Diane talks with Ruth Marcus, editor at the Washington Post. Her new book is "Supreme Ambition: Brett Kavanaugh and the Conservative Takeover."
California moved into the self-driving lane this week. With Google’s help, it passed a law that will allow computer-controlled cars on the road, at least on a test basis. Google has modified and tested a fleet of Toyota Prius hybrids that drive themselves using a range of video and radar technology. And that’s just the beginning. By some estimates, self-driving vehicles will make up 70 percent of the nation’s traffic by the year 2040. Proponents of driverless cars say their widespread use would reduce congestion and give elderly and impaired drivers new freedom. Others worry about safety, liability and privacy issues. Guest host, Tom Gjelten, and his guests discuss the future of driving.
- David Shepardson Washington bureau chief, The Detroit News.
- Azim Eskandarian Director, Center for Intelligent Systems Research and professor of engineering and applied science, George Washington University.
- Frank Douma Research fellow, Humphrey School of Public Affairs at the University of Minnesota.
- Anthony Levandowski Project manager, Google Self-Driving Car Project
MR. TOM GJELTENThanks for joining us. I'm Tom Gjelten of NPR, sitting in for Diane Rehm. A new California law now allows self-driving cars on state roads, at least on a test basis. Proponents promise computer-controlled vehicles will reduce accidents and congestion, maybe even eliminate the need for driver's licenses. Opponents worry there could be a loss of privacy for drivers. Joining me in the studio to talk about the future of driving in the digital age: David Shepardson of The Detroit News and Azim Eskandarian with the Center for Intelligent Systems Research at George Washington University.
MR. TOM GJELTENAnd for those of you who want to hear more about this fascinating development or have questions or concerns about driverless cars, our phone number is 1-800-433-8850. Our email is firstname.lastname@example.org. We'll be getting to your calls and emails later in the show. Of course, you can also join us on Facebook or Twitter. Good morning to both of you.
MR. DAVID SHEPARDSONGood morning, Tom.
PROF. AZIM ESKANDARIANGood morning.
GJELTENSo if you can just hold off for a few minutes because I want to start with Anthony Levandowski. He's the project manager for Google's self-driving car project, and he's on the line with us this morning from Mountain View, Calif. Thanks for being with us, Anthony.
MR. ANTHONY LEVANDOWSKIGreat to be here. Thank you.
GJELTENSo the -- I guess what -- driverless cars have been in development for many years. But the big news this week is this law that passed in California allowing some driving of self-driving cars. Tell us what's in this law exactly.
LEVANDOWSKIWell, this law is called SB 1298, and it sets forth the framework for the DMV to set up the guidelines that vehicles will need to adhere to in order to, in the future, be able to drive themselves with or without people inside them. It outlines the type of technical standards, the types of insurance requirements. Everything that you would expect a person to go through when they get their driver's license, you would expect a vehicle that doesn't have a driver in it to be able to have.
GJELTENAnd what vehicles will actually be on the road? I mean, these -- obviously, these vehicles are not available to the general public yet. So are we talking here about test drivers being able to do this? What exactly is going to change as a result of this law being passed?
LEVANDOWSKIWell, what's changing really is more on the fact that, currently today, the technology is ahead of the law. And at Google, we've been focused on developing this technology with safety being our top priority because we believe that it's unacceptable that 40,000 Americans die every year, 1.2 million people worldwide, and roughly 90 percent of these accidents are caused by human error.
LEVANDOWSKISo as the technology has been evolving in the last several years, we're reaching a point now where -- that we're seeing real benefits of having this technology installed on people's cars in our test slate. And I think it's a great move by Sen. Padilla to put forward measures that would close the gap between what the law says today, which is that a person doesn't need to be driving, which leaves the field wide open to something that's more sensible that has the proper guidelines in place.
GJELTENWell, tell us exactly how these cars work.
LEVANDOWSKISo these cars have a variety of sensors on them. They have cameras that look at traffic lights. They have laser scanners that measure the world all around it in three dimensions and radars that track other vehicles and their speeds. And all of this information is sent to a computer in the back of the vehicle that processes information and understands where the vehicle can safely drive on the road and where it's located. So you can imagine, just like your GPS navigation right now, it gives you turn-by-turn directions verbally.
LEVANDOWSKIWith these additional sensors on the vehicle, you're able to properly have the vehicle drive down the street and understand where the traffic light is. And if it's red, it's going to stop. And then it's going to wait for the pedestrians to get out of the crosswalk before it makes that turn and kind of, you know, remove the aspects of driving that are dangerous for people, which is when we don't pay attention.
GJELTENAnd do these cars know where they are because of the GPS technology? I mean, they know what road they're on, and therefore, they don't just react to obstacles. But they have intelligence about where they are and what they're doing?
LEVANDOWSKIRight. We don't rely on GPS fundamentally to keep us driving in a similar lane. But GPS is a really good indicator to know that you're in, you know, downtown D.C., you know, or in San Francisco. But the driving itself really relies on the sensors. Just like, you know, when you drive, you look out the window. You don't look at your navigation display. You need to see what's in front of you, what's in front of the world.
