The slow collapse of Amazon’s drone delivery dream | Hacker News Reader

The slow collapse of Amazon’s drone delivery dream

wired.co.uk

188 pointsTigeriusKirk265 comments

Open articleView on HN

I think drone delivery will find it's niche. The military started testing drone delivery from ship-to-ship [1].

flyzipline has been successfully doing drone delivery in Africa for a while. [2]

There's a need a huge role that could be filled.

However small drones zipping around your neighborhood dropping off tacos? I don't think so. Not any day soon.

[1] https://www.thedrive.com/the-war-zone/41838/drone-makes-firs... [2] https://flyzipline.com

flyzipline is so cool! I saw an AMA they did on reddit a while ago and their tech for the thing landing where it hooks onto the the wire is really neat! https://youtu.be/jEbRVNxL44c?t=284

As with everything, the buzzkill is regulatory and administrative rather than technical. We could easily make drone delivery a thing but the paperwork will kill it.

Regulating safety and noise are not a buzzkill. They are essential in preventing corporations, like most regulation, from killing people.

Tesla's autopilot is the best example that the hurdles are not regulatory in nature, no matter what musk is trying to tell us.

As of now, these machines will kill people on the regular. Even in situations that a human could handle. And that's the bar we're setting and I'm glad we do so.

Edit: I know that people are killing people on the regular in situations that a machine could handle. And introducing this stuff as safety systems (like autonomous emergency breaking) is not hindered by regulation.

The main way you reduce traffic fatalities, is by desgining streets and roads in a way that increases safety for all users.

One of the most effective ways is to get as many people to use something other than cars for transportation. You do that by making cycling, walking and public transport faster and saver.

This side of the equation is public policy though. But I agree, that this is the best mechanism we have.

It will also become a economic necessity.

Car dependence is not sustainable. Infrastructure costs need to be covered by the income they generate.

That is not the case with car dependent cities.

Here is an excellent playlist about how car dependence is so expensive: https://www.youtube.com/playlist?list=PLJp5q-R0lZ0_FCUbeVWK6...

Absolutely. If you want cars to go 30kph in a residential zone, you cannot make a straight, four-lane street and plonk a round sign with a red border saying "30". You need to make a narrow street with bumps and/or curves to make it physically impossible for cars to drive at unsafe speeds (and in the process make most of that space walkable/cyclable and safe for people.

YT channel suggestion: "Not just bikes"

As a pedestrian I am terrified by streets with occlusions and random obstructions in the name of traffic calming. There's a road in Oxford with build-out with a tree in the middle of it. You can't see cars coming and they can't see me. Same goes for removing lines from the road. I'd rather cars knew where they stand and focus on what's important like looking out for other road users.

I've always thought it was a perverse conclusion that the way to make someone operate heavy machinery more safely is to increase cognitive load of the people operating them.

Why not have a the straight roads you describe, with rigid automatic enforcement and penalities that take dangerous drivers off the road? And then put in place infrastructure to make sure that not driving a car is a viable option, so that stricter enforcement has public support. I see this option as a self-reinforcing route to reducing car use.

On the other side of this: there's a residential street I cross rather frequently in California which is a straight road with good visibility, and this being California, both unsignalled crossings and a 35mph speed limit. Everyone drives at or above 35mph, and almost no one yields for pedestrians at cross walks despite all the yield to pedestrian signs that are clearly visible. I've seen people wait for several minutes afraid to cross; I've had one driver turning right onto the road run into me while I was crossing at a crosswalk, despite their stopping and presumably checking the road before turning; and I've frequently seen cars have to swerve or heavily brake in order to avoid running into the cars that have yielded to let pedestrians cross, despite those cars being quite clearly stopped and visible.

In all my experience with high-visibility roads, even though you'd expect people to be better able to pay attention to everything on road with good visibility and clear markings, they actually seem much worse in those conditions. And given the ways in which they are worse, it's not clear how automatic enforcement would work; speed limits and traffic signals can be enforced, and perhaps crosswalks, with cameras, but often regulations seem focused on banning, rather than encouraging, this sort of enforcement.

Crosswalks with traffic lights and speed cameras seems like such a self-evident solution to this. That doesn't seem any more authoritarian or heavy-handed than literally digging up the roads to make them more miserable to use.

(Also, turn-on-red is such a weird idea. Especially as, from the sound of it, the rules vary from locale to locale.)

In my hometown they implemented a system that checks for your speed and if you’re over turns on the next red light. Works like a charm.

Sounds like a great way to train all the locals to run red lights from 11pm-5am or so.

No idea if that is universally true, but I feel like most people here (in Germany) have a lot of respect for red lights. It's not unusual to see cars waiting at a red light in the middle of nowhere.

Obstacles are only one of the ways you can calm traffic.

Planting trees near the streets, narrower lanes, speed humps, roundabouts are all measures with a proven record that they improve safety.

Crossings should also be one lane at a time.

> In all my experience with high-visibility roads, even though you'd expect people to be better able to pay attention to everything on road with good visibility and clear markings, they actually seem much worse in those conditions

This is nothing new. You shouldn't design a street like a road. You will end up with something called a stroad, that is bad at both beeing a street(a place where you can live, shop and whatever) and a road(something that connects two places)

Stroads are ugly, inefficient and dangerous. https://www.youtube.com/watch?v=ORzNZUeUHAM

> You shouldn't design a street like a road. You will end up with something called a stroad, that is bad at both beeing a street(a place where you can live, shop and whatever) and a road(something that connects two places)

Wow, I never saw this phrased like that.

Many streets in Shanghai don't provide much access between the street and the sidewalk -- instead, there are pedestrian overpasses or underpasses.

(It's not the norm -- there are more streets that a pedestrian can easily walk into -- but it's not uncommon either.)

You don't need to make driving a car a nonviable option. Making walking into the street nonviable is just as good.

>I've always thought it was a perverse conclusion that the way to make someone operate heavy machinery more safely is to increase cognitive load of the people operating them.

But the catch is that people doing things optimize for static cognitive load. In practice this means slowing down so "stuff" that demands thinking happens less readily when other things take up brain cycles. So you can get them to drive slower by adding "stuff" to the road. There are right and wrong ways to do this. Generally you try not to limit visibility.

Remember, the goal is safer, not slower. And slower is just a crude proxy for that. With big multi-lane roads your best bet is generally to try and normalize traffic speeds or segregate traffic based on speed (truck lanes, bike lanes, sidewalks, etc).

>Why not have a the straight roads you describe, with rigid automatic enforcement and penalities that take dangerous drivers off the road?

Because in a democracy with even enforcement they vote out the people who implemented such a system and with uneven enforcement it just turns into a system of extracting revenue from whichever groups nobody cares about. Orwellian crap like that simply doesn't fly in most of the world.

Regulatory compliance is, in many areas, a major cost of development. Sometimes the largest. Much of the regulation is in form, not function. Comply with box ticking and you pass, regardless of whether the product is safe in reality or not.

If regulation was focused on function it would be less expensive and yield better results for society.

> ... is not hindered by regulation.

It is effectively stymied by added total cost.

I wonder which regulation you're referring to. Most ticked boxes I can think of are indeed of the "does not kill a certain kind of person"-kind, or of the "can be used by a certain type of person"-kind.

And of course, regulation can not cover every case. And if it turns out to be a problem, we can cover the additional cases. This way (in Europe at least) we gradually require cars to be safer for passengers and the environment.

I think GP is right that there's plenty of "form, not function" regulation - of the form, "must be no longer than X / no heavier than Y", etc.

But I think GP is wrong in considering this to be a problem. Rules may be expressed in a non-functional way, but that's often just a way to embody functional rules in a heuristic that doesn't require expertise to apply.

For a computing example, you may have a processing algorithm that has a tendency to explode in computational complexity on some datasets, because of complicated algorithmic reasons. Instead of explaining the exact problem to users, most of whom will not be able to understand it, you could determine that the failure mode is very unlikely to happen for smaller datasets, and just say "the input should be no larger than 10MB in size".

Other reason for "form, not function" is coordination - there may be many good ways to do something, but all of them require everyone to do the same thing. So it makes sense to pick a reasonable default and tell everyone to stick to it.

You have still not given a concrete example of what you mean.

> If regulation was focused on function it would be less expensive and yield better results for society.

... and that would actually require that those in charge of regulations would be technical enough to understand the complexities involved. If you're that good with technical matters and can confidently reason about the wider second- and third-order effects, you are likely to have a bright - and better compensated - future in the industry at large.

If you are a regulator, you have to confront two facts. 1) that the most competent will either never be available or will leave at the first opportunity; and 2) that box-ticking is much easier to measure.

