Autonomous cars are not a theoretical, maybe-if proposition that depends on a lot of different pegs falling into place. They’re already driving themselves. You may have seen one beside you in traffic. (I recently saw one next to me at a red light, and it was weird.) There’s already a full-blown trial in Chandler, Arizona.
If you bought a car in the last few years, there’s a decent chance you won’t buy another one. In a decade—maybe less—you may subscribe to a car service the way you subscribe today to Netflix or Blue Apron. When you need to go to Whole Foods, you send for a car on your iPhone or tell Alexa to send you one. Or you’ll order Whole Foods online, and an autonomous car will bring the groceries to you.
The revelation that that comes through like high-beam headlights in Lawrence D. Burns’s Autonomy: The Quest to Build the Driverless Car—and How It Will Reshape Our World is how much of the work is already done. General Motors introduced an autonomous concept car in 2002, and engineers started racing self-driving trucks and SUVs across the Mojave Desert in 2004. In the decade and half since, a long list of companies—Google, Uber, Tesla, Ford, GM, and many others—have spent tens of billions of dollars improving the technology to the point that it’s ready for the world.
Burns, a former GM head of research and development and a current advisor to Google’s Waymo autonomous car company, sat down with The Daily Beast.
When will I be able to order an autonomous car to take me to dinner? Two years from now? Five years from now?
It depends a lot on where you live. Waymo is testing its fleet right now in Chandler, Arizona. Within the next six month, a group of consumers in that market would be able to request an autonomous car to take them to dinner. This is a very big idea, but companies are starting small and picking specific places to prove out their technology and their value to the customer. I think you’ll see it in a lot more markets over the next five years.
What’s actually happening in Chandler, Arizona?
There are 400 individual riders who volunteered to participate in Waymo’s program, and they have an app that allows them to request a ride anywhere within about a hundred-square-mile area, and an autonomous vehicle will come and take them to their destination.
Are the big things that need to happen in the next five years for widespread adoption of autonomous cars more on the business side—investment, manufacturing, distribution, etc.—than on the technological side?
We’re much closer on the technological side and closer than most people think, but we’re still not done. If the speech recognition is 99 percent there, that’s not good enough for everything an autonomous car is going to handle in everyday driving. What remains to be done on the technology side is the very long tail, the 99.99 percent capability. Americans drive three trillion miles a year, and the machine learning still has more to encounter on the roads to be ready for widespread use. It’s not just the software and the maps and the sensors and the vehicle actuation; it’s all of that as an integrated system.
Do you see the deployment of self-driving cars being based more on geography than use case? I would have assumed that commercial applications like long-haul trucks would come first and personal use in different cities would come later.
Those are both happening. An important breakthrough in this entire journey was Level 4 autonomous driving, which was a standard set by the American Society of Automotive Engineers and subsequently the National Highway Traffic Safety Administration. A Level 4 system is capable of operating in a particular region during certain times of day under certain traffic conditions, and those boundaries expand going forward. At the same time, over-the-road trucking, last-mile goods delivery, gated communities, campuses, etc., are all areas where you’re seeing progress in geographic expansion and use-case expansion.
You wrote in the book about figuring out three factors—shared-use vehicles, electric power and autonomous driving—that it would take for autonomous cars to succeed on a large scale. Is the shared-use part of that mostly about reducing consumer cost?
Right. It’s important for driving down cost because you get much greater utilization with shared-use vehicles. Americans only use their vehicles about 5 percent of the time, which is poor use of capital. When you’re not driving it, you’re parking it. And 80 percent of the trips we make are 1- and 2-person trips, so we could tailor shared-use vehicles for those kinds of trips.
Will the automakers be hostile to that? Ford and GM won’t be super happy about selling five cars to a fleet company instead of 10 to consumers.
Initially, it was hostile. Lately, it has been more embracing. I left General Motors in 2009 and had reached the conclusion as the head of R&D that the combination of electric vehicles, driverless vehicles, tailored-design vehicles and transportation services were going to really, really change how we get around. I led the research at Columbia University that concluded autonomous vehicles could disrupt $4 trillion a year of the U.S. economy.
