Tech visionaries may tantalize us with visions of instant gratification via drone delivery, but Silicon Valley has yet to deliver on such promises. Meanwhile, halfway around the globe in an African country barely the size of Maryland, drone deliveries have already taken flight—with more serious cargo than burritos.
Jeremy Hsu is a science and tech journalist based in New York.
Sign up to get Backchannel’s weekly newsletter.
In October 2016, Rwandan crowds cheered the launch and landing of delivery drones developed and operated by Zipline, a San Francisco-based startup. The locals call the Zipline drones “sky ambulances” as they soar overhead and swoop in low to drop off lifesaving blood supplies by parachute to remote hospitals and clinics located hours outside the Rwandan capital of Kigali. That may sound very different from the PR circus surrounding Google drones testing delivery of Chipotle fare to Virginia Tech college students—and it is. But Zipline and similar delivery drone pioneers have also learned some valuable lessons about what a large-scale delivery drone operation can look like—and whether Silicon Valley can ever realize the dream of drone delivery to your doorstep.
“Countries like Rwanda can make decisions fast and can implement new technologies in concert with new regulations fast, so we’re now in a position where the US is trying to follow Rwanda,” says Keller Rinaudo, CEO and co-founder of Zipline. “They’re not trying to catch up to US infrastructure. They’re just leapfrogging roads and trucks and motorcycles and going to a new type of infrastructure.”
In early 2018, Zipline will officially kick off the world’s largest delivery drone service in Tanzania, Rwanda’s much larger neighbor. The Tanzanian government aims to use Zipline’s delivery drones to make up to 2,000 deliveries of medical supplies per day. Those deliveries of supplies such as blood products, medicines, and snake antivenom will go to more than 1,000 hospitals and clinics serving 10 million people. An operation at this scale will dwarf anything previously attempted in the drone-delivery universe.
The Tanzania launch will fulfill the dream that led Rinaudo to found Zipline in the first place. In 2014, he met a graduate student named Zac Mtema while visiting the Ifakara Health Institute in Tanzania. Mtema had created a mobile alert system that could help doctors and nurses text emergency requests for medicines and vaccines to the government. There was just one problem: The government had no way of quickly delivering those medicines and vaccines via the country’s existing roads and distribution networks.
Today, Mtema is helping the Ifakara Health Institute evaluate how Zipline’s service affects health outcomes in Tanzania. Quantifying lives saved and medical conditions treated could go a long way toward convincing Zipline’s deep-pocketed backers in the international aid and development community—such as the Bill & Melinda Gates Foundation—that delivery drones can become a global force for humanitarian good. The for-profit startup has already raised at least $41 million in funding from investors.
By focusing on carrying critical medical supplies, Zipline has gotten off the ground faster and in a bigger way than other, more mundane delivery pioneers. It’s a lot easier to convince regulators to tolerate the potential safety risks of delivery drones falling out of the sky when those aircraft are making lifesaving deliveries to hospitals rather than carrying shoes or pizza.
Zipline isn’t the only delivery drone startup to latch onto the idea of carrying high-value packages in difficult terrain. Matternet, a startup in North Fair Oaks, CA, plans to launch a partnership with the Swiss Post before the end of 2017, carrying healthcare supplies between hospitals and labs in Switzerland. “In healthcare we’re targeting over 1,000 hospital groups with three or more facilities in our target markets, which include the main European Union markets, the United States, and Japan,” says Andreas Raptopoulos, Matternet’s CEO. “For the applications we’re pursuing in health care, [delivery drones] are clearly profitable for us while giving a 50 percent saving to hospital systems over the on-demand ground delivery methods they use currently.”
These drone pioneers have learned that if you’re going to provide reliable service delivering essential, life-saving goods, you may end up with technology that looks very different from the familiar demo videos of consumer delivery by air. All drones must contend with limited battery life or fuel tanks. But many early experiments with delivery drone services used quadcopters or other multi-rotor drone models similar to those available online or on retail store shelves. These designs usually have limited delivery range and speed; their less-than-aerodynamic shapes and vertical lift rotors limit the efficiency of forward flight.
