To be clear, they crashed into the vertical cable hanging down from the end of the crane. Not into the structure of the crane itself.
So it's not as bad as "they don't see cranes". But it absolutely raises the question of whether they can see cables, whether hanging from cranes or spanning telephone poles.
And honestly, cables are really hard to see in the air. That's literally why high-voltage power lines hang those big red-orange marker balls on them for pilots to see.
Genuinely curious what the solution here is. Hard-code some logic to identify cranes and always assume there's a cable dangling from the end? Never fly underneath anything? Implement some kind of specialized detection for thin cables if that's possible?
Flying machines are never to be flown near cables. It's not like human pilots on a helicopter can detect and avoid the cables in the first place.
Long-distance transmission wires are sometimes inspected with helicopters, so I guess there are exceptions and protocols, but outside those, flying machines just aren't supposed to fly near cables except for explicit intent to catch them. Especially across or under. You may only approach in slow parallel motions and/or back off.
Contra this assertion, drones are already frequently used around power lines, and as such, "finding hanging wires with a drone" is actually a very active field with fairly robust solutions. Not only are drones used for power line inspections (which are actually a somewhat easier variant of this problem, because the drone usually flies above or adjacent to the power lines in this scenario), but also for infrastructure inspections in direct adjacency to power lines. Power line detect-and-avoid is a headlining feature in one of DJI's newer enterprise platforms, the M400 (where it's based on LIDAR + mmWave Radar fusion).
Also of note, this isn't the first double-failure issue for the MK30 - they had an issue last year at their test facility where their LIDAR malfunctioned in the same way on two drones in the same weather condition (misting), the drones believed they were at 0.0AGL and powered down in flight.
My guess is that detecting wires is also a very active field of research in a military context unfortunately as nets become a common defense mechanism.
Yeah. Friend of mine was a news helicopter pilot and he had one of these systems that will cut a cable if you hit one by accident: https://www.youtube.com/watch?v=J6CsNqhAeeQ . Better than getting tangled, I guess.
It does protect, and is already certified and installed on many copters. Still, it's an emergency device, last resort, like parachutes on some small airplanes.
Here's an excellent video from Juan Browne around the challenges that wires present to aircraft operations [1]. Some of these are human factors for manned aircraft, like seeing a wire but then forgetting it's there, but one of his points is that it's simply safest to avoid flying below 1000ft AGL. That's not an option for drones today, and they presumably don't (yet) have the ability of humans to make inferences about the likelihood of cables near cranes and transmission line towers, making them particularly vulnerable.
That's not true, commercial and municipal drones operate above 400ft all the time. Non-commercial/civilian drones piloted by someone without an FAA Part 107 certification and a waiver cannot fly above 400ft. Also, you are allowed to fly 400ft -above- a vertical structure (like a tower) up to 1000ft tall to avoid exactly this type of collision as long as you are within 400ft horizontally of the structure.
EDIT: crud, I did not read your post carefully enough, sorry, you addressed this point exactly. My bad.
If you've ever driven through Seattle, as you come past Boeing Field on I5 there are red spheres on the power lines. These lines are on a hill across the highway from the airport, so there's no way any plane should be there except during an emergency landing due to power loss.
There are situations where aircraft and wires might come in close proximity. It's more accurate to say that be default we keep them way the hell away from each other, we make exceptions for special circumstances, and the exceptions tend to seem far more conservative than you would guess.
At the risk of stating the obvious, the drone shouldn't be flying anywhere near the crane. It's an active construction zone with a structure that moves and swings about in unpredictable ways with people and equipment moving about below. It shouldn't be delivering to the construction zone, and if it can't figure out how to stay out of the area, it doesn't belong in the sky.
There are some FAA requirements about cranes/temporary structures that would give pilots an appropriate NOTAM, but I don't know if all cranes require this. That said, I'd argue that if it isn't tall enough to require notifying the FAA, the drone is flying too low.
