Читать книгу Robot, Take the Wheel - Jason Torchinsky - Страница 9

For all the excitement and hype surrounding autonomous vehicles, it’s worth remembering that, for most of the history of mankind, we’ve been using vehicles that were capable of full autonomy. We call these vehicles “horses” or sometimes “donkeys” or “camels” or any number of other large, muscular mammals that we’ve coerced into taking us from place to place. All of these are, of course, fully autonomous, and have been for thousands and thousands of years before any horse ever even saw a human.

Generally, horses and other animals squander their autonomy wandering around, eating lawns, having steamy horse sex and making new horses to start the whole thing over again. Once employed by humans for the purpose of transport, animals like horses became, effectively, semiautonomous vehicles.

There’s actually an accepted system in place for describing levels of autonomy for cars, known as the SAE (that’s Society of Automotive Engineers, like the Freemasons but much worse dressers) automation levels. They break down like this:

Level 0: No automation, the human driver does all the driving.

Level 1: Driver assistance, an advanced driver assistance system (ADAS) on the vehicle can sometimes assist the human driver with either steering or braking/accelerating, but not both simultaneously.

Level 2: Partial automation, an ADAS on the vehicle can actually control both steering and braking/accelerating simultaneously under some circumstances. The human driver must continue to pay full attention (“monitor the driving environment”) at all times and perform the rest of the driving task.

Level 3: Conditional automation, an automated driving system (ADS) on the vehicle can perform all aspects of the driving task under some circumstances. In those circumstances, the human driver must be ready to take back control at any time when the ADS requests the human driver to do so. In all other circumstances, the human driver performs the driving task.

Level 4: High automation, an ADS on the vehicle can perform all driving tasks and monitor the driving environment—essentially, do all the driving—in certain circumstances. The human driver need not pay attention in those circumstances.

Level 5: Full automation, an automated driving system on the vehicle can do all the driving in all circumstances. The human occupants are just passengers and need never be involved in driving.

Based on our modern scales, I’d have to say a vehicle composed of a horse and cart is somewhere between Levels 3 and 4: the “vehicle” is in complete control, but some human intervention is required.

Of course, the manner in which a horse is autonomous is quite different from an electromechanical car. While the destination is pretty much a given for an autonomous car, thanks to GPS, the horse doesn’t necessarily know it. What a horse does know are the fundamental mechanics of driving. A horse inherently knows how to stay on a road, follow a path, avoid obstacles, stop if confronted with confusion or danger, make turns, look for potential hazards, and so on. What the horse relies on the human for are inputs regarding the desired speed of travel and guidance to maintain a proper path.

With a horse-car, you’re not “steering” the horse in the same way that you steer a car—the horse is handling those mechanics. You’re guiding the animal to your destination, and, in some cases, the horse may even know familiar routes and paths, so what the driver needs to do in a horse-car can be pretty minimal.

We forget just how much natural processing an equine brain is doing to drag a streetcar along a path—it’s essentially what we’re currently trying to make automated vehicles (AVs) do. It should remind us that getting a car safely to your desired destination requires a very specific set of skills, and there’s no reason to assume that, as humans, we’re somehow hardwired to know how to do it. In fact, the fates of the two earliest human-driven automobiles speak directly to how unprepared we were, and how difficult the basic task of piloting a moving machine really is.

An automobile is any self-propelled wheeled machine designed to transport passengers and/or cargo. What powers that car—as long as it’s mechanical in nature, somehow, doesn’t really matter. A steam car is as much an automobile as a gasoline, diesel, or electric car. I want to make that abundantly clear in case anyone reading over your shoulder decides to pedantically correct this book. If someone does, please tell them to get bent.

The first machine that we can really call an automobile—a self-propelled, mechanical, wheeled machine driven by a human—­was Nicolas-Joseph Cugnot’s 1769 steam dray.

(I know Mercedes-Benz likes to talk about how they invented the car; they cite the 1886 Benz Patent-Motorwagen as the first example. Don’t be taken in by this self-serving bit of historical revisionism.)

Cugnot’s steam dray was designed to be an artillery-hauling truck, basically, and as such was designed in a way that made handling pretty terrible. Really, you probably couldn’t design a worse vehicle to drive than Cugnot did, but, in his defense, no one had any idea what the hell “handling” was or what “driving” would be like, or anything at all like that. These problems simply didn’t exist before Nick-Joe fired up the huge, teapot-like boiler on the steam dray.

This first car, being designed to haul heavy artillery, cannonballs, and other massive iron things, was designed with all the mechanical parts (and weight) well up front, with a large, flatbed-like area at the rear. The solitary and massive front wheel was driven by steam pistons, and in front of the wheel hung the massive, heavy boiler.