In the end, you have to ask yourself who are you left with?

And insurance :).

Having 10-30kg high-speed flying machines, built by the lowest bidder and operated by companies that like to move fast and break things, being hurled around above densely populated areas full of squishy people and car traffic... how long do you think it will take before a first fatality? Do you want to live in constant fear of getting hit in the head by a low-flying brick?

No, a thick regulatory straitjacket is the only thing that can enable drone delivery in cities. But then, there's little point to even do that, until the noise issues are resolved.

On a related (and likely just as improbable) note, perhaps cities should start treating delivery as another municipal utility. Right now, most houses and apartments in a typical city are fitted with pipes that bring in water, gas and fresh air, and take away sewage and stale air. I can envision another utility - a "package pipe" - a conceptual blend between old-school pneumatic tubes and parcel sorters. Imagine each apartment having an outlet sized for a standardized container[0], through which packages could be sent and received. I bet it could cover upwards of 90% of delivery needs of any town - most things people send would fit, larger orders could be split into multiple containers, and bulkier things would be delivered by trucks on-demand.

An installation like this would deliver all the benefits and solve all the problems of in-city drone delivery, cheaper, safer, better, and without any of the noise issues.

The obvious drawback would be the costs of building it up and maintaining it. Someone would need to do a calculation, but I suspect it could be paid for the same way other utilities are (small fixed payment for access + pay per use + whatever other bullshit business shenanigans electricity providers usually engage in). Looking at the cities I've been in, most existing buildings could be retrofitted with it as an external attachment (imagine an external lift going to every apartment). The big problem I envision is routing the inter-building segments of the pipeline - it would likely have to be above ground due to size, and thus expose it to the usual problem of landowners trying to extract as much money as they can from the public to let the project go through their plot.

But hey, one can dream :).

(My bigger dream is that one day, we'll all have something resembling a Star Trek matter replicator, perhaps in a form of sophisticated 3D printer fed from a municipal "matter stream" supply.)

[0] - Say, UPS 10kg - "16.5in x 13.25in x 10.75in (42cm x 34cm x 27cm); weight limit 22lbs/10kg", or UPS 25kg - "19.75in x 17.75in x 13.25in (50cm x 45cm x 34cm); weight limit 55lbs/25kg". Via https://www.ups.com/us/en/shipping/create/package-type-help.....

>>Having 10-30kg high-speed flying machines

Much better than 2,000-4,000kg driving machines being operated at street level by the people willing to do the work for the lowest pay.

These machines kill thousands of people every year. This method of delivery is also very costly relative to autonomous drones, making a huge number of goods inaccessible to millions of people.

> Much better than 2,000-4,000kg driving machines being operated at street level by the people willing to do the work for the lowest pay.

No, it's not much better. Cars may be heavier, but they're constrained in their movements. They (mostly) follow the road network. It's easier to keep track of them, as you only have to scan the 360° plane around you. A drone could hit you anywhere, from any angle, so you have to pay attention to the whole half-sphere around you.

Additionally, cars are already here, and delivery services are only a small fraction of car traffic. Drone delivery will not displace any significant amount of cars from the streets, so drones will have to operate alongside car traffic. And I can't think of a better way for a civilian drone to cause large amounts of damage than crashing into traffic.

And, of course, car transportation is already in a regulatory straitjacket - one that was built gradually over the past century, rule by rule, one fatality at a time. It would be wise to not go through the same process with drones.

A car travels at grade, crossing numerous pedestrian paths. You need to cross streets that cars use. You do not need to cross flight paths that drones use. A drone can fail and land at grade, yes, but that's a major failure, whereas it only takes one misstep to walk into oncoming traffic.

Many years ago I proposed to my friend we should start the work on anti-drone technology, and went through the various technical means of doing so, but never followed through. I think the best work these days on topic is probably focused emp bursts. Regardless, we should all consider how we would track drones in real time if we wanted.

> I think the best work these days on topic is probably focused emp bursts.

I think a magnetron from a microwave oven attached to a directional antenna should just do the trick :).

> On a related (and likely just as improbable) note, perhaps cities should start treating delivery as another municipal utility.

Hmm, or maybe national governments should step in and provide this as a subsidized basic service to everyone. It could work for all kinds of post, not just packages. And we could go to great lengths to make it work also across borders! I suggest we call it something like a public postal service.

:-)

Haha, I know :).

But here I was thinking about technology and logistics. In-city drone-based delivery promises features that neither public nor private delivery services can... deliver ( :) ). The package stream utility idea promises those same features and more, without the drawbacks of having fleets of lawnmower/brick hybrids zipping around everyone's heads.

Why would I want government to do it, when private companies like FedEx or Amazon can do it better, faster and cheaper?

Private companies deliver network services "better, faster, and cheaper" by cherry-picking the economically-viable aspects of service, and leaving the noneconomic segments to either the public sector or to be unserved at all.

Universal service --- whether telegraph, rail, telephone, air, postal service, or data networking --- doesn't have that option. It serves both the dense, high-margin zones and the low-density, low-value outlying regions, because it has the interest of the common weal as a whole, not merely profit.

Worse, the private sector then attempts to change legislation and regulation to impose greater burdens on the public-sector services they compete with (as with USPS pension liability funding obligations).

Government-run postal services are tremendously efficient, effective, fast, and reliable ... at least until they became actively sabotaged from within.

> when private companies like FedEx or Amazon can do it better, faster and cheaper?

Because they don't.

> perhaps cities should start treating delivery as another municipal utility

This used to be called a postal service.

I look forward to the jurisdiction fights over city-run package delivery; I expect Fedex to make substantial donations to state legislatures to have it banned, as happened for city-level broadband.

If you accept regular drone accidents involving buildings, people, animals, vehicles and perhaps even aircraft, sure. I know regulations are heavy but self-piloted craft can do really weird things when they think they're right.

It's no wonder the difficult part was when the drones came in to land. That one washing line that's hard to spot, the overjoyed dog jumping up at what he thinks is a new toy, that grass field that was actually an overgrown pond...

Oh wow - Zipline is so inspiring. This video showing their setup in Rwanda is fantastic: https://youtu.be/jEbRVNxL44c

You mean Tacocopter was a joke?!?!

https://tacocopter.com/

Beer Drone is real if we all believe it’s real.

Here in the UK drones have been used to deliver medical supplies and mail to the Isles of Scilly [1] (south west tip of Britain) and Isle of Wight [2] ('big' island on the south coast).

Essentially these are operated like planes but at a fraction of the cost.

[1] https://www.youtube.com/watch?v=SHrcORbBo04

[2] https://www.bbc.co.uk/news/uk-england-hampshire-56322350

Yeah, there are clearly cases where you don't have an easy way today to take a package from point A to a relatively nearby point B.

But it's really hard for me to imagine what delivery problem drones solved for the average Amazon package. At first it really seemed like a PR stunt but, by every indication, Bezos was actually serious about it. Which I honestly don't get.

The problem of paying for delivery drivers and partners I reckon.

I'm not convinced it's an improvement over ballistic delivery with some kind of elastomere faring.

> “The best living metaphor for it all is that you had some bloke on the other side of the planet telling you what to do and then just leaving,” the source claims.

This is commonly known as seagull management: fly in, squawk a lot, shit over everything, and then fly off.

Nice one. Added to https://github.com/globalcitizen/taoup

https://en.m.wikipedia.org/wiki/Mushroom_management

'keep them in the dark and cover them with manure"

> “The best living metaphor for it all is that you had some bloke on the other side of the planet telling you what to do and then just leaving,” the source claims.

Some poetic irony to this happening to the UK.

I've heard it called drive-by management - same thing, different smell.

Describes my previous AWS manager perfectly.

I can't really see any wide-scale drone operation operating in a residential areas. The decibel output of a single delivery drone similar to a delivery van ~70-80 decibels (hard to get comparable decibel output). A delivery truck driver averages around 120 deliveries from one van. A drone can carry one (or maybe a few packages at a time). There would need to be many more round trips for drones meaning far more distance covered by drone. This would require more drones than vans to complete deliveries. Drones sound has a much higher frequency than a van, and to me, drones are more noticeable than the similar decibel output of a van.

To be fair, drones have one advantage and that is that they can cover most of the range between the pickup and drop-off points at higher altitudes that would decrease decibels at ground level.

Edit: Every 6 decibels sound aptitude doubles so 70-80db is a big difference. Would need to get more reliable and comparable dBA readings.

