Despite fear of self-driving vehicles, this expert says their expansion is inevitable
Download mp3 (12.14 MB)
Last month, a crowd in San Francisco destroyed a Waymo autonomous vehicle, eventually setting it on fire. That followed an incident in which an AV from a different company, Cruise, hit a pedestrian, also in San Francisco, and pinned her underneath the vehicle. That sent the woman to the hospital, and forced Cruise to pull its fleet off the road.
These incidents, and others, have led some people to question the safety of AVs and ask whether or not they should be allowed on the road in certain cities. Waymo has been operating in Phoenix for years, although Uber pulled out after a vehicle hit and killed a pedestrian in 2018; that car had a driver behind the wheel.
To talk more about the perception of this technology, and its potential future, The Show spoke with Alex Roy, a former self-driving executive and now principal at the firm Johnson and Roy Consulting. He calls himself a former AV skeptic who’s now on board with them.
Full interview
ALEX ROY: Well, look, it's important to distinguish between autonomous vehicles and driver assistance and the two approaches to each of them. So on one side, you have companies like Waymo and Cruise and Motional and Zoox, with more complex hardware and software, but generally, a more cautious approach than Tesla, which is using a camera-first approach and is putting a large quantity of vehicles in the street. I would not say they have been cautious at all.
BRODIE: All right. So like there have been issues with the public and, and with safety also, both Cruise and Waymo have had some issues, and it seems to be sparking some backlash among residents in, in the communities in which the these companies are operating. Why do you think it is maybe that, that people, at least some of them are so unhappy, so upset about the these autonomous cars driving around in their communities?
ROY: Well, I've traveled all over the world and ridden in almost every autonomous vehicle that is available to the public. And it is not clear to me that there is any real opposition to this technology. There is in San Francisco, and it's misdirected towards Waymo recently, when in reality, Waymo is the best actor in the sector.
BRODIE: Right. Well, I mean, and you know, we saw, for example, you mentioned San Francisco, there's a Waymo that was essentially set on fire by a crowd of people there. iI that case, are, are, do you think it's maybe just a stand in for, for Cruise or a stand in for the industry since Cruise isn't on the streets in San Francisco anymore?
ROY: There's no question that Waymo is the victim of sentiment that should be directed at Cruise and potentially at Tesla. Although Tesla's not fielding any autonomous vehicles right now
BRODIE: How do these companies try to turn the tide? I mean, how do they try to get, to the extent that there is opposition to this concept, how do companies try to overcome that?
ROY: One of the biggest problems, and it's been true for many years is that Elon Musk, you know, owns the language of all of this. You know, when he declared vehicles that one could buy quote unquote full self-driving years ago,, this really muddied the waters. I'm not sure what the companies can do other than keep fielding vehicles very cautiously and building goodwill in communities. There's nothing will substitute for a cautious, honest and transparent approach. And by that measure, Waymo is by far the leader.
BRODIE: Well, so let me ask you about that, because Waymo earlier had put out some safety data which showed that, you know, their vehicles are for the most part, very, very safe. But it seems like there's a decent amount of skepticism because the data came from the company itself, and it would obviously not be in Waymo's interest to put out a report saying, hey, our cars aren't that safe. So is there a way for maybe an independent third-party auditor or investigator to, to look at the, at the safety data and really say yes, these are OK. No, they're not as good as maybe the company says they are.
ROY: Well, one of the biggest problems is that there are very few experts if any who aren't working at these companies. But among the people who've looked at the data are people like Professor Philip Koopman at CMU, who himself is quite a skeptic, has acknowledged that, you know, some of Waymo recent data indicates both progress and that they are even with human-driven ride-hail drivers, if not slightly better. So what's critical here is that there is progress and demonstrable improvement.
BRODIE: So given that there aren't that many people out there, experts who don't work for one of these, one of these companies, does that make it harder then to maybe win over other skeptics?
ROY: There is no substitute for availability of the vehicles and actually using them. There's also no substitute for wisdom in when and where one deploys. You don't need to be an AI expert to know that San Francisco is probably the last major city in America you'd want to launch it. But it's an important market. If you go to a place like Miami, where I lived and worked on these vehicles, people love them. There was zero opposition even when there were some, you know, minor inconveniences. People thought it was cool. San Francisco from a cultural standpoint is a tough place to deploy anything.
BRODIE: So, what do you think is the future for this industry? I mean, do you anticipate that more carmakers will, will get into this market? Do you think it's maybe Waymo's to lose?
ROY: Of the companies that are taking the cautious and, let's do it all approach, I think Waymo is the most likely to succeed and thrive. I do not think more car companies are going to invest in or develop this than the ones we've already seen. If anything, I think that most of the companies are going to go away or get acquired. And, and the only question really is which business model survives. And autonomous vehicles are inevitable, which business model makes sense. My money is on Waymo. I think eventually we'll see this in passenger cars who license the technology from third parties.
BRODIE: Really? So you think, you know, you can go buy, for example, like a Honda or a Ford or something, and there will be some kind of autonomous setting in it, as opposed to a fully autonomous car like a Waymo?
ROY: Well, to be clear what fully autonomous means, fully autonomous means a vehicle in which no one needs to sit in the driver's seat. And if they do, they can sleep. But eventually you'll be able to buy a car that has a feature, and every vehicle will have two modes: fully autonomous, I can sleep in the front or anywhere in it; and collaborative or assisted driving, which we're starting to see in cars you can find today. And for that to occur, many years have to pass, many, many years.
BRODIE: Do you think that that feature on cars will face the same again to the extent that there's been pushback on fully autonomous? Do you think those, those features will face that or because carmakers maybe have a little bit of built-in goodwill, maybe people have owned cars made by those makers, will they be seen with a slightly different light at the beginning?
ROY: The irony here is that Tesla's selling vehicles with advanced driver assistance that have had countless incidents and fatalities, and yet people still buy them. I own one. But I think the vast, 99% of the anti-AV sentiment we see is a function of people seeing failures in driver assistance. And that if Waymo could hold out long enough to deploy at scale, time and culture will bring a lot of popularity to this technology.
BRODIE: And when you talk about other carmakers putting, you know, the the autonomous features in, you know, the the, the mode into those cars, like when you talk about being in the future, how far into the future do you think that might be?
ROY: It's not a question of when, but where. Because if you buy a Cadillac today or a number of Ford models, you can get a hands-off feature, which works on divided highways in most of the United States. So that is going to expand it inevitably to cover, you know, urban streets and then eventually complex weather. So think of it as their multiple dimensions: place, weather and road complexity. And if you can nail all three, you've got a fully autonomous vehicle. If you could do one or two, you've got a feature you can use part of the time, and that's gonna determine how long we get to every vehicle has this.
