Название | Robot, Take the Wheel |
---|---|
Автор произведения | Jason Torchinsky |
Жанр | Техническая литература |
Серия | |
Издательство | Техническая литература |
Год выпуска | 0 |
isbn | 9781948062275 |
The system was said to have performed well in simulations but Skinner complained that no one would take him seriously.10 There’s a lot of reluctance for people to entrust weapons of war to pigeons, a prejudice that still holds even to this day.
1940: Automatic Transmission
While not strictly a self-driving innovation, the development of the automatic transmission should be considered a step on the way to autonomy for automobiles. It was even referred to as “self-drive” in many early contemporary descriptions, because to people used to the near-constant shifting of gears, it felt like the car itself was doing a lot more work.
And the truth is it was. Determining the proper gear ratio for the needs of the car wasn’t trivial, and the automatic transmission freed the driver to focus on the more fundamental tasks of driving, speed modulation, and directional control.
First developed by General Motors (GM) (specifically its Cadillac and Oldsmobile divisions), the first commercially available automatic transmission was GM’s Hydra-Matic. Developed in 1939, the Hydra-Matic was the result of a number of earlier research efforts to make a self-shifting transmission, and used a combination of a planetary gearbox—a transmission type that uses a central sun gear and orbiting planet gears (you see why it’s called that, right?)—and a novel fluid coupling to achieve its goal. While heavy and complex, it was a success, and, for the American market in particular, regarding automatic transmissions, the industry never looked back.
1945: Cruise Control
Incredibly, automotive cruise control was invented by someone who didn’t even drive. In fact, it was invented by a blind man, Ralph Teetor, who came up with the idea after being annoyed at how bad his lawyer drove.11 Teetor was riding with his lawyer and having a conversation with him, and felt the car slowing whenever the lawyer was speaking, and accelerating when he was listening.
Nauseous but determined, Teetor wanted to find a way to keep a car’s throttle constant and free from the human driver’s fluctuation, and so developed cruise control, the first real semiautonomous assist device for cars.
Engine speed governors designed to keep an engine running at a constant speed have existed for over a century—you know those two spinning balls you sometimes see on old steam engines? That was a centrifugal governor, and also the origin of the phrase “balls to the wall,” since at maximum speed, those balls would be flung out nearly horizontally, or, you know, to the wall.
Balls or not, just keeping the engine at a constant speed really isn’t enough for a car cruise control system. Teetor computed the car’s actual speed based on how fast the driveshaft was rotating, and then used a bidirectional electric motor connected to the carburetor’s throttle to adjust the position of the throttle to keep the desired speed constant.12
Chrysler was the first to market the system, which was variously known as Speedostat or Auto-Pilot (foreshadowing Tesla’s name for their semiautonomous driving system), and later Chrysler came up with the name “Cruise Control,” which eventually stuck.
The cruise control innovation is important because this was the first taste most drivers had of any sort of actual driving automation. Sure, automatic transmissions—and before that, automatic spark advance and oiling and so on—took over many of the functions of the operation of a car that used to be manual, but those, including gear shifting, were less about the actual piloting of the vehicle itself and more about the technical requirements needed to get the car to drive at all.
Cruise control, though, was clearly different, in that it took one of the primary tasks of driving—speed control—away from the human driver and placed it under control of the machine. Sure, this was the simplest form of control, with no ability to independently sense its environment, but it was a start. Modern dynamic-cruise systems use radar to keep a set distance away from the car in front, and can brake automatically if the system determines it’s approaching an object in front of it too rapidly.
All of this is thanks to one blind man and his lawyer—who couldn’t drive and talk at the same time.
1956: GM Firebird II Concept Car
The first time a major carmaker presented the idea of real autonomy to the public was in 1956, when GM showed their concept car, the Firebird II. As the name implies, this wasn’t the first Firebird concept car, but it was the first of the Firebird concept cars to suggest a then still-fictitious world of automated driving.
The Firebird concept cars were designed by Harley Earl and were heavily influenced by jet fighter designs. The Firebirds were showcases for cutting-edge technology, and as such employed exotic gas-turbine drivetrains, and cathode-ray tubes (CRTs—you know, the old big tube kind) in its television- and camera-based rearview systems.
More important for our purposes, though, is the automatic driving system that GM imagined for the car. This wasn’t just a concept for the car itself, but rather an entire network for automated vehicles, including specially prepared roadways with integrated “conductor strips” and control towers in the “Autoway Safety Zone,” the name given to the automated highway system in the film.
GM describes the system in a brochure published for the Firebird II’s introduction at GM’s Motorama show at the 1956 World’s Fair:
This amazing concept places control of the motorcar in the hands of an “electronic brain”—actually releasing the driver from the wheel. . . . These include a Dashboard View screen which has two panels. The left panel is for “internal communication” between car and driver (information he would normally receive from visible instruments as to fuel supply, engine operation and temperature). It also reveals a radar pattern when he guides the car onto the electronic control-strip for automatic steering. . . .
Extending from the two engine air scoops on each side of the nose of the car are probes or antennas which pick up wave impulses from the conductor strip in the center of the control lane.13
Of course, none of these things actually worked, but it is interesting to see how the very sticky problems of computer vision could be avoided if fully autonomous operation is limited to areas where an infrastructure has been built to guide the cars.
No mention is made of obstacle avoidance or anything like that; presumably, it is the job of the singing gentlemen in the control towers to make sure everything is running smoothly and to warn drivers to stop if they’re approaching a broken-down vehicle or a coyote on the road or something else they don’t want to barrel through. I’m not confident that would have worked out so hot.
Really, this sort of system is more like a hybrid of a tram or trolley car type of vehicle combined with a conventional car. On roads with the proper control towers and guidance strips, the car cedes control to the roadway network, much like a tram or train. On roads without the necessary hardware, you’re just driving a normal car, even if it is powered by a kerosene-fueled turbine and has a clear bubble dome.
Because of the very significant infrastructure investments required in such a system, nearly all modern self-driving technology and research is designed to work without outside, physical infrastructure help, which is, of course, a much more difficult task.
One more thing about the Firebird concept, and, specifically, the film GM made to promote it. The film shows a family on a road trip in their Firebird, enjoying all of the considerable comforts of 1950s-envisioned 1976 future life. At one point, based on the recommendation from the guy in the control tower who assures them that the “hostess is a dream,” they contact a nearby hotel on their dashboard video screen, where they see a lovely woman who sings the praises of the hotel.