Workhorse has full on plans for drone delivery that use the van as a base:

https://workhorse.com/horsefly.html

Yeah, I don't see why drones and vans would be mutually exclusve. You drive a van down the main road and have it spin off a fleet of drones for individual residential deliveries. They do their thing in 1-2 minutes - far less time than it would take for the driver to get out of the van and walk the package to the door - and then return to wherever the van is currently driving to recharge and pick up the next delivery. The drones never need more than a 1-2 mile range, they recharge while the van is going to the next neighborhood. You avoid all the congested residential streets, the need to find parking or double-park, the risk of hitting kids that run out into the street to catch their ball, the difficulty finding houses, plus all the hard work drivers need to do to walk up driveways and carry heavy packages. Also is very amenable to self-driving trucks, since the delivery van only needs to navigate the main arteries and they have drones acting as the capillaries.

Truck + driver means you have a real person who can kick boxes around and find stuff in the back.

Add drones and you have to create an automated drone loading system which works on multiple drones simultaneously.

Otherwise you have truck + 2 (3? 4?) drones + person, and the person loads one drone while the other is/others are delivering.

That will save some time, but not enough to be a game changer. There will still be kicking of boxes.

Meanwhile drones will have to autodock to a charging cradle, land, be loaded, deliver, and then either repeat or autodock again.

The system breaks down badly if there's any kind of issue.

>and then return to wherever the van is currently driving to recharge and pick up the next delivery

Are battery charging times fast enough to support this? I thought we're still at the "5 minutes flight = a couple of hours charging" stage, but I haven't used anything more advanced than one of those miniature toy quadcopters so I'm not sure how more serious drones (DJI etc.) handle it.

They simply have easily replaceable batteries. For a major event, they’d have lots of batteries charging.

Simply worse for the environment?

Replaceable doesn't mean one-time-use necessarily - it might just mean that the recharge time is longer than the downtime expected between deliveries so you'd need to swap out a new battery after each run while you recharge the used one. There are a lot of assumptions here but it seems sane.

That just raises more questions - do you have some complicated automatic battery swap system, or have an extra human on-board to manage the drones? At least if it was a person they could also load packages onto the drones as you drive, else you'd need to automate that as well.

Do you charge batteries on the van, or just swap trays of them each time you return to the depot?

How much of the van's internal volume are you trading for drone-related subsystems rather than cargo storage?

It seems impractical.

if you have developed drones with the ability to return to a van and autonomously pick up the next package to deliver, picking up a battery at the same time shouldn't be a difficult step on top of that.

there's lots of things about drone delivery that seems impractical, but not sure that battery swaps are one of those things.

You can't build the battery-changing system into the drone due to size, power, and weight constraints, so it has to be part of the van. So you need something on the van that can

1) Secure the drone in place while the battery is being changed, and ensure drone can successfully land with sufficient precision for the battery changer to work, or to manoeuvre the drone into place.

Probably you would also want to keep supplying auxiliary power to the drone so you don't have to wait for it to reboot and reconnect to the van's swarm management when the battery is removed.

2) Physically remove and replace the battery, including manipulation of any connectors and fasteners required.

3) Convey batteries between the drone and the storage inside the van. I imagine this would work something like an LTO tape library, where the van tracks which storage slots are occupied, if they've been used or not, and have some internal mechanism which moves batteries between the storage slots and the battery changing mechanism.

I can see this being quite bulky, as with that many batteries you would need compartmentalisation and maybe even a miniature fire-suppression system in case one of the batteries "cooks off".

4) Must not interfere with whatever cargo-loading system is in place - you could have a nice flat battery tray in the roof, but you might need to punch a hole through it to fit a cargo elevator or so on.

Also, must not take up too much internal volume of the van that could be better used for profitable activities (storing cargo for delivery)

5) It would still have to comply with normal road requirements such as maximum heights, environmental resistance (rain, dust, etc.) and so on.

That's just off the top of my head, I'm not an engineer and I'm sure there are plenty more challenges that I haven't thought of.

I think it's easy to do #1 by having a dock station with the van actively reaching out to grasp and control the movements of the drone. For #2 I think we've already solved that by just having some sort of cartridge battery with exposed contacts - think something closer to an old flip-phone battery that is forced to slide into place correctly. For #3 if we assume the van is equipped to manipulate drones into properly docking the drone this really isn't a stretch - and inventory retrieval systems have been realistic long enough that they're actually featured in Hackers (1995)[1]. Regarding #4 I guess this might be a problem somehow? - but let's just assume the battery storage bank is on an external wall of the van - if security is an issue we're going to lose like a foot or maybe a foot and a half of the internal space to the retrieval system - again I don't think this is a problem. Lastly onto #5 I could see weight being possibly the most difficult road req to comply with here - batteries are heavy and generators are heavy - that's a lot of weight and ideally these vans would be able to drive on any road out there - including those with old wooden bridges or "seasonal pavement".

I think of your points above - #1 is the most difficult in that step you're still needing to contend with the environment which may involve strong winds - once you dock to the van (assuming you can limit yourself to a single docking station or a small number) everything else seems pretty trivial... That all said I don't program autonomous drones for a living so it'd be ridiculously hard for me - but I assume if we can get drones to reliably do pre-programmed flight paths we can probably tackle this.

Most of the problems I can think of are on the other end where stupid users will do all sorts of non-compliant things with the reception pads - like placing them in impossible to reach areas due to locked doors or too little clearance to actually get a package on the target - also just like... keeping those things in a functional condition.

1. https://youtu.be/2efhrCxI4J0?t=111

The answer to a lot of these problems is to flip the script and have the drones homebase be the delivery location or a nearby substation rather than the van itself.

That's also a really interesting idea - though the substation one probably. Right now drones capable of lifting a box of cans are well over 1k which is an uneconomical one time expense for slightly easier package delivery... and extremely uneconomical if occasionally some neighbor is going to hit your drone with a frisbee or if drone theft becomes more profitable than package theft (which it almost certainly would be).

While it's not meant for delivery, Airobotics has/had a very slick battery swapping mechanism for their multi-rotors.

https://www.youtube.com/watch?v=r3QWtXjWJTA

Battery seems eventually solvable with a swappable system.

The thing I don't see is "getting to the front door". As a machine problem, it seems so insanely ambiguous.

From a technical standpoint, a workable system might be to have an official "drone pad" of standard size and appearance, similar to a helipad. You might put one in your backyard if you have space, or if you live in a city block there's probably room for a shared one on the roof.

Of course, the social logistics of that would need a bit of work. It's one thing to make sure everyone's within walking distance of a drone pad, but it's another to make sure nobody gets their parcel stolen.

Not bad, definitely have to start somewhere. Rain would be a big concern though.

Sadly, people in apartments, those that lack a back yard (my parents bought a house w/o one so there's no maintenance, just a ravine) are possibly worse off since theres no "hiding" the package. Its definitely possible to say it's not for everyone though.

While I don't believe particularly drone deliveries as imagined today is a great idea, the drop off should be solvable with an physical token that you place where the goods should be dropped off. Something like an RFID-equipped mailbox.

The location seems solvable, the routing to get to that location not such a simple problem in many urban layouts, especially if people aren't very sensible about where on their property they locate their delivery box, or optimise for having their parcel delivered to locations which are sheltered from the rain

Glad to see this. I've been making the sound pollution argument since the original Amazon commercial.

My lab neighbors a drone research lab. The noise from a single drone is surprisingly distracting even with the door shut. The HOAs would be in the streets the first week.

Volume aside, a lot of the annoyance comes from the unnatural sound of four motors running at slightly different speeds, going in and out of tune with one another.

Unfortunately, noise pollution is at the very bottom of public agencies' priority list.

Even in Europe, with stricter regulators and less tolerance towards loud and obnoxious behavior, it's way too common to hear tuned motorcycles blasting 110dB engine screeches at 2AM, keeping everyone in the neighborhood awake. In theory it's not allowed, in reality the bikers can do it with impunity any time they please. Drones would at least provide some beneficial service.

It's a lot easier to get a small handful of delivery companies to comply with noise pollution regulations than it is to get a large number of motorcyclists to comply. It's also easier to write very strict noise pollution standards for new technology like drones than it is to write very strict standards for existing popular technology like cars and motorcycles.

Noise pollution may be a reason drone delivery _shouldn't_ happen, but don't confuse that with an actual practical obstacle. In the US, at least, the airspace (including drones) is regulated at a federal level by the FAA, and their mandate doesn't really include residential noise levels. For example, I've never seen noise levels discussed when it comes to private aviation, even though small planes can _definitely_ be loud enough to affect quality of life (they don't even have mufflers...) Drone regulations seem to be following a similar trajectory, with categories and rules based on mass and prop guards, and no formal measurements of perceptual noise at all.