For example, Amazon and Google have been testing delivery drones that usually top out at ranges of 15 miles or less. Shorter delivery ranges limit the number of customers drones can reach, and may also limit the profitability of future delivery drone services. Still, these short-ranged multi-rotor drones are by far the most common choice for drone-delivery innovators, even as some companies have tried creating hybrid drones with both vertical and horizontal rotors to improve flight efficiency. Most delivery drone prototypes still tend to hover like helicopters as they lower a package by cable or take the time to actually land.
That may work for the big city or suburbia. But Zipline needed to provide timely delivery of medical supplies across dozens of miles in Rwanda and Tanzania. So the company’s San Francisco team of engineers—drawn from organizations such as SpaceX, Google, Boeing, and NASA—decided to create the equivalent of small drone airplanes with wings. The current “Zips” in Rwanda can fly at speeds of up to 62 miles per hour and reach destinations within a 46-mile delivery radius.
Zipline’s airplane-style drones don’t waste time or battery power landing at their destinations. Instead, they simply swoop in low to drop off supplies by parachute before winging their way home. That’s easier done in the open grassy areas near remote Rwandan clinics and hospitals than in densely populated city blocks. Still, Amazon alone has several patent ideas around the concept of midair package drops or even folding parachutes within package shipping labels. Both Amazon and rival Walmart have even envisioned the possibility of someday using giant airships as flying warehouses that could deploy glider drones with packages.
It’s easy for almost anyone to create a slick video showing a drone perfectly delivering snacks or tech gadgets for a smiling customer on a sunny day. But when lives hang in the balance, drone deliverers must meet a higher service standard—one more like the US Post Office’s: “Neither snow nor rain nor heat nor gloom of night stays these couriers from the swift completion of their appointed rounds.” Mostly, quadcopter-style drones won’t cut it here. Matternet’s Raptopoulos says that a “key challenge is designing a platform that can handle dynamic wind loads during take-off and landing.” The Amazon Prime Air website currently states: “We are currently permitted to operate during daylight hours when there are low winds and good visibility, but not in rain, snow, or icy conditions.”
By comparison, Zipline’s rugged, fixed-wing drones have already made over 1,400 flights and delivered 2,600 units of blood, even during bad weather conditions involving heavy rain or wind in Rwanda. “Everybody and their grandma has bought an off-the-shelf quadcopter and in perfect weather flown three to four kilometers to deliver something under ideal test conditions,” Rinaudo says. “It’s hard to distinguish between that and a national-scale operation that can fly in any weather.”
An additional challenge is limiting the human staff needed to oversee the drone swarm. After all, every human engineer or drone operator on staff is another person on salary, points out Gerald Van Hoy, senior research analyst at Gartner. And the promise of delivery drones hinges in large part upon delivering packages more cheaply than today’s hordes of human bike messengers and delivery van drivers. “With delivery drones, if you’re not flying them automated, then it’s probably costing too much,” Van Hoy says.
The most obvious solution is for drones to mostly fly themselves instead of relying upon a human operator. But automation gets complicated quickly when you’re aiming to deliver to the doorsteps of individual home and business addresses. That could require a drone’s computer brain to track other flying objects in order to avoid midair collisions, fly to a random drop-off point where a given customer lives, and avoid getting entangled in tree branches or power lines when landing or lowering a package for delivery.
It’s possible that better artificial intelligence and the growing trend of edge computing could eventually make for smart drones capable of delivering anywhere without human supervision. But until then, delivery drone services will likely have the best luck sticking with pre-planned flight paths and deliveries to set locations. For example, Flytrex, a startup focused on cloud solutions for drone operations, recently started a delivery drone service involving one or two drones making up to 20 flights per day between two set points separated by a large bay in Reykjavik, Iceland. But Yariv Bash, Flytrex’s co-founder and CEO, says his company is working with the Icelandic Ministry Transportation on a next phase that could involve drone deliveries to Reykjavik street corners before the end of the year.
The point-to-point delivery drone system is also being used by JD.com, a Chinese e-commerce and logistics giant that is already living Amazon’s dream of owning its last-mile delivery. Beyond operating an online marketplace, JD.com has more than 70,000 delivery people getting those packages to paying customers in China. Since 2016, the Chinese company has also operated a fleet of 40 drones that have made “thousands of runs that have packages going to customers,” says Josh Gartner, vice president of international corporate affairs at JD.com.
