Safety Systems For Stopping An Uncontrolled Drone Crash

We spend a lot of time here at Hackaday talking about drone incidents and today we’re looking into the hazard of operating in areas where people are present. Accidents happen, and a whether it’s a catastrophic failure or just a dead battery pack, the chance of a multi-rotor aircraft crashing down onto people below is a real and persistent hazard. For amateur fliers, operating over crowds of people is simply banned, but there are cases where professionally-piloted dones are flying near crowds of people and other safety measures need to be considered.

We saw a skier narrowly missed by a falling camera drone in 2015, and a couple weeks back there was news of a postal drone trial in Switzerland being halted after a parachute system failed. When a multirotor somehow fails while in flight it represents a multi-kilogram flying weapon widow-maker equipped with spinning blades, how does it make it to the ground in as safe a manner as possible? Does it fall in uncontrolled flight, or does it activate a failsafe technology and retain some form of control as it descends?

Is One In Three Thousand An Acceptable Crash Rate?

With customers demanding faster delivery and low or no-cost delivery the industry is scrambling to develop delivery systems that undercut competitors. The Swiss postal service — appropriately named Swiss Post — are testing autonomous multirotors operated by the drone delivery startup Matternet. The design includes a parachute that can be deployed to slow an uncontrolled fall. But spinning rotors and string-like items don’t mix well. In their testing, a failure occurred almost immediately following deployment when the line securing the parachute to the machine caught in a rotor and broke/

In 3000 flights they had experienced only one other failure, when a machine deployed its parachute successfully and descended into Lake Zurich. Swiss Post are now insisting on some upgrades to the system including two lines to tether the parachute instead of just one, and metal reinforcement for the parachute lines. It’s unclear whether Matternet have developed their own parachute system or whether they used one off-the-shelf, but the incident provides a rare public examination of the technology.

Save the Drone: Parachutes and Lost Propeller Operation

There are multiple manufacturers offering parachute systems for existing multirotors, either triggered through the machine’s firmware or through their own on-board sensors. Just one example is SkyCat, whose spring-loaded parachute canon is demonstrated in this video. It’s interesting to browse their various sites and observe the different marketing tacks for the different communities, for professional systems the emphasis is on safety when flying over crowds while for the enthusiast flier the emphasis is on protecting the valuable machine from damage.

Of course, a parachute is not the only game in town when it comes to multirotor safety. There are systems that can keep a multirotor flying without a propeller. Clever algorithms in the flight controller can detect when a motor fails or a propeller is lost, and use the remaining motors to induce a spin and bring the craft to ground with some stability.

Normally, multirotor aircraft rely on the force from one rotor to keep that side of the aircraft level. Losing one propeller upsets a delicate balance between all rotors, but equilibrium can be regained in an emergency situation by using extreme yaw so that the force of the remaining propellers is be averaged over all sides of the aircraft. A team at ETH Zurich pioneered this technique. It has been used to great effect by the drone show provider Verity Studios who have produced a series of videos showing it in action. Even when the machine is assaulted in flight with a piece of timber and loses a rotor it regains level flight under some control and is able to descend safely.

We Have To Take This Technology Seriously

At the professional level, these technologies are essential for craft flying in proximity to an audience. Unfortunately these systems are absent from the hobby side of things. We are unaware of any open source or otherwise free drone firmwares that contain something like the ETZ algorithms, and we can’t find any parachute failsafe systems that we might have featured in the past.

Perhaps this doesn’t matter for a casual user with a small drone and an eye to safe flying rules, but should it be something that we think about? Those serious in the hobby are operating thousands of dollars worth of equipment worth protecting from a sudden and unexpected meeting with the cold hard earth. And we exist as never before in a hostile environment where the drone is the subject of a questionable moral panic, and anything to make flight safer and mitigate fear can only be a good thing. As always the comments are open.

32 thoughts on “Safety Systems For Stopping An Uncontrolled Drone Crash

  1. It is not a “flying weapon”. Nothing is a weapon until it is used with intent to harm. Are you using a weapon to pound a nail in? No, it’s a hammer. If you use it to pound someone’s head in, THEN it’s a weapon. Get it right!