For stationary noise concerns like a highway being built, it's possible for local communities to rally and be heard, but it's entirely possible that drones will be regulated purely on safety grounds with minimal conversation about noise levels.

This is the point where I should end with a call to action, but I'm honestly not sure what the path is to resolving this sort of structural blindness. If anyone with experience working in government or advocacy wants to chime in, I'd be curious to hear.

noise pollution laws already exist. why wouldn't drones be subject to them?

The FAA's immunities do not apply to noise sources less than 400 feet above non-airport property.

But the FAA does not have rules against it, nor do many localities.

> " For example, I've never seen noise levels discussed when it comes to private aviation,..."

Then you must not have looked very much. Noise was the primary reason cited why the Concorde was not allowed to travel supersonic above the continent. But if you want something fresher then you only have to look at the Airport Noise and Capacity Act (ANCA) of 1990. Or the many airports where they have nighttime curfews because of noise concerns.

It is also an everyday concern while you are learning to fly. I was doing my PPL in Europe, so my anecdote is not directly related to FAA, but I remember a ton of "local rules and customs" around noise. Think about things like "at this airport we only do engine run-up at the far end to not disturb those people", or "at this (other) airport we only use the left pattern to avoid spooking the horses on that field there"

> Every 6 decibels sound aptitude doubles so 70-80db is a big difference.

It’s every 3db the sound intensity doubles. At 6db, the sound pressure level doubles. Loudness (aka the perceived volume to a human) however only doubles every 10db, so the difference in this context is 80db is 2x louder than 70db.

Not the best link, but skimming it it seems to do an ok job explaining the difference:

http://www.sengpielaudio.com/calculator-levelchange.htm

It correctly explains that loudness is an arguable concept, that depends as much on the subject as it does on other qualities of the sound (for example the frequency makeup)

Thanks for the link and the explanation.

Yup! Happy to help. To further break it down:

sound intensity increase of 10db: 10x increase

sound pressure level increase of 10db: ~3.5x increase

loudness increase of 10db: ~2x increase

It sucks that there’s so many variations that all use decibels as their unit of measurement with different rates of change. I had to research it years ago because I’d always been told 3db = 2x, yet found it odd that 3db didn’t seem 2x louder. Now I understand.

The major expense for deliveries is the man-hours needed, NOT the truck (or drone). You also get time savings from traveling in a straight line between point A and point B.

A drone's only in earshot for 10-15 seconds if it's doing a drop off, and will usually be flying high enough to be inaudible. At 200m you can barely even see them.

200m high object you can't hear or see loses power....

> At 200m you can barely even see them.

But you can certainly still hear them.

Do you actually know how high 200m is? Hint: If you've seen a consumer drone it's probably not at 200m because in most regions their ceiling is 120m.

Yes, I do know how high 200m is.

It's only today I read a story suggesting that delivery bikes (electric) would be better for the environment and throughput than delivery vans:

https://www.theguardian.com/world/2021/aug/05/cargo-bikes-de...

This only holds, if it holds at all, for the delivery of low-weight packages in the city centre of major metropolises like London. And everyone in London already knows it's often faster to cycle than drive short distances.

I worked for a drone start-up that focused on a drone in a box / dedicated landing pad type solution. Seeing the big names and well funded competitors not succeed validated the technical and regulatory challenges for me.

The biggest technical challenge by far was getting the drone to land back in the box reliably over 10's,100's,1000's of attempts. Other annoying issues included buggy open source flight controller firmware, motor bearings only rated for 50 hours of flight (industry has since improved), and attempting to recharge batteries in hot ambient temperatures.

I always thought the biggest impediment to drove dileveries would be a shotgun blast in certain parts of the country.

I guess multiple cams, and very quiet drones, would solve the problem.

I'm actually for drones. The sound of delivery vans delivering one package seems wasteful? Unless you are the driver?

Oh, there's no shortage of possible hazards for drones. Ever see a hawk take one down?

If the cameras are streaming back video that would be a very easy legal case against the person shooting. I'm guessing after it happens a few times and people see the repercussions they would stop.

Or a a legal case FOR the person shooting. Having a floating spy cam in my yard is shotgun worthy.

If you requested it and were informed of when and where from it was coming could wouldn't be able to make that argument.

This could be an interesting case - how high do property rights extend into the sky above your land? Obviously planes are allowed to fly over, even small private planes like Cessnas that fly pretty low. Is it actually the filming aspect that makes it "illegal"? What if the guy in the Cessna has a GoPro pointing out the window? What if it's not regular filming but say thermal imaging, do the same rules still apply?

In the US at least, I beleive the current legal definition is 500 ft in uncongested areas and 1000 ft elsewhere. Above that is considered "public highway" in federal law.

US v. Causby is the only ruling I could find (that was 83 ft).https://supreme.justia.com/cases/federal/us/328/256/

I think there may have been one or two other--possibly state level--cases. It's pretty ambiguous but loose consensus seems to be that you probably have property rights up to 400-500 feet. On rural land, a drone at 50 feet would seem to be fair game for skeet practice.

This actually happened, I believe the guy was in the wrong because the drone wasn’t on his property or something. The details are hazy.

Assuming there is a camera pointed in the right direction

Just like the all the successful legal cases against all those porch package thieves caught on amazons ring cameras.

I'm assuming that if you destroy amazons property they will be a lot more litigious than if you steal some guys headphones.

> people see the repercussions they would stop.

I doubt it. I think it's more likely that they'll do the shooting from cover so they can't be seen by the drone.

> I always thought the biggest impediment to drove dileveries would be a shotgun blast in certain parts of the country.

Bezos, in an all hands meeting was asked about that. "If we reach a point where people shooting down drones is our biggest barrier, I'll call the project a success".

What do you think of open source projects like iNav, Ardupilot, and PX4?

(I don't have enough open-space to test their navigation / RTH functionality, I use them only for poshold & althold in horizon mode)

For everyday users, I think they're OK. We were using Ardupilot and had very specific problems with the state estimation algos (kalman filters) for guidance, navigation, and control.

As part of our landing r&d, we tried RTK GNSS, which was generally OK in rural environments but a complete nightmare in built up urban environments. When the RTK GNSS receiver would lose fix, the system would fall back to standard GPS, thus causing a step change in position/state estimate of the drone, causing it to misbehave in flight. Fighting the "features" of the wrong tech is a slog, but I was not CTO... ce la vie.

Ultimately, we wound up licensing some tech for vision based landing.

That's fascinating, thanks! I knew navigation and control systems would be complex, but when I tried writing my own transfer functions, I got a taste for just how complex it really is. I admire what the USAF was able to do in the pre-integrated circuit era!

I guess I'm not surprised GNSS is so tricky for the final approach. Wide open spaces work great with averaging GPS, but as soon as some bogus data comes in shit goes sideways fast. I just assumed the experts figured that out, but it sounds like not the case if vision-based is the solution at that point.

Cool!

New L5 GPS should be able to help, at least in terms of not being thrown off by receiving reflections off of buildings (multipath). We'll see!

It's worth noting that while many of the comments here involve the interesting problem of if drone delivery is viable, the article itself focuses on what appears to be a mismanaged project by Amazon. That is to say, their failure here seems to be separate from the difficulty of the problem space itself.

Sure, but I mean, I literally always thought this service was a joke. It never even occurred to me that this was a serious project, I thought it was like tacocopter for how silly it was (don't get me wrong, I love tacocopter).

If Amazon's leadership did too... limited resources and lack of support make a lot of problems impossible to solve.

Yes, this is not a technology failure but a top to bottom management failure. Here's a juicy quote:

"Another source says their only contact with the central office was an American executive, who would visit every few months, buy the team pizza and then ask them to double their workload"

I guess if they were following the infamous Bezos rule, it would have been just one pizza as well.

So, yes there are some challenges with drone deliveries; but flying stuff from A to B with an autonomous drone is pretty much a solved problem. Substitute stuff with people and that's going to be a reality soon as well. A couple of autonomous drone manufacturers were showing off their products at the Oskosh air show recently.

There are challenges related to non technical issues like certification, where to fly, when to fly, how to price, how to identify suitable places to land and drop the packages, etc. But that should all be solvable.

Why do deliveries need flying drones? A truckload of wheeled delivery vehicles that could be released each time the truck stops might be more practical in suburban environments (and flying drones don't make as much sense in a skyscraper environment either).

There are some domains in which delivery is highly time-sensitive. Food is one canonical example, best illustrated here: https://idlewords.com/2007/04/the_alameda_weehawken_burrito_...

