On the wall of the robotics company Starship Technologies’ headquarters, a collection of handwritten letters has been pinned up. Almost all of them are written by kids; conveying, in their cutesy drawings, hearts and smiley face emojis, the kind of wonderment that would normally be reserved for notes posted to Santa. This is the generation of kids who grew up seeing robots as benevolent rather than malevolent: more Pixar’s WALL-E than Terminator. And it shows.
“Thank you!” reads one letter. “We love your cute robots.” Another is addressed to a “Mr. Robot.” “You didn’t speak to us [when you visited],” it reads. “Hope we didn’t upset you.”
Starship, the recipient of these letters, is the maker of six-wheeled delivery robots which look a little bit like a toy-friendly sidekick robot from some as-yet-unmade Star Wars movie. While only a few years old as a company, and operating in just a fraction of its possible eventual market, the thousands of deliveries it has already made means that its robots are among the most visible seen in the real world on a daily basis.
As with many successful companies, Starship’s sales pitch is the epitome of simple. Don’t have time to head to the store for anything from batteries to books? Hanker after a hamburger, but don’t want to hop in the car to get it? No problem: for a small delivery fee, you can select your exact location on an app and one of the company’s robots will pick up the item you require and trundle autonomously along the sidewalk to deliver it to you. All you, as the customer, have to do is unlock it (again, using your app) and retrieve your order.
“We want to make deliveries a seamless experience,” Henry Harris-Burland, Starship’s VP of Marketing, told Digital Trends. “Right now, if you’re ordering online you have to think about where you’re going to be. Which address are you getting something delivered to? Are you going to be at home? Are you in the office? Then there’s the problem that the places packages can be left in are liable to theft. Right now, customers are paying for that inconvenience. We want deliveries to be so seamless you don’t even have to think about it.”
Hanker after a hamburger, but don’t want to hop in the car to get it? No problem.
Harris-Burland calls the letters the company has received “really humbling.” But he also doesn’t expect them to last. Over the coming months, the cards and kindly notes will almost certainly diminish in number. Within the next few years and, certainly, the coming decades, there’s a good chance that they will disappear altogether. What fate does Harris-Burland thinks will befall Starship’s delivery bots? What reason does the company’s marketing VP lie awake at night anticipating will cause the sort of backlash that stops kids writing notes to thank their friendly neighborhood delivery robots?
As it transpires, he isn’t predicting people will suddenly turn against the company’s delivery robots. Instead, he thinks — even actively hopes — that people will simply stop noticing. And, contrary to what you might expect from a publicity-seeking tech startup, that’s a good thing.
The Skype of delivery robots
Several years ago, Ahti Heinla and Janus Friis, one an Estonian and the other a Danish entrepreneur, decided to start a robotics company. Heinla, then just a couple of years into his forties, had already worked with robots. Building them in his spare time, he had spearheaded a proposal for a NASA competition asking for help building an autonomous rough-terrain robot which could be used to find and retrieve rock samples on Mars. His design was, ultimately, not selected — but the idea lingered. What if, he and Friis considered, such robots could be used not for wheeling around the Red Planet, 33 million miles away, but instead for carrying out door-to-door deliveries here on Earth?
The idea might have sounded crazy, but Heinla and Friis had a proven track record. Friis had been a co-founder of Skype, while Heinla was one of the lead software engineers who developed the software. Skype sold to eBay just a couple of years after its founding for $2.5 billion. This, combined with a clearly defined business strategy for their new company, was enough to convince investors. By 2017, the startup they called “Starship Technologies” raised $17.2 million in seed funding. One year later, in June 2018, it pulled in a second round of seed funding, this time worth $25 million.
Compared to backflipping or parkour-performing robots, deliveries are not an especially sexy area of robotics to focus on. But it was a smart decision. As countless robotics companies have sadly proven, making money from robots is supremely difficult. Startup costs are high, but despite massive technical advances in the field there aren’t all that many ways to currently generate cash flow. This is especially true if you’re operating outside of the current hotbeds for commercial robotics companies, which typically involve warehouse or military applications. In these scenarios, robots are cloistered away from regular citizens, the same way that computers were in the decades before the personal computer.
Starship Technologies’ delivery robots use some of the same technology found in today’s autonomous cars.
To Starship’s founders, one of the key insights which led them to embrace deliveries was the discovery that up to 50% of total transportation costs are weighted in the last couple of miles. Another promising data point? That 95% of the things people order online are sufficiently small that they could, in theory, be transported via a smallish delivery robot.
“There was nothing [that Starships’ co-founders could see which was solving] one of the biggest outside problems, which is last-mile delivery,” Harris-Burland said. “This is a big challenge. You’ve got vans stopping hundreds of times each day outside houses, missed delivery slips being put through doors, big cars being driven around with small pizzas in. It’s incredibly inefficient. They really thought that they could solve this problem.”
This early vision has seemingly been borne out. Today, Starship Technologies has its HQ in San Francisco, another office nearby in Silicon Valley, others in London, Germany and Washington D.C., and an R&D lab employing more than 100 engineers in Tallinn, Estonia. Its delivery robots have travelled a combined 200,000 miles, carried out 50,000 deliveries, and been tested in over 100 cities in 20 countries. It is a regular fixture not just in multiple neighborhoods but also university campuses. At the start of 2019, Starship Technologies teamed with food services company Sodexo to launch a food delivery service for George Mason University’s 40,000 students, faculty, and staff. Its fleet of 25 robots makes it the single largest implementation of autonomous robot food delivery services on a university campus.
To carry out their driving, Starship Technologies’ delivery robots use some of the same technology found in today’s autonomous cars. Their sensor suite includes cameras, ultrasonics, GPS, and inertial measurement units, among others. They do differ in one crucial way, however.
