for in the world, US and former soviet, stingers and their equivalents.

            So now, you wanna get something into a football pitch from 5000ft plus, more softly than just dropping it like a brick, because friendlies/neutrals are down there doing what they’re there for, how many parachutes do you think you should drop to have a 95% chance of getting one in there? Lots is going to be inevitable…. So, something a good percentage more accurate than a parachute can still cost more than a parachute and be worthwhile.

            Even if your guys fully control a wide swathe of territory, no interference in retrieval… the parachutes are essentially random, some percentage will drop in lakes, rivers, get stuck high in trees, and other inaccessible places. Plus may get spread over a wide area… you may not want your team on the ground to devote a whole freaking day to chasing around after them and rounding them all up. That’s also potentially fuel, rations and water wasted that will need to be re-supplied. Obviously this is going to be treed over or mountainous territory because if it was simple to use a low level skid drop or helo, then that would be used. Helis have some operational restrictions, range, altitude, certain shapes of territory reflect their own downwash and flip them, downwash on the leeside of hills/canyon walls etc can create no-go zones. Anyway, if the territory is difficult for low level drop, or dangling on a string from a chopper, then it’s going to be extra difficult to round up a bunch of widely spread parachutes from.

        3. gps <$10
          controller <$10
          two servos ~$4

          its the motor controller (mosfets,caps), actual motor and the battery that are expensive parts, still you can make 'wing with a propeller' RC model for <$50

          1. Lightweight parachute fabric is less than $5.00 a sq.yd. (probably less in quantity) – still less expensive.

            Look the claim here is that a disposable drone is a cost-effective way of delivering goods by air, and I question this assertion. If the mission is to deliver a high-value cargo with high accuracy then any off-the-shelf heavy lift drone can serve and be considered ‘disposable.’ If the mission is to deliver a lot of lower-value cargo to a general location, parachutes will serve. So we are looking at a somewhat narrow mission profile where there is both a need to deliver with high accuracy, and in volume such that the cost of the type of drones being considered here are competitive with parachute airdrops when the potential losses of cargo on the target doing it this way are factored in as unacceptable. My argument is that there are very few missions that fit that profile, such that scale would reduce manufacturing costs to the point where unit pricing would make this a competitive option.

            This is the crux of the matter: anything is disposable if the situation warrants it, but if disposability is being contemplated as a cost-saving feature then it has to be competitive with other disposable modes. I just don’t see that here in a large enough number of situations to make this viable.

        4. On cost? A GPS receiver is currently under $10. A servo for this is around $3. A chip to coordinate them and the board for it is between $3 & $5. So, it is true – it costs more than something that won’t work. Your solution is also more costly than wrapping the item in bubble wrap and chucking that out. Which also won’t be useful.

          Come up with an idea that provides a significant standoff distance and less than 30 foot delivery radius.

          How about a 3 mile range from a 1 mile release height into an 8mph wind?

          1. The question I have is with the emphasis put on the disposable airframe aspect of this which is obviously an attempt to reduce costs. However whatever is saved is going to be taken up with navigation and control expenses and a limited payload for the sorts of range that the stand-off aspect becomes useful. Everybody keeps answering my question on cost-effectiveness by listing the technical advantages, which I am not arguing. But show me that the number of potential missions where this mode is needed to deliver the volumes of cargo that would justify the cost over chutes. No one can, because critical missions are few enough that the cost of the vehicle is not an concern and if targeting isn’t as important, standard airdrops suffice.

            As usual on these pages, everyone gets caught up in how cool something is and fails to look at the practical aspects critically.

        5. Not just the GPS and controller. Don’t forget the power supply and some sort of actuator. Even if muscle-wire were used, that requires more power.

          Still, this might be VERY useful. TFA mentioned dropping medical supplies to an area under quarantine. Even a $10 BOM (which might be possible in a few years) would be a deal-killer for Amazon dropping off a $10 phone case, but would be great for emergency supplies.

          1. Depending on the available load-factor. In other words: how much weight/volume can be delivered over what range. Again for special missions, cost is not a factor, but if the idea is to deploy these in mass (such that being cheap and disposable is a benefit) then they indeed have to be inexpensive. As described, I cannot see them carrying significant load, or given the internal equipment needed, all that low-cost.

    1. DARPA is funding this project so there are obvious military applications. But I doubt its ability to effectively deliver medical supplies. For that parachute delivery makes far more sense. Unfortunately medical supplies are either rather bulky like vacuum sealed bandages, or heavy like any IV fluid. A single liter of IV fluid weights one kilo, and a single patient could need multiple bags of IV fluid.

        1. That is actually a great idea.

          No more $100,000 missiles wasted on taking out one target and his 20 closest. Smaller ‘blast’ means less collateral damage.