This honestly seems like the obvious approach. Even if we suppose you have perfect sensors flying underneath something still means something might be dropped on you... why risk it when you can just fly above it?
You may have a tall mas or an antenna and massive cables stretching at angles around it for support. The distance between the base of the mast and the base of supporting cables can be quite large, so even a simple logic like "stay 100m away from tall structures" can be insufficient.
It would be interesting to see what comes out of this investigation. Hopefully the injured person will be alright.
But then how do you deliver to the upper floors of vertical buildings? That must be half the near-term market for these kinds of drones: people in dense, urban areas well-served by local droneports, who are looking for convenience above all else.
If you can't safely manage urban canyons—you can't manage. It'd be like selling self-driving cars that are only approved for private racetracks.
Here's a curious article I read the other day, that underscores the market factor:
https://news.ycombinator.com/item?id=45445406 ("What It Takes to Get Lunch Delivered to the 70th Floor in a Shenzhen Skyscraper (nytimes.com)" / "An informal network of last-mile runners close the gap between harried delivery drivers and hungry office workers in a Shenzhen skyscraper")
> But then how do you deliver to the upper floors of vertical buildings?
Elevators, dumbwaiters, baskets and pulleys, or just go downstairs and get it yourself.
Also, how tall are we talking? Where have you been where the upper floor windows actually open?
edit: the more that I think about this, the more it bothers me. Getting close enough to a building to deliver a package through a window with a drone sounds incredibly improbable and dangerous. A small gust of wind and the drone crashes into the building. You'd need some kind of pad at a distance from the building, or use the roof. Looks like there's a design for that[1].
And what if a package or drone falls from such a height?
I think the concept on its face is flawed, but then again, this exists[2].
> But then how do you deliver to the upper floors of vertical buildings?
Maybe asking the obvious, do you need to? Why not drop the package downstairs, people can use the elevator like normal people? Assuming there is some sort of hand-off with identification.
Many parts of Europe solved it in a more low-tech way: street-side parcel lockers unlocked with your phone. Massively reduces delivery cost (i.e., one driver can deliver far more packages per hour), is pretty convenient and safe (no packages left unattended), and best of all, doesn't require a fleet of UAVs.
You can pretty naturally extend this once you have self-driving vans.
I think Amazon has a proprietary version of this in some parts of the US, but at least where I live, the lockers are a car drive away, which defeats the purpose.
The last tower I worked in wasn't some really big tower (only about 16 stories) but in the lobby near the security booth there was a set of shelves operated by a food delivery company. Every day there would be a few restaurants listed, you'd place your order, and they'd deliver everyone's food around noon to the shelves. You'd just go down and grab your meal.
Seemed to work well enough for the times I used it. Honestly though I really valued the time to take a break and go for a walk and have lunch away from sitting at my desk.
Japan too, convenience stores are everywhere, packages go to lockers in the convenience store, unlock with phone app.
(Of course you can still choose to have them delivered to your door, but I find the delivery people don't ring the doorbell and then mark the delivery as missed, even with instructions to leave the package in front of the door. But that's a separate issue.)
The chinese too, afaik, in shanghai drones deliver parcels in specific parcel stations. Flying drones delivering stuff (through the windows?) to upper floors of buildings sounds like sth between scifi and madness right now. Having specific pickup locations solves a lot of problems like the ones here, as drones can just have to follow specific, predetermined routes that can be more easily monitored, instead of having to go to some random, different address.
Have you ever been inside a tall building? How are you thinking a drone would deliver to the upper floors of vertical buildings? The windows don’t open. There are occasional balconies or terraces, but these are more the exception than the rule, especially for “hungry office workers in a Shenzhen skyscraper.”
Residential buildings usually have operable windows (and often balconies). My condo building is only 17 stories tall, but my windows open all the way. On the other hand, it's an old building and there might be new rules for new buildings, since for those I usually only see the windows open a little.
Still it would be silly to deliver through the window. The rooftop might make more sense (can just drop off package and not accessible to random people on the street).