LEVANDOWSKISo the GPS part is just to guide you in terms of saying you need to be going northbound. But what the car picks up from the ground is really the important part of the driving.
GJELTENAnd what have you at Google already done with this? You know, I read somewhere that you actually were secretly doing some test driving -- in advance, that is -- of getting official permission to do so.
LEVANDOWSKISure. So we haven't been doing secret testing. But, you know, the project started back in 2008, and we've been testing our self-driving technology in California. And the number one thing that, of course, motivates us is the ability to improve people's lives dramatically. And so far, we've done about 300,000 miles of testing. And, you know, our vehicles, A, they look a little bit strange 'cause they have all this technology on them, and, B, they have, you know, Google's self-driving car labeled on the side of them. The...
GJELTENWhere was that done, Anthony, that 300,000 miles that were logged?
LEVANDOWSKIIt's been done on a variety of highways and surface streets in the Bay Area, in California, Nevada. We've driven in downtown D.C., actually, near the Capitol, and we've driven also in Florida.
GJELTENBut how did you do that? How did you do that if the laws weren't on the books yet to allow it?
LEVANDOWSKIWell, so the laws that are on the book currently in California were requiring a person to have a driver's license for the person to drive but no requirement for a computer to be able to drive the vehicle. And that was the same in Nevada, Florida and D.C. today. So it's more -- it's not that we're looking for permission. Permission has kind of always been there. What we're really looking for is kind of setting a sensible guideline so that everybody can do this safely.
GJELTENYeah. So what's -- I mean, you said quite clearly that, you know, what your interest here is in safe driving and promoting safety. But what is Google's interest as a company? In other words, how did your company in particular come to this interest? How did it grow out of other products you've developed, other research you've done, you know, other business lines that you've been in?
LEVANDOWSKISo Google's really a technology company, and we love innovation. And we love solving big problems. So, you know, where we see our ability to build new software to have a huge impact in the world, to save lives, make people's lives richer, we see it as a huge opportunity for us to explore. And so we don't have a commercial plan on how this technology will come to market, but we just think that, you know, in our innovation spirits, we can't pass this up. This is something too important for a society to have for it to lie dormant for a long time.
GJELTENYou know, I think it's really easy to get excited about this development. Are there any caveats that you want to sort of lay out there right now so that our listeners don't get perhaps too excited about, you know, what they're going to be able to have?
LEVANDOWSKIAbsolutely. So it's important to remember that, unlike things in Hollywood, technology, you know, takes a long time to develop. And right now, we're at the early stages. The things that we're working on the most are making sure that the system is more reliable than people are. And in many case, we've shown that that's true for some scenarios but not all scenarios. So while I'd love to have everybody be able to have one of these in their driveway next year, it's going to be a lot more longer before the technologies mature enough to go out.
GJELTENAnd will Google actually manufacture cars or work with manufacturers, provide the technology that then car manufacturers will incorporate? What's going to be your relationship with car manufacturers?
LEVANDOWSKIWell, we love technology, so we're really interested in building the technology aspect of this, and we're looking forward in talking to all the manufacturers on how to incorporate this into their vehicles. So how it will come out is undecided yet, but we're talking to everyone.
GJELTENWell, finally, Anthony, how quickly can we expect actually to be able to buy a computer-controlled car?
LEVANDOWSKIThe -- it's a handful of years, right, so less than the number of fingers on one hand. But more precisely than that, we just don't know today.
GJELTENAll right, very good. Anthony Levandowski is the project manager for Google's self-driving car project. He joined us on the line from Google headquarters in Mountain View, Calif. Thanks so much, Anthony, for sharing this exciting news with us this morning.
LEVANDOWSKIYou're welcome. Thank you very much.
GJELTENAll right. Thank you. David Shepardson, The Detroit News, you have actually ridden in one of these vehicles. How realistic is this idea? Same question that I just put to Anthony Levandowski, you know, how important it is for us to kind of constrain ourselves here a little bit in terms of our excitement?
SHEPARDSONIt is realistic in certainly the medium term. But in the short term, we're going to see more uses of computer assistance to drive cars. We already see some of those now with devices like adaptive cruise control, which your car keeps you at a set speed and a set distance away from the car in front of you. There are lane departure warning systems, which will alert you if you leave the lane.
SHEPARDSONAnd probably the most -- the biggest example of the computer actually taking control of your car is in the case of you're not paying attention to a stopped car in front of you, and in some cars, especially luxury cars, the car will actually brake the car and stop it for you. So in the short term, you're going to see more examples of the computer, you know, jumping in when you're not paying attention when you should be 'cause, as Anthony pointed out, at least about 90 percent of accidents are human error related.
SHEPARDSONBut in the long term, I think, you know, 10 years or so, I do think this is a realistic possibility simply because, you know, you're talking about 32,000 deaths, $300 billion to society. This is a big cost, and computers will likely be able to do a better job.
GJELTENSo this is a very incremental development. We have -- my car, for example, warns me when I'm getting too close to a post or...