        1. So close… but still a weapon. Just like an assault rifle is a weapon even not when in use. Besides, the author said it “represents” a flying weapon. So not a literal weapon, but rather an object that can bring harm, although unintentional.

          1. So is a knife a weapon? I mean I use it to cook but I could just as easily use it to stab someone. Is it ‘not in use’ when I’m cooking because I’m certainly using it just not to harm others.

            The same is true of your “assault rifle” example (I’ll give you the benefit of the doubt that you meant the M16 or other which has a select fire and not the AR15 which is objectively not an assault rifle), yes I could use it to kill people, maybe shoot a fox to defend my farm, or I could use it to punch holes in beaver damn that’s flooding a local town (probably not the most effective tool for the job but one that would achieve it’s goal). In that last example is it a tool, a weapon, is it a weapon being used as a tool?

            As for the nuclear bomb example. https://en.wikipedia.org/wiki/Nuclear_Explosions_for_the_National_Economy It was definitely researched for peaceful applications for the device, it in fact was used a few times to put out natural gas fires. https://www.world-nuclear.org/information-library/non-power-nuclear-applications/industry/peaceful-nuclear-explosions.aspx

            So again I ask, is my chefs knife a weapon? Or does it depend on the application which is exactly what the original poster was saying.

          2. @anon My point was that the intent behind how an item is being used at a given time is not the sole determiner of whether something is a weapon or not. And also that there really is no point in nitpicking given the context of the article. It’s safe to assume most people understand the article.

            An assault rifle was designed specifically for the purpose of bringing harm to living beings. You can use the weapon as a tool, but it is still a weapon. A bayonet was designed to be a weapon. You can use the bayonet as a tool, but it is still a weapon. Of course you can use a tool (or almost anything else) as a weapon. So yes, sometimes the intent is used to determine if something is/was a weapon, but not always. In conclusion; “Nothing is a weapon until it is used with intent to harm.” is incorrect.

        2. Respectivey Project Plowshare and Project Sea Dragon.
          And I agree. Drones, hammers, knives are not weapons, they’re tools that can be used as weapons.
          Aso in my native language “a weapon” is literaly transated to english as “something that defends. So weapons are not ingerenty bad things. Bad people are.

        3. Respectively Project Plowshare and Project Sea Dragon.
          I agree that objects like drones, knives or hammers can be use as weapons, but let’s not forget that they are far better TOOLS.
          Also weapons are not a bad thing. Eg. i n my native language “weapon” is “broń” (litterally something that defends).
          These items are not bad. Bad people are.

        1. Kind of missed the point though. ““Nothing is a weapon until it is used with intent to harm.””

          Something can be a weapon without being used.* It’s intent as a weapon is clear in it’s design. In other words an extreme example to make a subtle point.

          *MAD

    1. You have a point that this is not a superb choice of words. In changing it, it’s really difficult to figure out what to call it. I went with “widow maker” since it describes an objects passive potential to cause harm.

      1. Still overly negative. People aren’t describing Camrys as widow-makers, yet they have far greater history of actually being involved in people’s deaths. While the military does have drones, the weapon systems onboard are what cause destruction. Mounting a SAM to a moped doesn’t make the moped bad (poor example, as mopeds are inherently bad), it would simply be acting as a platform. Additionally, comparing a military drone that is the size of a manned plane, with that of recreational or commercial drones would be like comparing a toy car to an MRAP.

          1. A cruise missile is a rocket with a warhead. We use rockets for scientific research all the time. Some of the first scientific rockets were captured V2s used for atmospheric research.

      2. I’ve always understood “widowmaker” to be something which is really dangerous even when used properly, and in particular a mains lead with a 13A (male) plug on both ends, typically used to inject power into a circuit when the supply is dead (e.g. from another phase or next door).
        It’s dangerous firstly because when you plug it in, you’ve got a plug with exposed mains on the pins. Ah, you say, I know what I’m doing and I’ll stay safe. Yes, but the reason it’s called a widowmaker is:
        Usually in this situation a utility engineer will come out to fix something. They’ll safely isolate the supply side, then unexpectedly encounter mains voltage on the consumer side which shouldn’t be possible.

