When will fully Autonomous Driving become a reality? That’s the trillion dollar question. It may come down to which company is willing to offer it first, but there are some very clear hurdles to that.
The easy answer is that Autonomous Driving is almost here. All the component technologies are in place. Various companies have have been testing semi-autonomous cars since the 90’s. Google’s prototypes have been on the road for years.
Rumors suggest Tesla could offer full autonomy now, if only the law allowed. The truth is that even if the regulatory and technological hurdles are crossed, no manufacturer wants to be the first to sell you a fully autonomous car.
Why? Because secretly they are utterly terrified of the third hurdle, which is psychological.
You don’t need to be Nostradamus to know that manufacturers believe in the Autonomotive Singularity, which will come the day the last person gets into a car and relinquishes all input other than destination. They’re investing billions in self-driving technologies all on the assumption that this day will come.
No company wants to be left behind on the road to this theoretical pot of gold. A pot filled with shared services and subscription-based revenues forever. Things about which they know nothing other than headlines in Wired.
And yet they watch jealously as Silicon Valley spits out one disruptive product after another, annihilating old paradigms. They wait as Tesla moves first on high-end EVs, they watch as Google has driverless cars touring America’s streets, and they pray Apple doesn’t enter their market with a transformative product. They intuit that car-sharing apps will intersect with their Autonomy-capable vehicles, someday. They hope and expect to provide cars for such services, and some have even invested on the periphery of a market that has just begun to stir.
What are they waiting for? Someone to lead them down into a dungeon they think is their only hope? They’re like a bunch of World of Warcraft newbies, awaiting their first raid. They know winter is coming. They know everyone has go on the quest, but no one wants to be first. So they assemble at the mouth of the cave, chopping wood and adding semi-Autonomous features as their party grows, waiting for someone to take the lead.
I recently attended a fascinating panel hosted by Volvo at the Swedish Embassy in Washington, D.C. on Autonomous cars. The bagels were fresh and both the lox and panelists were wonderful.
The latter included Brian Soublet from the California Department of Motor Vehicles, Nathaniel Beuse, the Associate Administrator for Vehicle Safety Research at the National Highway Traffic Safety Administration, Alain Kornhauser from Princeton University, Ron Medford from Google, and Erik Coelingh from Volvo. The event was called “Is it Safe?”, and by the end of the third entertaining hour, it was clear this question had far more to do with when manufacturers would enter the dungeon than whether your autonomous car would kill you.
Imagine two Dungeon Masters who’ve never met trying to agree on the rules of a new expansion pack. That was NHTSA and the California DMV. Now imagine a group of veteran players, none of whom have ever met each other or the DMs, disagreeing over what weapons to bring along for an expansion they both want to play. That was Volvo and Google.
Now add Princeton’s Kornhauser - the veteran DM there just to watch, an academic with the soul of Jay Leno - telling everyone that it sounds like it will be a good game, if only they would stop arguing and grow up.
Google, Volvo and Kornhauser were united in the certainty that the technological hurdles to AD would be crossed, if it hadn’t been already. Volvo’s Coelingh cited how Sweden was the ideal testing ground because of its four seasons, after which he turned toward Google’s Medford.
The audience knew where this was going. Medford defensively admitted Google’s current prototypes were not capable of going “everywhere, anywhere, all the time,” that they can’t yet handle real-world weather extremities.
But that will change. Kornhauser appeared to think any debate over the first hurdle was a waste of time, because it would be crossed.
The panelists basically agreed on resolving the regulatory hurdle: national guidelines for Autonomous Driving are necessary. There was some lively discussion of the push-pull relationship between regulators and getting products to market.
But if you have any doubts as to how long it will take for fully Autonomous Cars to become commercially available, here are some quotes from Brian Soublet of the California DMV, the largest agency in the state overseeing the majority of such cars in this country, namely Google’s:
“We’re not in the business of regulating safety.”
“We’re not an agency filled with automotive safety experts.”