GJELTENAnd, you know, bit by bit, we're going to see more of these technological innovations in that direction.
SHEPARDSONAbsolutely. Just for the simple reason that computers do a better job. They don't get tired. They don't, you know, drink too much, and they, you know, they stay on focus, on message.
GJELTENIt's pretty exciting. David Shepardson is Washington bureau chief for The Detroit News. Also with me here is Azim Eskandarian, director of the Center for Intelligent Systems Research. He's also professor of engineering and applied science at George Washington University. And, Azim, I'm going to go to you in a minute or two. But we're going to take a break right now. Stay tuned.
GJELTENWelcome back. I'm Tom Gjelten, sitting in today for Diane Rehm. And our subject is computer-controlled cars, driverless cars, something that we used to see on "The Jetsons" that actually is now becoming, step by step, a reality. And my two guests here in the studio are David Shepardson, the Washington bureau chief for The Detroit News and Azim Eskandarian, director of the Center for Intelligent Systems Research at George Washington University.
GJELTENAzim, I apologize we didn't get to you before the break. We got too wrapped up talking to the guy from Google, which is the company that's really doing a lot of the research for this. But I read somewhere, Azim, that the DARPA, the Defense Advanced Research Projects Agency, the Pentagon outfit that funds so much leading-edge research that was instrumental in developing the Internet, for example, actually had a role in the first wave of innovation that lead to this.
GJELTENAnd as I understand it, Azim, DARPA was interested in developing drone vehicles for use in combat, like drone aircraft that are very well-advanced, very far advanced and in widespread use in war fighting right now. Is that correct?
ESKANDARIANThat's correct. Yeah, DARPA had a major role in developing autonomous vehicles. Earlier, I think around 2005, they put out a major challenge called desert challenge, which required cars to be driven autonomously for long distance in a desert environment where you had the possibility of GPS signals missing and through rough terrain. And then later they had another major challenge called Urban Challenge, which was the same challenge, similar challenge but an urban environment, requiring autonomous cars to be driven in an urban environment with traffic lights and other cars.
ESKANDARIANSo, of course, autonomous vehicles will have both commercial applications as well as military application. That was the -- DARPA's role in enhancing the military capability as well as the commercial capability.
GJELTENAnd where does the technological innovation actually stand right now? I mean, are the obstacles at this point more legal, as David and Anthony were both talking about, economic in terms of, you know, dealing with cost of manufacturing? Is the technology -- are the ideas actually already there to do a lot of this stuff that we're dreaming about?
ESKANDARIANI believe the technology is already there. My personal research is in the area of advanced driver-assistance systems, which is some of the technologies that David already alluded to in terms of adaptive cruise control, lane departure warning, lane departure control, blind spot detection and many other technologies that you see in the vehicle right now. So as you mentioned earlier, these technologies can be collectively put together to drive the car autonomously eventually.
ESKANDARIANNow, the sensor technology is there, and the actuation technology in driving the car by itself is also available. But I think the sensor technology and the collective of these programs that run the car together is expensive with today's standards. But after mass market production of these systems, I believe the cost will not be an issue. So I think, from an engineering perspective, the automation technology is currently available.
GJELTENAnd how much of the technology is in the vehicle itself as opposed to the environment? I'm wondering if, you know, whether we're going to see something like what trolley cars used to have, which was, you know, something embedded in the pavement that guided the trolley car. Are we going to see technology embedded in roads? Are we going to see, you know, traffic lights that are part of this network? I mean, it would seem to me that it seems like a lot of the technology has to be outside the vehicle.
ESKANDARIANWell, it's a collective of both technologies. I think in intelligent vehicles by themselves, autonomous vehicles by themselves, that most of the technology currently is inside the vehicle. If you look at the driving scenario, what the autonomous vehicle is trying to do is to duplicate or replicate what humans do when they're driving. We sense the environment. We respond to the environment by trying to keep -- caught at the center of the lane. We accelerate and brake. And we are replacing all of those by machines.
ESKANDARIANSensors will sense the environment. The computer program will do the thinking of navigation and instructing the different actuators in the vehicle, and the actuators will drive the vehicle within the lane. Now, this is all inside the vehicle. But there is a trend of a new technology coming up which is called vehicle-to-vehicle and vehicle-to-infrastructure communications.
ESKANDARIANAnd that vehicle-to-infrastructure communication will require that the infrastructure be equipped with communication technology and part of the computer programs that can interact with the vehicle and send signals to the vehicle for safety applications. And you can imagine various applications, for example, in intersections, which is a major concern, you can have traffic signals that are equipped with this technology to interact with the drivers.
GJELTENAmazing. David Shepardson.
SHEPARDSONIn fact, just last month, Transportation Secretary Ray LaHood announced in Ann Arbor a 3,000-car test of vehicle-to-vehicle research to actually get these cars on the road, getting cars talking to each other so when they come to an intersection, if one car is going to speed through, it alerts the other car. And they had done testing on infrastructure. The reason the infrastructure has not been as popular is just the sheer cost.