Burritos are particularly robust to delivery (along with Pizza, Chinese takeaway, and curries). Most finer meals, and even items such as sandwiches or hamburgers, aren't. I don't see a large market for duck a l'orange avec quadcopter however.

Even food deliveries, though, probably don't pass the sniff test.

For me, seeing some new fancy tech prototype or proposal that looks to be unsuited to general commercial or residential use, I tend to see a submarine military application. Drone distribution in a combat environment could conceivably make sense (though it would risk blowing cover in many cases). Emergency response, research, fieldwork, construction, and disaster response are a few other potential fits. Much less price sensitive, high time sensitivity.

But mass delivery of random consumer stuff in an urban or suburban environment? Not very likely.

I suspect the problem just sounds too cool and is sufficiently engineer-catnippy for anyone to say "no" in time.

Not that this is very relevant to drones, but I've found over the past ~year that burgers survive being delivered surprisingly well, and the burritos I've gotten have been disappointingly soggy by the time they got to my door.

In Korea McDonalds has a long history of delivery. A typical franchise will have 10 red and yellow electric scooters parked outside. Typical delivery time is 15 min or less.

Tacos are more drone-friendly than burritos.

https://tacocopter.com/

Says the inventor of the pizza menu. You have a thing about food!

As soon as you say "food" I imagine cans of beans being dropped into windshields or onto heads.

Because a particular vision of the future requires everything that can be done to be done via "Flying X." Even if they're literally worse at everything than the alternatives, cost more, are less reliable, etc.

We've had flying cars almost as long as heavier than air flight - they're not some future tech that has yet to be invented, as much as everyone pretends that they're some magical thing just around the corner. We've had them for over 100 years, and they've sucked as cars, and sucked as airplanes for 100 years. Any materials breakthrough that will "make flying cars possible and too cheap to meter!" can also easily be applied to both cars and airplanes to make them better (see carbon fiber).

Drone delivery is worse than wheeled delivery in just about every option, but "flying is the future," so people have gone down that path, dorking around with it for a few years (which, honestly, sounds like a ton of fun). The results end up being quite reliable, which is "There is no way in hell we can actually make this work in any practical, economical sense."

I don't really agree that wheeled delivery is superior when you start to consider how dangerous and varied sidewalks can be - does your wheeled vehicle have the ability to easily manuver stairs or circular stairs? Can it make it over sidewalk fringes without issue - what about when there is snow and ice everywhere? Do you want your drones to have to navigate the social interactions required to ask someone for right of way on a sidewalk or avoid getting in the way (and tripping) a couple that's moving a couch?

I think that when you think of sidewalks you're thinking of how they look in your town (assuming you actually have sidewalks) they look quite different in different areas - in the southwest they're wide, expansive and quite level - in new england they're broken frequently by tree roots and frost heaves and, honestly, frequently just don't exist at all in a lot of the cities.

I do agree that people fetishize "Flying X" but saying that drones are worse than wheels in just about every way is pretty ridiculously hyperbolic - they have tradeoffs and it's quite possible that the market winner will be whoever sends out a van with wheeled delivery vehicles for most runs that can call in a drone if it runs into difficulties.

> but saying that drones are worse than wheels

I meant wheeled drones, not flying ones.

Oh sorry for the confusion - but yea I read you right... I just didn't want to type out wheeled drones each time. I definitely understood the point.

> Even if they're literally worse at everything than the alternatives, cost more, are less reliable, etc.

Drones may be better for very quick delivery or where road network is very poor.

See https://news.ycombinator.com/item?id=28070084 with amazing https://www.youtube.com/watch?v=jEbRVNxL44c (delivery of medical supplies in Rwanda) and mentions ship-to-ship transport.

Wheeled delivery vehicles do not make sense to me, because they can be easily ran over by cars (limited view bc of very low height). Many suburban roads do not have sidewalk on both sides, but bushes can be seen almost everywhere. Actually, bodies of dead animals ran by cars are so common...

Add a flag on top.

I think there are far more obstacles and safety considerations on the ground than in the air. Kids running around, branches off trees, long grass, gates, other vehicles, etc. An autonomous ground vehicle couldn't reach my front door, but a flying one would have little problem.

Step 0 of improving package delivery is to make sure every location has a simple package locker available.

Police have run bait car programs, and youtubers have run successful bait package schemes...

Don't see why this couldn't be scaled up to a large number of bait packages at pretty low cost... you just need a small GPS tracker embedded in the box.

Then what does law enforcement will do with the perpetrators? If they don't investigate this [0] will they really go after package thieves?

[0] https://youtu.be/arss9V3A1VA?t=95

I thought you were going to suggest a simple drone landing pad at every location. Plus, you'll need to provide a way for the drone to connect to power for a bit of a recharge. Your Alexa will be able to share wifi with the drone so it can map out its next route while it's charging.

You are way ahead.

Think 3 story apartment building. Where do the packages go? Dedicated locker, lobby, porch, sidewalk?

For the housing prices in SF,SV,Seattle,NYC, I just assumed they all come with their own helo pads

Passe.

Starship ports.

Gotta move to a container house in Boca Chica for that

And here I thought drones might be loud.

A pilotless rocket is a drone.

If a rocket launches and no one is around, does it make a sound?

Matternet was working on a home drone docking station for a while although I think theyve pivoted.

https://mttr.net/

My building has a very safe package delivery location: The roof.

If we could get our packages delivered there, fewer would be stolen, and there wouldn't be drivers trying to get in the front door and failing all day.

If only 10-20% of delivery addresses have such a location, it's enough to start meaningful delivery to those places in cities.

I think 10-20% might be an overestimate, at least in my experience in the US. I don't think I've ever been in a building that has an accessible roof. Maybe if you could precision target second-story balconies you'd get 10-20% in some regions?

It depends on your locale - in southern cities this number is going to be quite a lot higher than 10-20% since a decent number of even detached houses have flat roofs because snow isn't a thing - but heading into northern altitudes only apartment/condo towers will tend to have a flat roof (and some of them even have peeked roofs due to the snow).

By "such a location" I mean a reasonably safe and drone accessible spot to drop packets. Not only my specific roof situation.

If one drone can drop it on the roof, can't another one take it away ? Pirate drones!

Captain Drone Sparrow!

This would actually be a thing.

Roof-based drone counter-measures. See? An unnecessary, job-creating industry is born.

How do I invest in the nascent combat drone industry?

See the recent Armenia vs Azerbaijan conflict for a list of combat drone companies to invest to ? ;-)

As someone who lives on the -1st floor of a forty floor building I am strongly opposed to this approach - I'd rather have installable balcony landing pads be the required solution.

Then you should not choose that delivery mode for your orders.

Think of Amazon's drone delivery program the same way as you would about Blue Origin: an extension of the PR machine, not something that is meant to move the needle in a practical way.

I doubt many people really want obnoxious drones flying through their neighborhood anyway. I know we already have cars and trucks which are noisy and potentially hazardous, so maybe people would just get used to drones too - but drones are just inherently unsettling at some basic level. I'm not sure why.

I would predict the issue of interference from third party ground lifeforms to be the largest obstacle to over come. Heck, even a kid with a slingshot could do a lot of damage.

Windows seem to do fine, despite being equally being vulnerable to slingshots and being sitting ducks.

Windows aren’t despised, though

(I love drones)

Nor is there a challenge to shooting a window. Nor is there potential for getting a "prize" for shooting a window. Playing a game amongst friends with slingshots (or other drone-incapacitaing devices) where the one to disable the drone gets to keep the delivery. Hell, it doesn't even have to be a game played amongst friends.

I was out flying with some friends in a spot on an abandoned golf course. Being super respectful of the houses around the edges, flying pretty much in the middle of the course. Some dumbass decided our drones were good skeet targets. The police were very unkind to him and slightly peeved with us... it was a number of years ago.

Windows don't deliver random gifts if you break them. [1]

[1] Does not apply to professional window breakers.

At some point, the local Corvidaes (e.g. crow, raven) will figure out that some drones are carrying food.

Or birds of prey becoming pissed and more territorial.

If they do start allowing drones to do deliveries in cities. Please, please, someone get a grant to study if the bird start attacking any particularly colored drones or drones that originate in certain places. I really want to know if the Walmart (probably carrying food and other supplies) gets attacked as much as say a GrubHub drone.

I like the blimp mothership idea.

> I'm not sure why.

Small drones are 80+ dB (louder than a vacuum cleaner). Delivery drones are even louder.

> but drones are just inherently unsettling at some basic level. I'm not sure why.