I was doing this work in 2011 and 2012, and the auto industry were still saying in 2014, 'Hey, this is interesting, but it’s probably 20 or 30 years away.' The progress by Waymo, Tesla, Uber,’ and other companies finally converged, and the auto industry realized that the business model of people paying $35,000 to buy a car, insure it, finance it, drive it, maintain it, pump gas, look for parking spaces is ripe for disruption. The automakers didn’t get it initially, but they get it now.
What will the business plan for an automaker be? Will they supply autonomous cars to fleet-management companies, or will they be fleet-management companies?
I think you’ll see both of those. General Motors is going to build its Chevrolet Bolt with a self-driving system and is positioning its subsidiary, Maven, as a car service. Ford has indicated that it’s going to have autonomous vehicles by 2021 that would be tailored for particular use cases. I think you’re gonna see a range of approaches.
What will the emblems on the backs of cars be 10 years from now? Volvo and VW, or Uber and Lyft, or Waymo and Tesla, or something else?
The driving service could become the thing that gets branded. Take BMW, which has branded itself for a long time as the ultimate driving machine. The vehicles of the future will be the ultimate riding machines. We’ll prize how smoothly the vehicle rides, how safe they are, where they can go, what the service costs. The branding could be built around those factors.
Do you think owning a car and using a car service will coexist for a long while like print books and ebooks do now?
I like that question, and I think the answer is yes. Autonomous vehicles and human-driven vehicles will both be on the road. One of the most important decisions Google made with Waymo was deciding not to wait until every car was capable of being autonomous before putting autonomous vehicles on the road. An autonomous vehicle is safer for the people in the autonomous car, but it’s also safer for everyone else on the road.
The speed of that shift from driving to self-driving will have a lot to say about whether new buildings have parking garages and where people will live. Are those the kinds of impacts you foresee?
One example I’m seeing is in construction. If you’re building a high-rise condominium in a city that has a requirement for a certain number of parking spaces that could become obsolete in the future, builders are starting to anticipate with ceiling heights, wiring, plumbing, etc., how they could cost-effectively convert that space to condominiums. If cars don’t occupy as much real estate, I think you’re gonna see increased density that will actually make the cities more walkable and make many cities more livable than they are today.
Do you see a big difference in the way dense cities like New York and Boston adapt to this world vs. driving cities like Los Angeles and Atlanta?
This future will not be one size fits all. There are a lot of different transportation solutions emerging that will be able to serve different cities in different ways. In cities that have good public transportation systems, the last mile will become very important. That could be a two-person pod with a top speed of 32 miles per hour as a way to get from the transit station to the destination. A more suburban area will need a different solution with higher-speed vehicles.
You note in the book that 1.3 million people are killed annually in automobile accidents but that only 37,000 of those are in the United States. How is it possible that only 3 percent of worldwide automobile deaths are in the United States?
I’m glad you picked up on that. I started traveling to China in the ’90s when General Motors was starting operations there, and the fatality rate per mile in China was 20 times what it was in the United States. There were a lot of pedestrians and bicyclists on the road, and a lot of cities weren’t using intersection-design solutions that took those people into account. It was a serious situation. Less developed countries have many more traffic fatalities.
Do you see evidence that other countries will be faster to adopt autonomous vehicles than the United States?
The United States is ahead right now. That said, China, India, and Singapore are doing some pretty impressive things. GM developed a two-person, autonomous, electric vehicle called the EN-V for the Shanghai World Expo in 2010. There are huge opportunities around the world for autonomous cars to leapfrog the current transportation systems like mobile phones leapfrogged wire-line phones in many countries.
Given the advances in battery life and reductions in production costs that you’d expect over the next decade, will we need thousands and thousands of charging stations across the country for autonomous vehicles?
The people who subscribe to autonomous-car services will expect the services to make sure the vehicles are charged, so people won’t have to charge those vehicles at parking structures and gas stations. The services will do that at depots, which will give them some economies of scale.