SHEPARDSONYou know, you're talking about, you know, millions of intersections, lights, roads. And it's more cost-efficient, I think, now to have devices in vehicles. We can get vehicles talking to one another and sensing, you know, lights and so on. It will be far cheaper than having to put this on every intersection in the country.
GJELTENHow realistic do you think are these claims that are being made that this will dramatically reduce traffic accidents, casualties?
SHEPARDSONI think a lot. I mean, number one, you're talking about 90 percent of accidents are human error. And there's so much money that can be saved and so many lives saved. You're talking about $100 billion we spent on congestion costs, $300 billion on just the cost of crashes, the 2.2 million people a year who are injured in accidents. And, yes, the fatality rates come down dramatically. It's 32,000 people who died last year. That's 1.1 people who die per 100 million miles traveled. By comparison, in 1950, it was about seven people who died per 100 million miles traveled.
SHEPARDSONSo we've made dramatic progress in reducing deaths, but it's still the number one cause of death for people aged 5 to 34. And there is, you know, just a huge demand to see improvements. And, you know, you're not going to ever build a car so safe that the driver is not going to make mistakes. And the only way you will avoid long-term drivers making mistakes is if you give the computer more control. It sounds like, you know, a science fiction movie, but...
GJELTENAzim, you know, sticking with this safety issue, Morgan writes us an email, "What happens when the computer crashes? Is it game over?" All computers have glitches. We know that. What happens if you're driving down the road 70 miles an hour, and, all of a sudden, your computer freezes?
ESKANDARIANRight. Well, the major challenge in these autonomous driving is obviously to have robust and reliable sensor systems and actuating systems. Now, if the computer crashes, you can always design a system which would -- will be far tolerant, and it will have enough redundancy built into the system that will bring the vehicle to a safe stop. And that has been experimented with. And you can model different failures and try to come up with countermeasures for those.
ESKANDARIANSo that's the role -- even now, we can have driver-assistance systems that can do that. For example, in emergency maneuvers, take over the control of the car when the driver fails to maneuver the car properly.
SHEPARDSONShe's exactly right. I mean, that's why Google has already put 300,000 miles out. That's where they're going to put at least a million miles on before they actually try to test this. I mean, the public relations nightmare of a single accident being caused by a computer, I think, could set back the program 10 or 20 years. And, I think, keep in mind the first five, 10, 15 years, you're not going to be sleeping in the back of your car when this driverless car is driving around.
SHEPARDSONYou're going to be in the driver's seat having to pay attention. Maybe you're going to be able to not always look forward, but the car is going to alert you to more dangerous situations. And, you know, initially things like getting on and off an exit ramp, I mean, the computer -- the technology is not there yet in terms of the most complicated maneuvers. I mean, initially, you're on a long stretch of highway. That's a type of thing where we're going to feel more comfortable allowing the computers to take over rather than piloting a more difficult...
GJELTENSo we start out in a rural area. And driving in an urban area is far more difficult, far more challenging.
SHEPARDSONPedestrians, everything else.
ESKANDARIANWell, I think even in the urban area, there is a role for autonomous vehicles because, just this morning, I spent about 45 minutes in a stop-and-go traffic coming to the studio here. So that's a scenario where you're going at a very slow motion, and the sensors will be able to detect the vehicles in front and the pedestrians and whatever else that happens in the traffic in a regular traffic situation.
ESKANDARIANSo you can imagine the increase in productivity where people are sitting in a congested environment but the vehicle is driving by itself because that's one of the safe areas where autonomous vehicles could be driven.
GJELTENDavid, what's the -- you know, Google is not a car manufacturer. Google is a technology company, an innovator. Car companies are, you know, facing some pretty stressful times right now. I mean, GM just about went bankrupt. What is the attitude in Detroit toward these cars? 'Cause I can imagine, you know, it can vary from excitement to nervousness. I mean, are people going to sort of no longer see cars as toys, which is a big part of the advertising for automobiles for the last 80 years?
SHEPARDSONIt's twofold. Number one, car companies want to be with it with technology. That's why you see a lot more technology in your cars now, being able to read your emails to you, being able to send text and Facebook updates, you know, by voice. So car companies want to do that, in part, because they're worried, frankly, that kids are not going to want to drive as much. In fact, studies show younger Americans are not getting their driver's licenses.
SHEPARDSONThey'd rather, you know, text their friends. So they are looking for ways to keep cars rolling. And, you know, in theory, hey, if you could still do the type of things you want to do -- you could text, you could do whatever while the car gets you from point A to point B -- they like that. But on the other hand, there's a tremendous liability issue here. You know, just this week, the auto companies, the trade groups said, hey, we're worried that, you know, once we sell our car, you know, Google or whoever can modify it, but we don't have any control over what happens to our car.
SHEPARDSONWe're going to be held liable. They're going to sue us if someone, you know, modified our car and used part of the systems to do this driverless car. So I think, generally, they're pretty cautious. But on the other hand, the major auto companies are doing research on this. I mean, GM in 2008 thought that driverless cars can be possible by 2018. So they're definitely all interested in this. The question is some companies are a little more eager to get it done sooner, and others are going to kind of wait and see.