          Very similar to the balloon or bat bombs of WWII, more precise.

    2. It needs to deliver in various wind conditions; the problem with parachutes is they have limited up-wind capacity or range. I would guess that the goal is to keep the main delivery vehicle well back from those who might shoot at it. Parachutes are also easily entangled in trees and brush, making recovery difficult.

      1. I didn’t say there wasn’t conditions where this mode would be better, only that the claim of being inexpensive was questionable. As for this type of craft doing much to solve the issues you mentioned I suspect that there will be very limited circumstances where it would make a difference.

        1. You must be aware that no matter how cheap an item is, if it doesn’t deliver, it costs too much.

          As long as there are -any- circumstances where this is better there is no reason to develop both systems. There would have to be some circumstances where this is notably worse to balance that out.

          This design is also amenable to adding a motor and battery to extend the range; something that is not possible with a simple parachute. This design is also much cheaper to assemble than a guided parachute.

    3. I would imagine the advantage over parachute or other air drop is precision. A lot of times, vital supplies are captured by the local warlord rather than the people who most need them. Then either sold to them for a high price if they have any money, or sold elsewhere. Also “the mob” in general can be a problem. Magic western pills from the sky can cure anything. They’ll be swallowing handfuls of malaria treatment for their gammy shoulder if they lay hands on them.

        1. Undereducated, desperate and fearful …. and lack of adequate nourishment doesn’t do anything for critical thinking either.

          Also “the mob” in mentality will exist in 1st world countries a couple of weeks after SHTF, looters find a case of tetracycline they’ll be popping it as soon as they get the sniffles.

    4. Parachutes work great for dropping needed supplies out in the open, right in the middle of the enemy’s sight line.

      see also: if the glide ratio is good enough its even feasible to fly it several miles from where it was dropped, avoiding having to fly into an area covered by Anti-Aircraft or Radar systems. (it may be hard to detect considering the airframe is a non-metallic material and the board could potentially be very small.)

          1. From a Hacker in their garage standpoint this is a really cool project.
            Research is great but there are some experiments you don’t need to do to find out how feasible they are.
            Efficient research isn’t about finding answers to random questions, it’s about finding answers to the -right- questions. There’s enough data out there to run the numbers on this project without building a single drone/glider.
            How many medical emergencies happen where delivery drone is the best way to get medical supplies to them? Someone somewhere has this number. I assume since DARPA is involved they think this number is significant enough to warrant a project.

          2. The thing about research is that you don’t get to know what the right question is in advance of finding the answers. DARPA’s reason for being is to push boundaries on things that people don’t think are possible.

          3. >>The thing about research is that you don’t get to know what the right question

            True but you can frequently discern if it’s a good question (note I said ‘efficient research’ sure you can just throw stuff at a wall & see what sticks, but that’s not efficient). Grant proposals have to spell out the problem they’re trying to solve. ‘Deliver supplies to isolated groups in need on a biodegradable platform’ is a bit broad and answerable w/o working prototype. This problem reads more like a naiive kickstarter or cool senior design thesis than a government funded project. It’s like when people use a 3D printer to print out a $0.50 gear available from any parts catalog, sure it’s a great exercise but it seems like they’re (DARPA) viewing this problem as a nail to hit with a drone hammer.

            ‘Can we use drones to deliver supplies’ – Resounding yes.
            ‘Can they be made out of cost effective biodegradable parts’ – Easily answered by a spread sheet.
            ‘Are drones the most effective delivery mechanism’ – depends on the situation, again, something the bean counters can answer in a spreadsheet.

            This is a cool project with interesting restrictions (I know my harsh criticisms may seems as though I don’t think so), but IMHO not something that needs a grant to develop (though, you’d be a fool to turn down someone else funding your hobby). Building a GPS drone as a hobbyist is still novel, but GPS guided aircraft are old hat for the military.

          4. Leithoa: The 2nd and 3rd of your questions can’t actually be answered without doing the things and finding out. A spreadsheet analysis can give you a model of how you think the system will work, but without actually testing the model, you don’t know the answer. This is -why- DARPA uses the “just throw stuff at a wall & see what sticks” approach to funding projects for defined use cases like this.

          5. @darren
            Sure they can. The glider design isn’t new. People have been flying wing designs for decades built from foam. We know it works. There are dozens of other similar, proven, designs that fly just as well. There’s no R&D behind the aircraft design. All you need to do is calculate the amount of materials needed & see if they have biodegradable substitutes.

