When Will Driverless Cars Arrive? They Just Did

California's decision to give robotaxi services free rein in San Francisco will lead to a host of next steps, and quickly.

Man in Self Driving Car

Mark the date: Aug. 10, 2023, was the day when the starting pistol was fired for the rollout of autonomous vehicles. 

Everything to this point was testing. But the California Public Utilities Commission voted 3-1 last Thursday to allow two robotaxi services -- GM's Cruise and Google's Waymo -- to operate at all hours of the day in all parts of San Francisco and to charge riders.

What happens now?

A lot, and as quickly as Cruise and Waymo can manage -- but only step by step. Let's call the next phase radical incrementalism.

It has taken quite a while to get to this point in the rollout of AVs, certainly longer than General Motors expected when it bought Cruise in March 2016. I'll 'fess up to some excessive optimism myself, having written a book with Chunka Mui in 2013 called "Driverless Cars: Trillions Are Up for Grabs."  

Part of the issue is the sheer complexity of the technology. Glitches don't appear often -- but often enough to cause skepticism. In 2021, Waymo cars developed the odd habit of pulling into a particular cul-de-sac in San Francisco, then making a three-point turn and leaving. Residents complained that as many as 10 cars a minute were buzzing up and down their quiet street, waking many people up in the middle of the night. Just last week, as many as 10 Cruise driverless cars simply stopped in the middle of the street in San Francisco's North Beach area. The crowd at a concert was using so much cellphone capacity that the cars lost contact with their home bases when they neared the venue and shut down. More problematically, driverless cars can get confused in the presence of emergency vehicles -- a complaint that the San Francisco Fire Department made at the PUC hearing.

The glitches have happened often enough that resistance has developed among San Franciscans, even though the city may be the most technologically aggressive in the world. Some protesters learned that they can confuse an AV simply by putting an orange traffic cone on its hood, and they do, paralyzing cars with indecision.

The broader problem is that the technology has to fit into an existing system full of human drivers and pedestrians and adapt to all -- as in, every one -- of our odd little behaviors. That's double-parking, opening car doors into traffic, cycling the wrong way on a one-way street, darting out from behind a parked car... you name it. And driverless cars are trained to be cautious, so they're weak about left turns in traffic and no good at all on a crucial bit of human driving. You know the one I mean: You have to merge into a steady stream of traffic, so you make a quick move and hope drivers think you're just reckless enough to pull into their lane, even though you're just feinting. Then, when one of them blinks, you pull out in front of them.

Cruise and Waymo both say their cars haven’t caused any traffic fatalities. Waymo says its self-driving vehicles, in fully autonomous mode, haven't even caused a collision with another vehicle in their first million miles.

But now they have to prove it in full-on commercial operation. That means roughly 300 robotaxis for Cruise in San Francisco and 250 for Waymo. Waymo says it has a waiting list of 100,000 signed up for rides in its AVs, so the cars should get a workout.

Getting San Francisco right will be step one in the radical incrementalism of the AV rollout.

Step two will be expanding into other cities. Waymo has been gradually rolling out service in Phoenix and will continue to expand there -- the environment is far more favorable than in a city like San Francisco, which is why I count the California PUC's decision as the start of the real rollout. Waymo is also beginning service in parts of Los Angeles county and Austin, Texas. The company says it currently provides 10,000 fully driverless rides per week and will soon reach 10,000 a day. Cruise has announced plans for robotaxi service in Austin, Dallas, Houston, Phoenix, Miami and, most recently, Nashville.

Other companies will likely jump in, too. For instance, Amazon's Zoox has been testing self-driving technology in San Francisco and is developing a boxy vehicle specially designed for taxi services. (I'm not discussing Tesla because its approach to driverless technology is very different -- it mostly consists of Elon Musk claiming every year for the past decade that full self-driving is months way, only for it never to arrive. He may get there eventually, but, for the foreseeable future, "self-driving" in a Tesla requires a human driver with eyes on the road and hands on or right near the wheel.)

Step three will move self-driving beyond cities. At the moment, self-driving of the non-Tesla variety requires extremely detailed maps, so the car can triangulate and figure out exactly where it is even in rain or snow. But Cruise says it wants to get beyond its home base in San Francisco and serve all of California. If it can do that, then the handcuffs will truly be off. Cars will be able to drive even in areas that haven't been carefully mapped -- meaning AVs can go just about anywhere. 