        So I propose “whirly blades of death from the sky” as the official term for a falling multirotor.

        1. A friend of mine was a lumberjack.
          Yes, large and strong, and an all around great guy.
          “was”
          A widowmaker is a tree that drops a large branch when it is being cut down.
          He left a widow and a couple of kids.

  2. My half-baked idea would be airbags to slow down the fall and cushion the landing. Independent power source and instrumentation, perhaps with a gas cartridge to avoid omg explosives on board.

    1. Like the mars rovers. I think it makes sense. Two issues I can think of: Accidentally setting off the airbags while in proximity to the ground/people (and the accompanied flying shrapnel) and the penalty to payload of carrying such a safety feature. Still, i think we should investigate the possibility.

  3. There are so many corner cases, I don’t think there is a way to prevent all injuries from a multi copter in flight. So much energy in motion.

    Not saying they shouldn’t try, maybe there will eventually be an acceptable risk level.

  4. Reminder that we’ve had consumer radio-controlled helicopters since the 70’s with no serious incidents–it’s not new. Most drone safety hysteria is just neurotic fear driven by sensationalism and won’t ever actually be assuaged by any possible safety mechanisms or design. It’s largely psychosomatic. It won’t ever be safe enough for laypeople fearing something new and novel, even if it’s far safer than most everyday activities already.

    1. There have been deaths resulting from high power RC helis, just google search “death rc heli”. But I agree with your point as to the hysteria surrounding drones. Unfortunately, ignorance is rampant these days.

    2. radio controlled helicopters in the 70s (And up to the late 2000s) where expensive and difficult to fly. People that controlled them usually did so far way from obstacles and people, most often at their local RC clubs field. Few people were silly enough to fly them over large crowds like outdoor concerts/festivals. When accidents happened, few people will have ever seen or heard of one flying near them, so generally people didn’t care much. However with drones, they’re MUCH more common. Many people have seen them fly, a lot more people actually own one. The barrier to entry (both financial and in skills) is MUCH much lower. Flying a drone overhead of people (Roughly a 45 degree cone overhead) is just NOT safe. It will never be safe. It should not be done. Below that, people should either be aware of the drone and able to get out of the way, or people should be far enough away (and this is much further than you think) or protected by something. However, that is a lesson that will probably have to be learned in blood. Like most of the lessons that have been learned in aviation.

      Drone safety is NOT hysteria. It IS an issue and few people have put proper thought into it. The awareness is there in most operators of very large and heavy professional drones. However, there is a lot of people out there that just chuck a DJI Mavic or a Phantom up and pay no attention to their flight path or what is beneath them. And all it will take for this to blow up in everyones faces (including the actually responsible operators) is for an idiot to drop his Phantom drone into a crowd of festival-goers and maiming or killing someone.

  5. The ultra risk adverse society in which people are growing into sucks the fun out of everything…
    You assume a risk every time you get out of bed in the morning, that is if you survived the night – because you have an 8 times higher risk of dying to your own bedsheets strangling you in your sleep than to dying of a terrorist attack. SO, are you all going to throw away those 800 thread count murder machines you’re sleeping with every night vs. banning religions of people that you don’t understand that are 99.999% peaceful? You have to calculate the risks in everything, mitigate what you can, understand the residual risks, and determine whether those leftover risks are acceptable. I for one am keeping my bedsheets and don’t see the point in attacking entire sects of people due to the actions of a minute few. If that were the case we ALL should have been wiped out long ago…
    So going all belligerent over drones, which are not going away, is silly. It’s still an infant technology that is only going to get better. Its no secret that general aviation has a major safety problem for years and years, but no general aviation pilot will own that. And some people are trying to lower the dollar and training threshold for small aircraft. But I see small planes flying every day and don’t bat an eye.
    I also see some old guys on here whining about RC planes and how drones have crashed their party and that drone pilots lack the skills they have. I’ll tell you what, go and buy a 5 inch freestyle quad copter and tell me how “easy mode” it is to fly… I’ll grant you the DJI and knock offs can be easy to fly, but most of those companies are trying to make things safer too.
    But TLDR; we can’t go backwards just because of some residual risks!