GJELTENYou know, mentioning -- with reference to legal liability, we have a listener who wrote, who's wondering: "One would assume that if the state of California is going to permit driverless vehicles on its roads, that the details of legal liability have been defined and will be implemented." But I think I remember Anthony basically saying that this is at the very first stage, at the root, that the legal liability issues have not all been worked out yet.
SHEPARDSONRight. Basically, your own insurance policy you have to have as a driver will cover you. The question is, will the insurance companies -- and that's the talks right now -- will they require additional coverage for essentially the hardware? We have to have a separate policy if the car is driving as opposed to you in the driver's seat.
GJELTENAnd what is the attitude of the -- I mean, is the -- the car manufacturers have to also worry about what their legal liability would be if this technology breaks down. And, you know, it's one thing if that -- if a driver messes up and, you know, causes a serious accident. If the computer messes up, does that mean that the manufacturer of the computer is liable?
SHEPARDSONSure, and the parts. I remember in 2008 when there was a huge issue about Toyota and sudden acceleration where electronic glitches took the blame. I mean, people are worried. I mean, we have very large, complex computers already in our cars, running lots of systems. And so there's already a question of the liability, you know, for the computers in cars.
GJELTENDavid Shepardson is Washington bureau chief of The Detroit News. I'm Tom Gjelten. You're listening to "The Diane Rehm Show." And we're talking about the prospect of driverless cars in the short-term future, not the long-term future like we might have thought. You can join our conversation. Our phone number is 1-800-433-8850. Our email, as you know, is email@example.com.
GJELTENAnd just a few minutes, we're going to get to our callers, but I want to go now to Frank Douma. He's research fellow at the Humphrey School of Public Affairs at the University of Minnesota, and he's joining us on the phone from his office in Minneapolis. Good morning, Frank.
MR. FRANK DOUMAGood morning.
GJELTENLet's pick up with you because just before the break here, we were talking about these legal liability issues, which is something that you have been looking at a lot. What do you...
GJELTENWhat is the state of the current legal system? Is it ready to deal with these very new liability issues that are coming along?
DOUMAI would say that it probably is not. The way most laws are written governing how people drive their cars on the road assume that there is a human in control and ultimately going to be responsible for most things that happen when things go wrong. And as the computer takes over more and more of these duties, that becomes a defense for the person who is in the driver's seat and otherwise probably having to face the liability, as everyone assumed it's going to be the case.
GJELTENKevin, one of -- I'm sorry, Frank, a listener, Kevin, wonders if self-driving cars will fix the DWI problem. How about that?
DOUMAWell, it would seem that they would functionally, but, legally, that is very, very unclear. The closest analogy that I've been able to find have been cases where someone has been over the limit of intoxication, and so they chose to park their car off the road and fall asleep and try to sleep it off.
DOUMAAnd they were found guilty of DUI because they had the key in their pocket and therefore had the potential, at any time, to drive off while they were still over the limit. So, in this case, if there's even a potential for somebody to take over the car again while it's driving itself, they probably would also face DUI liability.
GJELTENAnd what do you expect will be some of the specific regulations that are written into laws that limit what can be done with computer-controlled cars? I think you've already done some research into what the state of Nevada is doing in this regard, as well as California. What are some things that drivers, or that occupants, of driverless cars can, cannot do?
DOUMAWell, most of the laws -- you know, California's is very much sort of a guideline to sort of, OK, let's get the regulations going law at this point. It's quite general. The Nevada law has -- had some regulations come out, and so you can see a little bit more. It's still not overly specific, but probably the -- one of the most interesting things I've seen coming out of Nevada is they specifically exempt self-driving cars from their texting while driving ban. They specifically say that someone in the driver's seat of a self-driving car while it's in self-driving mode can be texting.
GJELTENThat's amazing. So they've already anticipated that and decided that these cars are safe enough to drive that you can actually text while you're driving, which everybody else is telling us is a really bad idea. But if you have a computer-controlled car, you can do it.
DOUMARight, right. And in Nevada, the way -- one way to help enforce that is that self-driving cars will have a very different-looking license plate. And so somebody can -- officer can realize that, oh, this is somebody that is texting, but they're in a car that could be driving itself.
GJELTENWell, Frank Douma, please stay on the phone with us. We're going to come back to you after the break. He's -- Frank is a research fellow at the Humphrey School of Public Affairs at the University of Minnesota. He's been looking into a lot of the legal liability issues around the use of computer-controlled cars.
GJELTENI'm also joined here in the studio by David Shepardson, the Washington bureau chief of The Detroit News, and Azim Eskandarian, director of the Center for Intelligent Systems Research at George Washington University. And after the break, we're going to be going to our phones. Please stay with us. This is "The Diane Rehm Show."
GJELTENWelcome back. I'm Tom Gjelten, sitting in today for Diane Rehm. And we're talking about the prospect of computer-controlled cars, driverless cars that -- where you can sit there and read emails or watch TV -- I don't know if you can take a nap yet or you'll be able to take a nap, but it certainly is a -- an exciting development.