JD.com currently uses delivery drones in the rural areas of four Chinese provinces. Those drones are “fully automated” and fly “fixed routes” between warehouses or to the backyards of certain “village promoters” employed by JD.com in each country village, Gartner explains. The village promoters then distribute the packages on foot to customers within each village.
Zipline employs a similar logic of using automated drones to boost the productivity of individual human workers. Zipline’s Rwandan distribution center can operate with just three to four human flight operators potentially handling hundreds of drone flights per day. The flight operators spend their days loading packages onto the Zips, placing the drones in the launch catapult, and then recovering the returning drones that land by using a tailhook to snag onto a line so that they can plop safely onto a giant cushion. (Rinaudo proudly points out that the Rwandan operation runs entirely on Rwandan engineers and flight operators rather than outsourced foreign labor.)
The predictability of automated delivery drones’ flight paths also helps them pass muster with safety regulators. For example, Zipline has provided air traffic controllers at Kigali International Airport with heads-up displays that allow them to track each Zipline drone with centimeter-level accuracy using advanced GPS. That has ensured smooth daily operations, even as the number of flights increases. In fact, Zipline’s Rwandan delivery drone operation is on track to become one of the busiest “airports” in the world based on flight volume by the end of 2018, Rinaudo says.
Still, the upcoming Zipline launch in Tanzania will push the company’s current model to its limits. Tanzania’s sprawling land area—including geographic highlights such as Mount Kilimanjaro and the Serengeti National Park—is 36 times larger than Rwanda’s. That means Zipline will need both more distribution centers and more capable delivery drones to cover a country that is bigger than any US state except for Alaska.
For Tanzania, Zipline eventually plans to roll out four distribution centers each equipped with fleets of up to 30 drones capable of making up to 500 deliveries per day from each center. The startup’s engineers have also been developing an upgraded drone that can carry up to 4.4 pounds, fly at 68 miles per hour, and deliver within a 99-mile radius. That system would enable healthcare workers to simply place orders by text message and receive their packages within half an hour.
Don’t expect to see a Rwanda- or Tanzania-style national-scale delivery drone service coming to the US anytime soon. For one thing, the US Federal Aviation Administration has suggested regulations for delivery drones will not be ready until 2020 at the earliest. By that time, delivery drone operations may only account for one percent of the global commercial drone market, according to a report by Gartner.
Another factor is that the economics of delivery drones make less sense in cities already crowded with many competing delivery services and where safety concerns are more abundant, says Josh Gartner of JD.com (no relation to the Gartner research firm). Indeed, the Chinese company is considering ground-based delivery robots for Chinese cities instead of delivery drones. Similar delivery robots have already been rolling around certain cities in the US and European countries, where delivery drone services mostly remain grounded.
Some companies may seek a middle ground in suburban or rural areas by testing delivery drones as robotic partners for delivery van drivers. In February 2017, UPS—the world’s largest package delivery company—joined forces with the Ohio-based company Workhorse Group to conduct a much-publicized test of a delivery drone deployment from the top of a “Big Brown” van. “Our HorseFly delivery drone can handle packages of 10 pounds and under,” says Mike Dektas, a representative for Workhorse Group. “With the truck and drone delivery system this is a good weight limit, and the 30-mile [drone] range works as well.”
Similarly, Matternet has teamed up with Mercedes-Benz to try out the combination of delivery drones and vans. Matternet’s CEO Raptopoulos also envisions solo delivery drones as becoming profitable for both his startup and logistics companies such as FedEx or UPS, with a price point of around $5 for delivery within an hour. He adds that a company such as Amazon could make the delivery drone service even cheaper or potentially free for customers who have already signed up for the $99 Amazon Prime subscription.
Silicon Valley’s magical vision of delivery drones—Harry Potter-style owl messengers for each of us Muggles—remains seductive as ever. But it’s worth paying attention to what has already been accomplished in the more remote parts of the world—lessons likely to be applied as delivery drone services slowly take off in developed countries. By the time 2018 rolls around, Zipline will surely have more insights to offer would-be winners in the game of drones.
“The vast majority of people thought this was crazy and stupid and there was no chance it would happen in Africa,” says Zipline’s CEO Rinaudo. “Now our entire distribution center is run by a totally driven and brilliant team of Rwandan operators and engineers who are not only working 12 hours a day and 7 days a week, but [also] doing things that the richest tech companies in the world haven’t figured out how to do yet.”