Driverless Cars and the Future of Transportation

Wednesday, April 5, 2017
Peter Nicholls/Reuters
Hans-Werner Kaas

Senior Partner and Cofounder, Automotive and Assembly Practice, McKinsey & Company, Inc.

Jack Weast

Principal Engineer and Chief Architect, Autonomous Driving Solutions, Intel Corporation

Karlyn D. Stanley

Adjunct Senior Researcher, RAND Corporation; Professor, Pardee RAND Graduate School 

Daniel Runde

William A. Schreyer Chair and Director, Project on Prosperity and Development, Center for Strategic and International Studies

Experts discuss innovations in driverless cars, the costs and benefits of autonomous vehicles, and the regulatory, ethical, and policy concerns that need to be addressed with the implementation of the technology.

The Emerging Technology series explores the science behind innovative new technologies and the effects they will have on U.S. foreign policy, international relations, and the global economy.

RUNDE: OK. Thank you all for coming. I’m Dan Runde. I hold the Schreyer chair at a different think tank, but I’m a member here at CFR. I really appreciate being asked by the Council to preside on this very interesting topic—I think quite timely. The conversation’s going to be about driverless cars and the future of transportation. We’ve got a really great panel to help us deal with this issue of driverless cars.

And I would consider driverless cars as part of the concept of a fourth industrial revolution. I wrote a blog about this in Foreign Policy about a year—I did a book review of Klaus Schwab’s book called the “Fourth Industrial Revolution.” I’m not a techie. I suspect not everyone in this room is a techie, and one of the reasons you came was to get smarter and understand what is this stuff.

And the so-called fourth industrial revolution is described that way because we’re going—we’re undergoing—and even if all of us in this room don’t really fully feel it yet as a technological revolution that’s going to change how we work, how we live, and how we relate to one another. This is not an extension of the third industrial revolution because of its speed, its breadth and depth, and its systemic impact. And there’s a whole series of components of the fourth industrial revolution which could include a whole series of more than a dozen phenomena, but I’ll just list a couple here: smart and connected machines and systems, quantum computing, driverless cars, and the sharing economy—things like Airbnb and Uber, to give you an example.

I’ve got a series of questions I want the panel to help us unpack in about 40 minutes, and then I want to turn it over to this very thoughtful groups of folks of members so that—and some of the questions I want to get are as follows: So what are driverless cars? What does this mean for the United States? What does this mean for how we live in the future?

And the other thing, frankly, is what will the disruption look like for folks in affected industries? For example, in the United States we have 3 million full-time truck drivers in the U.S. It’s the largest employer of white males, is the truck driving industry. And they’re between the ages of 20 and 60. We just had a disruption election in November in Washington. So think about what happens if those 3 million folks are put out of jobs in swing states, just to give you a concept.

So, when driverless trucks arrive, what does it mean for all of them? You can extrapolate to Europe, or China, or Brazil, or India, where there’s literally millions of people or tens of millions of people who are in the truck driving business or in the taxi driving business. So I think this is something we need to get at.

Overall, I think we should be optimistic about this technology. I think we should be optimistic about the fourth industrial revolution. But there are going to be these challenges that we need to think about, and there are ways in which we can shape the conversation.

So—but before I get to the panelists, I want to do a survey of both the panel and this audience. No such thing as a free lunch. So I want to ask the panel and the audience the following question. Today, if I offered for you to get in a driverless car and drive from San Francisco to New York, today, on this date, would you get in that car and take that ride today with nobody controlling the steering wheel? Complete autonomous control, San Francisco to New York, who would take that ride today? OK, panelists—I’m watching the panelists who know what they’re talking about. OK. (Laughter.) OK, so kind of 60 percent. So I’m going to come back at the very end because I want to ask a different question at the end after we’ve had this conversation.

OK, so we’ve got really three thoughtful panelists for us to—and you have—I believe you have their bios in front of you. We have—we have my friend Hans-Werner Kaas, who’s a senior partner and co-founder of the Automotive and Assembly Practice at McKinsey & Company. Thanks for being here, all the way from Michigan.

KAAS: Thank you, Dan.

RUNDE: Thanks so much for being here.

KAAS: Short, around the corner. No long way.

RUNDE: Just around the corner, exactly. Just one driverless car ride away, right? (Laughter.)

KAAS: Exactly, exactly.

RUNDE: So let me start with you, and then I’ll get to the other panelists. But can you—when we had this conversation, I said driverless cars, you said something else. So what—there are a couple of questions for you, but the first one is what—when I say driverless cars, what does that mean to you? And then second, what does—how is the automotive and the auto parts industry thinking about driverless cars, or whatever you’re going to call it?

KAAS: Thank you, Dan.

So, first of all, a quick and additional survey. Who of you has ever been driven in an autonomous vehicle?

WEAST: Wow, that’s impressive.

KAAS: So maybe 15 to 20 percent. That’s good. So I’m sure you are also around the group that drive—being driven from New York to San Francisco.

RUNDE: Exactly.

KAAS: So there—but that means there are also quite a few risk-open people here.

RUNDE: So risk—yeah, exactly, risk-takers, right.

KAAS: I mean, they are because these people raised their hands to be driven from New York to San Francisco.

So, Dan, I think first one, a driverless car, by definition, is always fully autonomous. And without deviating too much in technology, there are five different levels of what the Society of Automotive Engineers defines as a progression towards driverless cars. Level five is truly driverless and fully autonomous, which means under any condition, whether road conditions, traffic conditions, this car can operate without a driver behind the wheel. In fact, those cars do no longer have a steering wheel. Anything else are progression levels towards that level. So level four are autonomous vehicles, but those autonomous vehicles still need to have a driver to take control back just in case. That just in case, by the way, can be automated or, you know, you can yourself take back over again because you don’t trust the vehicle. So there’s, I think, a very important understanding.

This is also important to understand projections of adoption rates of autonomous vehicles in 2025 or 2030. You can go—obviously, we are seven to 13 years out.

Point two, the auto industry—and I would say globally, not only in the United States—sees it absolutely more as an opportunity than a threat. Now, with that opportunity, a few caveats are coming. Leave alone the maturing of the required technologies from sensor systems, hardware, but also the computing capability and barriers—actually, barriers of the software—all that has to be resolved.

The more important point, in addition to technology, I personally feel very confident—and I think Jack or Karlyn can actually comment on that—I feel confident we can actually solve the technological questions over time—not day or night. But business models that the service providers or the offerors of vehicles for autonomous applications, they also need to have a case for quote/unquote “earning their cost of investments or capital.” Otherwise, it’s not going to happen. And the consumer needs to have some advantage, whether it’s saved time, freed up time, I mean, some version of, you know, joy being operated or being driven in autonomous vehicles—which it can, by the way. There needs to be a business model case.

And I think this is where the research, and frankly, many private sector companies are today quite engaged on: What are the business models which are promising? And think about it. There’s a whole spectrum of business models. Private application—you as a private owner, would you maybe buy as your second or third vehicle, let alone a first one, an autonomous vehicle?

Then you have private mobility fleets, for which obviously think about the Lyfts, the Ubers, et cetera, taking a driver out of the equation in terms of cost per mile traveled. It has an important impact, actually. It’s a very important impact. The biggest cost input: driver.

But then as you mentioned, Dan, there are obviously commercial applications for autonomous trucking, and those pilots are already going on as we speak. In some industries, we have autonomous truck applications today already. For whoever among you have been in a mine in some remote areas, we have large autonomous operated mining trucks today already. So there are significant not only pilot applications out there, quite some what we call used cases.

Your third question, what does it mean—and frankly, whether it’s not truck drivers, but think about taxi drivers, and we are really talking about more mid- to long-term. Nothing is going to happen quickly within two, three years until 2020 or 2021, but there will be disruptions. There will be jobs eliminated. There will be jobs created.