GJELTENAnd my guests are David Shepardson, the Washington bureau chief for Detroit News, Azim Eskandarian, director of the Center for Intelligent Systems Research at GW University here in Washington. And joining us on the phone is Frank Douma. He's research fellow at the Humphrey School of Public Affairs at the University of Minnesota. Couple of emails here -- did raise a similar question.
GJELTENTerry wonders whether this will be something that can help blind people get to work. It's always been an issue of blind people, public transportation and so forth. And Jason says he lost his ability to drive in 1991 due to Leber's disease, which resulted in him being legally blind. Azim Eskandarian, does this technology offer promise for blind people, either totally blind or partially blind, to actually be able to drive?
ESKANDARIANSure. We have to see what -- how we look at this technology. If you're talking about totally autonomous vehicles, the vehicles that do not need the driver intervention at all, then of course, this technology can help blind people as well as any other disability. Basically, what is expected of the driver is to specify the destination and the computer program will use a navigation system, plan a trip -- optimal trip, and then execute the commands that would drive the car to that destination.
ESKANDARIANSo, regardless of the status of the driver, the car can be driven. That requires that the vehicle will be driven in totally autonomous mode that will allow no driver intervention. Of course, the discussions we had earlier, we had scenarios where we said the driver can intervene and interact with the vehicle and take over the control. For this particular application, the driver will not be allowed to do that.
GJELTENFrank Douma, I have one last question for you before you take off. Google is involved in this, and Google, of course, is associated in the minds of many people with privacy issues. Are any privacy issues raised by the prospect of this technology becoming widespread?
DOUMAYeah. This is something that definitely needs to be looked at. I'm not saying that we're going to all just to have to give up our privacy to enjoy it -- enjoy self-driving cars. But we've got an issue here of, whereas most drivers in a non-self-driving car or just overly dumb car can decide that they are, you know -- they'll make their decision about where they're going to go and how they're going to get there. And that decision and those other points about where they're at all stay in their own head.
DOUMANow, you are sharing that information, at least with the car. And when you're talking about the vehicle, the infrastructure of vehicle, the vehicle situations, you have that data being shared with other parties as well. And we need to have a legal structure that will identify how and when that data can be used beyond the particular interactions that are needed to make the car drive itself.
GJELTENLots of legal issues to be worked out here. I'm not sure that my -- I think my car would be offended at being called a dumb car, Frank. But in any case, thanks very much for joining us.
DOUMASure. My pleasure.
GJELTENFrank Douma is research fellow at the Humphrey School of Public Affairs at the University of Minnesota, and he's been looking carefully into a lot of the legal issues around the introduction of computer-controlled cars, driverless cars. I want to go now to Dennis, who's on the -- Dennis, there you are. Dennis is on the line from Allegan, Mich. Good morning, Dennis.
DENNISGood morning. Question is, is what do you do about malicious hackers? You know, in any system that can be developed is going -- can be overcome by something who's at least as logical as the designer. A really easy one would be if somebody wants to stop your car in traffic, they could use a Doppler shift unit attached to a receiver and a transmitter that transmitted back on the same frequency as the radar emitter in the car, thus would lock the car up in traffic and stop it, you know? Malicious hackers are going to do what they do just to -- you don't have to apply logic. They just do it.
GJELTENWell, Dennis, it sounds like you know what you're talking about. And do you think that this -- from what you've heard, do you think that this technology, like other computer technology, would be pretty susceptible to hacking? I guess you do, or you wouldn't have said that. We actually had an email from Grant, who raised the same question. "Is hacking a possibility when cars are operated by computers?" David.
SHEPARDSONAbsolutely. You have two issues here. Number one is making sure that data is heavily encrypted so that as it's being shared between vehicles and to the infrastructure that hackers can't, you know, can't get to that data. But the other more troubling question that Dennis raises is exactly right, that people, malicious people could send signals toward cars, especially as the number of driverless cars grew, you know, to essentially trick the cars into doing things they shouldn't.
SHEPARDSONIn fact, we had an example of a car, because of some metal in the road -- in fact, the government's investigating this -- was coming to a dead stop when the metal in the bridge basically set off a sensor in his car because of where the metal was. So we already have, you know, non-malicious examples of where something can, you know, can tell the computer, you know, give it inaccurate data. And it is something we'd have to figure out, how could you prevent hackers from doing it. So it's a very real issue down the road.
GJELTENAzim, is it -- would it be possible to design this technology, to build this technology in such a way as to make it less susceptible to hacking?
ESKANDARIANOh, absolutely. I think, just like any other computer technology, there is potential for intrusion by outsiders. But you can design the system such that it has enough encryption incorporated in the whole system that prevents that.
GJELTENGood. Eric is on the line with us now from St. Louis, Mo. Good morning, Eric.
ERICHey, good morning. Hey, I must say I am a little bit skeptical of even the -- solving these technical issues we're talking about. But I'd like to mention a larger human implication here. And if you look at the history of technology over the last 200 years, it's often a history was new technology is brought in to solve problems that previous technology created.