Drones are a perfect surveillance or weapons platform. You could really mess someone up with a drone and most people intuitively get that.

> but drones are just inherently unsettling at some basic level. I'm not sure why

I'll give two reasons, though doubtless there are more.

1. They add a third dimension of visibility, thus decreasing or eliminating outdoor privacy. Currently, I can sit on my back porch and watch the kids play in a suburban, fenced yard. With drones overhead, the privacy we currently enjoy is lost.

2. They remove what reasonable boundaries for "safe zones" we have left. For the entire lives of everyone now living, "don't play in the street" has been equivalent to "play in the grass where you're safe." With drones, instead of "look left and right before crossing the street," it's "look left, right, and up, and estimate the trajectory of any drones at all times."

The privacy point may be somewhat valid, although 2-story houses in an average U.S. suburb already easily put most yards in view of neighbors' upper story windows. And any drone specifically used to spy on neighbors will either be relatively close and thus fairly loud and noticeable, or have nice camera gear that would already enable a lot of spying even without a drone (e.g. in upper story windows, again).

The safety issue doesn't strike me as compelling. I don't think drones introduce much physical danger beyond, say, playing baseball in the neighborhood. Obviously if the sky became very crowded with delivery drones, or crash landings became at all common, this would be an obvious reason for heavy regulation, but I don't see this posing much of a danger.

I worry about falling drones. How far above ground did the drones fly? How fast were they flying? How heavy were they when you added payload? What material were the blades? Was there any protection or cushioning? Where do they intend to fly? Above streets or above homes or above sidewalks?

Worrying makes sense. I worry the same for much, much larger delivery airplanes that probably pass overhead multiple times per day. But there's a difference between worrying and outright opposing.

Large commercial drones have parachutes.

I've been under the impression that it takes a considerable distance for parachutes to come into operation--while drones fly fairly close to the ground.

From what I’ve read, main parachutes are packed so that they open slowly, and BASE jumpers can design for a rough ride that starts below 200 feet.

Maybe a drone could constantly drag an open chute because it’s not trying to get anywhere near terminal velocity?

> or crash landings became at all common

Many things can cause a drone to crash. Motor failure (from gunk/over-use/loss of power/birds/bugs/...) to unseen obstructions (eg: branches/power lines) to user error. My bet is that like all things Amazon, there will be a race to make everything as cheap as possible which includes the labor watching over the operation of individual drones. This makes the perfect recipe for balancing crash statistics with money invested... which means crash landings will become common.

At some point, the question will be, "Is it cheaper to maintain a fleet of cheaply constructed drones or just replace them through attrition?"

I don't doubt that a delivery company would want to do things as cheaply as possible, but I also don't oppose regulations to prevent them from skimping on certain safety features. We do, for example, require Amazon delivery vehicles to follow all vehicle safety regulations (or, if there are cases where they're not enforced, they obviously ought to be).

It might be reasonable to worry that delivery drones will suddenly take over cities before regulations can even be put in place (although that apparently hasn't happened yet). But it's not reasonable to assume that there's something unique about drones that will make them inherently more impervious to reasonable regulations and thus inherently much more dangerous than, say, existing delivery automobiles and airplanes. In fact, I would expect any remotely competent deployment of delivery drones would be much safer than the delivery automobiles they replace, simply due to the baseline accident rate of automobiles.

> I also don't oppose regulations to prevent them from skimping on certain safety features.

> I would expect any remotely competent deployment of delivery drones would be much safer than the delivery automobiles they replace, simply due to the baseline accident rate of automobiles.

Why would you expect that a brand-new, mostly-unregulated technology would be safer than one that is over a century old with a lot of safety regulation in place? Even over the last several decades, safety regulation has reduced the per-mile as well as overall death rate of motor vehicles. [1] Regulation takes time and is often reactionary to large-scale problems.

How do you expect to define a "remotely competent deployment of delivery drones" given the general lack of experience in doing so?

Given the employee churn described in the article, do you think there is a culture of safety in this organization so deep that finding a fundamental+fatal flaw would be grounds for ceasing all drone-deliveries until the flaw is provably fixed?

From the article: “There was a lot of decision making made in the moment without long term thought to it. It was almost slinging shit at the wall and hoping stuff would stick.”

How can you justify better safety from something developed with constant engineering churn, short-term thinking and unreasonable time goals?

[1] https://www.iihs.org/topics/fatality-statistics/detail/yearl...

> Why would you expect that a brand-new, mostly-unregulated technology would be safer than one that is over a century old with a lot of safety regulation in place?

Because they have a very different risk profile. Regulations are trading heavy risks with cars. Drones are smaller and don't share space with humans on the ground during normal operations. That means they "only" have to coordinate with other air traffic while not falling out of the air at too high a rate.

It's easier to understand typical politics concerns in the US through the lens of peoples dreams.

A modest and common dream is to live in a place where you are free to do whatever you like in your yard without concern for neighbors e.g. A house in the woods/rural area. A simplification of that dream is to have fences in yard.

If you remove a dream from possibility Americans will often react negatively even if the reality is that only a portion of them will achieve that dream in their lifetime.

I think that's a pretty universal reaction in that case.

Socialism never took root in America because the poor see themselves not as an exploited proletariat but as temporarily embarrassed millionaires.

- John Steinbeck

Well, that and socialism sucks for the poor. But at least it also sucks for everyone else, too!

50 pound drone seems inherently more dangerous than a baseball.

Especially with spinning scythes.

It's interesting tech, but a nightmare for the liability lawyers.

Inherently? I'm not so sure. Of course if you're going to get hit in the head by one at full speed, you might choose the baseball (although not a great choice either way). But if you're choosing between a competently-operated delivery drone delivering a package to your house, or children playing baseball in your backyard, I'd probably guess that the drone is less likely to cause injury or property damage.

Well, drop a 50 lbs dumbbell on your car from 2nd floor and see what happens. Now if that thing falls on someone's kid running in the neighborhood...

And then you probably don't need a whole lot of drone deliveries to outnumber the kids playing neighborhood baseball.

Kids around here are barely playing catch in their yards.

But when we say “inherently more dangerous” I assume we’re incorporating the likelihood of that actually happening. A dining room table dropping on a kid’s head would be very bad too, but we don’t generally think about tables as being “inherently dangerous.”

Tables don't fly through the sky above urban areas as part of their normal operation.

The neighbor's window is a fixed consideration: you mentally map out who can see into your yard once and you're done. The possibility of a drone requires that mental calculation to be played out on a constantly repeated basis.

What about the danger of drones falling into streets, or dropping their payload into streets? Or windshields? Or onto people's heads? Think of gusts of wind. Think of a sudden rainstorm. Think of how rain changes the reflectivity of everything, making navigation much more difficult. Think of snow. Think of sprinklers. Think of a faulty battery dying. Think of batteries catching on fire midair, drone on fire dropping onto a home, or an apartment building. Think of the drone losing power midflight. The fact that they threw people at this means it was clearly a marketing tool, and not a serious engineering project.

That privacy is already lost with renconaissance, you only just have the illusion of privacy. You always only owned the land, not the airspace above it.

I never find this sort of argument convincing. There’s what’s legal and there’s the de facto circumstances. Yeah, can fly drones and peep in people’s windows right now, and that’s legal, but people don’t actually do that. Amazon has the actual money and logistics to indifferently blanket your neighborhood with noisy and nosy drones.

Privacy is and always has been an emergent legal construct to counter advances in surveillance and intrusion capabilities.

I'd rather formulate that as privacy is a fundamental concept, but its legal implementation is reactionary.

That's similar in notion.

Privacy is inherent in a nontechnical world, and created a set of practices and expectations generally. Yes, there could be an immediate intrusion by those sharing the same place and time, but improvements in information technology created an ever-growing set of methods to violate privacy remote in space, remote in time, and pervasive in both what it could perceive and record.

If you look at how expanded notions of privacy developed, largely beginning in the 19th century (though with obvious precursors, see the US Bill of Rights and some earlier legal protections), those emerged in response to specific technological threats: printing and publishing, photography, telephony, improved optical sensing, recording, and increasingly over the past half century, digital monitoring, data storage, and analytic tools.

There's a substantial if fractured literature on this, much of which begins by observing that there's no generally-agreed-on definition of privacy. "An emergent response" is my contribution to that.

Warren & Brandeis, Solove, Nissenbaum, and Lepore are amongst the authors I've found useful.

Not to alarm you, but I'm pretty sure that satellites can see you and your kids in your back yard - and they're taking pictures.

