While Amazon makes plans to begin drone-based package delivery in the developed world, a start-up called Matternet is plotting to build a drone-based package delivery service in low-income countries. Many rural roads in developing countries become inaccessible for months during the rainy season, cutting villages off from nearby cities. In those cases, lightweight, autonomous aircraft could be the fastest and most cost-effective way to move high-value cargo across the countryside.

I talked to Matternet CEO Andreas Raptopoulos on Monday. He explained how his company's technology works and how it could revolutionize the future of package delivery. The transcript has been edited for length and clarity.

Timothy B. Lee: Can you describe the vision behind Matternet?

Andreas Raptopoulos: The thing we've been seeing as the big potential of this technology is to create a physical Internet: a network that would allow us to transport packages between locations in an autonomous way. It would be a new type of transportation infrastructure.

Imagine if this type of network could be as big as mobile telephony. Imagine if you had transportation networks able to solve the last-mile delivery problem. It could be frictionless, cost-efficient, energy-efficient, and very scalable, with very little infrastructure investment.

In the future, this is something that belongs to cities and megacities in the rest of the world. [The technology can work in] places of high need like places in the developing world without infrastructure, [but also] places in the developed world where we have congested urban centers, where roads to exist but they're very inefficient because of traffic.

The economics are compelling. A lightweight vehicle can deliver a lightweight payload. When you have a small vehicle like a small unmanned aerial vehicle that is only 3 kg in its weight and it's transporting a 2 kg payload, it solves the most inefficient part of delivery today. If you commission a van, a car, a motorcycle, or even a bicycle, much more energy will be expended.

Consumers in the developing world don't have very much money. What do you expect to ship that will justify the costs of your system?

I'll give you an example where it makes sense to pay for a system like ours. There is a small country in South Africa called Lesotho that is suffering from an AIDS epidemic. They don't have a way to transport diagnostics from remote clinics to hospital labs. Transportation is one of the issues they're facing. We're working with them. It wouldn't be the final user, [i.e. the patient] it would be that organization whose mission it is to solve that problem [that would use Matternet].

In the Dominican Republic, we're working with the Ministry of Health and if you think about infrastructure investment, this is like orders of magnitude less than normal infrastructure investment. Our vehicles cost $3,000 to $5,000. Our stations cost from $100 up to $500. For a network of 10 nodes with 20 vehicles, we're not talking millions of dollars, we're talking hundreds of thousands.

And there are lots of other savings. You don't need people on the streets driving the vehicles. You have a level of reliability that countries with bad roads have never seen before. Eventually we'll have urban delivery with cost savings. It can be automated to a very large degree.

How is the development of the Matternet system coming along?

We have a solution that comprises three parts, the first is the drones. They have similar characteristics to what Amazon talked about yesterday. Second, there are ground stations where the vehicles land to swap vehicles or pick up a load. That's similar to what Amazon has demonstrated. Third is [software to manage the drones.] These are the elements we need to have in place, we're working all three of those.

We're developing technology, now getting ready for our next big pilot. We've done a trial in DR last summer. We want to do a larger pilot where we'll be on location for a few weeks.

What's the most important obstacle to putting this system into practice?

We have to get those vehicles reliable enough and safe enough to be able to make a claim that it doesn't pose a threat to public safety. And we have to operate them at a level where we can show enough data for the regulatory authorities to be convinced that that is the case.

We've come to expect a very high level of reliability. The technology will surely catch up with that, but it's going to take a few years.

We, as an industry, will want to get the technology working reliably in more low-risk environments before we take it into a complicated high-risk environment. Our strategy is to operate those networks first in places in the developing world or in rural areas where the risk of failure is diminished.

Are authorities in developing countries more open to experimentation than those in the United States?

The general answer is yes. The reason is because if you target an application that is a big problem, people are willing to take a higher risk. When people have no alternatives, they'll try a new technology and see how that works. So far, we've been greeted with a positive reception from governments.

There's a difference between operating in a trial mode versus operating in commercial operations. I think FAA is taking a very comprehensive look at this. We'll see where they land. They have already made a few assumptions about how the issue will be regulated. We have a framework of regulations that we can present and say this is the way we think it's going to work, that's why we think it's going to be that way, this is the type of operation we'd like to run, scale it up and so forth.