ERICAnd the -- what this is gradually leading to is a future in which humans are becoming obsolete and the docile captives of their technology, which is a complete reversal of the promise of technology, which is to give us autonomy. It's actually giving autonomy to our technology.
GJELTENEric, are you thinking of any things that have happened in particular that sort of fit that pattern that you're describing where, you know, some new technology has resulted in people becoming less responsible, less active, more docile? What are you thinking of in particular?
ERICWell, another perfect example is right in the automobile itself with the GPS. People, they turn over the -- they turn over the map reading skills to their GPS system, and, as a result, they are clueless about where they are. If you take away that GPS, they have no idea where they are.
GJELTENThat's actually a really good point. David.
SHEPARDSONEric is exactly right about what the car, over the last century, has stood for, you know, freedom, the ability to go wherever you wanted to go, get on the open road. And, you know, that's how Detroit, you know, became one of the wealthiest areas in the country, sort of, you know, giving sort of this American dream, empowering people to go where they want to go. And, you know, certainly by abdicating, you know, the ability to drive in some or all situations, you turn the car into something that it's -- that it hasn't been.
SHEPARDSONAnd certainly, as a society, we have to decide, do -- is it -- do we want to let people drive to enjoy the, you know, racing down the road? Or are we so concerned about deaths and injuries that we'd rather let our computers do it? And it's a question society's going to have to answer.
GJELTENWell, it becomes -- the car, in that case, becomes much more a utilitarian thing and not something that you can identify with personally, that you have pride in, that's got personality.
SHEPARDSONRight, a washing machine, basically.
GJELTENA washing machine. But what's going to happen to Corvettes? What's going to happen to Porsches? What's going to happen to all these cars that people love to drive and not just sit in, Azim?
ESKANDARIANI think a little differently because I think you could still have your freedom to drive and then enjoy the safety that the new technology provides because, as we see this -- all of these technologies that are considered active safety systems or driver assistance systems, they're really helping the driver to drive safer. So what you're trying to do is to make sure that the technology is helping the driver, not totally taking over the drivers' control.
GJELTENLet's go now to John, who's on the line from Lafayette, Ind. Good morning, John.
JOHNGood morning. Do you -- can you get the cell call here?
GJELTENYeah. Are you calling me from inside your car?
JOHNI'm in the car.
ESKANDARIANI hope he's hands-free.
GJELTENI hope it's hands-free. Is it?
JOHNOh, it is.
JOHNActually, it is.
GJELTENAll right. Good.
JOHNThe thing that I wanted -- I enjoyed the comment about people getting lost, letting GPS decide where they are because that's what happens to all of the passengers in the car, you know? But my comment is I have a 2011 Cadillac with active cruise control. It's a great device. You can follow a car at 300, 350 feet on the highway at highway speeds. And the only problem is, if -- you know, it will speed up, slow down, it will follow the car in front of you or the truck.
JOHNAnd it's a wonderful device. The only problem is if somebody falls in between you and the car that you had targeted on, then, automatically, your car brakes hard and surprises anybody that's following you at highway speeds because they can't see any reason why you would brake.
GJELTENHow hard? How hard? Has this happened to you, John?
GJELTENHas this happened to you?
JOHNWell, you're breaking up now, or I'm breaking up on your signal.
GJELTENAll right. Well, that's a really interesting point. And, David, what John is pointing out is that this is not all that futuristic, that, you know, this is a perfect example of computer-controlled driving already in use.
SHEPARDSONAbsolutely. And there are going to be plenty of situations where the computer reacts to a sensor or data that, you know, is inopportune, like the car, you know, weaving in between traffic. So, you know, we design these safety systems, and they, you know, for the vast majority of instances, they are doing exactly what we want them to do. I mean, you know, think about, you know, not having to speed up, slow down. You know, you're not paying attention in the cruise control. You know, basically, you ram into the back of the car.
SHEPARDSONBut none of them are foolproof. And that's why, at least for the short term -- you have to think 10 years or more -- you're still going to have a driver in the seat. You're not going to feel comfortable enough, I don't think, you know, handing over total control of the car until we've had a lot of data and a lot of tests where the driver didn't get involved but could have got involved.
GJELTENAzim, I have a question for you. This comes from an email from Joe. He wants to know, "What would be a likely price tag on this thing?" How costly would this technology be?
ESKANDARIANIt's very hard for me to predict what the price tag will be, but I'm sure it would be more expensive than the normal cars that we have. I'd like to just add one comment to the previous question...
ESKANDARIAN...that, really, the -- how aggressive the car brakes or how moderate their braking is is a design issue. And also, another issue is the drivers getting used to this technology. Most people are not used to automatic braking, so when the car does that, they, you know, they feel uncomfortable about it.
ESKANDARIANOne of the challenges is, for example, when you're approaching a potential collision, and if you have a system that swerves around that obstacle or around that car, it has to take over the control of the steering. And people do not feel comfortable with that. It's one of the areas that we are conducting research to see the acceptance level of such systems.
GJELTENAzim, have you driven one of these cars?