That doesn’t make it better. The cable hangs down from the crane and thus the rest of the structure is still nearby. The drone should be well clear of any obstructions precisely to avoid this sort of thing from happening with hard-to-see ancillary obstructions. Something went really wrong here with the tech.
For manned flight instrument approaches the FAA has very nuanced math that defines this which typically comes down to a few hundred feet. That translates pretty well here too. Amazon will need to explain to the FAA why they were flying anywhere near this crane let alone that close and below the hight of its support structure. There’s no real defense for doing something that stupid.
A human operator would see a moving crane and say "that's a construction site, I'm going to go around". They would not fly directly under a crane even if it visually looked clear. In this case the crane was actively lowering something, so the drone not only missed the cable but it flew directly in between the crane and a visible object hanging in the air below the crane.
For concrete numbers, I would say stay 50 yards away from construction equipment, and always laterally or above, not below. Honestly these drones are enormous so I think "don't go under" can just be a blanket rule. They can't be going under trees or bridges or overpasses either, they're too big.
Edit: Also, the drones themselves should be far enough apart that if one crashes the other has time to react and stop or change course. I don't have a concrete number there, it depends on their speed and acceleration, but they shouldn't be flying so close that if one crashes they all will.
In non-drone aviation, we require vehicles to be separated from each other by 5 nautical miles horizontally and 2,000 feet vertically. Additionally every area of the planet they fly over has an MSA figure - Minimum Safe Altitude - which is supposed to guarantee 1000 feet clearance over any obstacles or terrain.
Both of these allow healthy margins of error, whether that error is from a human pilot or ATC, or from computer systems - either in the vehicle or the ground.
I'd argue these would be a great place to start for drone aviation.
If such limits make drone burrito or toilet paper delivery expensive, that seems fine.
At least in the US, minimum separation is for when you're talking to air traffic control, and MSA is for flying on instruments. Minimum separation when you can see outside and aren't talking to ATC is "don't hit other planes." Minimum altitude is 500ft, or 1000ft over "congested areas," plus 500ft distance from any obstacle. Unless you're flying a helicopter (or powered parachute or hang glider) in which case the only requirement is "the operation is conducted without hazard to persons or property on the surface."
It would make sense for a quadcopter to follow helicopter rules. Obviously it does not follow the "without hazard" requirement if you crash into cables, though.
Fair enough, I am not a pilot, just a nerd so I wasn’t aware of this!
I’d agree the helicopter rules seem most appropriate, though I guess I’d still feel like that would still rule out operating anywhere near a building under construction.
That said, a regular helicopter that suffers a loss of power or other fault, still has options like autorotation to at least attempt a landing without killing anyone on the ground. Do drones have any equivalent ? I.e. if battery is below x% it returns to safe landing spot?
"Without hazard" is pretty subjective, but I agree that it probably shouldn't include casually flying around cranes.
I don't know how drones are programmed, but landing immediately if the battery gets low certainly sounds like a sensible precaution. Electric motors might be reliable enough that you don't have to worry about gracefully handling failure of those. I hope so, because I don't think a quadcopter has much hope if any motor fails.
Was this particular operation careless and wreckless? Could be.
Someone else in the thread said the weather conditions included mist. I'm skeptical misty conditions also permits a minimum 3 miles of visibility, but what do I know I'm just a pilot.
But also, it's possible the waiver I assume Amazon is operating under could include visibility. I assume this because Part 107 requires visual line of sight operation, but Amazon's operation sounds like beyond visual line of sight (BVLOS). I don't know anything about that. I'd like to think the waiver and the operating requirements are public information, but I don't know that either.
Thanks for the context! Makes sense for traditional aircraft to be super conservative like this, especially given they tend to travel very long routes and sometimes have nothing more than a pair of human eyes paying attention to obstacles.
Do you know what are the rules for helicopters in a city? That seems like a closer analogue.