“We know how to test a person.” (I laughed out loud at that one.)
“We don’t have the capacity to [test] an automated vehicle.”
That sounds ludicrous, but to be fair, the California DMV’s position - however awkward - is not their fault. To wit:
“Someone thought it was a good idea… to introduce regulations that cover the safety of self-driving vehicles. We didn’t ask for it. We didn’t ask to be in the business of regulating technology. We haven’t been in that business before. It was forced upon us by legislation.”
Something tells me Google had something to do with that. Google is based in California. The power of aggressive lobbying, perhaps, which highlights a coming problem. Only four states currently allow testing of Autonomous Vehicles. And that’s testing.
Unless or until a national policy is developed, manufacturers will have to lobby state-by-state for legalization, first for testing (because most states will probably want to independently verify neighboring states’ assertions, states rights and all), and then for sales, which will be...well, no one knows what that will take. Not until there is a national policy. Friction lurks in the possibility that states will have seemingly close but functionally different policies - and therefore different experiences - for consumers. No manufacturer wants to offer an AV whose functionality changes when crossing state lines.
Is this hurdle surmountable? Inevitably, given enough time and money, but it will be tedious and expensive, and it pales before the third one, the one manufacturers fear the most.
The panelists were united in the assertion that fully Autonomous Vehicles were already safer than Human Drivers (HD). Hold that thought, Luddites, because I know what you’re thinking. AVs will make mistakes. Of course they will, but the “experts” will tell you that no matter how many they make, they will make up to 90 percent fewer mistakes than humans, for a savings of some 30,000 lives a year in the United States.
I’ve never heard anyone, anywhere, ever suggest that AVs might increase accidents. That would be absurd, of course. If that was even remotely possible, no one would ever consider building them. Given that even a 10% improvement in safety would save 3,200 lives a year in this country, and that the myriad benefits of reducing traffic and pollution are self-evident, what possible reason would manufacturers have to hesitate going balls deep into Autonomous Cars? Any one of them could afford the lobbyists. They have the technology.
Herein lies the conundrum behind manufacturers’ fear of being the first to sell a fully Autonomous Vehicle: If control must be sacrificed for safety, safety is no longer a selling point.
One death is a tragedy. A million is a statistic. - Joseph Stalin
We, you, I, every living person who has ever driven a car - do not exist as part of a hive mind. An ant thinks nothing of sacrificing itself for the good of the hive. Human beings are not ants. Years of military training might compel soldiers over the parapet, but it can’t force them. They must believe. This is the logic and folly of human nature, and this is what manufacturers fear most.
If Autonomous Cars could reduce accidents by 100 percent, the fourth and final hurdle would merely be cultural. Some people like to drive. All manufacturers would have to do is wait for enthusiasts to die out (if ever!) and lobby governments to mandate autonomous vehicles. But most people will want to drive not merely for driving’s sake, but for the sake of control over outcomes, however illusory. People want to retain agency.
Even Google, the biggest proponent not only of fully Autonomous cars but of removing steering wheels, admits that accidents will occur. The 90 percent reduction figure is self-justifying, debate-ready shorthand for up to 90 percent. There will still be accidents by human drivers, but, more importantly, Autonomous Vehicles will have a failure rate. Call it 1 percent. Or 5 percent. It doesn’t matter what it is. Technology fails. Things happen.
One might argue that we get on planes and elevators all the time, and that we assume some insignificant level of risk each time, and therefore a self-driving car that reduces our risk makes sense.
But there’s a critical difference between planes and cars. The average person will never learn how to fly a plane. We know how to drive. Maybe not well in most cases, but we generally know how. We have accepted, assimilated, assume a level control and agency over skills and tasks within our reach. Flying is beyond the reach of most. Driving is not.
This agency, once assumed on an individual level, comes to define who we are, and becomes nearly impossible to relinquish without a sense of loss. We therefore accept, however irrationally, 32,000 deaths a year. Why? Stalin knew. Thirty-two thousand is a statistic. Your friend was killed? A tragedy.