And I would, as I mentioned, always look at the net-net effect of a system. There will be jobs created. Think about the services you will need to provide. Autonomous vehicles, by definition, will have utilization rates—it doesn’t matter whether it’s commercial or private applications—which will go above 50 or 60 percent. Today’s utilization of a private vehicle—vehicles you and I, we all drive, is around 4 to 5 percent—4 to 5 percent. That’s already at the high end, by the way. It’s sitting in parking, in your private garage, wherever. So we need to have higher utilization, but with higher utilization comes wear and tear, replacement needs, services, spare parts, et cetera, et cetera. How would you actually offer those services and infrastructure points? You don’t want to always drive it back to the dealership. You need dedicated infrastructure points. So there are job opportunities. And frankly, like you would advise your kid when it goes to college or to graduate school, apart from what they like and feel passionate about, you always have a lens towards what can actually earn a livelihood. And you need to have an anticipation of those disruptions, some view of what will be new job fields and what will be jobs which are eliminated.

RUNDE: OK, thank you very much.

Karlyn, I’m going to just jump over you and I’m going to get to Jack, because I want to—you wrote a—Jack, you are the principal engineer and chief architect of autonomous driving solutions at Intel Corporation. I saw your name on this when they asked me to do this. I said why is Intel on the stage? I thought you guys made microchips, Intel inside. What is—what does Intel have to do with driverless cars?

WEAST: Thanks, Dan. Yeah, it’s a great question. And by the way, I’m very happy to be here, so appreciate this opportunity to speak to you all.

So, at Intel we look at the driverless car opportunity as an end-to-end or really car all the way to cloud opportunity. So, yes, you need those traditional chips that go inside the vehicle so that the vehicle can understand the world around it and then make decisions on how and where it drives, but you also need connectivity. You need cellular communication with things like LTE and 5G so that your driverless vehicle can talk to infrastructure.

RUNDE: What’s LTE and 5G?

WEAST: Wideband communications for your cellphone that you have today, imagine your car having that same sort of connectivity so that as your car enters a city your car can talk to the infrastructure and understand where parking spots are available, or in a scenario where you want to group up with other cars on the freeway and form a platoon or kind of a caravan.

RUNDE: Sorry, what’s a platoon? Because we are having this—what is platooning? I’ve never heard that word before.

WEAST: It’s a concept where multiple cars can get together on a freeway in a lane very, very close together for increased fuel economy if they’re all traveling 150 miles in the same direction. So connectivity plays an important role.

RUNDE: When I took drivers safety—(laughter)—in the 1980s there was a 1960s film that said watch out for wolf packs. That was like groups of cars on highways, and that was bad. Now in the future it’s good.

WEAST: Yeah, exactly. Yeah, we’ll have to update those driver manuals.

RUNDE: We’re going to have to update those videos.

WEAST: Yeah, so for Intel it’s the in-vehicle compute, it’s the cellular connectivity, and then it’s the data center. These cars generate an incredible amount of data, and that data needs to be processed somewhere. You need to learn from that, and you need to train your car, train your virtual driver to learn from new experiences and new scenarios just like we as human drivers learn from new scenarios that we encounter on the road. So, from Intel’s standpoint, we’re fortunate and feel honored to be able to bring a breadth of technology to this industry that spans from the car, cellular connectivity to the cloud, and then all the way back again to be able to support the industry’s efforts.

RUNDE: You all just bought a company. Just tell us about what that—why I’m raising that.

WEAST: Sure, yeah. We announced the intention to acquire a company called Mobileye, which brings a computer vision element to the solution. And this is really important. Now many of you may have this technology in your car today. It’s a little thing that sits on your dash, and there’s a camera that looks forward. So it has eyes just like you have eyes. So it reads speed signs, it reads traffic signals. It can detect pedestrians. It can help see the world around it. So camera sending combined with LIDAR, which is a laser scanner that shoots lasers out and brings those back, traditional radar that’s been around for a long time—the combination of these different sensing technologies allows your car to have a virtual representation of the physical world that’s around it. And with that virtual environment model, is what we call it, then your car can figure out, OK, where is a safe path in the environment so that I can keep the humans inside the car and the humans outside the car safe as the autonomous vehicle navigates an environment.

RUNDE: Can you just—co-evolution of different kinds of sensors, and you called—you used a term called “sufficient degree of fidelity.” What is that?

WEAST: Yeah, what—good question. What that means is that you need a sufficient degree of fidelity, which is a quality of sensing. So, if you think about a black-and-white camera versus a color camera, or today your digital cameras that you have on your phone today are much higher-resolution and higher degree of fidelity than they were just even a couple years ago. And so the better the sensors can see, the better the sensors can understand the environment around the car.

So those advancements in sensing technology is what’s bringing us to a point where the sensors can actually see and perceive the physical environment in many cases better than we even can. A computer-vision-based sensor can see things that human-based eyes cannot see in terms of light spectrum and things like that.

So it’s quite exciting to combine these technologies together to deliver on this vision of autonomous future—

RUNDE: So let me ask this question of you, and then I want to come back to the panel in a second round to talk about this question, but I want to start with you, which is who—of the countries in the world, who are the—who’s the leading country on, let’s call it, driverless car technologies today?

WEAST: Yeah, I think—great question. So, from a technology provider standpoint, the U.S. continues to lead. Companies like Intel and many of our competitors and partners are continuing to deliver the most advanced pieces of technology that go into delivering these experiences. I think where we should discuss further with—

RUNDE: I want to come back to it.

WEAST: —the other panelists is, what about deployment of automated vehicles? It’s one thing to build the technology, but are we doing the right things from a policy standpoint to support the deployment of these vehicles in the real world? And that’s where you see efforts in Germany, Japan, China, and elsewhere that are creating a policy environment, particularly around the deployment of cellular 5G technology, that’s really encouraging the quick deployment of this technology so that we can get it tested and proven. And that’s, I think, maybe where the U.S. is falling a little bit behind, but I’ll certainly defer to my esteemed panelists with more experience here in D.C. to comment on that.

RUNDE: So the bottom line is America has got the lead now but could lose it because of a—of a bad policy environment or an incomplete policy environment. Is that a fair—is that a fair bumper sticker for this audience to take away?


RUNDE: Okay, good.

Karlyn, thank you for being here. Karlyn, you’re the adjunct senior researcher at the RAND Corporation, and you’re professor at the RAND Graduate School. And you were one of the coauthors of a really interesting report—I think it’s called “The Crash Economy.”

STANLEY: No, it’s called “Autonomous Vehicle Technology: A Guide for Policymakers.”

RUNDE: So even better than what I—a better title than what I came up with. But actually you dealt with a whole series of interesting implications, policy implications, implications for how we’re going to have to rethink a whole bunch of assumptions in how we live and work and how we make policy in the future because of autonomous cars.

So thanks for being here. Could you—so let me first start with—we’re going to have to rethink legislation. We’re going to have to think regulation. What kind of—what kind of an approach should we take to that? Should it be heavy hand? Should it be light hand? What do you think?

STANLEY: Well, thanks very much, Dan. Well, first of all, let me just give you the bottom line on the report that you mentioned and our RAND report. And that is that technology is necessary but not sufficient to achieve the societal benefits that we are looking for from this technology. Good policy is necessary. And so let me give you just a few highlights—

RUNDE: Please.

STANLEY: —of our—of our report. We’ve looked at both the positive outcomes potentially from this technology and some of the disadvantages. And that term you use, the crash economy, that’s one of the downsides. But first let’s look at the upsides since we haven’t really focused on those. I’ll just go over them very quickly.

First of all, this technology has the potential to save lives. In just the past year, traffic deaths in the U.S. were up 6 percent, to over 40,000 people dying on U.S. roadways. So, with this technology, we may be able to dramatically reduce that number.