You estimated that the cost of long-haul trucking would drop by half once autonomous vehicles are implemented. Will a truck going cross-country stop at designated interstate exits to recharge?
When you get the driver out of an over-the-road truck, a lot of things change besides saving 64 cents a mile in labor costs. You can double the use of the vehicle because you no longer have a driver who is spending half the day not driving. The cost of vehicle design and parts will start to come down when you get rid of the windshield, the seats, the doors, the air conditioning, etc. The vehicles likely won’t crash as frequently, so insurance costs will also come down.
I’ll just be glad if they stop changing lanes every time I get behind one.
[Laughs.] I drive between Ann Arbor and Chicago a lot on I-94, and it’s unreal how often trucks change lanes and how that feeds road rage. I’m on the board of a company called Peloton Technology that works to optimize the distance between trucks to optimize safety and traffic flow, and organizing the traffic flow of over-the-road trucks is going to help considerably.
In theory, a truck that doesn’t have to stop half the day won’t need to drive as fast or change lanes as often.
It also doesn’t need to be as heavy. One reason the loads are as big as they are is that you’re trying to spread the cost of the driver over as many pounds of freight as you can. That becomes much less of a factor when you no longer have a driver.
Do you see companies that have an enormous stake in transportation and delivery—Amazon, FedEx, Kroger, Walmart—being early movers right now? Are they all doing the same things to get to a more automated mode of moving goods?
Those companies are doing a lot of the same things, but they’re also looking for ways to differentiate themselves for an advantage. The companies also recognize that not all interstate highways are the same—some are flatter, straighter, have different weather and different levels of traffic—so you may see freight corridors develop first in the areas that are the most conducive to autonomous transportation.
You have a quote from composer John Cage at the beginning of the book: “I can’t understand why people are frightened of new ideas. I’m frightened of the old ones.” Is that too cavalier about the potential unintended consequences of autonomous vehicles?
Unanticipated consequences are just that: they’re unanticipated. It’s paralyzing to spend all of our time brainstorming unanticipated consequences instead of stepping forward and learning. The potential benefits of autonomy are extraordinary measured in lives saved, emissions reduced, climate change averted, land-use improvements, and improvement of access.
The only way we can manage unanticipated consequences is to start learning what those consequences will be. When we come up with a new drug that cures a disease, we’re weighing the lives saved by the drug against every possible adverse reaction to the drug. When we have an opportunity to bring a huge benefit forward, we should start slow, learn fast, and then scale up as smartly as we can.
You think we should move faster to adopt widespread use of autonomous vehicles?
The biggest risk is not going forward as fast as possible. With 1.3 million people a year dying on the roadways and the potential there to eliminate 90 percent or more of that, if we get to full maturity of this one day sooner, we save 3,000 lives. That’s a big deal.
You sound enthusiastic about where things are headed. Does that come from being a kid putting together model cars? Does that come more from seeing the public safety benefits?
I had a life-changing event in the early ’90s when I lost my hearing and was completely deaf for a year. And then I received a cochlear implant, which is a technology not unlike what we’re talking about with automated cars. It’s digital, it involves speech recognition, it requires batteries. As the technology has improved, I can have a conversation with you that’s fairly normal. If it wasn’t for an Australian doctor named Graeme Clark in the ’70s ignoring everyone who told him that you couldn’t stimulate the auditory nerve with electricity, I probably wouldn’t be talking to you. I can’t help but be an optimistic technologist.
Plus, we’re almost there with automated vehicles.
Right. The other part of my optimism comes from leading R&D for 12 years at GM and seeing what we can do with improvements of lithium ion batteries, communications systems, and everything else. It has been remarkable to see what engineers can accomplish through fast learning cycles. And my enthusiasm comes from the fact that the automobile industry has been around for 130 years and is still extraordinarily wasteful.
Only 2 percent of the energy in a gallon of gasoline moves the driver in the car. We’re not utilizing the capital of the car 95 percent of the time. We have three parking spaces for every car. All of that waste has been embedded in the transportation system for a century, and so many companies profit from selling waste. Autonomy gives us an opportunity to get out from underneath all of that.