At a mere (say) 240mph, 5 miles is 75 seconds. Talk to an experienced pilot about how short a time that is, when a bottom-10% pilot is trying to figure out some problem with his instrumentation, or has set his radio to the wrong frequency, or whatever.
100mph is probably around the top speed for home built FPV quadcopters with 5 inch propellers; that's neither uncommon or expensive. Amazon's drones are probably built for efficiency though so they will go slower than that.
5 miles at 50mph would give only Amazon 360 seconds to fix the bug that caused the first drone to crash. Or figure things out enough to get the second drone into manual override mode.
Why do people always get a hard on for bizarre drone rules? Some amateur pilot goes up in a fireball every other week, I think we put up their safety record to that of drones and there are going to be some hard choices to make - I think it's over for our amateur pilots out there.
TBH my concern is less for the fate of people who choose to fly planes or drones - they are the ones making the choice so if they hurt themselves that sucks, but none of my business. However I am concerned for the innocent bystanders who might be under them, who did not choose to risk death or injury by an aircraft, drone or otherwise.
The template for this ofc is how we handle (or don't handle) the danger posed by people inside cars, to people outside cars. In aviation we will do a lot to avoid even one death, in the air or on the ground. But in cars we mostly don't give a crap. It took decades for drunk driving to become unacceptable, but outside of that, we are still pretty ok with death by car. The only survivor of the collision just has to claim the person outside the vehicle "darted" and we all shrug and move on.
I would just love it if we could apply lessons of the past to new technology. Instead we just hand wave it all away. Then in a few decades, if enough people die and their surviving loved ones invest enough time & energy, maybe we'll make a few tweaks to the formula.
So no, we don't need 5nm separations for 20 lb drones. But we do need some sort of structure that recognizes the people under them didn't sign up to be part of the beta test. For bonus points it should also recognize that externalities exist and should be priced but I am not holding my breath.
I had a look at the video... if that's the crane that was in the incident then the drone was simply way too low for cruising. This isn't a tower crane with a flight restriction. They were moving equipment on the roof of a single story building.
I suppose recognizing that there is a cable even when we don't clearly see it, but we know it is there because we know the concept of a crane, is exactly the amodal completion of our brain's top-down perceptual inference that CNNs and whatever else those drones use are currently still lacking?
It shouldn't be necessary to hardcore such things if the goal is to build something resembling intelligence.
Of course for a drone it might be more feasible to do so though.
The easy answer is to follow the same rule that you have for every other certificated aircraft and operator: Never fly under a structure. When humans do stupid shit like this, we take away their license.
I don't think they will take away their license, but AMZN should have to explain exactly how their drones managed to crash by flying themselves under an obstruction twice in a row.
Neural nets in drones are only used for object recognition. Beyond that, drones (and other autonomous vehicles) aren’t doing any sort of reasoning or decision making, they follow rules, they’re just robots.
Although I hear that Tesla is thinking about using AI for decision making as well, which I find quite scary. Frankly I think it’s safer if vehicles don’t have concepts and intelligence, and just follow the rules.
> raises the question of whether they can see cables
Should the drone's vision be comparable to a humans though?
I feel like drones can either see or don't. If we go and try to tackle every corner case then nothing would come of it.
Also, do I - as a citizen - have to bear the externalities of Amazon's beta testing?
The video perspective might be misleading - but if the area surrounding the crane is as totally devoid from other obstacles as it appears? I wonder if the operator maybe didn't see the crane. How is that possible? No idea.
14 CFR 107 covers visual line of sight commercial UAS operations. My two cents is the operator should fly around or over and well clear of the crane. They're given a wide latitude. If within 400' of a structure, they can fly up to 400' above the highest point of that structure.
However, it seems probable these operations are BVLOS (beyond visual line of sight), which requires a waiver from the FAA. In which case most of 107.205 applies. I'm not sure if the operating agreement between Amazon and FAA is public information.