Number two, mobility. We’re going to be able to give people who are disabled or elderly the opportunity to be out and drive and participate in society again.

Number three, potentially we could reduce fuel consumption and traffic congestion.

Another thing that we could potentially do is to change the urban landscape. Why? Because if you don’t have to have cars parking in the center of town, you could have cars in a satellite parking or maybe even not parking at all if they’re using—if you’re—we’re using a shared fleet. We looked at 30 cities, and we found out that one-third of the inner-city area was dedicated to parking.

Well, imagine if we could use that space—

WEAST: That’s amazing.

STANLEY: —more beneficially.

And then the final area of potential benefit is to reduce congestion. You mentioned platooning and how that might affect congestion.

But now let’s look at the disadvantages that we found. There’s a flipside to the congestion question. It could be that if people want to live further because they can be productive when they commute into town, maybe we’ll have more congestion. So we’re not sure.

Secondly, we looked at something we called the crash economy. So think of it. With all the deaths and the injuries we have from car crashes every year, we have auto repair shops and the people who work in them, we have the hospital trauma centers with the doctors and nurses who repair the people as opposed to the cars.

RUNDE: I was thinking the ambulance chasers.

STANLEY: And we have—and we have the auto insurers. It’s a big business, $220 billion a year and 300,000 claims adjustors and investigators. So there are a lot of people involved in car crashes. So that—things are going to change not only with truck drivers, but also with this whole crash economy.

It’s not going to happen immediately, but—and so there are also potentially impacts from that ability to change where cars are having to park, if at all. And that is that many cities and municipalities depend on the revenue that they derive from parking. For example, in New York City alone, just parking tickets—not parking lots—New York City gained over $500 million last year from parking tickets. And that they use to fund their other infrastructure. So think about it. Cities and municipalities are going to need to find another source of revenue.

So there are upsides and downsides.

RUNDE: Thank you. I’m just thinking about the story—the saddest day of an—of an elderly person’s life is when they lose their driver’s license.


RUNDE: So this will be an—that would—could change that. I’m sure everyone in this room has known someone who’s died in a car accident. Imagine if you could take instead of 40,000 deaths a year—I don’t know—let’s say there were 500 deaths a year. So let me—but let me—even if you—let’s say the good news is we reduce it from 40,000 to 500 deaths a year. If there’s an accident in an autonomous vehicle—and there will be—who’s responsible?

STANLEY: Well, what an excellent question. Right now we’re preparing a seminar. It’s an automated vehicles symposium in July that’s—Transportation Research Board is sponsoring. And we’re going to look at this. And I can tell you right now—let’s look at the stakeholders. The stakeholders are in the event of a crash, first of all, law enforcement. They have to come. They need probably data to help figure out what happened. Then there’s a manufacturer. Manufacturer will need also data. There’ll be the insurance companies also looking for data and information and then products liability lawyers and others.

So there are many stakeholders who will be looking at what caused the crash. And what we are foreseeing and what we talked about in the report is that liability may be shifting from the traditional model of a(n) individual driver of a car to the manufacturer. And so that is one of the key questions, Dan.

RUNDE: So when a car crashes, it’s GM’s fault; it’s not my fault.

STANLEY: Well, that’s the—

RUNDE: Think about that.

STANLEY: That’s the critical question that will be looked at as we evolve.

RUNDE: I blame Jack, you know, technology. (Laughter.) No, but—so I think—I think this is going to be a really thorny issue.

So let me ask another—who owns the data? So let’s say I go for a drive in a car. Is that—my driving habits—is that my data? Is that a technology company’s data. Is that—is that an automotive parts data? Is that General Motors’ data? Who owns that data?

STANLEY: Great question. It depends on who you ask. We surveyed many, many people in research that I did for the Texas Transportation Institute on data considerations for automated and connected vehicles.

RUNDE: And there’s a report that the state of Texas did—or Texas A&M did a report on the issue of data privacy related to driverless cars.

STANLEY: Correct. And for the people we interviewed, it was really, really interesting. We asked each and every one of them, who owns the data from your perspective? And surprisingly, we had insurance companies say, well, if we have a subscriber agreement that allows us to do certain things, we own the data. The OEMs said, oh, well—

RUNDE: Who’s “OEM”?

STANLEY: Oh, pardon me—the car manufacturers said, well, owners of the vehicle always own the data. And then there were many other opinions expressed. So this is an area that can be really contentious. So I think it’s who you ask.

RUNDE: OK, so just one last thing. We talked in the pre-call about black boxes. All these cars are going to need a black box like an airplane. When there’s—when there’s an airplane crash, you say, let’s go see what happened in the black box.

Could just—someone just explain—there is—they don’t call it that, but there’s in essence the equivalent of a black box if there’s an auto accident and a driverless car. Is that—

STANLEY: Right. Right. On all cars currently, there is something called an event data recorder, EDR.

RUNDE: Event data—EDR.


RUNDE: So that’s the equivalent of the black box.

STANLEY: Right. But there are discussions going on right now at SAE and other organizations looking at what will be needed in the future to capture the data from—

RUNDE: What—I’m sorry, what’s the SAE?

STANLEY: Oh, I’m sorry. That’s the Society of Automotive Engineers.


STANLEY: And they’re a neutral body, just engineering, looking at key questions around autonomous vehicles, among other things. So they’re looking at what will be need in the future to gather that data to learn what went wrong, what caused a crash, et cetera. And it’s probably going to be very different that what is—than what is currently in the EDR that’s in your car.

RUNDE: So let me—this is now—thank you very much for that. So let me put this to the group, which is, I think the good news for this—for our members here at CFR is that the United States has a technological lead on driverless cars, or autonomous vehicles. But we could lose it because of policy issues.

So I would like to spend a few minutes talking about policy-related issues. So if you were—if you were in front of the Department of—the secretary of transportation, what—well, first of all, who are—who’s going to—who can—who can do something about driverless cars in the federal government? Who are—who are those folks? And is—currently is there a patchwork of regulation about this right now? I think there is.

And if you were in a first term of a Trump administration, what would be a couple things you could do in short order in the next 12 to 24 months? Because this is a rapidly evolving—this is a rapidly evolving landscape. And so even if we’re not in driverless cars or we’ve had one, what are the things that people in Washington could do right now, and what are the sorts of things that people in the states could do right now?

So any and all—any and all takers on that.

WEAST: Go ahead, Hans.

KAAS: Yeah. So first of all—

RUNDE: Appoint an—appoint an NHTSA person, right?

KAAS: Certainly we don’t have a NHTSA, quote/unquote, “leader” right now. So that is certainly—

RUNDE: We haven’t named somebody. They need to name somebody and appoint somebody.

KAAS: So they need to name somebody, hopefully very quickly, and a qualified person of course. (Laughter.) True for any appointment, right?

RUNDE: Yeah, of course, with any appointment.

KAAS: For any appointment. With that said, first of all, what is the purpose of regulation? Number one, you want to protect the consumer. Number two, you actually want to create a fair playing field for market-based competition. And number three, you obviously don’t want to favor certain technological choices, one over the other. But what you should mandate is certain achievements.

RUNDE: Outcomes.

KAAS: Outcomes, you can call them. So I think against that playing field or against those definitions, that is what the, quote/unquote—the “guardrail” should be of any future regulation. I do think we need a much bigger regulation—in the United States, there’s no attempt—maybe thinking to actually define such regulation on a national level. Certain states are actually engaged in it.

RUNDE: And who are those states?

KAAS: Nevada, Michigan, where you have actually initial tests and pilots taking place. But as Jack said, when you look beyond the boundaries of the United States, which we need to do, Germany, Japan, China, Korea, and even Sweden are defining—

RUNDE: They’re ahead of us.

KAAS: —ahead of us—defining national-focused regulation for autonomous vehicle operations.

