Random Thoughts on Drone Package Delivery

By now most people have seen or at least heard about Amazon’s Prime Air service. Supposedly, still in the R&D phase Amazon unveiled its drone delivery plans on 60 Minutes back in December. I heard many people dismiss Prime Air as a publicity stunt with no real chance of ever… ahem, getting off the ground. While I do think announcing it at this point is almost certainly a way to garner publicity, I don’t think that there is any reason it can’t be done.

Of the handful of companies that I think can make this sort of thing happen, at the scale they are proposing, Amazon is one of them. The incentive for Amazon is certainly there, not only would they be able to deliver orders faster but they’d be able to eliminate the middle man — USPS, UPS, FedEx, etc. — for a large majority (86%) of their orders.

That’s not to say that a drone operation of this sort is not without its challenges. In Amazon’s case the challenges are exacerbated because of the scale that it would be operating at.


Up until recently it was generally understood that FAA rules forbid flying any remote controlled aircraft (i.e. RC planes, helicopters, or “drones”) for commercial purposes. It’s a silly status quo. If, for example, you’re a real estate agent who wants to get aerial pictures/video of your listing and I were to fly my drone for you as a favor, that’s perfectly legal. If money were to exchange hands, however, then it’s a commercial flight and I could get in trouble with the FAA.

In 2012 the FAA was asked to create rules that would govern the private and commercial use of small drones. Most people thought that until those rules were written, commercial drone use risked FAA fines. A recent NTSB ruling, however, declares that the FAA does not currently have the authority to regulate the category of aircraft that drones fall under. That category is basically any unmanned aircraft that flies under 400 feet. The legal issue is far from over and the FAA will almost certainly seek to have its mandate extended to include unmanned aircraft.

The challenge for the FAA will be to craft regulations that are restrictive enough to maintain safety and sanity, without being so restrictive that it makes it impossible for a budding industry to form. It would be a travesty to see the industry stifled by regulation, or so regulated that it makes it impossible for all but the Amazons of the world to enter.

The fact is that people have been flying RC aircraft for commercial purposes for some time. What’s different now is the convergence of ease of use, relatively low cost equipment, and sophisticated control systems have made it a viable means of entertainment and commerce for even the average Joe or Jane.

In the case of Amazon what catches people’s attention is the massive scale of the drone operation. Imagine hundreds of small quadcopters flying around delivering packages.

Privacy, Safety and PR

The public is definitely going to voice concerns over the privacy and safety issues that the commercial use of drones presents. Any company that uses drones for package delivery, or any other reason is going to have to address those concerns ad nauseam.

In Amazon’s case they will have drones flying overhead, by necessity going over houses, through neighborhoods, etc. People are naturally going to be concerned about privacy, especially if the drones are equipped with cameras.

I think that Amazon can squash this concern immediately by simply not equipping their delivery drones with cameras. There is no technical reason why such a drone would need a camera. Autonomous drones of the sort that Amazon would use don’t need cameras for navigation — GPS, proximity and other sensors should be able to handle the task just fine. Besides, cameras add weight and would be difficult to use remotely at a distance anyway. The only reason I could see to have cameras on-board would be to, I don’t know, snap your picture as you stepped out to grab your package. Or, perhaps to transmit video if the craft was experiencing problems and had to be flown manually.

Even if the drones don’t have cameras, people may still have concerns about drones flying over their back yard. Some people are not going to believe that the drone isn’t watching them no matter what you tell them. Others are going to complain about the noise. Some may consider a drone flying over their house trespassing. Any company that is going to operate drones is going to have to be sensitive to these concerns, no matter how silly some may be. The quick and simple way to avoid a lot of these issues would be to have the drones fly at, or above 500 feet. That’s the minimum altitude the FAA says a plane can fly in uncontrolled airspace (i.e. the airspace above your house). While trespassing issues are mostly moot (even below 500 feet) flying the drones as high as possible would help avoid some of these issues. A small drone 500 feet in the air would probably go unnoticed, at least until it decided to land. Being as unobtrusive as possible will be paramount to garnering public acceptance of these drones.

Safety. For Prime Air, or any similar service, to be viable the drones would have to be at or nearly 100% autonomous. It does not make sense, and is certainly not scalable, to have people manually fly every drone. For a large scale operation like Amazon’s, there would certainly be a monitoring center where human operators would keep an eye on each drone, but human intervention would likely only occur if there was a very specific and narrow situation that necessitated it. Even then, it’s unlikely that any human operator would, or should, have any real control over a drone. The only thing that a human should be able to do would be to command the drone to come home, or land safely.

Not only does removing the need for human interaction make the service viable, it also helps eliminate and answer privacy concerns. If a human cannot operate the drone beyond a couple of simple commands (home and land) then there is no concern that a rogue Amazon employee could use the drone for nefarious purposes. I could imagine the need to reroute a drone that has either not launched yet, or is far enough away from the original destination for it to be practical, but any such change would be done through automatic systems, and only at the request of the customer.

The intelligence of the drones will have to be well beyond just being able to go from A to B and back. The drones will have to be able to self diagnose, handle in-flight emergencies, and navigate with exacting precision.

The fact is we don’t know how safe drones at such a scale are. Most of today’s drones are not even really “drones” in the sense that they are not completely autonomous, even billion dollar military drones are not flown completely autonomously. I don’t think it’s out of the realm of possibility to make drones safe though. Off the top of my head I can think of a few major operation parameters that would need to be addressed for a service like Prime Air to be viable:

  • Battery life that far exceeds the intended range of the craft. Or the operation of the craft well under the battery’s limit. If the battery is only good for 5 hours, you only operate it for 2.5.
  • No overloading of the payload capacity. If the drone can carry 10bs, you only ever have it carry 5lbs.
  • Ability for the drone to run under at least one motor or prop failure.
  • Ability of the drone to autonomously self diagnose issues, to detect and recover from faults. Like a prop or motor failure but also things like loss of communication, or navigation systems.
  • Ability of the drone to autonomously assess flying conditions. Is it safe to fly? Has the weather suddenly made it dangerous to continue flying, etc.
  • Ability to land safely in the event of a catastrophic in flight failure. Shut motors down and deploy a parachute to soften the landing perhaps.
  • Ability to land safely in populated areas. Can’t have the craft landing on the kids or pets in the yard.
  • Secure and limited communications. My instinct tells me that Amazon type drones will communicate via satellite or cellular networks. Perhaps both for redundancy. In any case the communications will have to be secure, encrypted and verified.

The Achilles heel in commercial drone use is that first accident. People will be looking, waiting, for an example to confirm their fears and get the endeavor thrown out. Since that accident is inevitable, it is imperative that systems be designed to mitigate its effects.

While there are some major technical challenges, I don’t think any of them are insurmountable in the timeframe that Amazon has specified (by 2015 if the FAA gets it’s act together by then). I think that the biggest issue that Amazon (or any other company doing this sort of thing) will face will be from a skeptical public. Again, and again, the number one concern people are going to have, naturally, is safety. Amazon will almost certainly have to run a huge campaign demonstrating the safety of their drones. Still, it will take thousands of hours of operations to truly demonstrate that drones of this sort can be operated safely at the sort of scale that Amazon is proposing.

For me personally, removing the bias of the “cool factor” as much as I can. I can see the benefits of Amazon’s Prime Air, I think that the benefits outweigh the risks and think that it’s a worthwhile endeavor.

Leave a Reply