From what I’ve read, the highest commercially available satellite imagery is ~0.3m

Which would make one’s head take up a whole pixel.

Now, I don’t know what the best military satellites are capable of, but I’d wager there’s a reason many companies, governments, and armies still find airplanes and drones (and even blimps sometimes) very relevant today for ground imagery.

Most “satellite” imagery is actually aerial photos taken from an airplane which can easily get to 10cm. At least on maps websites; local governments typically give away the imagery they pay for. Of course that’s not taken all the time, so it’s not exactly spying. We do have satellites that film the whole hemisphere constantly, but GOES only has 0.5km resolution or so — depending on latitude, of course.

Do you suppose people felt the same way about automobiles before their use was widespread?

Not really. Horse drawn carriages were also dangerous to get near, iron shod hooves on cobbles were loud, and they left big piles of unsanitary dung.

I recently watched this video which made me think maybe we're taking the loudness of traffic for granted:

https://youtu.be/CTV-wwszGw8

It seems that there are a number of proven solutions to reduce automobile noise when we put in the effort, which IMO is a huge benefit for quality of life.

I find the noise of drones much more annoying than trucks and cars. There's something about those small rotor noises that really grates on me. I really hope this never kicks off.

Yes, the trucks can be pretty bad for noise already. The payoff of convenient delivery makes it just about acceptable. Taking the noise to another level won't go down well in many locations and doesn't offer that much more for the consumer.

I find them to be annoying too but I imagine it's the same as the car superseding the horse on streets i.e. "get off my lawn" more than anything else. When they really bug me, for the same reason you have, is when I'm away from people, roads and industry and I hear them.

Same here, the high-pitched whine reminds me of a mosquito buzzing around my ear - it's just so intensely irritating!

> The systems required to make drones land outside people’s homes were heavy and Amazon’s drones ballooned to about 27 kilograms…

> “The hard bit is the last two metres off the ground. It’s astonishing what machine learning can do, but it’s also astonishing what it gets wrong,”

First child killed by a delivery drone and it's over.

This. People will be using them for skeet shooting practice after that.

In the US maybe.

The Chinese will keep perfecting the system regardless.

> First child killed by a delivery drone and it's over.

I am doubtful, given that there are 4000 kids killed annually in the US by cars and a further 3000 by firearms. These are relatively easy to legislate (compared to disease, drowning etc) yet are the two highest causes of death.

https://www.usnews.com/news/health-news/articles/2018-12-19/...

The sound of a drone and the sound of a car or truck, even if the volume is the same, is very different. Drones put forth a lot more high frequency noise, which is far more irritating (and harder to block out) than the relatively low frequency of cars and trucks.

I know a few people involved in drone delivery projects. Their employers are always just about to launch commercial services, and it has been that way for several years. There was to be a big trial of food deliveries within a university campus, but the local aviation regulator stopped it from happening due to a load of unresolved safety issues.

The problem is that drone delivery in urban areas is difficult. Accurate landing issues, noise, kids throwing stones at them, overhead wiring, managing safe landing when there's a fault etc. Oh and the big one, air traffic control. The authorities want drone operators to co-ordinate their flights so that they don't crash in to each other. The clients of the drone operators are obsessed with commercial confidentiality and don't want that to happen.

I'm in Ireland the only type of drone delivery that seems to be successful so far is delivering medicines and the like to small islands off the west coast where most of the above issues don't apply. It seems to me that drone delivery will end up being a niche thing. Ground based robot delivery systems will probably succeed where drones fail.

I think this is the first time Amazon made people redundant after they ramped down a project. Not that people aren't leaving after that, but being made redundant? Usually people either leave "voluntarily" or go to ther functions within the company. Not sure what to make of that, IMHO that's untypical for Amazon.

It sounds like the people that are being made redundant were mainly tasked with labeling data (which also matches them being temps instead of full time employees, and them training replacements in Costa Rica), so I don't think they're software engineers that can easily switch to other teams.

These articles take a pretty short term view to amazon's efforts.

They are a global company. Is it possible that the UK regulatory environment was a bit less amendable to drone delivery? What about switching focus to areas with poorer roads / delivery networks etc.

Ie, go to africa, go to costa rica, where a drone will be going over trees instead of houses (who will complain?).

Do not underestimate how much Amazon spends on shipping alone. I'd guess $50B easily per year.

So when we say their delivery dreams have collapsed (even in an area) we need to benchmark against what they are doing / focusing on. Where I am Fedex Home Delivery is TRASH.

One week delays easily - sometimes two.

Meanwhile Amazon delivers. So they are outcompeting a company focused on delivery. That's a bit crazy?

Now let's look at innovation. Amazon are doing lots of weird things that get negative writeups. Their key program where they deliver INTO folks houses (!). Locker deliveries into lockers. Car trunk deliveries. I'm curious if the companies 100% focused on logistics (Fedex / USPS etc) are trying out these ideas?

They are going a bit crazy with warehouses and went from 2 day to 1 day to now same day deliveries - https://www.amazon.com/b?ie=UTF8&node=21382124011

Where I am they deliver till 10PM at night in waves.

They seem to have contracted up an airline network for their own package backhaul and consolidation.

With billions per year in shipping - I'm not writing them off yet - even their drone program (though maybe it morphs into ground based or other approaches). That said, I do think that the US and UK markets may not LIKE drones flying over houses nonstop even if things are technically possible.

> Ie, go to africa, go to costa rica, where a drone will be going over trees instead of houses (who will complain?).

It’s a funny thing I only realised after spending a couple months in West Africa: Amazon is pretty much non-existent in Africa. They ship only to to more or less a dozen countries there, and a third of those are North-African countries.

And I’m not sure it’s a logistical problem. In 2014, less than 15% of West African countries’ populations had a bank account, let alone a credit or debit card.

And while mobile money definitely is taking off, cash is still very present and… well, most people probably don’t earn enough to make it worth operating there for Amazon.

Just a quick search gave average salaries for Senegal at less than XOF100k (~$180) / month.

In Ivory Coast, it would appear to be ~XOF300k.

Just looking at the prices on Amazon makes me seriously doubt it would be worth the effort.

And that's without even taking into account how different doing business can be in some countries.

I had to look it up so I thought I'd save future readers a moment or two:

<wikipedia> The West African CFA franc (French: franc CFA; Portuguese: franco CFA or simply franc, ISO 4217 code: XOF) is the currency of eight independent states in West Africa: Benin, Burkina Faso, Guinea-Bissau, Ivory Coast, Mali, Niger, Senegal and Togo. </wikipedia>

https://en.wikipedia.org/wiki/West_African_CFA_franc

Quite thoughtful. Thank you!

> Meanwhile Amazon delivers. So they are outcompeting a company focused on delivery. That's a bit crazy?

This is what I don't get. Where Amazon will enter and who it will need to defeat has been abundantly clear for well over a decade. But all of their competitors have just chosen to lie down and slowly die.

Automation can't deliver packages either, lol

My naive assumption is that regulations (appropriate or not) is what prevented this from getting off the ground.

sure

> UK regulators also fast-tracked approvals for drone testing, which made the country an ideal testbed for drone flights and paved the way for Amazon to gain regulatory approval elsewhere.

The only hurdle was that the drones had to be so heavy to handle packages that it put them in a higher weight class subject to more rules. But other than that, the article mentions that the UK and the US regulatory bodies have both approved the project.

I wonder what the stats were on weight of packages vs required drone weight. Ie where is the cutoff of package weight, to ensure that the money they invest in drones would affect X percent of sales?

Eg i imagine if they made drones for a package weight of, say, 10 ounces, they could.. hopefully, make smaller drones and perhaps have an easier time with regulations. However the package size limit is so small that the drones aren't likely worth it.

So as we increase package weight, we include more packages and percentage of sales... but also increase drone size and regulatory issues.

I imagine there isn't a "sweet spot" otherwise they would have done it. Nevertheless i'm curious what that axis looked like.

I don't know about the UK, but in the US the FAA required that each drone be operated by a licensed pilot, which changed the economics.

There's some talk of autonomous software being granted a pilot's license, so that might shift things again.

It is literally regulations that keep unproven aircraft grounded, yes

Only if they’re uncrewed. Any idiot can legally fly an ultralight experimental aircraft with basically zero oversight. I don’t think you even need a pilot’s license. But God forbid you try to do the same thing without putting a pilot at risk; that’s effectively illegal.

That makes some amount of sense. If you put the ultralight pilot in the craft, they literally have skin in the game. Further, ultralights cannot be flown over “congested areas” of a town/city.