But a tragedy most people think they can avert, if only they retain control. People will almost always choose to remain in control, even if they are likely to fail.
No one wants to be a victim when things “just happen” in an Autonomous Vehicle. And no manufacturer wants to have their name on the first AV that kills someone.
If and when that happens...
I heard this from several people at the Swedish Embassy, but Kornhauser was the only one who would go on the record. He’s right. Consider the lifecycle of corporate disasters. Unintended acceleration cost Audi 15 years and it wasn’t even real. It cost Toyota a hell of a lot of money, too. Consider the Ford Explorer, and the Tylenol scandal, and the exploding Pinto. It doesn’t matter whether the products were actually defective or not. Cultural memory is long.
The first inevitable failure of Autonomous Driving technology - should it happen in its infancy - won’t doom the sector, but it will delay its cultural assimilation. Manufacturers intuit this more clearly than any threat coming out of Silicon Valley. All are waiting for someone to take the hit, then, in the wake of the inevitable and undeserved PR bloodbath, grow their AD strategy in proportion to the public’s trust, however long it takes. All of this has happened before, and will happen again.
Audi knows that, and VW is just starting to find out.
I never thought I’d write that. Yes. Tesla and Volvo, in the same sentence, leading the way. Think about it. Of all the auto sector headlines since the VW catastrophe, only two matter, because these are the companies leading the way. Whether they are at the leading edge of autonomous technology is irrelevant. They are leading in the marketing of Autonomous Driving. They are the first to attempt the jump over the psychological hurdle.
Tesla may have grabbed all the headlines this week because of their new Autopilot software update, but this is merely an incremental step up the autonomy ladder, hovering just above Level 2. The reason those with similar technologies have remained silent is because all boats rise when Tesla foots virtually the entire bill for educating the public on autonomy. Their volume is insignificant, and should a fatality occur in an Autopiloted Tesla, however much the Autonomotive Singularity may have been set back, no traditional manufacturer will shed a tear.
Which leaves Volvo, the only legacy manufacturer with the courage to raise their shield, blow their horn and enter the mouth of the cave. Volvo, doubling down on the safety for which their brand is known, is the only company to announce a meaningful policy intended to build trust in the autonomous technology. “We’ll take the hit if something happens,” they said. In a world where brands are notoriously craven, it’s kind of admirable.
Volvo’s assumption of liability in the event their AD technology fails proves they are confident in technologies they’ve yet to release, which is more than you can say for companies far larger. To be honest, I expected this announcement from someone else, and not this soon. Combined with their Vision 2020 project, which is intended to eliminate all fatalities in Volvos by 2020, it suggests they have not only a better understanding of marketing autonomy than anyone besides Tesla, but a better understanding of psychology.
Kornhauser suggested that Volvo’s assumption of liability “wouldn’t cost them anything, because they know their cars will be safe.” He’s half right. It cost them a lot. It cost them countless kronor and years of R&D in safety. But it was the right move. Little, conservative Volvo, leading the way into the future. Not so conservative any more. And, in the future, not so little, I bet.
I really expected one of the car companies embodying power, passion and aggression would have been first to show such confidence in their future products. As a lifelong Porsche and BMW I owner, I hoped it would be one of them.
What a disappointment.
Roy is really proud of coining the phrase “Autonomotive Singularity.” Roy is President of Europe By Car, the founder of Team Polizei, a columnist for Jalopnik, a host on /DRIVE and author of The Driver - which depicts his 2006 NY-LA Transcontinental Driving Record, accomplished in 31 hours and 4 minutes. He also the Producer of The Great Chicken Wing Hunt & 32 Hours 7 Minutes, was Chairman of The Moth from 2002-2007, won The Ultimate Playboy on Sky One, has competed in LeMons & the Baja 1000, and holds a variety of driving records which must still remain secret.
Top photo credit Volvo