Those are just the initial steps for robotaxi services, but car and tech companies won't stop there. They didn't spend tens of billions of dollars developing the technology, with tens of billions to go on the rollout, just to be a cheaper Uber. Step four -- a huge step -- will be an attempt to replace your car with an AV, whether you own it or just summon it from time to time. That means a whole set of other steps to watch for.

Fortunately, my friend Chunka laid out those steps for us in a series of LinkedIn posts here, here and here in 2018.

Let's look at the key issues he raises that are still in play and see how AVs are progressing.

--Mass production. Basically, the question is: Even when the AV technology is ready to go, will car companies be able to manufacture enough? I'd say this is a pretty clear yes, especially because the slower-than-expected rollout has given car companies time to prepare. GM has said the ability to turn technology into cars is a competitive advantage that it intends to exploit.

--Charging infrastructure. Again, the slow (in relative terms) rollout for AVs has given companies time to recognize the issue and prepare. We're a long way from having enough charging stations, but companies are building out the infrastructure quickly. A key is that companies are rallying around the Tesla plug as a standard -- it just doesn't work if you pull up to a charging station and find the plug doesn't fit your car.

--Fleet management and services. This will be one to watch as Cruise, Waymo and maybe others scale up. The issue with AVs isn't just whether the technology works -- as high a hurdle as that is. You also have to position the cars so they're as close as possible to those people who want to commute into a city in the morning and then back to the suburbs at night. You have to be able to recharge the cars and to clean them whenever a couple of kids covered in sand roll around in the back seat -- or adults engage in other forms of recreation. You need to ward off criminal behavior, such as having people use your cars to make drug deals. You have to have a backup when someone puts an orange cone on the hood or when the cars lose cellphone connection because of a nearby concert. You have to assume that major competitors such as Uber and Lyft will respond to your entrance into their market and be ready to react. There is a lot of potential for trouble here.

--Customer service and experience. Cruise and Waymo say customers settle into a driverless car and begin to trust it within minutes, but that first ride will still be disorienting, and many customers will always need care that they can no longer get from a driver.

--Security. Women may feel more comfortable traveling alone because they won't have to worry about a male driver, but it's possible to stop a driverless car just by standing in front of it -- you know it won't run you over -- and the cars are a huge target for cyber hackers, so plenty of security issues will surface. 

--Regulation. It's going to take regulators a long time to get their heads around all the issues with AVs -- how can it be otherwise? In the meantime, anyone dealing with driverless vehicles will have to be prepared for twists and turns.

--Public acceptance. This is what Cruise and Waymo are testing now, and we'll know soon enough. Based on my rides in AVs, starting a decade ago, I think Cruise and Waymo are right, and people will quickly become comfortable. But we'll have to wait and see. I learned long ago not to trust my personal impressions.

--Business viability. Early on, AV companies seemed to underestimate the complexity of operating robotaxis as a business. They didn't think in terms of the positioning of cars, the recharging, the cleaning, etc. They now seem to be more realistic and have even benefited from some changes in the economic environment -- the soaring cost for cars, including auto insurance, and the fact that Uber and Lyft are finally charging realistic prices, rather than heavily subsidizing fares as a way of capturing market share. Even if AV makers pass this first test of viability, the much bigger one will come when they try to move beyond today's robotaxi services and into the market for personal auto ownership in a few years.

--Stakeholder resistance. One person's cost saving is another person's revenue -- so all the benefits from AVs will face pushback. That could be from car dealers, insurers, personal injury lawyers, oil companies, truck drivers, transit unions and so on. Potential losers include some of the most influential policy shapers at federal, state and local levels.

--Congestion. When services become cheaper, people buy more of them, right? Well, if driverless cars become as cheap and simple as proponents believe they will, that could mean an awful lot more cars on the road. Various studies have found that Uber and Lyft have increased congestion in pursuit of convenience. We'll have to see what AVs do to the broad traffic picture.

If you put on your insurance hat and look at all the steps between last Thursday and a transformation of the world's transportation systems, you can rest easy for a while. It'll probably take a couple of years for robotaxis to be rolled out in metropolitan areas, even if all goes well. Then it'll take a few more years to get outside cities and on to the open road. Even if AVs become viable replacements for private cars, it takes a dozen years or more to replace all the cars on the road. So, personal auto insurance isn't going away any time soon. 

But I hope I've given you some sign posts. You can take your thinking today about how quickly AVs will roll out and either move up or postpone your estimations based on how well things go in San Francisco, how quickly the rollout to other cities occurs and so on down the line. 

Just remember: The starting pistol has been fired, and some massive corporations are now running as fast as they can.