Trying to apply those rules to drone deliveries while ignoring those two pretty fundamental elements doesn’t strike me as evidence that the FAA is being arbitrary or capricious.

Idiots flying ultralights are a problem that’s going to self-eliminate quickly.

Which actually makes sense.

When you're risking your own life, you'll likely be at least somewhat careful.

When you're just flying an empty test vehicle you'll be much more reckless, and thus way more likely to crash it on other people's property or other people.

Hence, much more tightly regulated.

Part 103 aircraft don’t need a license of any sort. There are of course a bunch of regulations about how/when/where you fly it, but you don’t need any official training or certification.

> Any idiot can legally fly an ultralight experimental aircraft with basically zero oversight. I don’t think you even need a pilot’s license.

But they have limits:

No passengers allowed.

No flying over towns or settlements.

No flying at night or above (or in) the clouds.

No flying in airspace around airports with control towers and certain other airspace without prior permission.

No commercial operations (for hire) except instruction.

Ultralights must yield right-of-way to ALL other aircraft.

Yup. Zipline in Rwanda and Ghana have made over 150,000 commercial drone deliveries. And have done so safely. (Their autonomous winged drones have a range of over 300km on a charge.) https://www.morningbrew.com/emerging-tech/stories/2021/06/30...

Only token demonstration projects in the US, unfortunately.

They've found a niche there where drone deliveries can work and be worthwhile. Getting lightweight but vitally important payloads (medical supplies) quickly to locations hard to access by road.

Amazon is in the business of delivering bulky and heavy packages, often of low value and importance, and poorly-but-cheaply packed into oversized cardboard boxes. I can't imagine a very large percentage of Amazon deliveries would be in any way suited to drone delivery, especially when you've got to limit it to destinations with a suitable landing/drop zone, within a certain range of a depot but away from dense urban areas (too high-risk to operate in)

Yeah, I live in a relatively rural part of the US and can drive hundreds of pounds of stuff at 60 miles an hour to almost anywhere within a few hundred miles.

That doesn't eliminate rapid delivery of small packages as a business model, but it really constrains what is worth shipping using a drone that can carry a couple of pounds.

Right. But a lot of Amazon deliveries are single packages, and probably most of them are within the ~4 pound weight limit of current Zipline drones. 80mph, straight-line one-way delivery without requiring a dedicated human driver is pretty effective for rural delivery (and the drones have a range of 300km per charger, currently). But it really is more like Starlink than fiber. It's better suited to underserved rural areas where it's impractical to do single-package delivery.

And I think the current ~$15-20/delivery price for Zipline is too high to be terribly practical for most items, but in a rush (and given 24/7 availability), it's absolutely worth it. I can think of many personal instances where a $15-20 delivery charge would easily be worth it to trim hours off of delivery time.

I think at this point it's likely safer (i.e. to pedestrians, etc) than delivering the same item via truck. I really do think this has a future, and I'm glad that they've reached a sustainable level of operations in Rwanda and Ghana.

My former startup got Jersey City to draft legislation that fully paved the way for delivery drone to operate in the city, they even went as far as to have the FAA clear their draft legislation, basically on the premise that drone companies would take it seriously, talked to Prime Air and they just shrugs and said sorry, not interested.

Can we all allow us the freedom to fail after trying than not having tried at all?

For non-food in cities there is no fundamental reason why drones should be any more efficient than ground-based delivery.

Flying takes in general more energy than driving and ground-based delivery is not more dangerous or much more expensive.

Energy could be disregarded as (with renewable sourcing) it would be essentially free in the amounts needed for drones.

Drones would offer increased speed of delivery within boundaries, plus scaling without the tricky problem (for Amazon it seems!) of directly employing more humans. (Obvs humans needed behind the scenes.)

Energy will not be free for quite a while, if ever.

And while energy for a single flight of a single drone is minimal, having tens of thousand of delivery drones, operating 24/7 (or 16 hrs a day) is another matter.

Even if cost of production of energy is 0 (and it's not. even for renewable, solar panels need cleaning and occasional replacement, wind turbines need regular maintenance, etc.), there is a large cost of maintaining electric grid.

In some places on earth, local solar installations, might be enough, but for most of the planet, we will still need normal grid, and probably even expand them, if we plan to migrate everything to green electricity.

I don't know how much electricity you pay where you live, but for me, only around 50% of the bill is actual usage.

And since other forms of energy(like gasoline and natural gas) provide significant tax income, this will eventually mean governments will raise electric taxes to compensate, when and if electricity replaces them.

So I would not be surprised if cost of electricity stayed the same.

I’m surprised there isn’t a startup doing ballistic delivery. Puts all the heavy components in the end that doesn’t leave the ground and gets reused.

Looking forward to having Bezos shooting missiles around my city

Ballistic delivery would likely be unpredictable due to changes in the wind, which could result in dangerous or destructive accidents. I suppose there could be some ballistic projectile with a low-power guidance system.

There's an American company[1] in Ghana sending medical supplies around the country with small launchable planes.[2]

1: https://flyzipline.com

2: https://www.youtube.com/watch?v=jEbRVNxL44c

That's not at all ballistic though.

that was an amazing watch. thanks for sharing the links. !

I recall reading about narcos doing ballistic delivery over the border.

There's SpaceX

https://www.cnbc.com/2021/06/04/us-military-rocket-cargo-pro...

T-shirts are already frequently distributed using ballistic methods at sports events. Just upsize it.

Why drones? I’ve been seeing small four wheeled robots old traveling around SoCal sidewalks. These small delivery machines don’t need to fly.

musk should rebrand his boring stuff as internet of things underground

Delivery drones was always ridiculous on its face. I don't know why so many took it seriously. I blame the media miseducating people about what was technically and economically feasible.

Drone delivery will never be economically competitive with other sources of transport based on the materials energy balance woes of lithium ion batteries or even lithium ion phosphate batteries.

The Lithium LCOE(which is already a perturbed metric) is several times that of other sources of energy which makes the whole economic action moot.

I also imagine buccaneering teams hunting drones for their batteries and cargo. It will be a poetic justice: after years of mislabelling copyright infringement as 'piracy', Amazon will get a taste of real piracy.

Delivery drones are constructed for fairly narrow usage scenarios, and will get more specialised with time. For short hops with light loads air drones are already better than traditional solutions if they are allowed to be used as intended.

Air drones span orders of magnitude in cargo capacity, speed, distance. Drone ships, drone cars, ...

Don't use a hammer to drive a screw.

Drone delivery looks cool and does have real potential in some cases. But when looking at 'standard' deliveries in residential areas drones don't look practical or attractive at all.

Oh, we can but hope. I was not looking forward to having to put up with the sounds of drones flying around the neighborhood.

They should target employees who live and operate in remote areas like O&G.

How practical was the idea? Anybody thinks this is the future?

I think drone delivery is the future for certain narrow use cases.

I don't think it's the future for common delivery in towns/cities, at least unless someone can figure out how to make them quiet enough so that people don't get mad. No matter what the FAA says about it, cities and towns have the authority to disallow them, and if the townsfolk are mad enough, they will.

> an employee drinking beer at their desk in the morning

What’s wrong with that ?

If you write down every step of drone delivery you should see flying drones can't do general delivery. But there is a lot of cool stuff they might do, but you have to be specific, and also write it down using numbers not hand waving.

"an employee drinking beer at their desk in the morning"

How TF is this news, on Wired. How has the scene become a bunch of accountants and proud.

And it seems it was lame anyway -

"It was about 11am or 12pm and this guy just had this open can of Stella,” explains one former employee. “He took it out of a fridge and popped it open at his desk.”"

Wall Street did coke off hookers in their offices if movies taught me anything. Now we are down to one beer by one employee at 'lunch' shows morale has been destroyed. It is good proof removing lead from gasoline as severely disrupted society, perhaps we should put it back, not 80's wall street level which was problematic but at least something, half the amount of lead.

> How TF is this news, on Wired

Wired was always a gossip magazine, but now it is just a gossip magazine. It's sad.

I sent AOC a dm on Twitter when she had them open asking how on Earth she doesn't pursue a drone delivery ban, considering how many delivery drivers it would put out of work. No answer.

In a previous role, I worked with one of the leads of the prime air team, a European man whose name is curiously that of a popular Asian food. Anyways, he impressed me 100% as a schmoozer/ladder climber, so I can't say that I'm particularly surprised by the result if he is the kind of person who was running the program for 5+ years. He also (before prime air) led a team of like 4 or 5 very senior(principal+) engineers who basically stole the credit for a certain project that was actually done by my team of 8 mid-level/fresh out of college SDEs

Reply on news.ycombinator.com