The driver of this thing was expected to steer with a set of handlebars that looked like a steampunk bull’s horns, and that driver would likely need the strength of a steampunk bull to be effective. The vehicle was designed to be balanced when piled high with cannonballs or towing cannon. Laden, the balance would likely have been better, but the whole thing would have been so heavy as to be deeply ungainly. Unladen, it would have been lighter, but with all the weight on the one front wheel, steering would have been a nightmare.

Cugnot not only invented the automobile, he invented lethal understeer.

Understeer, when a car turns less sharply than desired, is what happens with nose-heavy, front-wheel drive cars because they naturally want to go in straight lines. Cugnot’s steam dray was a ridiculous caricature of this design, and as a result, the first test ended up with Cugnot driving it into a wall, which he partially demolished. The second test didn’t fare much better; the truth is that I doubt the steam dray could have been driven effectively. The design was far too unforgiving and difficult and, what’s more, nobody had any idea how to drive.

The next attempt at an automobile was built by William Murdoch in 1784 and seemed to recognize the layout issues that Cugnot’s vehicle had, and pretty much corrected them. Too bad it only existed as a subscale working model. If it had been built to human scale, it’s likely it would have been far more drivable than the Cugnot car.

In 1801, the invent-cars project was renewed with the help of a Cornish man named Richard Trevithick who built a crude but full-scale test vehicle, the Puffing Devil. In 1803, he built a much more realized vehicle, arguably the very first passenger car designed to be a passenger automobile from the start, the London Steam Carriage.

The Puffing Devil was a proof-of-concept test of locomotion; it didn’t really have any steering mechanism, or a real passenger compartment. It wasn’t “driven” in the sense we understand driving today, which is why we should focus on the London Steam Carriage, which had an actual steering mechanism and a place for passengers. It was a real car, and as such could be driven. Its steam engine was set low in the tall chassis and toward the rear, controlling the rear wheels. A lone steering wheel up front was turned via a simple tiller. It was a basic design, but it was effective. The center of gravity was pretty low for such a tall vehicle and the steering mechanism worked, even if it required two people—one to stoke and manage the engine at the rear, and one to steer up front.

This division of labor necessitated communication between the two parties—as if you, while driving, had to yell at your feet to get off the gas pedal and get on the brake. Even with the task of driving divided between two people—who didn’t know how weird that would one day seem because nobody had ever done this before—the act of driving proved difficult.

To Trevithick and his team’s credit, they did manage to drive it a bit on the first try, about 10 miles, at speeds between 4 and 9 mph, but the next night they managed to wreck it.

There is a pretty good recounting of the wreck in the Life of Richard Trevithick: With an Account of His Inventions, Volume 1.1

They kept going on for four or five miles, and sometimes at the rate of eight or nine miles an hour. I was steering, and Captain Trevithick and some one else were attending to the engine. . . . She was going along five or six miles an hour, and Captain Dick called out, “Put the helm down, John!” and before I could tell what was up, Captain Dick’s foot was upon the steering-wheel handle, and we were tearing down six or seven yards of railing from a garden wall. A person put his head from a window, and called out, “What the devil are you doing there! What the devil is that thing!”

What we see here is that people were starting to learn just how much attention and processing is involved in driving, something horses have understood for centuries. A horse pulling a carriage the same size as this 1803 car would not have made this mistake. From what the accident reports state, it looks like the driver misjudged the distance to the garden fences and sideswiped them. It’s a pretty rookie mistake, but, to be fair, the driver of that steam carriage had more driving experience than anyone else on earth.

I mean, if you really think about what was being asked of these early, early drivers, the demands were decidedly nontrivial. For the first time, a vehicle moving at speeds significantly faster than a walking pace had to be controlled through city streets. This means that people had to make many new and unexpected decisions at a pace greater than they’d been used to.

The vehicle itself was tall and ungainly, with extremely skinny, metal-clad wheels that likely had very poor grip. Wheels like these over cobblestone or macadamized streets wouldn’t be easy for a modern driver with decades of experience, let alone people new to the fundamental concept of motorized motion. Everything must have felt unfamiliar and strange; the single-wheel steering couldn’t have been that confidence inspiring, and understanding how to follow the track of a road is the sort of thing that’s only really learned by visceral, physical experience. It comes quickly, but it’s not necessarily instant, and in a vehicle as ungainly as the London Steam Carriage, there’s a lot about how the car feels and behaves on the road that has to be learned.

All of this is to say that I’m not the least bit surprised that the first automobile drive of any length ended up in a wreck.

Keep in mind that these early cars even predate trains, which diverged from the automobile evolutionary line the year after the London Steam Carriage, with Trevithick’s rail locomotive of 1804. The fact that Trevithick, who was part of the second automobile wreck in human history, decided to eliminate most of the driving skill required by running his automobile on rails, is telling. Driving, even though it has become second nature to most of us, isn’t easy.