            And you certainly can figure out if this is the most cost effective way to resupply.
            This problem didn’t arise in a vacuum there are real world scenarios they envision using it in. Factoring in supply weight, manpower, firepower needed to complete the mission & probably a few other things will easily tell you if this is the best solution. Historic review of the highest casualty or most resource intensive resupply missions would be another way to compare this solution to current ones. Then frequency analysis, how often will 1-2kg of supplies make the difference in a battle zone? If you have to send 30 gliders to get the job done it’s not really competitive to just rolling in hot with the cavalry, tipping them out of a helicopter, or dropping supplies by parachute.

            The analysis isn’t that hard if you have the data. If they don’t have the data, why is ICARUS even being funded?

        1. Haha. DV82XL is absolutely correct, but doesn’t realize he’s trying to herd cats. I find it funny because I suffer fools too.

          For a given payload, a parachute will always be cheaper than a stiffer version of that amount of matterial using significantly more labor to build and attach actuators and electronics.

          Also LOVE that these people developed a smart bomb err unrecoverable payload delivery device for DARPA.

        1. You don’t even need a servo: rudder control with a solenoid only, full right, full left un energized middle position. PWM’ing could give even more precise control. Or go full 1940’s RC style: Use a wound rubber band to supply the torque to the rudder and only have the solenoid control the escapement mechanism like in a grandfather clock. Then it could fly off with a button cell as only power source.

          1. Yah the solenoid is engaged very briefly to skip to next position. In the later iterations with superhet transistor receivers, late 60s say, you could get a few flights out of a 150mA 3.6V nicad pack, though popular name for them then was DEACs after a brand name, if you’re looking at the old stuff. I was screwing around with this in the late 80s for lightweight miniature RC models, because submicro receiver and servos, even just for 2 channel was something like a months wages at the time.

            Button cell might not like peak current though, so might have to have a capacitor in there.

    5. This is a research project – there are already self-steering parachute systems in use (JPADS system – link below) that have pretty good accuracy and glide ratio. They were supposed to be able to have 100m target accuracy but the test data I’ve seen shows they’re close-but-not-quite-there despite what the Wikipedia article says. The final purpose of this kind of thing is to avoid getting the delivery aircraft shot down since they can drop with accuracy at 20,000′ or higher.


      Good article on the general system and astounding acronym glossary [pdf]: http://faculty.nps.edu/oayakime/ADSC/PATCAD%20-%20Benney%20-%20The%20New%20Military%20Applications%20of%20Precision%20Airdrop%20Systems.pdf

      My guess is that they’re having a look at the economic/performance space for the gliders – is it a useful platform for small, targeted weapons, swarm surveillance cameras and the like. I doubt they’ll be dropping off your Chipotle order anytime soon, particularly in bad weather.

      FWIW I built a series of corrugated R/C gliders many years ago using the deeply-cambered Jedelski airfoil and they performed wonderfully (for a limited period of time).

  2. flitetest ( http://www.flitetest.com/ ) has been making RC planes from foam for sometime now. A great and low cost way to learn about flight, physics, planes e.t.c. Once a Chassis is busted up and is unsalvagable, you could always reuse the electronics. I see no reason why one can’t put a pixhawk or arduiPilot board in one of their larger foam plane models and turn it into a low cost foam drone.


    1. “out of cheap biodegradable products like cardboard.”
      “and then be easily disposed of.”

      You can decide whether this project is truly about “medical supplies to a war zone” and they think biodegradable trash is a priority in said war zone, or if it’s really more like “spy supplies to secret agents/special forces behind lines/whoever”, and those are code for “ability to dispose of the evidence easily and without the stench of burning plastic”, but either way, while foamies are cheap enough, the disposal issue is a reason to use cardboard instead.

  3. i think some of the cost cutting will involve replacing servos for something cheaper (not that you cant get a servo for less than $2 these days). ive been wondering if you could use muscle wire to actuate control surfaces. single transistor for control via pwm, feedback directly from a gyro ic.

    i did some science ™ on some coil based actuators and they worked well and only used up a little wire, epoxy, and a small neodymium magnet. the problem was it was a little power hungry. im curious if muscle wire would prove a better no budget actuator.

    the plane here looks like it just has a couple of elevons. im curious if you can do a plane that can maintain a natrual glide slope on its own and you could just have a rudder, or ailerons on a single actuator. you could shorten or extend range with s turns to help bleed off altitude. its worth looking into at least.

    1. Galloping ghost multifunction off one rotary actuator :D

      Actually would probably be more workable than the old button mashing method of controlling it, presuming the onboard smarts had some positional feedback.

      The way it works is to have basically a crank acting on yokes to both rudder and elevator, but could use mechanical mixer to do same for elevons. Control asserted by holding it in the position for longer, for instance 12 o clock of the crank would be up elevator, 3 o clock right turn or roll, 6 o clock down elevator, 9 o clock left turn or roll…. with cheapy cheap not many pole motor you’d probably still have a few intermediates you could hit as well, because you’d probably want slight up elevator in the roll to turn with ailerons.

