Self-driving cars, it seems likely, are transportation’s future. When they finally arrive, autonomous vehicles and the new technologies involved seem poised to deliver profound societal and economic changes. It will alter how, or even whether, we buy personal automobiles. Yet, it’s a topic rarely covered in much depth outside of the nerdiest automotive publications; for most of us, it’s something you read about occasionally, most likely after there has been an accident with one of the self-driving prototypes dotting the country’s roads.
But it’s something that deserves to be discussed more. The technical aspects and implications of this forthcoming transition are complex. The timeline for full autonomy — which stretches from “next year” to “never,” depending on who you ask — can be confusing.
So for some clarity, we decided to talk to an engineer who knows something about it: Ford AV LLC’s chief operations officer John Rich. Here’s what he had to say about autonomous driving, which he termed “the most ambitious technical endeavor since the dawn of the automobile.”
Autonomous driving will be safer
Safety is the most critical concern with autonomous vehicles. Safety is the reason for autonomous driving to come into existence in the first place, in order to reduce the number of accidents — so it’s no surprise that it’s the most critical concern when it comes to developing self-driving cars. It stands to be the major hurdle between autonomous driving and public acceptance.
But Rich says the industry is still defining what “safe” means.
“We are steadily unifying into industry working groups that seek to define the standards around these things and start to look towards how you define good enough,” Rich said. “I don’t know the answer to that yet. I do know it has to be much, much better than a human being.”
But he believes the industry will get there.
“Smart people are going to be wrestling with that question and trying to define this in a way that brings confidence to consumers, regulators, and ourselves,” Rich said. “We can do this in a manner that is really, really helpful for society, which deals with an unacceptable number of traffic deaths every year. We know we can do better than that human-driven system. We’re in the process of defining the hurdles and the qualifications and the assurance to the public that this is a good and safe thing to do.
You probably will never own a fully-autonomous car
Autonomous driving will hit the commercial realm first — and most prominently. The technology is expensive; it’s easier to amortize those costs across fleets of vehicles. You’ll likely see it used for tasks like delivery and freight-hauling initially; well-defined, low-variance routes are simpler to program into autonomous vehicles.
Rich does not see autonomous vehicles as being privately owned, at least, not anytime soon.
“What most of the well-funded, advanced efforts are proposing is not a private endeavor by any stretch of the imagination,” Rich said. “It is a situation where the fleets will be owned and operated. The technology is so sophisticated, and the needs of that fleet are so great that it is not something that will be in the realm of private ownership.”
The basic layout of a car may not change too much.
Autonomous driving will allow vehicle designers to rethink car interiors, conceivably reorienting them toward productivity and comfort versus the act of driving. But Rich advises not to let your imagination run too wild. Safety concerns will keep cars from changing too radically.
“I don’t want you to completely reimagine the vehicle as a gondola with periphery seating,” Rich said. “There are things that have evolved to this state for a lot of very good reasons around safety.”
Ford is skipping straight to Level 4 — full autonomy — instead of the intermediate steps.
Ford is focusing its investment on vehicles reaching Level 4 autonomy, which the NHTSA describes as “capable of performing all driving functions under certain conditions.”
Rich believes Level 4 is where self-driving technology becomes transformative.
What Are the Levels of Autonomy?
The Society of Automotive Engineers defines six levels of vehicle autonomy roughly as such:
Level 0: No features that can actively steer the car
Level 1: Simple features that help steer the car (such as lane-centering assist) or speed up/slow down (active cruise control
Level 2: Features that combine active steering assistance and active speed control
Level 3: Features that can drive the vehicle under limited circumstances, but require the driver to constantly ready to take over
Level 4: Features that can drive the vehicle under limited circumstances, but will never require you to take over under those circumstances
Level 5: Features that can drive the vehicle at all times under all conditions
“There’s a massive gulf between [Levels] 3 and 4. It’s the economically transformative chasm to cross,” Rich said. “Instead of being a driver-enhancing feature, it becomes a business game-changer.”
The scale and complexity involved mean that reaching Level 4 is not a natural, linear progression from Level 3.