RUNDE: So let me just—what is NHTSA, and why does that matter for this conversation? Karlyn, maybe I’ll ask you. What’s NHTSA, and why does it matter for this conversation?

STANLEY: Well, NHTSA is the part of the Department of Transportation that focuses on safety. And so they have been very deeply involved in looking at the safety of this new technology—technologies, I should say. And they recently issued—well, at the end of 2016 they issued guidance to the states on how they might want to create model legislation. Again, it was a light touch, just a model. It wasn’t anything mandated.

RUNDE: That the states could copy/paste.

STANLEY: Well, that the states could certainly use as a framework.

RUNDE: OK. Because you have to—in essence, right now, Karlyn, if I understand it, much of the energy around this is going to be at the state level. Is that fair?

STANLEY: There’s a lot of energy at the state level because of interest, and then there’s not anything being dictated or mandated from the federal side. So, yes, the states are very, very active.

RUNDE: But the model law—the reason you would have a model law is you need to have all 50—ideally, you want all 50 states to have the same set of laws?

STANLEY: Right. Well, the concern is—and we talked about this in our report—that there not be a patchwork of regulation state to state, where states will require something—one state would require something that another would not, so that the manufacturers would have to, instead of dealing with 50 states and making an automobile that is—can be certified in all 50 states, they would have to go state by state or several states. And so that’s really a concern, and that could hold up deployment.

RUNDE: Has there been any federal white papers, either by the secretary of transportation in the last administration or from the Obama White House? Because this is a very new thing. Really, driverless cars sort of were born around 2003 or this—what we know as driverless cars started with a series of DARPA challenges between 2003 and—has there been a paper on sort of a strategy or a blueprint for the United States on driverless cars? Has anyone done that out of the executive branch? Have you seen anything like that?

STANLEY: Well, USDOT certainly created white papers and a whole structure around connected vehicles, and has talked about that at great length. So I don’t think that the administration, other than what I mentioned, which is the framework that was sent out for states by NHTSA.

RUNDE: So it sounds like there’s probably a need for something like that.

WEAST: Yeah. And to reflect what I said earlier, I think autonomous vehicles, I mean, if you think about it, is three elements, right? It’s the vehicle itself, where electronic or event data recorders are important, and policies and guidelines around that. You have the connectivity piece, in terms of how do you connect to a cellular infrastructure so that these connected vehicles can communicate with each other and with infrastructure. But then, also, there’s the data and the data center that’s storing the data. So I think we need to be looking holistically at the entire sort of car-to-cloud infrastructure, from not only an infrastructure-investment standpoint but also from a policy standpoint. And that’s where we’ve had some focus on—with DOT on just a piece of it, but not really holistically. And that’s where other countries are really leading.

RUNDE: So I want to bring Jack and Hans more into this conversation about, OK, so let’s put aside NHTSA and the Department of Transportation. What are the other focal points for this to happen at the U.S. level? What are the other agencies that are going to be involved? Is it the Department of Commerce? Is it the intel community? Is it—is it the Office of Science and Technology Policy in the White House? Who are the other folks that are stakeholders in the policy conversation in Washington? Who would you—who do you—other than Department of Transportation, which it seems to me it seems is going to be a big part of the future of this, and certain parts of the Department of Transportation, what are the other—what are the other policy stakeholders, if you will, in the executive branch?

WEAST: I think a topic area—and I’ll defer to my colleagues in terms of the specific agencies—but certainly something that’s going to crosscut, you know, almost the entire federal government is the new transportation funding, you know, discussions that are happening. So, as we look at significant investments in infrastructure, are we just laying new concrete or are we creating—are we—

RUNDE: Are we future-proofing it?

WEAST: Yeah, wonderfully said. Are we future-proofing our infrastructure to support the deployment, you know—

RUNDE: Of this stuff.

WEAST: —and advancement of this stuff, you know, not just paving new roads. So—

RUNDE: Hans-Werner, could you just spend a minute. We had a conversation earlier, so we were talking about security in cars. So when I think about my security, I have a car alarm, I have a Carjack thing for the steering wheel. (Laughter.) But in the future that’s not what car security’s going to mean. It’s going to mean something else. And so could you talk about that—A, what does security mean? And then could you spend a minute explaining—we had a conversation earlier about there’s different—there’s different formats of talking to the car, and it’s kind of VHS and DVD standard of formats of how to talk to the car. I’m being a little simplistic, but it works, I think.

KAAS: It’s communication protocols.

RUNDE: (Inaudible)—have. So could you just spend—could you spend a minute on the issue of security, and then spend a minute on this issue of communications standards for how the car talks to the cloud?

KAAS: Sure. Take the analogy of your smartphone, your wireless-capable laptop. Think about now you have a car which is obviously able to send and receive communication, hopefully very advanced bandwidth communication like 5G for example.

RUNDE: OK, what’s 5G?

KAAS: 5G is basically the newest wireless technology, which is actually getting deployed in other Western countries to enable—call it wireless communication, which could also enable not only smartphone capabilities but also cars talking to themselves or to the cloud, downloading data, storing data for different type of purposes.

So cybersecurity is, quote/unquote, “the security of today’s and the future age of vehicles,” especially with autonomous vehicles. But not only with autonomous vehicles. Even long before we have fully autonomous vehicles, most of our vehicles today—not most—most of the by certain OEMs sold vehicles already—

RUNDE: OEMs is like the carmakers.

KAAS: Original equipment manufacturer.

RUNDE: Yeah.

KAAS: By the way, we owe you that abbreviation: OEM, original equipment manufacturer. They have already today modems. And by 2020, you can assume that most vehicles, new ones sold, will have a modem like your laptop and your smartphone have a modem.

RUNDE: Yes, and I’m sorry to interrupt, Hans-Werner, but there was a different format from like 20 years ago that’s currently being used. What is that format called? You talked about 5G, but there’s another one. What’s the other one?

KAAS: And Jack, obviously, is in an excellent position to actually talk about that. That’s a short—dedicated short-range communication protocol which actually was created in the end of the ’90s.

RUNDE: And what’s the funny acronym for it?

KAAS: It is short—dedicated short-range—


KAAS: DSRC is the—

RUNDE: OK. So there’s 5G and DSRC are the two competing standards for how the car talks to the cloud right now.

KAAS: Or among themselves.

RUNDE: And so right now the DSRC was created in the late ’90s and 5G’s about to be. And so currently, the current American regulation as of walking out the door from the Obama administration was everyone should use the old technology. Is that right? Is that the current law—regulation on the books? That’s not a criticism of the Obama administration. I’m just saying as of right now the state of play was—is that correct, that’s the rule?

WEAST: The proposal, yeah.

STANLEY: It’s proposed.

KAAS: Proposal.

RUNDE: That was the proposed rule by the Obama administration when they walked out the door, right?

KAAS: Yeah.


RUNDE: OK. So if you—if you’re worried about regulation, what you want is to be able to—you can’t necessarily—this is in any administration, whether it’s any—you need—you can’t anticipate the changes in technology. And so right now we’re using VHS when we’re going to go to DVD or streaming, right, or the equivalent, or 8-track tape. We’re using 8-track tape technology to talk to the computer. Is the 8-track—is the whatever it is—DRSM (sic; DSRC)? Is DRSM (sic; DSRC) more hackable than 5G?


RUNDE: DSRC. Is DSRC—(laughter)—

KAAS: Well, that’s—whatever the four-letter—

RUNDE: OK, whatever the thing is, the old one.

KAAS: Whatever the thing is called.

RUNDE: The old tech, DRSC (sic; DSRC), is that more hackable than 5G?

KAAS: I think the answer there is more hackable. But keep in mind—

RUNDE: It’s more hackable. It’s more hackable.