    2. Given that what you care about is the torque/force on the elevon rather than the angle of it, you’d probably just use a cheap-as-possible DC motor/gearbox and drive it through a H-bridge. To return to center you would just cut power and let the flaps self-center.

  4. For mass production I would have thought that a starch based foam construction would have been optimal. You only need two major parts, the entire body is the largest followed by the payload container cover and finally the rudder. If it is a glider it does not need any other control as it can just spiral down to the target.

    1. Mold them from the same stuff those water soluble packing peanuts are made of, but with a thicker skin that’s coated with something water resistant for deploying in inclement weather. Put the electronic module and a pair of micro servos on a small PCB. Mount it with plastic barbs stabbed through the body of the wing. Control rods could be cheap, unplated steel wire that will quickly rust away when tossed. Load cargo into the compartment, seal with moistened strips of gelatine film.

      To retrieve contents, break it open, break it up and scatter the pieces. For faster disposal dump in a body of water. In dry country? Pile up the chunks and pee on them. Save the electronics module for re-use or smash it with a rock so it can’t be used for nefarious purposes.

      1. You only need one servo, for a rudder, because as a glider it just needs to aim a the target GPS coordinates and if it is high to orbit that location until it has lost altitude, i.e. spiral down. You can have the rudder and the payload lid as one part too if you are smart about the design, just rotate the entire thing. if the body-wing incorporates vertical stabilisers the rudder can be in such an atypical position. Even the servo concept should not be sacred, how many positions does a rudder need to be in anyway, three? Left centre right, or only two because 45 right is just 360-45 left. Analogue controls are just convenient for flesh blobs, silicon has other options. Just pulse the rudder position between two positions centre and left and with the right control software you can still point the thing at the target. Loft the foam with hydrogen, if you are brave, and gain a bigger payload/range value, then set it on fire to make it decompose almost instantly when it’s job is over. However I suspect that not having them disposable could bring much joy to the children of that war/disaster zone.

  5. I would still use elevons, only cost is one more servo and much better control and you don’t need a vertical stabilizer as I fly my wings all the time without fins or vertical stabilizers which means you can stack and store these wings without wasting space and worrying about broken fins.

    1. Elevons for sure on the wing shown. Just a rudder would leave no way to control the attitude or speed. CG is touchy on the wings.
      This operation needs to be smart too. Just add a little wind or weather and the target becomes much harder to hit.
      I love the assumption that this needs to be cheap because it’s disposable and biodegradable because of the environment. It just needs to go away so none of it can be reused. Cost is not object for the military. Think high tech, self destructing, mission impossible, flying tape player. Oh, and here is a band-aid.

  6. Screw the electronics, what you want is pigeon guidance…

    Then if homing pigeons used, you can sent a message back saying “Got it, send regards to me mum.”

    But also an operating mode that comes to mind is, carry the pigeon, onboard sensor data or cameras record to a touch memory, serial memory or other easily connected/disconnected device on pigeons leg, and when craft done, release pigeon to return home with the intel.

  7. Hey there, It seems a lot of the discussion on here is the value of this by way of $ and v’s other current deliver means such as parachutes. As a ex-paratrooper [with the Australian military], who has been working in the aid sector on and off for the about the last couple of decades and who has an interest in this type of thing. Here’s my two cents, per unit cost for the airframe is only going to be a little more than a couple of pizza boxes – the folding will be the biggest capital input. Electronics would come down to a few dollars – my guess would be <$10-15 at most. Even at 5 times this amount it it far less expensive than a parachute deliver system (canopies are not made from a meter of clothe as a comment above inferred, but what is referred to as silk, not inexpensive and again, you have the packing costs that isn't cheap either. But as we are discussing targeted deliver, the real current equivalent would be a parafoil which are used, but while able to be sized up or down cost much more than this delivery system. And yes, they require electronics as well; and don't like wind gusts. The very low cost option are little more than cardboard tubes with a four extensions at the top that make a cross and rotate on the way down that reduces the velocity a bit. Or more commonly, that just drag out a palate of bags of grain and they free fall to the ground. But neither of these last ones nor parachutes are very accurate so you need a larger DZ. or Safe for those on the ground. And expose the aircraft to threats as they need to fly a predictable path at low level. Also it seems the target applications are very different. This – at this stage, is for smaller payloads.

Leave a Reply

Please be kind and respectful to help make the comments section excellent. (Comment Policy)

This site uses Akismet to reduce spam. Learn how your comment data is processed.