“Ourselves and the Waymos of the world and a few others really made a decision that we believe the progression was not a natural progression beyond Level 3, and that the suite of technologies and approaches were so fundamentally different, that investing in building through the levels was flawed logic,” Rich said. It’s “no longer about individual vehicles. It’s about large fleets of vehicles and intense investments in specific geofences.”
Level 5 — the Holy Grail of self-driving cars — is a long way off
Apple co-founder Steve Wozniak, long a proponent of autonomous vehicles, has said he doubts he will see Level 5 autonomous vehicles during his lifetime. Rich agreed with Wozniak’s comments — at least, as they relate to Wozniak’s lifetime.
“I think Wozniak was right,” Rich said. “There were a lot of snide comments after he made the comment that he just turned 69, and how much longer of a lifetime does he have given he’s not necessarily thin — fat guys like me took offense to that — but I think he was absolutely correct in his statement that the true Level 5 vision of autonomous driving is a long, long way off.”
But Level 4 should happen sooner.
Rich believes Level 4 will arrive far quicker than Level 5.
“We’re going to see this in a modest amount of time. It’s not tomorrow. But you’re seeing elements of it with Waymo’s efforts to pull a driver on a very controlled, limited route. You will see expansions of these capabilities over a modest time frame that allows this to come into fruition,” Rich said. “What we don’t think we’re that far away from is a geofenced endeavor.”
Geofenced areas will be big.
Geofences are the specific areas where Level 4 vehicles have been cleared to operate autonomously. Rich believes these areas will be large enough to make an economic impact.
“Geofences are not postage-stamp-sized plots where tourists can move around the city center,” Rich said. “Geofences are massive metropolitan areas where the majority of GDP for that region takes place. By geofences being significant in size, it enables scale and economically transformative things to happen.”
Sensor technology is a major technical hurdle.
Camera and sensor technology remains a significant obstacle toward autonomous driving. Any system needs to be able to spot dangerous situations at a great distance and at high enough speed to react to them faster than a human would. And that technology needs not just to exist, but be cost-effective enough to put in vehicles.
“The sensing is, of course, difficult and extremely important,” Rich said. “The thing that [Elon] Musk chides is LIDAR. And LIDAR is an immensely important sensor. But there’s a suite of sensors that have to work together to solve a variety of different sets of challenges. None of those are easy.”
But technical challenges are only part of the equation.
Rich is keen to point out that solving the technical hurdles is only one component of a more complex problem.
What Is LIDAR?
LIDAR, which is an acronym for “LIght Detection And Ranging, is a sensor system that uses invisible laser pulses to map an area in great detail at high speed. Read More
“It is the most ambitious technical achievement that the industry has attempted. But it also has enormous societal adoption concerns that existing systems have to be morphed to,” Rich said. “If you were clean-sheeting this, it would probably be a lot easier. You would have more infrastructure, you’d design the streets to be more compliant, and you’d probably establish rules from the start. Unfortunately, we have regulations, we have roads, we have behaviors, and we have a lot of other cars on the road.”
Autonomous driving will work in all types of weather…eventually.
Adverse weather can befuddle the sensors needed for autonomous driving technology. (That’s true even for the driver assistance features found on current passenger vehicles.) There’s a reason a lot of autonomous vehicle testing occurs in places like Arizona. “The challenge of dry weather is great enough right now that we tend to like situations with happy weather,” Rich said.
Ford, though, is beginning to branch out to get more data. “We are purposely picking cities like Miami and D.C. that do get a variety of weather conditions so that we start to get that type of information,” Rich said. “We will learn over time how to deal with varying rain conditions, and then eventually we will learn how to deal with the fluffy stuff.”
Rich believes that autonomous technology will, eventually, be better at handling weather than humans.
“Autonomous vehicles, in the end, will be much better drivers in snowstorms than humans are — that doesn’t take much if you watch what happens here [Detroit] when the first snow comes — but in the near term, we’re taking data on it certainly.”
Note: Purchasing products through our links may earn us a portion of the sale, which supports our editorial team’s mission. Learn more here.