Original URL: https://www.theregister.com/2013/12/02/amazon_drone_matternet_analysis/

Drone expert: Amazon's hypetastic delivery scheme a pie in the sky

Flying robot authority spills beans on challenges of tending a drone flock

By Jack Clark in San Francisco

Posted in On-Prem, 2nd December 2013 20:13 GMT

Analysis Amazon's plan to use semi-autonomous drones to offer fast delivery of superfluous consumer items has garned the e-retailer a lot of press, but the scheme may be more fantasy fluff than feasible.

In trying to ship products from distribution centers to consumers' homes via the just-announced "Prime Air" service, Amazon faces a number of critical technical and regulatory challenges that could cause headaches for Bezos & Co.

"The problems we encountered and we believe are common for anybody who would try this, is how can you make each vehicle very reliable for a cost that is attractive," says Andreas Raptopoulos, the chief executive of drone transportation firm Matternet.

Matternet has built a technology for drone-based transportation that pairs an Amazon-like copter with a distribution system, and a beacon to help the drone zero-in on its final location.

The company has run two field trials so far in Haiti and the Dominican Republic, and focuses on medical transportation such as the shipping of medical samples from remote clinics to hospital labs. Its system is spec'd for payloads of around 2 kilograms, which is similar to Amazon but with the added benefit of saving human lives rather than encouraging the never-ending consumption of Chinese-made crapola.

In rolling out the tech, Matternet has run into some of the technical challenges that Amazon will face, and has begun to figure out how to work around them.

"When we went out to Haiti it was a humid environment, quite hot, [so] we had a few equipment failures," Raptopoulos notes. Dust is bad as well, he said, because it can get into motors and cause problems.

"These are all problems that can be solved and in fact have been solved by manufacturers of vehicles that cost $100,000," he explains.

The rub is getting this price down to a point where it's feasible for the speedy shipping of lower-cost items such as medical samples or books, rather than munitions.

Matternet is targeting a price of between $3,000 and $5,000 per vehicle, he explains, though he thinks the business model is actually feasible if vehicles cost as much as $10,000.

"It's a no-brainer that one can design a vehicle that's very expensive and has very good availability – we've had proof-of-concepts of that from the military – but of course it wouldn't make economic sense to transport small items," says Raptopoulos. "You want to keep the cost down, [so] it becomes an engineering challenge."

Matternet's drones use a local beacon to help them find the final location, as this has better accuracy than GPS, he explains. It also helps create predictable routes that can compensate for the dangers of flying a drone into an unknown environment.

"In the long future, you could imagine these things have amazing situational awareness, [but] right now the tech is not there so they cannot see, for example, an electric transmission line, they cannot see another obstacle that may be bringing them down," he says. "The ground-stations, because [they are] pre-determined, offer biggest advantage in our concept of the system."

Raptopoulos imagines Amazon may want to deploy a similar technology, though there was no indication of any ground-station in the info-lite FAQ and video published by Amazon on Sunday. He suggests consumers could set up a landing pad-cum-beacon in their back gardens to help guide the drones into place.

"Amazon has a very sophisticated view in how they think about technology," he says. "I'm sure they've analyzed this well enough to know what they're talking about."

Another problem Bezos & Co. will have to tackle is minimizing the risk of injury from a drone going haywire. Earlier this year a man in New York was almost decapitated – he later died of his injuries – by the rotors of a remote-controlled helicopter. A similar event would be a catastrophic PR problem for Amazon.

"I think we are on the quadrocopter hype curve right now."

Matternet is already working on a host of technologies to stop its drones killing the people they're meant to help, and is trying to develop a parachute system that can decelerate the robots in the event of a critical failure.

But the likelihood of Amazon getting its system into production anytime soon is minor, due to the immense technical challenges and the glacial pace at which regulators tend to move.

"I think we are on the quadrocopter hype curve right now," Raptopoulos says. "We assume the tech is more ready for primetime than it actually is.

Given the extremely tough regulatory landscape both in the US and UK, he thinks it far more likely that drone shipping tech will make inroads in countries with less red tape, such as China and other emerging countries. Matternet's next trial may take place in either Lesotho or the Dominican Republic, Raptopoulos explains.

"The way we see it working [is] a system like this will happen outside the US first – it will happen in the South and East of the world first and then come to the developed world when the tech is already proven," he says.

"The reason why that will happen is there are places in the world where [it's] not about saving cost but about making something happen when you don't have any alternatives. High impact [like] medicine. We think in those cases the increased adoption risk of a new tech is rectified much more easily." ®