KAAS: But keep in mind hackable is obviously a real bad outcome because people with bad and evil intentions, for different purposes. But even the reliability of any communication protocol between vehicle-to-vehicle communication or vehicle-to-cloud, or cloud-to-vehicle communication, do not only think about hacking or not hacking. You want to have highly reliable vehicle communication.

Think about it. You’re operating not even a fully autonomous vehicle with a certain set of sensors. And some sensors, by the way, have limitations because of fog, for example. Neither camera sensors nor LIDAR sensors are capable of detecting fog because of the light, and the LIDAR laser will be broken at the water molecules in the fog for those who are technologists among you.

RUNDE: Right. So if you’re driving in the fog—

KAAS: So you will actually—you can download data in a cloud and say, wow, the next five-mile stretch I’m really hitting here a foggy area; what if that communication from a weather-forecasting station never actually gets down from the cloud to your vehicle—(laughs)—and you keep on driving, you know, at 80 miles per hour?

RUNDE: Like right now when—

KAAS: But you should be slowing down already because you will hit a fog wall. That is why we need highly reliable communication bandwidth and protocols and (some, like, fallback ?).

RUNDE: It’s like when you lose signal on a cellphone. It would be—

KAAS: That’s exactly right.

RUNDE: And in this case it would be really bad.

KAAS: Yeah.

RUNDE: Like, if you lose signal on a cellphone, you lose a phone call. But if you lose signal on the equivalent of a cellphone and, like, you crash into a tree or something because you lose signal, right?

WEAST: I think—and I think this is a really important point, though, because I think sometimes when we think about this future of autonomous vehicles and we think about connected vehicles, connectivity to these vehicles with each other or to the cloud is important, but you cannot—you cannot rely on that connectivity necessarily always being there. And so the vehicle needs its own ability to be fully situationally aware of the physical world around the environment to detect and understand adverse weather conditions, where the car might just need to say, you know what, I’m going to pull over and wait for this sandstorm or whatever it is to pass because you might be in a rural area and you don’t have connectivity. So I think connectivity is a wonderful addition to it that has important security implications, but the car needs to be able to sense all on its own.

RUNDE: OK. So let me come back to my security point. So it’s not about carjackings in the future; it’s going to be, is my car going to be hacked? Is the—so the paradigm of security now is, is someone going to break into my car and steal it? Now is someone going to break into my car and hack it?

So let me just also paint another picture. Let’s say that the North Koreans decide they’re going to hack all our driverless cars and have them crash into one another. Isn’t that—isn’t that a possibility?

WEAST: Yeah, I think—so two things. I think, first, in that specific example, that’s the value of the car having its own sensors. So just because you received a message over the airwave which said, you know, turn right and head off a cliff doesn’t mean you should do so. Your own sensors—

RUNDE: Right, but if you don’t have a steering wheel, you know—

WEAST: Yeah, but your own sensors that are seeing and sensing the environment around you would say, well, I received a message that’s asking me to do something that would put my passengers in harm, you know, or injury, so I’m not going to do that. So that’s where the vehicle needs to be kind of intelligent on its own.

From a security standpoint, as the industry transitions from physical security—which is how do I prevent theft of my car—to cybersecurity, it requires a different approach and a different mentality. And that’s where we’ve been able to bring security technologies from the information technology space and bring that to the automotive industry to help them understand, you know, how do you protect and remediate against potential threats and attacks.

And I think one of the unique things that we’re going to need to do in this industry is certainly we’ll design the safest car that we possibly can, but you need to expect the unexpected. You know, secure—things that are secure today may not be secure 10 years from now. And if a car has a lifetime of 10 to 15 years, the car needs to be designed in a way that its security capabilities can be updated in the field. And so Intel has some unique technology in this space called programmable hardware, essentially, that would allow us to update your security protocols and your security capabilities of the car for the lifetime of its deployment.

So I think it will require a combination of—a balance of, you know, trust but verify when you—when you receive a message or when you are interpreting sensor data, but also a different mind-set in terms of what security means for automated and connected vehicles.

RUNDE: OK. I’m going to come back. I want to—one of the things I want to come back and redo with questions—I want to ask a question—a survey question for this audience, but I would like each of you to comment. What is the one thing that the Trump administration could do to help advance the—in addition to naming and approving a NHTSA administrator, what is the one thing they can do? So you’re going to think about that and come back to me. It’s like in two sentences. That’s going to be your closing point, because I think that’s important.

So I know—we now are going to move to the Q&A session. So I want to invite members to join our conversation with questions. This meeting is on the record here at CFR. I need you to wait for the microphone, speak directly into the microphone. Please stand, state your name and affiliation. You get extra credit if you’re short, and you get extra credit if it’s a—you only can do it in the form of a question because we want to get as many members as possible. There are a lot of smart people here, and I’d like to try and call on at least six to eight people. And for me to do that, I need you all to cooperate with me.

So this gentleman here in the back with the glasses, and we’ll call on some other folks. I’d like to bunch several questions together, I’d like—and then we’ll—yeah. Go ahead, please.

Q: Rob Quartel, NTELX.

So you’ve talked about—this is a fascinating conversation. What you haven’t talked about is the psychology of people, both personal psychology and contextual. It’s one thing to decide whether you want to do it—and I drive a Porsche; I don’t. (Laughter.) But the other is you may also not want the truck driver—and I look at the Metro system here in Washington, and it started as autonomous and they had to put a driver on it. So who’s looking at the psychology of the introduction of these systems? And what are they thinking?

WEAST: Yeah, that’s a great question.

RUNDE: No, let me—I want to bunch these together.

WEAST: Sorry.

RUNDE: I want to get several others. So I’d like this gentleman here and this gentleman here. We’ll get three of them.

Q: Tom Petri.

It’s on a related question. We’ve seen how drones have changed airpower. How will autonomous vehicles affect military operations?

RUNDE: Great. And this gentleman here.

Q: Irving Williamson, International Trade Commission.

I was—well, you should discuss what is going to be the vehicle for international collaboration? Because the U.S. competitiveness is very much going to be considered. It can be affected by how we work with other countries in terms of setting standards and things like that. Thank you.

RUNDE: That’s a great question.

OK. So let’s use those three, the psychology, the military dimension, and how do we cooperate with other countries on things—this standards question. I got to one of those topics having to do with these different forms of standards for how their cars to talks to the—to the cloud.

OK. Let me start with you, Karlyn.

STANLEY: OK. I’m glad you asked about the military question because I’ve been working on a project for the U.S. Department of the Army concerning technical requirements for automated and autonomous vehicles. And the U.S. Army has a very different view of how it can use automated and autonomous vehicles than our commercial world. They look at in a primary way force protection—if you can take people out of the automobiles or the platooning trucks, you can potentially save lives.

But these vehicles are going to have many more challenges than on urban roads that have good markings. They’ve going to have to go off road. And so there are definitely challenges. But the U.S. Army—and, I know, other militaries as well—are looking to incorporate autonomous vehicles, particularly for logistics, for delivery of fuel, ammunition, water—things that get people, the troops, out of the vehicles and out of harm’s way.

WEAST: I’ll take the human psychology question. It’s a great one. And first—

RUNDE: I like driving my car. (Laughter.) I don’t want to give that up.

WEAST: Me too, actually. I grew up in a big car family, so I hear it all the time from my father, what are you doing taking away my car? (Laughter.)

No, in all seriousness, though, the human psychology aspect of this I think is really important. And at Intel, one of the things that we’re doing is we have a team of sociologists and psychologists and anthropologists that look at the human element of how we introduce technology. And so we’re doing research in a space that we call trust interactions. And what we mean by that is we’re trying to understand, what are those interactions between a human and a machine, in this case an automated vehicle, that will engender trust, that will allow us to step into that vehicle or trust our kids inside that vehicle?