Railroads are a form of mechanical driving semi-automation; the rails take over the steering, navigation, and lane-keeping duties of a vehicle, a significant portion of the driving task. We went from the semiautonomy of animal power to a brief flirtation with entirely manual driving, then quickly retreated to a new, mechanized form of partial autonomous travel.

Sure, there were plenty of other reasons why railroads became the first widely used system of mechanized travel—poor road networks, economies of scale, centralized ownership, and so on—but the fact that no one knew how to drive is an under­appreciated factor.

I know it feels like we’re on the fringes of a revolution in driving, where we’re finally free to relinquish control over to a competent, well-trained machine, but the truth is that we’re really just going back to where we’ve been throughout most of history, just in a much more technologically refined way. It is full human control of a body- and ability-enhancing prosthetic—a car—that’s the really fascinating development, and it’s possible that this past century or so of widespread driving may be the anomaly.

This idea that a human-driven car is essentially a body- and ability-enhancing prosthetic is a concept that’s especially important to reflect on, now that we’re on the verge of transitioning to a new paradigm of automotive transport. The core of this idea is illustrative of what makes human driving so special, and one aspect of what we may stand to lose in an all- (or nearly all-) autonomous era.

Just think about how driving works right now: you get into a vehicle, and using physical motions of your body, you cause it to move, steer, stop, everything. Good drivers know how the car is balanced and gripping and moving on a gut level. They don’t assess these things by looking at the instruments and doing bursts of math in their heads, they feel it in the same way they feel their body’s motion and balance. The same goes for how people who know their cars really well can understand how their cars are performing and operating by feel as well. As you and a car grow used to each other, you begin to learn how it sounds and smells and performs and behaves, and when those behaviors or smells or sounds change, you immediately pick up on that and become aware that something may be amiss.

We can all tell if our car is idling too high, for example, or if there’s a change in brake pedal pressure. These are subtle things, but to someone familiar with their own car, they’re obvious and can be quite alarming. My very own 1973 Volkswagen Beetle that I drive as of this moment needs a new coil, I think, because I can feel the subtle pulses that mean that at higher speeds/engine revs it’s missing in at least one cylinder. I haven’t tested all the components yet to confirm this, but I can feel very clearly that something is going on.

Because our physical actions are what control a car and because our bodies directly interpret information about how the car is performing, both on the road and internally, I don’t think it’s too far a leap to say that automobiles are, fundamentally, prosthetic devices.

Think about how you feel when you get behind the wheel and pedals of a powerful car; you feel powerful yourself, because all those 700 or so insane horsepowers are directly controlled by your very own body; sure, you can get other people and many bags of groceries in there, but the car is really like a mobility suit for you, and the feeling of that can be intoxicating. Riding as a passenger in a powerful car does not give you that same feeling. In fact, many people who love to drive and drive aggressively are the most uncomfortable being driven fast and aggressively because they’re no longer in control, and it feels wrong, somehow.

Over the decades, humans have adapted remarkably well to using these motorized prosthetics to move around; we’ve proven that we’re capable of making complex decisions at speeds of well over a mile-a-minute, something that was by no means certain in the earliest days of motoring.

That empowering feeling of driving a car, that satisfaction and excitement, that rush of adrenaline or that unique relaxing feeling from a leisurely drive, these are just not the same when one is a passenger. No one cares about the handling or performance of a city bus beyond the basics of will it get me to work on time and unmangled because as passengers, we’re not in a position to enjoy such traits.

Once autonomous cars start to become common, we will be passengers, and our nearly two-century-long experiment in mechanical body enhancement as personal transportation could come to an end. As passengers in autonomous vehicles, we won’t have the opportunity any longer to experience the significant joys of being in control of a machine capable of remarkable speed, of handling that satisfies something deep in your gut, or of the ability to traverse terrain far too difficult to walk.

Will future generations look at us and our tedious daily mile-a-minute commutes with disbelief and awe? Driving could become one of those lost arts that nearly everyone once possessed, regarded by generations that grew up with self-driving cars the way we regard people who make their own soap or who know how to butcher a rabbit. There are still people in modern societies who can do those things, but not many, and for those who can, it’s a pretty assured way being considered a badass by your friends.

What are we giving up by relinquishing control of our vehicles back to something other than ourselves?

More than we think, and we’ll talk about that much more, don’t worry.

1 Trevithick, Francis, Life of Richard Trevithick: With an Account of His Inventions, Volume 1 (Cornwall: E. & F.N. Spon, 1872), p. 143,https://archive.org/details/liferichardtrev00trevgoog/page/n172.