Avoiding structures is pretty basic to operational control and responsibility. The fact they hit a seemingly obvious and avoidable structure, the ensuing loss of control being inevitable, which can (and in this case did) lead to on the ground injuries. Pretty remarkable operational failure in my opinion.
We’ve completed our own internal review of this incident and are confident that there wasn’t an issue with the drones or the technology that supports them, Terrence Clark, an Amazon spokesperson told CNN. Nonetheless, we’ve introduced additional processes like enhanced visual landscape inspections to better monitor for moving obstructions such as cranes.
Correct, never fly underneath anything. Telephone poles? Trees? Assume there is a solid wall between the highest point on any two objects except within a few yards of the delivery spot.
Aerial Lidar is pretty good at detecting power line cables. Power line mapping is a major use case for it. However, that's in large part because at the scan distance, the beam already has quite a big diameter that is likely to hit the cable. Maybe a higher beam radius scanner could work out for close-distance cable detection
The company that I work for builds power line detectors for helicopters. They sense the electromagnetic fields generated by the lines and alert the pilot when the field strength exceeds a threshold. I would imagine this tech could be easily adapted for a drone.
Obviously that wouldn't work for a crane though...
For commercial deliveries I would expect them to designate a landing zone guaranteed to be free of obstacles vertically. I'm guessing that installing radar detailed enough to see swinging cables is nearly impossible.
The same problem actually already exists for non-drone planes, because they must be able to operate in poor visibility conditions. FAA issues notams for construction cranes if they pose a risk to nearby airports. One solution for drones would be to extend these notams to all cranes/other obstacles, and the drones must subscribe to these notams to operate in the airspace.
The fact that these things are flying without rock solid “avoid this giant fucking thing” logic is asinine. The solution is don’t fly like a child playing a flight sim for the first time. Don’t zip around anything let alone construction cranes. Use common sense flight paths, decks and ceilings like everything else in the air.
The police is not qualified to investigate this. The only people that should be investigating is people who understand the code that the drones run.
Accidents will happen as long as we, as a society, agree and desire to have new tech. The investigations and bug fixes should be left to people who understand the tech.
The Feds will quickly arrive and take over. Aviation issues are
Federal matters. The only role of local law enforcement and emergency response is to provide any first aid and then secure the scene for the Feds. Unless lives are at risk local police shouldn’t even touch anything. They put up yellow tape around the scene and keep it secure until the FAA and/or NTSB arrive.
I'm not sure why this is getting down voted. Indeed, the FAA is the correct investigating body here as the local police department has no jurisdiction over aviation accidents. They should have immediately called in the FAA to investigate.
"The NTSB will retain far more employees than during prior shutdowns when it had to furlough 90% or more of its workers. In 2019, the agency did not send investigators to 22 accidents because of the funding lapse. But it made the case to White House budget officials that it needed more personnel for critical functions."
They should call in a bunch of robotics and PyTorch experts to investigate the code, actually, and preferably also submit a pull request to Amazon. Amazon should be required to pay this "squat team".
The FAA does not have the expertise to diagnose this.
And if it doesn't perform as expected in the world, all human efforts should be on fixing the actual code. Anything else, especially non-technical conversations, is a waste of human time.
Not at a distance. They're basically transparent even to our naked eyes at aircraft speed and distance scales. Let alone to digital cameras on flying robots. They'll probably have to either use really good active sensors(ITAR), or infer possible areas of danger from visually cable-end-like features.
Yes human also cannot see them from far away and flying at high speed but what I wanted to point out is that robots usually get the cheapest and lowest quality sensors.
So it's not as bad as "they don't see cranes". But it absolutely raises the question of whether they can see cables, whether hanging from cranes or spanning telephone poles.
And honestly, cables are really hard to see in the air. That's literally why high-voltage power lines hang those big red-orange marker balls on them for pilots to see.
Genuinely curious what the solution here is. Hard-code some logic to identify cranes and always assume there's a cable dangling from the end? Never fly underneath anything? Implement some kind of specialized detection for thin cables if that's possible?