And what we’re finding is that it’s often the same kinds of things that engender trust amongst people. So as you meet someone new, do you feel hurt? You know, do you have an ability to communicate bidirectionally in an open and clear manner? If you ask a pointed question or ask for change, do you see a response in that person? And so what we’re trying to understand is how do we take those human concepts of trust and apply those, you know, to an automated vehicle.

And so we’re doing some research, and we’ll be publishing for the entire industry to benefit from so that we can understand that as the interior design of these systems and vehicles get developed by many of our customers and the industry, that we’re keeping these foundational trust interactions in mind because if you don’t do that, you don’t establish that human-machine trust relationship, and then maybe we have an issue in terms of the adoption of this technology in the market today.

RUNDE: Hans-Werner, could you take on this issue of standards?

KAAS: I can. I would love to comment quickly on the first two questions.

RUNDE: Yes, please.

KAAS: I think in terms of your question of standards, they will be critical for the advancement of the technology because if we do not have common standards, we easily will have winners and losers.

I think second point I would make—

RUNDE: And just on that—we’re going to have to, like, build them into our future trade agreements?

KAAS: Well, you’re anticipating my answer. That’s exactly what I wanted to say. (Laughter.)

I think they need to be part of future trade agreements. And I also think—(laughter)—

RUNDE: Great. Maybe we must have, like—we must have, like coordinated—

KAAS: It’s all about fair trade. We just need to put a definition around fair trade, right? And there are different criteria and objectives, obviously, what fair trade really constitutes, both for the consumers but also for the players.

But that said, I think it’s very important to actually look at our current trade agreement and take advantage also of what I would call bloc agreements because it is much easier to actually negotiate—agree with the European Union, which still entails—leave Brexit alone for a moment—27 countries versus bilateral agreements country by country. If you do bilateral agreements, you will defy the purpose, actually, of standards because I promise you, you will come up with at least X number of different agreements, maybe not 27 different agreements with European Union, but maybe you come up with six different ones. And that, again, would defy the purpose of common standards across different region and countries.

In terms of the psychology question, I think there are different aspects to that, and I think what Jack outlined makes a ton of sense, we look at a very similar way about it, that you need to overcome certain trust issues. And keep in mind even though some of us have been driven in an autonomous vehicle—but using it in a more constant way as your personal vehicle may be still easier because you own it, you would take care of it. But now think about it, you would be driven around in an autonomous vehicle by mobility provider in a city. Do you trust him that he always has actually the latest, greatest software on it; that the sensors, for example, like camera sensors, the lens is actually cleaned and clean? So there are aspects of who owns and operates on autonomous vehicles and what that does for your psychology.

The second point is many still love to operate and drive a car. So, one, depending how many people you have in your household and how much you, indeed, like or love cars, you still want to have, quote/unquote, “normal” cars, or call it L3 or Level 4-capable cars where you really can still take control. We should actually never forget that.

The last aspect on psychology I would mention—and we actually did quite a bit of surveys across different countries ranging from Japan, China, U.S., Germany, et cetera, a critical question consumers actually do give us and the way they answer that is that the provider of autonomous vehicle technology is critical for them. There are companies and plans they trust more and they trust less. And I will not mention what the specific result of those surveys are because I can’t favor any specific company or brand, but there are differences, I can reassure you that. So, like with anything you do, you have trust in a certain brand promise and the technology provided.

And the point around the gentleman regarding military application, in addition to transport purposes, which I agree with, there are also battlefield operations possible. Think about fighting in house-to-house fights, which, again, we all hope that we see less and less of that in the future even though a lot of political—military-political issues we need to face and then solve. But there are very different ways you actually can conduct battlefield operations. And there is technology out there which, you know, leading U.S. defense manufacturers already engaged in today, what it actually would entail from, obviously, guiding and controlling different deployments of forces. But it also, in the spirit, it can save lives at the end of the day. Obviously, at some point, you might argue, what if the other side does it? We have autonomous vehicles, ultimately robots, fighting against each other. Hopefully never comes that far, and we have other ways to solve conflict issues. But there are significant applications on battlefield operations, so I’ll just leave it at that.

RUNDE: OK. This gentleman here.

Q: Jeff Bialos, Washington lawyer but not ambulance chaser. (Laughter.)

So my question relates to the degree of market penetration of these things in the future, notwithstanding all the challenges. And by analogy to the cellular phone industry, I remember in the ’80s, when it came out, and I had occasion to hire—to talk to three different consultants we were thinking of hiring about the projected market. Two of them said, you know, the growth of this is going be arithmetic, lot of challenges, costs are high, niche product. The third fellow, he was from McKenzie, said, you got to throw out all that stuff. The prices are going to come way down. People are going to build service models around it. It’s going to penetrate and be ubiquitous. What are we talking about?

RUNDE: Right. So is this a—maybe—how about a way to describe it would be as, is this going to be adopted in different ways, in different models? We actually had a conversation about this. What are the different—what are the different kind of business models that we’re going to see around driverless cars?

KAAS: Yeah. So, on the one hand, yes, costs will come down. The question, obviously, is the cost progression curve. And that is a question of application, scale, et cetera—what you see in other technologies, by the way, as well.

I think we will see, in my view, a variety of different business models. Again, be clear what are actually the applications behind it. Is it your privately owned autonomous vehicle? Is it a mobility fleet-owned vehicle, autonomous vehicle? Is it a commercial fleet-owned vehicle? There are different applications. And do we transport passengers? Do we transport goods? So think about spectrum of different applications you have.

And then you even talk about business models. There will be a need for providers, companies like Intel, car-makers, both hardware and software. But then there is a huge space, in my view, for service providers. And service providers, think about it, what is a service somebody is really willing to pay for? I gave a couple of examples, even very simple examples today, even without autonomous vehicles. All of you today, mandated by regulation, have a so-called stability control in your vehicle, succession generation of ABS—many of you will be wondering, where is it, actually? It sits actually a little deeper under the hood, and it does good things for you, which sometimes you do not even notice because it avoids that you lose control of your vehicle because loss of traction.

Now, what if you would even today get a monthly report with data extracted from that stability control which says, you know what, whenever you actually drive in those hot spots in your neighborhood, always stability control engages because of your driving style, not because of wet surface on the road or so, would be quite valuable. And you might say, you know what, maybe I’m willing to pay $2 for such a monthly subscription because it’s my driving conduct and driving behavior, and I can actually protect myself. Just one example which even today you could—nobody is doing that, by the way. Even today an opportunity for a service. Now, think about autonomous vehicle applications, what additional opportunities you have to come up with service ideas. So that’s where I think innovation actually will come to the forefront. And we have a lot of smart people in this country. We have a lot of smart companies in this country. I’m sure they will figure it out, and they will get help to find those service applications.

RUNDE: OK. Jack or Karlyn, how fast—so when are we going to have in—when we walk out this building—and is it five years from now, or is it 10 years from now, is it 15 years from now—when are we going to have an all-driverless vehicle Washington, D.C.? So, when I walk out this door 17th and F, when—is it five years from now, is it going to be all driverless vehicles? Probably not. Is it 10?

STANLEY: What Nissan has said is that they will have a fully autonomous vehicle on the streets in 2020. Tesla has said that it will have a predominantly automated vehicle as soon as—I think it’s next model year. So—

WEAST: And we have announced a similar partnership with BMW, which is also committed to delivering a fully autonomous vehicle designed by 2021.

But I think the hard part about picking down a date is, as Hans mentioned earlier, the different levels of autonomy. You know, you could have a fully autonomous vehicle with no steering wheel and no pedal, but if you wanted to drive on every single road and every single weather condition in every single country around the world, that’s probably a ways off. But I think you’re going to see because the end of the decade limited-scale geofence-type deployments where a municipality could say, you know what, these 20 square blocks of this downtown city environment are off-limits to human-driven vehicles, and you only have a robo-taxi service. And in fact, we’re seeing this—speaking about international competition, we’re seeing this happen already in Singapore, China, Japan, Korea and places like that where they’re really leading with these kinds of concepts around sort of constrained deployment environments that is going to result in seeing deployments of fully automated vehicles, possibly before the end of the decade.

