Skip to Content

The Woman Engineering Uber’s Driverless Trucks

Nancy Sun is trying to get driverless trucks on the road and convince truck drivers to trust Uber, too.
December 6, 2017
Uber ATG

Uber is publicly dealing with a lot of nasty problems: a revelation that it paid hackers to conceal a huge data breach, a lawsuit filed by Alphabet’s self-driving-car unit alleging the theft of autonomous-car technology, a corporate culture deemed so toxic the company replaced founder and former CEO Travis Kalanick with a new leader, Dara Khosrowshahi.

It’s also working hard on getting autonomous vehicles on the road. And Nancy Sun is one of the people heading this effort, as the engineering lead for self-driving trucks. Uber hopes that eventually these vehicles will make it faster to move goods from one place to another and easier for drivers, too, by taking over during highway driving (the plan is for humans to keep navigating the trickier routes on city streets). Sun, who also led engineering at autonomous-truck startup Otto before Uber bought it in 2016, spoke to MIT Technology Review about what it’s like to hang out in one of these vehicles and how Uber is working to gain truck drivers’ trust.

It’s been more than a year now since Otto—then newly acquired by Uber—made its first delivery via an autonomous truck, but there was still a human in the cab and the company has yet to announce any totally driver-free trips. What are some of the biggest challenges you see in getting these trucks on the road, hauling cargo, and driving regularly?

The challenges are really going from “What’s the product vision we want for our technology?” and then working backwards from there to help define all the various interactions and things that need to happen in hardware and software, and then diving one level deeper into that, in the base vehicle platform, to the sensor kit, to the computer platform, to the perception algorithms, to how we think about maps and localization, down to the details of, like, how do we scale this technology from a back-end perspective? The sensors are receiving tons of data that’s coming into a computer platform. We analyze all that data—that has implications on the infrastructure in terms of how we store that data and how we process it, and even to how we make a good developer experience. I think the biggest challenge I try to tackle bit by bit every day is how we make all of that come together in a way that ends up with all these various vectors tying in to one finished product.

How close do you feel you are at this point to being done?

What a loaded question! We have internal metrics for how we think about this. There’s a lot of data analysis that’s gone into what we think the definition of “done” means. And then there’s a lot of engineering work that goes into supporting that. The metrics are not something we want to talk about publicly at this time.

Two of Uber’s autonomous trucks waiting for their next drive.
Uber ATG

How does it feel to be in one of these enormous trucks while it’s driving itself?

The way that most commonly goes is, the first two or three minutes everybody is sitting there, and they’re paying attention, and “What’s that? What’s going to happen, like, if somebody opens the door and you drive by; are you going to nudge around the door?” And after the first three or four minutes happen, you forget you’re in a self-driving vehicle and you start having a conversation about something else. Maybe it’s the technology or maybe it’s a tour inside the truck; maybe it’s what the weather is like or what you had for breakfast that day.

A big concern people have about AI and automation is that it could put people out of jobs—in this case truckers, if the trucks become good enough at driving themselves.

It’s not like you’re going to flip a switch one day and you’re going to have self-driving trucks everywhere doing everything for you. It’s going to be a long journey to get to a point where it’s even going to be a majority-of-transit-happens-by-self-driving-vehicle world. It’s not in the next three years. It’s not in the next five years. We’re talking like five, 10, 15, 20-year future, beyond.

And for us it’s a matter of, basically, working with truck drivers, working with unions, with people who depend on that as an industry for their daily lives. To work on, like, evolving the nature of that work and working with people on understanding their needs and making sure our technology is able to support that industry as well. It’s not something that we’re looking to do is eliminate people’s livelihoods. We’re looking to supplement and help improve their livelihoods.

The long-haul trucker model—I think there are people who like that, and those jobs are going to exist for a really long time still. And there’s people who want to be able to go home to their families every night. And I think the model where we can have self-driving trucks automate a lot of the long-haul routes and have drivers continue to operate going from a transfer hub to a distribution center, driving more on the local routes near their homes, is a route we’re super-interested in pursuing.

In general, how do you get those in the trucking industry to trust you? Frankly, Uber doesn’t have the best reputation with a lot of people.

The first inkling of that was probably the [recent] Medium article. I think you’ll start seeing a lot more of that in the coming year as we start expressing not just our top-level vision but narrow down to what we mean when we talk about wanting to be more safe, when we talk about wanting to be more efficient, when we talk about wanting to improve people’s livelihoods.

Keep Reading

Most Popular

This new data poisoning tool lets artists fight back against generative AI

The tool, called Nightshade, messes up training data in ways that could cause serious damage to image-generating AI models. 

Rogue superintelligence and merging with machines: Inside the mind of OpenAI’s chief scientist

An exclusive conversation with Ilya Sutskever on his fears for the future of AI and why they’ve made him change the focus of his life’s work.

The Biggest Questions: What is death?

New neuroscience is challenging our understanding of the dying process—bringing opportunities for the living.

Driving companywide efficiencies with AI

Advanced AI and ML capabilities revolutionize how administrative and operations tasks are done.

Stay connected

Illustration by Rose Wong

Get the latest updates from
MIT Technology Review

Discover special offers, top stories, upcoming events, and more.

Thank you for submitting your email!

Explore more newsletters

It looks like something went wrong.

We’re having trouble saving your preferences. Try refreshing this page and updating them one more time. If you continue to get this message, reach out to us at with a list of newsletters you’d like to receive.