  6. Now for the facts… The last fatality anywhere across the globe due to a model aircraft was 6 years ago – 2013. Roman Pirozek Jr. was fatally struck by his own RC helicopter in a freak accident. There has yet to be any fatality from a multirotor (drone). The notion that drones are “flying weapons” or “widow-makers” is ridiculous based upon real world statistics. Instead, drones are actually the safest aircraft in the skies bar none! By comparison, helicopters account for an annual average of 20 fatalities and millions of dollars in property loss in the United States alone. Again, that is ANNUALLY. And the helicopter fatalities are a drop in the bucket compared to the numbers of fixed wing aircraft (Cessna, Piper, Jet) each any every year. If your goal is to “mitigate fear” then your article is doing exactly the opposite. You are stoking fear and calling for safety measures where there isn’t a problem. Again, let it sink in, there has NEVER been a fatality from a multirotor (drone) and the last model aircraft fatality anywhere was 6 years ago.

  7. Think a lot of people are really missing the point. It’s not whether or not a drone can kill, or how often. It’s about there being some real potential for serious injury or death, which could have been prevented. I’m sure some comment-Nazi, will flag this, like pretty much all my attempt to share my thoughts and opinions here. Not sure what they are afraid of, maybe the truth? People go to baseball games, lot of fans. Pretty much every year, someone gets seriously hurt, even killed, by a foul ball. Most of the time, people are paying attention, and trying to catch a souvenir. Seems, like there are more injuries now, than when I was a youngster, and we didn’t have the netting…

    YouTube is full of drone stunts, that could easily have gone badly for someone, in the wrong place, wrong time. I’ve never tried, but mine is suppose to go 40 MPH, and it’s sort of low-end, old and outdated. I keep it slow, since I’m recording video, the whole flight, rather not just zip by everything in a blur. That’s just me, there are many others obsessed with speed, range, altitude, even though we are supposed to stay under 400 feet, and have line of sight. My eyesight might not be as good, but 600-700 feet, is a very small dot, to keep in sight. Since most drones have a screen these days, many fly FPV, even though that wasn’t the intent of including a live view. The screen is used as a viewfinder, usually some slight lag as well. Tend to believe there are quite a few people, who don’t follow the rules, and use the hardware to their limits.

    I believe any added safety features, will mostly just encourage more disregard for the safety rules we already have. I’d rather see better lenses, a motorized zoom, would make it safer, since there would be less need to get in close. Would spook the wildlife as much either. Longer battery life, would slow people down a little, still get to their destination, and have plenty of time filming. Don’t see any way to stop people flying FPV from almost a mile away though…

  8. I dislike the parachute idea very much. You are attaching a device that *should* react to an impending crash, where for one I’d say you can better put the energy in actually preventing the crash. So, use quality parts, get your operations in order, and above all test test test in a safe environment. But, I actually think a parachute hurts safety.
    The nice thing about a quad is its simplicity, that’s also its main safety feature. Adding complexity is detrimental to safety, and should only be done if it really adds safety on top of that. I.m.h.o. parachutes do not. Not only are you attaching a device that is simply much harder to test compared to the other the critical bits, the quad will most likely be damaged in these tests. So, in practice it may be tested once or twice and that’s it. In the mean while many things will change at the drone (you know, progress), leaving the parachute system virtually untested. So the chance of a successful parachute deploy in practice is quite low. Secondly, the chance of a false positive with a parachute system on a drone is quite high, too lengthy to go in to but I speak from experience.

    So, a parachute is fine if you are doing a very high risk flight, say when you want to speed up development work. You can then cut a corner by using a parachute (airplane test pilots also often have a parachute). But for scaled up and daily use, just make sure the drone is so reliable a crash virtually does not happen (airliner passengers do not get parachutes).

Leave a Reply

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