RUNDE: OK. Other hands? Oh, this woman here please.

Q: Hi. Barbara Matthews.

So your example is great, except that many of the countries you just mentioned aren’t exactly known for having open democratic societies that care about liberty. Where I’m going with that is the American car culture—and the Americans created the car culture, and it was all about freedom. And as you look at the psychology, as you look at the space for individual freedom, there are a number of issues associated with data, the privacy—there are a number of issues associated with the ability to go wherever you want without being tracked, and there are issues associated with going wherever you want in your own vehicle. And I think that those are really difficult issues for a society that will change our culture if one becomes dependent on a fleet of vehicles that one doesn’t own. I don’t have the answer to that question, but, you know, for the last hundred years, the American culture has been all about just go out onto the open road whenever you want. And if you haven’t—I haven’t read your report, but I will—if you haven’t identified this as a negative potential outcome, there is no way we will ever figure out a way to mitigate it. And so I love the gizmos, I love the opportunity that it provides in the advancements for society, but I worry because we will lose that freedom.

RUNDE: So let me just—I think that’s an excellent point. Thank you.

So someone told me 15 years ago that those—you know, those—when you drive through those toll plazas, they have those smart blocks. They use them now in divorce cases. (Laughter.) So somebody will—so what were you doing at 3:00 in the morning driving in a place you weren’t supposed to be? So this is the issue of privacy. So I—you know, as the—but let me—let me take a different example, which is, OK, let’s say you have an autocratic society like China or some other bad actors, to your point. So let’s say, instead of in the case in the United States, if you break certain laws, they take your driver’s license away. Well, how about instead they say, well, we’re just going to take your ability to have any mobility? You can be under virtual house arrest because we’re not going to—you want to call up your driverless car. We’re going to take that away from you.

So issues of privacy and the issues of personal liberty regarding—welcome any of you to reflect on both those issues.

STANLEY: Well, I take your point. But do you take your smartphone in your car?

Q: And my car communicates with it. (Laughter.)

STANLEY: Right, because your smartphone has the capability to do just what you were describing. And it’s much easier to grasp that data, whereas it would be more difficult to extract it from your vehicle. But you’re quite right. You would have origin and destination information and the possibility that, at any given time, you could be located.

But there are other devices, as you point out, Dan. You’re the—

RUNDE: Whatever those things are called.

STANLEY: Yeah, the chip that allows you to go through toll booths. Smartphones and other digital devices are all using that kind of data as well.

RUNDE: So others on that point of—let’s call it the privacy. So privacy about if I—you know, do I really feel the need to have law enforcement know my—you know, my driving patterns, of driving my kid to soccer practice? I don’t—you know, is that an issue of privacy? And the issue of privacy and the issue—but let me press a little further about this issue of autocratic regimes, because I think your point is a good one.

Let’s say—you know, what does it mean if someone says we’re not going to give you the permission to use cars, because that’s part of sort of, you know, pressuring you for some reason, for political reasons?

KAAS: You could already do that today, by the way. You don’t need autonomous, connected vehicles. There are different ways, call it probably not always morally or legally justified. But you can exert pressure already today. And I think we can actually find fair agreements and laws to protect privacy. I’m actually less worried about that. Especially in Western societies, I’m less worried about it.

And keep in mind, even as we speak now, there’s conversation going on about the privacy Internet-related in our own country. So let’s deal with the issues we have at hand today before we think about fully autonomous vehicles and whether they can track me where I drive or not.

The second point I would make is as long as the freedom of others gets negatively impacted, we should have regulation and laws preventing that from happening. And when I’m speeding all the time with 80 miles in a 40-mile-an-hour—miles-per-hour zone—absolutely I should be pulled over and I should pay a ticket or lose my driving license for six months; no doubt about it.

So I think we need to be thoughtful about it. Whenever the freedom of another individual is then negatively impacted, I think law and regulation is justified.

WEAST: From a technology standpoint, I can offer that one of the advantages that you have with increased computing capability within the vehicle is you now have an ability to process, you know, what is personally identifiable information in terms of who are the passengers and how are you behaving and the personalization of that experience entirely within the vehicle and not need to necessarily send them back up to the cloud.

We’ve done some research on concepts where, you know, let’s say you still have your smartphone. But perhaps on your smartphone is stored your driver persona. So as you step into the vehicle, that persona, you know, communicates with the car, takes you where you want to go. And then that comes back to you, you know, when you exit.

So there’s a lot of security technologies out there that I think could come up—could support a technical-based solution to preserve privacy and, in fact, potentially even improve it, you know, compared to solutions that we have today.


STANLEY: But one other thing.

RUNDE: Yeah.

STANLEY: You might be interested to know that AAA, the Automobile Association, did a survey of over a thousand adults about two years ago in which they asked the question, do you feel that your automotive information should be protected by laws or regulation? And they asked that question a couple of different ways. But over 80 percent of people said yes. They strongly agreed with that. And a recent survey by the Canadian version of AAA, the Canadian Automobile Association, did a very similar survey just last year with very similar results. So people are definitely concerned about that issue.

RUNDE: This gentleman here.

Q: Thank you. Jonathan Chanis, New Tide Asset Management.

The panel touched on this slightly, but could we hear more about the impact of driverless cars on energy consumption, and in particular on fossil-fuel transportation?


WEAST: I think typically when people talk about improvement, enhancements in energy consumption, it’s two things. It’s expectation that more automated vehicles, particularly fleets, are going to be electric vehicles. They’re not going to be fossil-burning—fossil-fuel-burning cars.

RUNDE: Do you expect that to be the case?

WEAST: For the most part, customers that we’re working with that are planning on deploying fleet vehicles, yes. They’re intending them to be. For example, the BMW partnership that we announced, that’s been announced to be an electric-vehicle fleet. That’s what that will be designed as.

From a technical standpoint, for a host of reasons, it’s actually easier to develop an automated vehicle on an electric vehicle than it is a traditional combustion engine in terms of how the vehicle is designed. But I think that’s one of the big reasons why there’s typically a link between automated vehicles and reduced fuel consumption.

Karlyn, you also mentioned the platooning aspect as well.

STANLEY: Right, because cars or trucks can follow each other and get the benefits of the physics—

WEAST: The aerodynamics, yeah.

STANLEY: —and the aerodynamics of following each other. There’s a reduced fuel consumption. And also I’ve heard that there could be reduced tire wear. This is for truckers especially; and third, less trucker exhaustion. And so that’s why you may well see platooning in the trucking industry being one of the first real deployments of autonomous vehicles.

KAAS: The only thing I would add is there will also be more localized energy consumption in an autonomous vehicle, be it a passenger or a truck, simply because the electrical energy consumption for multidomain controllers we have, for feeding electrical actuation devices, sensors, there will be a significant bump in electrical consumption. And we will need not just a 20—sorry, a 12-wattage network in the vehicle; you will need probably 48-wattage networks in the vehicle.

And given where our efficiency limitations are today with electric vehicles, largely driven by the cost of the battery pack or the cost per kilowatt hour, the first applications probably of autonomous vehicles will not only be purely full electric, because you need to have really the immediate infrastructure for charging around; it might, indeed, also be plug-in hybrid electric.

WEAST: That’s true.

KAAS: So that’s a very technical question, but I thought you wanted a little bit better insight—

RUNDE: This gentleman here in the front. Yeah.

Q: Amitrai Accione (ph), George Washington.

You talked a little bit the lag between technological development and policymaking. But—you talked about one area, but obviously artificial intelligence is entering practically everything—(inaudible)—into medicine. The next issue is a military review into all kind of autonomous weapons. Is there—where you go from here if the policymakers will fall ever farther behind?