ESKANDARIANI've driven the car -- not totally autonomous cars, but cars that have a lot of the driver assistance systems.
GJELTENVery interesting. Azim Eskandarian is director of the Center for Intelligent Systems Research. I'm Tom Gjelten of NPR, and you're listening to "The Diane Rehm Show." Let's go now to Edward, who's on the line from Chicago, Ill. Good morning, Edward.
EDWARDGood morning. I appreciate this discussion. It's very interesting. I understand that the technology is advancing at record-breaking speeds and everything else. And this is -- branches off of maybe a subject that the previous two callers were talking about, the one where he had the GPS in his car and he doesn't know where he's at. He just looks at the screen.
EDWARDAnother one is the depends on the braking of the car, and the driver feels uncomfortable about that. I think that you're missing an entire premise that's underlying this entire thing. I'm not trying to give anybody a hard time about their research in -- do -- going forward with these kind of systems, but you are dumbing down the driver.
EDWARDYou are literally dumbing down the driver. You are causing the driver to become dependent on these systems. If he knows it's there -- and maybe it slightly brakes one time -- the next time it happens, he's going to be less dependent on his own responsibilities to react and maintain control of his vehicle and maintain his driving skills up to par.
EDWARDAnd he's going to become more and more dependent on these electronic devices, just like the GPS and just like the other things you talked about. I think this entire thing has to be looked at from a different perspective. The driver is dumbing down themselves.
GJELTENAll right. You know, Edward, that's a really important point. And Jim in Florida, who is a listener, has written basically the same thing in an email. He says, "Does your panel see any danger with this technology, creating drivers who are less able to safely operate a vehicle manually after relying on automation? Driving isn't always, as the phrase goes, like riding a bike." David.
SHEPARDSONThere's two issues. Number one, a big concern of the government and researchers has been, will people use these technologies to do more unsafe behavior like text or make calls or do other things they shouldn't be 'cause, hey, the car is going to, you know, going to save me anyway? But I think the bigger issue is we're already a society of pretty bad drivers. I mean, you know, there's 90 percent of the 5.5 million accidents a year are the result of human error.
SHEPARDSONI mean, everybody who drives down the road sees inattentive drivers, people doing things they shouldn't be doing, you know? And are you more comfortable having the computer do it, or are you more comfortable with the people around you who are not always the best drivers?
GJELTENAzim, how important is government? We talked at the very beginning of the program about the role that DARPA, the Pentagon's research agency, played in this at the front end. How important from here on is government investment, government support, government patronage?
ESKANDARIANI think government plays a very, very important role, especially in terms of technology, that the role could be a standardization of the technology because you don't want different manufacturers producing technology that is totally different. And the drivers will not be used to this technology from one manufacturer to another. So standards plays a major role, and promotion of safety, of course, the safety habits is another role -- major role that the government can play.
GJELTENAnd, David, very quickly, what would you say right now is the prevailing attitude in Detroit towards this technology, positive or negative?
SHEPARDSONOh, I think it's positive. That's why you see them rolling out all these technologies, like the Cadillac, all these other ones. No, they're positive long term. It's a big question mark about, will the driver, you know, be able to get out of the driver seat.
GJELTENThank you very much. You know, I want to end with this thought that -- again, a kind of a caveat, a warning to our listeners not to get too carried away with this. I read this in one of these articles about self-driving cars, "Tomorrow's car owners can settle back in the comfort of their ride to get a start on work emails, browse the news headlines or catch up on a favorite TV show all without worrying about getting lost or switching lanes on the highway."
GJELTENI think on the basis of what you guys have said to us today and what we've heard from listeners, we shouldn't think that that moment is just around the corner.
SHEPARDSONDon't sell your current car just yet.
GJELTENDavid Shepardson is Washington bureau chief, Detroit News. Azim Eskandarian is director for the Center for Intelligent Systems Research. He's also professor of engineering and applied science at George Washington University. I'm Tom Gjelten, sitting in for Diane Rehm. Thanks for listening.
ANNOUNCER"The Diane Rehm Show" is produced by Sandra Pinkard, Nancy Robertson, Denise Couture, Susan Nabors, Rebecca Kaufman, Lisa Dunn and Megan Merritt. The engineer is Toby Schreiner. Natalie Yuravlivker answers the phones. Visit drshow.org for audio archives, transcripts, podcasts and CD sales. Call 202-885-1200 for more information. Our email address is firstname.lastname@example.org, and we're on Facebook and Twitter. This program is a production of WAMU 88.5 from American University in Washington, D.C. This is NPR.
Most Recent Shows
In 2015 journalist and author Evan Thomas set out to get inside the troubled mind of President Richard Nixon. Using dozens of interviews and what was then newly released archival material, he paints a portrait of the complex man he calls “fantastically contradictory.”
What makes dogs so unique? Animal psychologist Clive Wynne says their capacity to love.
From Diane's archives: A 2002 interview with Fred Rogers, and a 2017 interview with Tom Hanks, the actor who masterfully brings him to life in the new film, "A Beautiful Day in the Neighborhood"