Self-Driving Cars and the Question of Mapping

One of the most pressing issues facing self-driving car technology today is the lack of detailed, accurate maps of complicated city areas.

In this segment from the Industry Focus podcast, host Sean O'Reilly and Fool contributor John Rosevear explain how companies are working to solve this today, and how the potential popularity of self-driving vehicles could present a solution in itself. Also, they take a look at what auto giants General Motors (GM 4.37%) and Ford (F 0.47%) are doing today toward making self-driving cars a reality.

A full transcript follows the video.

This podcast was recorded on Aug. 25, 2016.

Sean O'Reilly: One of the problems, and the solution -- it's really tricky to know where everything is for a car, even with the best cameras, the best systems. It's all unpredictable. Where do you think these systems are with mapping cities? Because my understanding is one of the solutions that's been offered is just 3D-mapping major cities, and that would make it a little bit easier.

John Rosevear: Well, that's what everybody's doing right now. Some people are buying this, some people are doing it themselves. Obviously, Google is out there, has been mapping everything for ages. You need a highly detailed 3D map. But the thing is, over time, once you have a fleet of cars out on the road all sharing data...

O'Reilly: They're all mapping.

Rosevear: They build it. So we should talk a little bit about the definitions of self-driving. What they've said in this deal is that it would be a level four/five. This is from the Society of Auto Engineers sorting out driver assistance systems. Level four is full-blown self-driving in some circumstances. For instance, if you're in a city where it has a map. Level five is full-blown self-driving everywhere all the time, don't even need to think about it, don't even need a steering wheel in the car.

O'Reilly: And that's going to be a while.

Rosevear: Right. But here's the thing. If you have a level four car, maybe you have a map of Pittsburgh, and you drive around Pittsburgh, but then you drive to Chicago to visit grandma -- the car is recording all that. Enough people take that trip, and the system's official map that's shared among all the cars that use these systems, it scales.

O'Reilly: So usage is going to lead to scale.

Rosevear: Yeah.

O'Reilly: So, Uber is in Pittsburgh. What model was it that they were using? Do you know?

Rosevear: I don't remember offhand.

O'Reilly: It was an SUV crossover thing. It was not a car.

Rosevear: Right. They also have Ford Fusions out doing this, by the way. Their first one was a Ford Fusion.

O'Reilly: Do you know what GM and Lyft are going to use car-wise for their stuff?

Rosevear: Chevy Bolt.

O'Reilly: Oh, boy!

Rosevear: That's the whole deal, it'll all be the electric Bolts. We've actually seen self-driving Bolts on the streets in San Francisco. They're already testing them. They're not officially rolling the Bolt into full-blown production, I think, until October or November, but they've made enough of them that they have them out there testing with these self-driving --

O'Reilly: This is awesome. And of course, Ford is going to be mass-producing self-driving cars in 2021. Although, you noted in an article, or maybe it was the show notes, the 2021 line that Ford's going to come out with is only going to be for ride-sharing and ride-hailing at first.

Rosevear: Yeah. I think that's their way, in part, of finessing the level four/five thing. Mark Fields, Ford's CEO, came out and said, we don't need a steering wheel, we don't need gas pedals, we don't need anything like that. But by saying we're making it available to only ride-sharing services, like Zipcar, and ride-hailing services, like Uber and Lyft, they can probably fence it in geographically while they build out the maps. I think that's what that's about.

It's also, to some extent, hedging their bets a little bit around whether there's going to be mass-market consumer demand for this at the price they'll have to charge for it quite yet.

O'Reilly: That way, if they do pull it off, they look like heroes or something.

Rosevear: But not just that. They get several thousand of these cars out. Ford is building cars for robot Uber or whatever, some new competitor, possibly, maybe Didi in China. If they're building these cars, and they're building several thousand of them, the costs start to come down, they gain more expertise, the maps get bigger, etc., over time. Then, maybe in 2025, they can bring a real low-cost solution out after a few years of running these things with companies like Uber and Lyft. They can bring it out to everybody. Maybe by then, enough people will have tried it that there's real consumer demand for this thing.