The End of Driving: The Promise and Pitfalls of Autonomous Cars

Sometime recently—it was likely the first week of October, 2019—a Chrysler Pacifica minivan slowed on a street somewhere in Phoenix to stop and picked up a passenger who’d requested a ride. Unlike a normal car with an Uber or Lyft sticker on it, nobody was behind this one’s wheel to greet them. Just an empty seat. And as it drove away, the passenger could watch the wheel twirl on its own as it entered traffic.

In that moment, Waymo’s tentative commencement of driverless rides for its Early Rider participants quietly bisected American’s personal transportation into two species of automobile: human-driven ones and driverless robotaxis.

You have to flip back 126 years to find anything comparable. Several years ago, I spent an afternoon in a in Springfield, Massachusetts, library scanning through microfiche to find the local newspaper’s account of a one-cylinder Duryea creeping through the city’s streets on September 20, 1893. This was the first account of an automobile on an American public road (until that moment, everything was literally horse-powered). Ironically, Springfield was the nation’s buggy whip capital at the time, and not a single one of its citizens would have predicted how upside-down their world would be in less than 20 years.

However there’s no shortage of research projecting the rise of the autonomous car. A few months ago, I attended the Autonomous Vehicle Sensors Conference (my vacation!) where Dimitrios Damianos of Yole Développement (Yole) (a market research and strategy consulting outfit) presented this interesting slide that predicts how they think things will unfold:

See the brown bars (Level 0)? That’s old-fashioned, no-gizmo driving the way Instructor Doug taught you when you were 16. It’s disappearing. And being replaced by increasingly capable automated assistance (adaptive cruise control and lane centering). By the late 2020s we’ll finally see Level 4 (fully automated but geography- or conditions-constrained), and in roughly 15 years, the emergence of Level 5 (the Full robocar Monty, able to go anywhere with no steering wheel).

“Ah,” you ask, “didn’t he just say that the Waymo Pacifica—a Level 4—had just started operation?” The delayed emergence of Levels 4 and 5 in Yole’s graph pertains only to privately owned ones. For robotaxis, which Yole distinguishes from Level 4 and 5 robotic cars, it’s a whole different story.

More from MotorTrend on autonomous cars and the future of driving:

Oddly, though, there’s an unlikely voice in support of hanging on to at least some steering wheels—Waymo’s CEO, John Krafcik. As he told CNN Business: “I imagine people will still drive cars, but we’ll see an incredible array of really interesting cars. Day-to-day commuting will include self-driving cars, which we hope will make our roads safer and give people back their time.”

What does 119 degrees feel like? Like opening the oven door to get your DiGiorno pizza and the heat roils out and blast-roasts your face. Bad for the eyebrows but useful for stress-testing vehicle senor suites and cooling systems, which is why I drove out to Death Valley to spend a hot afternoon with Waymo engineers.

What was Waymo learning? If I told you, Google’s package-delivery Wing drones will swoop down like flying monkeys and Toto me away. But maybe I can get away with describing three of the problems they were studying. (Still, I’ll be walking under an open umbrella for a while to be on the safe side.)

Problem One: cool thinking. Waymo’s silicon gray matter is located under the cargo floor (mounted, I noticed, by some lovely production castings), where it generates quite a lot of heat. Actually, autonomy’s energy-intensive computing is regarded as a good match for EVs because of the intrinsic electrical power that’s available. The downside is that it sucks serious range out of their precious batteries and produces problematic heat—so much of it that they require their own liquid cooling system. In Waymo’s case, it isn’t being radiated directly to the air but instead, transferred to the Jaguar I-Pace test vehicle’s standard cooling system, and ultimately, to its existing radiator via a heat exchanger (with two pumps for redundancy as failure literally isn’t an option). Does the I-Pace’s existing cooling system have enough extra capacity to handle this? That’s what they wanted to know.

Problem Two: lidar aerodynamics. You don’t think of those two words having anything to do with each other. But the characteristic spinning lidar on the roofs of Waymo vehicles is hollow and fitted with internal blades to elegantly draw air through its center for cooling. At a certain speed, though, the air passing over the roof piles up with enough static pressure against the lidar’s base to partially block the cooling air’s path. How serious is this in extreme, desert temperatures? They wanted to know that, too.

Problem Three: brake heat. Waymo’s fender-mounted cameras and lidar units are situated above the Jaguar I-Pace’s wheelwells. While the car is stopped at a red light, rising brake heat can degrade the cameras’ image quality. And for the object recognition software to best identify its surroundings, it needs to perceive everything at the highest possible definition. How badly does brake heat blur the sensors? In Death Valley, Waymo even built a plywood enclosure around the car to block any breeze while recording the data.

Notably, the Waymo team’s next stop was another data-gathering week in stop-and-go Las Vegas traffic (where Lyft and its autonomous-tech partner, Aptiv, have already tallied 55,000 safety-driver trips). Waymo’s testing in severe conditions—it was also in Florida rain storms a few months ago and Detroit blizzards earlier this year—suggests it could be aiming well beyond the simpleton, super-mapped, easily navigated geo-zones often talked about.

I’m going to unfold that umbrella now.

Notice that vehicle sales on Yole’s graph don’t climb forever. By about 2043 it peaks and starts to decline. Why?

A reality that we all tend to repress is that cars are typically used about 4 percent of the time, then sit, doing nothing but depreciate, for the other 96. So tech-intensive, hyper-expensive, privately owned Level 4 cars make for dizzyingly bad economics; they’ll emerge only as their cost subsides. But robotaxis like Waymo’s, will be tireless, all-day, all-night trip-making busybodies (with no gig-drivers) turning the economic calculation upside down. With little downtime between paid rides, fully automated taxis—and the people or companies buying them—can quickly make up the cost individual buyers can’t.

Eventually, there will be enough of them displacing private car rides (many multiples of their slim-looking sales numbers) to cause total sales of vehicle to start declining, marking the moment of Peak Car. Where the total sales numbers will settle in the Level 4/5 era is anybody’s guess, as is the exact ownership model.

By the way, an interesting consequence of this frantic use rate is that vehicles will wear out a lot quicker. Most guesses put it at about five years, which radically changes the durability necessary for all the components that aren’t wearing out so soon—for instance, things like rust resistance or UV degradation.

Three years ago, I sweated through a hot, blue-sky Palo Alto afternoon as Ford’s then-president, Mark Fields, made a ribbon-cutting speech at its just-expanded Research and Innovation Campus:

“The next decade will be defined by automation of the automobile, and we see autonomous vehicles as having as significant an impact on society as Ford’s moving assembly line did 100 years ago.” He then declared that Ford would skip Level 3 autonomy because of its tricky driver/car handoffs and leapfrog straight to Level 4 by 2021. We all wrote that down.

Now we’re in the final weeks of 2019, three-fifths of the way to 2021. And the best that privately owned autonomy can show for itself is Tesla‘s Navigate on Autopilot and Smart Summon (both distressing Consumer Reports) and Cadillac‘s Super Cruise (a terrific system that’s restricted to a single lane of traffic).

Autonomy’s infamous “corner cases” are indeed trickier than we thought, and back in 2016, legacy auto execs were prone to aping Musk-speak to hopefully boost their tech image and stock multiple into Tesla territory. So now we’re back to the Level 2 to 3 then 4 and 5 incrementalism where it all started. But oddly enough, Mark Fields’ prediction of Level 4 by 2021 wasn’t laughably premature, after all. In fact, it was more than a year too late. It was just that the car he was talking about was a Waymo Chrysler Pacifica robotaxi, not a privately owned Ford.