After Amazon's Jeff Bezos announced that his company wanted to deliver packages with small unmanned aerial vehicles, many people have questioned the viability and wisdom of the idea.
Yesterday, we got one optimistic perspective from Andreas Raptopoulos, an entrepreneur who founded Matternet, which is developing drone-delivery technology.
But there are many other ways to answer the questions that I posed to Raptopoulos. So, today, we bring you an interview with the University of Washington's Ryan Calo, who has become a leading authority on the ethical and policy implications of emerging technologies. Specifically, he's focused on the problems at the nexus of drones and privacy in recent months.
To offer the most intriguing parallels, I tried to keep my questions to Calo as similar to the ones as I posed to Raptopoulos as possible.
What rationale do companies like Amazon and Matternet give for creating drone-delivery networks?
Some people are saying Mr. Bezos made this announcement with no intention to carry through. I disagree. I think companies like Amazon and Matternet like drone delivery for a few reasons. The trend is toward immediate consumer gratification. We are actually behind China, for instance, where I understand hundreds of delivery services translate into fast, low-cost delivery in urban environments. A second, underappreciated reason Amazon might incorporate drones into their business model is to attract talent and investment. There is no denying that innovation is a currency in the tech world today.
Is there actually an efficiency bonus? (Or can one be imagined with some sort of system in the future?)
Today, drone delivery would not be efficient. Even if operated by a human being, drones cannot travel far with a payload. Amazon and Matternet are banking on gains in performance and lower costs over time.
Given your research, how close are we to this kind of future? How quickly could the Amazon Prime AIR future happen?
I am a law professor, not an engineer. But I do talk to roboticists quite a bit. There is a range, but many believe we could see what Mr. Bezos described within four years. Note that the first DARPA Grand Challenge involving driverless cars was in 2004. No team completed the challenge that year. Fully autonomous cars came within a half-decade, and are now within a couple of years of being commercialized (if you believe the statements of multiple car-markers).
What are the major challenges?
I see three categories of challenges. The first category, as we've alluded to, is technical. For drone delivery to work, you would need better energy sources, better software, and likely improvements in physical design. But note that companies and universities are working on these problems, as are thousands of hobbyists (ask Chris Anderson of DIY Drones).
The second category is regulatory. Mr. Bezos appears to have autonomous delivery in mind. The Federal Aviation Administration's roadmap to integrate commercial drones into domestic airspace specifically says (at page 33) that "Autonomous operations are not permitted" outside line of sight of the operator. Amazon and others would have to work with the FAA on this and other issues. There are related challenges around litigation including trespass, negligence, nuisance, and other potential lawsuits from angry citizens.
Which brings me to a third category: social forces. Many people find drones unsettling. They understandably worry about being hurt or watched. How people come to see this technology will drive adoption and legal risk. I will say that Mr. Bezos probably did the image of drones a favor just by delinking the technology to a degree from targeted killing.
What about range?
Range is a big issue, both because of the energy source, and due to the complexity that attends navigating more, and more diverse terrain. The further you have to go, the more that could go wrong.
How about reliability?
I think this is a very real, but again, likely solvable issue. The FAA is about to select its testing sites for drone use. Hopefully these six sites will help drone manufactures and operators make considerable gains in reliability. Note that technology need not be perfect to see widespread adoption. Hundreds of people die every year from falling out of beds.
How should we think about litigation around commercial drone flights?
I believe there will be two waves of litigation attending commercial drones. The first will involve tort claims related to various commercial applications. People will say a drone user trespassed on their land, violated their privacy, created a nuisance, or even physically injured them. I think the common law, being versatile by design, is up to the task of addressing these claims. I described the second wave of litigation in my article Open Robotics: drones will one day function as the equivalent of flying smart phones. Consumers will buy a drone from one party and download an app from another. This will pose a significant challenge, I argue, for product liability law.
Should certain reliability measures be required?
Yes, and this could take several forms. The FAA itself could require a certain amount of testing before issuing a commercial license to use drones, the way the state of Nevada does with driverless cars. Or courts could look to industry standards in determining whether a given accident falls below the requisite standard of care.
Won't people shoot them out of the sky?
A few people in Colorado might. I don’t think the phenomenon will be widespread, any more than vandalizing a vending machine is. Shooting drones out of the sky, except in defense of a person or maybe property, is illegal. The shooter could face criminal charges or a lawsuit from, in Amazon’s case, a well-resourced company. That is a pretty good disincentive. The greater fear may well be hacking drones—this technology will have be secure enough that bored teenagers cannot go shop-downing.
What does the US regulatory situation look like to you?
The use of drones for a commercial purpose is not allowed today. But Congress has directed the FAA to fashion rules to integrate commercial drones by 2015. The FAA has yet to give details on how a company like Matternet would go about securing permission to operate drones for delivery or another purpose. They have given some hints, however, about what technical challenges would need to be address and what restrictions might be in place, including around civil liberties.
What are the implications of this kind of drone network, if it scaled up?
I think it is important to consider the bigger picture. Drones won’t just deliver goods to your door. They will be used to deliver goods over a long haul. Or between stores. Imagine if you’re shopping at Best Buy and you ask after an out-of-stock item. How great would it be if Best Buy could fly one over from the next nearest store while you wait. If this proves to be economical and a competitive advantage, then I think the marketplace will (try to) deliver it.
You've specialized in privacy and drones. How do retail delivery networks figure into the overall privacy-drone nexus?
There are problems, but maybe not unique ones. Drones greatly increase the capacity for surveillance, including by corporations. But delivery drones are not necessarily architected to observe. Still, you could imagine abuses. What if, for instance, law enforcement were to ask Amazon for all of the drone footage in a given area? Or even require Amazon to alert law enforcement if its drones saw something unlawful? My hope is that the FAA will follow the advice of the Electronic Privacy Information Center (where I’m on the advisory board) and require meaningful privacy safeguards before issuing certificates of authorization. And I’m encouraged that the recent FAA road map does repeatedly mention privacy, albeit without offering too many specifics.
We want to hear what you think about this article. Submit a letter to the editor or write to letters@theatlantic.com.