RUNDE: Is the regulation—how are we going to—how is the regulation and the laws going to keep up with all of these challenges?

Q: Assuming they will not keep up.

RUNDE: Yeah. Assuming they won’t keep up, what do we do about this?

STANLEY: Well, I think that’s the best argument for a very light touch to regulation, because one of the grave concerns around regulation in this area is that as soon as you have a regulation that creates a certain standard or certification or requirement, that technology will change; or, worse, that you can’t have technology improve because you’re trying to meet certain—

RUNDE: Or we mandate 20-year-old technology like the DRSC (sic; DSRC), right?

STANLEY: Mmm hmm. So that’s part of the policymakers’ challenge for sure.

Q: (Inaudible)—regulation in different states.

STANLEY: And the patchwork issue. That is also a very big issue.

RUNDE: Could I—if we could have Jack and Hans-Werner. Could you each comment on this issue of the—what do we do about regulation that’s never going to be able to catch up with the galloping of technology?

WEAST: I think Karlyn said it well. I think we want policy to encourage investment, to accelerate deployment, and support testing of this technology in the real world. And I think that’s what we’re seeing around the rest of the world. I think it’s time for a fresh approach to how we look at development and deployment of automated vehicles here in this country with a light touch, to not, you know, pick technology winners and losers, but let us engineers, let the market figure out what’s the right—

RUNDE: Let a thousand engineers bloom.

WEAST: Yeah. Let us figure out the right technical solution, you know, to keep us all as safe as we can be. Let’s not have policymakers sort of pick those technical choices for us.

KAAS: The only addition I would have—and I agree with what Jack and Karlyn said—any type of regulation entity of which formally it’s regulation needs to talk with the different stakeholders and syndicate. You cannot define regulation in an ivory tower, because you will easily fall prey to some version of bipartisan interest. So you will need to talk to technologists. You need to talk to private-sector players. You need to talk to people who protect the end consumer as well. And it is an outcome of a multi-stakeholder dialogue; by the way, true for any type of regulation, not just for regulating autonomous vehicles.

RUNDE: OK. This gentleman in the back.

Q: Hi. Puneet Talwar with Crest International.

I have a couple of technical questions. You talked a little bit about 5G and LTE, but also the notion of cars talking to infrastructure around them. So I’m wondering, what are one or two areas that could be real accelerants in terms of, you know, other types of infrastructure, let’s say, in an urban or not urban site, and even that could really help the technology along?

And then, in terms of, you know, we’ve heard about the levels to the end state of fully driverless, do you anticipate any of those levels to be sort of weigh stations or plateaus that we might settle in for a little while, you know, where the mass of people might be using and sort of satisfied with that, you know, taking into account things like psychology and so forth, before we sort of jump up to the next level? Or do you anticipate this kind of a smooth ride all the way to driverless?

RUNDE: Technologies.

WEAST: Right. I’ll take the first one.

RUNDE: Yeah, start.

WEAST: So, again, great question. Thank you.

And one of the key advancements of 5G cellular technology is, unlike the cellular technology that we have in our phones today, which only talks to, you know, a tower somewhere, the fifth-generation cellular technology will support that traditional-use case, but would also support the ability to talk device to device.

And so, for example, in the platooning example, let’s say you were the very last car in a line of 15 cars, you know, cruising from New York to San Francisco, you know, but you’re two inches away from the car in front of you. But wouldn’t you like to see what that first car is seeing, right?

Well, then a localized, you know, 5G cellular vehicle-to-vehicle link could transmit video information to all the cars behind so you can all have sort of a view of what’s coming in the future. So that’s an example how we apply it sort of locally, vehicle to vehicle.

Another example would be high-definition mapping technology, which is an important part to the deployment of these vehicles. Intel made an investment in a company called HERE, which you probably all have in your car. They’re, like, 80, 90 percent of the market for navigation systems.

But what we want to be able to accomplish is have live maps. So when one of you drives down a road here in D.C. and there’s temporary construction cones that are set up, you don’t want every single other person that drives down that road to have to deal with the inconvenience of three lanes blocked off and it taking longer for you to get to your destination.

But if you have that 5G cellular link, you can capture some of that information. You could even take a picture or two maybe of what the cones look like, send that up to the cloud, that can analyze it, get the report, a similar report from different vehicles, and then distribute that information back down so that you can all—other passenger or vehicles that would come in after you could benefit and avoid that street altogether, because now you’ve been updated with sort of a live, you know, kind of high-definition map. So both, I hope, are good examples of how 5G technology, both from a vehicle-to-vehicle but also vehicle-to infrastructure standpoint, can help in these use cases.

RUNDE: Plateaus?

KAAS: And, by the way, this is a great example for value providing service—

WEAST: Mmm hmm. Sure. Yeah.

KAAS: So this is just another example of a business model which you can develop, whoever develops it.

In terms of other elements in the infrastructure, I think digitalization of road infrastructure and operating is critical, quote/unquote, “under the bigger theme of Internet of Things,” that also applies to autonomous driving and the—

RUNDE: So future-proofing the infrastructure is going to be critical.

KAAS: Exactly, that is very critical. And we will prioritize our investments. We might indeed say, in certain geo-fenced areas, we want to actually enable, with certain sensor technology, on road conditions, for example, to actually make our investments first, or certainly frequently call it use long-haul interstates, where we actually want to really future-proof the technology. So there is a smart way of prioritizing our infrastructure or digitalization of our infrastructure.

The second question you had, I do not think it will be a sequential path. Let’s first prove out L2, then L3. L3 is basically controlled, automated driving. L4 is autonomous driving, but the driver is still capable. L5 is fully autonomous. We will see parallel pathways to L3, L4, L5—and we have already L2, actually, today—simply because they will be different applications, different geo-fenced areas. So it’s much more parallel and iterative, not sequential.

RUNDE: OK. So I’m sorry we’re not getting to all the questions. So I’m going back to my question. So if you were in front of the secretary of transportation or you were in front of the president’s chief of staff, what does the Trump administration need to do in the short run to advance American interests regarding driverless vehicles over the next 12 to 24 months?

WEAST: I’ll start first. I think look at what’s happening internationally in terms of the light-hand touch. It’s encouraging the development and deployment and testing of automated vehicles. I think focusing on a framework to support that here in this country, where most of the technology companies are that are—

RUNDE: Do we need legislation?

WEAST: I’ll defer to my colleagues on whether we need—(laughter)—formal legislation, being the engineer in the room. (Laughs.)

RUNDE: Karlyn.

STANLEY: Well, what I would suggest is collaboration. And we actually make this point in our report. And that collaboration would be between states that are already moving on looking at some legislation and the federal authorities who are involved. And also I think the multi-stakeholder model is very important. Hans, you mentioned that.

KAAS: Yeah.

STANLEY: But that collaboration among the different stakeholder interests is critical.

RUNDE: OK, Hans.

KAAS: I will reinforce both comments and just use slightly different words. We are not living on an island. That’s why we need to collaborate beyond our boundaries, both for standards, but also for co-development of technology.

And the second point is I would indeed be quite inspired if we could have incentives for technology investments—not mandating outcomes, but to invest in technology, to invest in education. We will very different—we will need very different type of skill sets in the future. So what type of enabling from an education standpoint and an investment standpoint can this administration do? And I think they can do a lot. But it requires indeed a multi-stakeholder dialogue.

RUNDE: All right, thank you very much.

Please join me in thanking the panel. (Applause.)

KAAS: Thank you very much.


This is an uncorrected transcript.

Top Stories on CFR



The closely watched elections on July 28 will determine whether incumbent President Nicolás Maduro wins a third term or allows a democratic transition.

International Law

The high court’s decision could allow future U.S. presidents to commit grave abuses of power with impunity, with serious implications for U.S. foreign policy and national security.