“Cars are looking for 100% autonomy,” Harris-Burland continued. “That’s because a car traveling 70 miles-per-hour on the freeway doesn’t have the luxury of calling up a human operator to ask a question if it’s unsure of something. With a delivery robot we can. We understand there is a long tail of situations, just through the laws of probability, that will occur that the robot may not be able to handle. It’s much easier and more cost-efficient to be able to ping a human operator, who may be overseeing 100 robots at a time, to ask for assistance.”
While most of their driving is done autonomously, any one of Starship’s robots can, at any time, be taken control of by a remote human operator. These operators, who watch every delivery, don’t even have to be in the same country as the robot they are monitoring. But the important thing is that they’re there if needed.
The right time and place
Any consumer-facing technology which truly takes off needs to do more than just offer the right tech. It needs to offer the right tech at the right time. Successful entrepreneurs must therefore have a sixth sense for diving through the small window during which technology is up to speed, few or zero competitors have entered the market, and, crucially, the public is ready for what you’re offering.
Starship Technologies didn’t doubt that its technology worked. But its employees, rightly, questioned how the general public would receive wheeled robots driving up and down their local sidewalks.
“The social acceptance has been amazing around the world.”
“No-one knew what was going to happen when delivery robots were put on sidewalks with humans,” Harris-Burland said. “What are humans going to say and do? Are they going to accept these robots or not? That was a big question for us. People thought surely these things were just going to get stolen or vandalized. [People from outside Starship have said], ‘this would never work in [such-and-such] town.’ I’ve heard it all. But the reality goes against our first impressions about human nature. The social acceptance has been amazing around the world.”
According to Harris-Burland, the reaction is “give or take exactly the same” in each place Starship has rolled out or tested its robots. “What normally happens is that the first time a robot is rolling around, and no-one’s ever seen it before, you get a few people taking pictures and pointing,” he said. “Lots of people look, but only around 1% actually interact directly with the robot. The second, third and fourth time people see a robot, they don’t care. It’s boring. Onto the next thing.”
Does he worry about what will happen if Starship proves its model so well that the sidewalks are filled with delivery robots? Is there some hypothetical tipping point at which the robots go from cute new arrival to invading horde pushing humans off the sidewalk and into the gutter?
“Back in the early days, we did do some modeling to look at what the sidewalks might look like when we’re at capacity,” he said. “Capacity in this case would mean all the deliveries which could potentially be done, which is not likely to be the case. Even in that case, we’re still not talking about thousands of delivery robots in every suburban neighborhood. In [the U.K. test bed city of] Milton Keynes, for example, we currently have 60 delivery robots covering around 15,000 to 20,000 homes. We’re doing thousands of deliveries. Robots are a common sight, but they’re by no means hindering pedestrians.”
Ultimately, Harris-Burland said, this is a crucial area for the company. Playing nice with humans isn’t just an added extra, like getting a protective case for your new smartphone. Starship’s robots are like rolling autonomous billboards — and you don’t have to be a business major to realize that annoying potential customers isn’t exactly a recipe for success. (For this same reason, Starship has also carried out research into the impact of its delivery robots on other sidewalk users, such as guide dogs.)
“If these robots suddenly started hindering humans, becoming obstacles and a nuisance, our entire model doesn’t work,” Harris-Burland said. “Because then they wouldn’t be socially accepted. Our robots need to blend in to the infrastructure of cities, campuses, neighborhoods, and suburbs. They need to go almost unnoticed. If that’s not the case, we’ve got a problem.”
It’s for this reason that, to return to the wall of letters, Harris-Burland doesn’t expect these responses (as endearing as they are) to necessarily last. He doesn’t expect the good-will to go away, replaced by some consumer backlash against delivery robots. Instead, he thinks the company’s delivery robots will simply become so commonplace that they will become wholly unremarkable. When, after all, was the last time your kid thanked a mailbox or a light bulb or any one of a number of technologies which once seemed astonishingly futuristic?
Here come the rivals
One place that Starship’s robots certainly haven’t gone unnoticed, however, is by other companies. Since Starship Technologies made its debut, a number of other companies are joining the delivery robot party. These aren’t rookie upstarts, either.
Courier delivery giant FedEx is bringing its half century of expertise to the space that once solely belonged to Starship. It has already announced plans to work with retailers including AutoZone, Lowe’s, Pizza Hut, Target, Walgreens, and Walmart for same-day door-to-door deliveries. Oh, and its delivery bots can hop curbs and climb stairs, too.
“We see it as a compliment, because it shows that we’re definitely onto something.”
Amazon, the great white shark of retail, is also jumping into the market with its Amazon Scout service. Harris-Burland refrained from commenting on competitors, other than to say that, “We see it as a compliment, because it shows that we’re definitely onto something.” Both of these, however, will present significant challenges in their own right — with deeper pockets and all the right connections.
For now, though, Starship Technologies has the high ground. It has carved out an impressive name for itself in just a few short years. Regardless of what happens next, Starship believes that it is offering something that is a whole lot more than a short-term gimmick.
“We’re not interested in novelty,” Harris-Burland said. “Of course, it’s an interesting technology, and it’s going to be novel at first. We’re the first company to be doing this. As someone who works in marketing, I also appreciate the novelty of that first order that customers make. That’s going to help make customers order the first time. But it’s definitely not going to help on the second, third, fourth and fifth time — never mind the fortieth or fiftieth time. If it’s just about novelty, we’re going to churn through customers within weeks. We now have a number of customers that have ordered from us over 200 times! I’m guessing the novelty has worn off on them — and they’re using us because we genuinely make life a little bit easier for them.”
Starship Technologies aims to be in it for the long run. For now, it has done more than enough to back up those ambitions. Roll on the worldwide rollout!
