Self-Driving Cars Just Around the Corner: 1958, 1969 and 2016

IEEE Spectrum: (also here and here)

In the late 1950s, the Radio Corporation of America thought it had a lock on the self-driving car. The January 1958 issue of Electronic Age, RCA’s quarterly magazine, featured its vision of the “highway of the future”:

“You reach over to your dashboard and push the button marked ‘Electronic Drive.’ Selecting your lane, you settle back to enjoy the ride as your car adjusts itself to the prescribed speed. You may prefer to read or carry on a conversation with your passengers—or even to catch up on your office work. It makes no difference for the next several hundred miles as far as the driving is concerned.

“Fantastic? Not at all. The first long step toward this automatic highway of the future was successfully illustrated by RCA and the State of Nebraska on October 10, 1957, on a 400-foot strip of public highway on the outskirts of Lincoln.”

Two and a half years later, reporters got to experience this “highway of the future” for themselves, on a test track in Princeton, N.J. Cars drove themselves around the track, using sensors on their front bumpers to detect an electrical cable embedded in the road. The cable carried signals warning of obstructions ahead (like road work or a stalled vehicle), and the car could autonomously apply its brakes or switch lanes as necessary. A special receiver on the dashboard would interrupt the car’s own radio to announce information about upcoming exits.

Pictured above is an experimental autonomous car from General Motors, in which the steering wheel and pedals had been replaced with a small joystick and an emergency brake. Meters on the dashboard displayed the car’s speed as well as the distance to the car in front.

The system was the brainchild of renowned RCA engineer Vladimir Zworykin,who was better known for his pioneering work on television. In a 1975 oral history interview with the IEEE History Center, Zworykin explained his motivation for the autonomous highway: “This growing number of automobiles and people killed in accidents meant something should be done. My idea was that control of automobiles should be done by the road.” (Earlier inventors were similarly motivated by traffic fatalities; see, for example, IEEE Spectrum’s recent article on Charles Adler Jr., who came up with an automatic speed-control system for cars in the late 1920s.)

According to the RCA vision, it would be just a decade or two until all highway driving was autonomous, with human drivers taking over only when their exit approached. Well over half a century later, we’re of course just starting to get comfortable with autonomous vehicles on highways, and the problem of reliably transitioning between autonomous and human control still hasn’t been solved.

As for RCA’s demonstration, one contemporary account, in the Oxnard, Calif.,Press-Courier, made it sound much like the autonomous car experiences of today:

“We stayed about 80 to 100 feet behind the other car, and when it stopped we pulled up smoothly to a halt 20 feet behind it. Then the demonstrator had the first car go around to the other side of the track and stop. Then he activated our automatic works and we started zooming around the single-lane oval.

“ ‘Now,’ said the demonstrator, ‘let’s suppose we’re cruising normally down the highway around a blind curve and we don’t know a car is stopped in front of us and—oops!’

“Our automatic auto apparently didn’t know it either. The demonstrator said later he probably had forgotten to flip a switch. As we sped dangerously close, he had to flip back to manual operation and apply the brakes the old-fashioned way, with a foot. We didn’t hit.

“Nobody screamed. But the age of automation can have its moments.”

In July 1969, IEEE Spectrum published an article called The Electronic Highway, by Robert E. Fenton and Karl W. Olson, two engineers at Ohio State University who were working on ways to make vehicles operate autonomously when traveling on major highways. Nearly 50 years have passed, which is practically forever in a technological context, but what’s striking about this article is how many contextual similarities there are between the past and the present.

(For more about the history of intelligent transport, make sure to read our feature on Charles Adler, who was working on intelligent traffic control systems in the 1920s.)

The specific solutions that Fenton and Olson propose are a bit outdated, of course, but the problems that they discuss and the future that they look forward to have a lot in common with those peppering current discussions on vehicle autonomy. IEEE members can read the entire article here. We’ll take a look at some excerpts from it, and talk about what’s changed over the last half century, and what hasn’t.

As you read these excerpts, try to keep in mind that the article was published in 1969, and that the 1980s, a decade away, represented the distant future:

An examination of traffic conditions today—congested roadways, a large number of accidents and fatalities, extremely powerful automobiles—indicates the need for improvements in our highway system. Unfortunately, conditions will be much worse in the next decade, for it is predicted that the total number of vehicles registered in the United States in 1980 will be 62 percent greater in 1960, and 75 percent more vehicle miles will be traveled. If one should look further ahead to the turn of the century, he would see vast sprawling supercities, with populations characterized by adequate incomes, longer life-spans, and increased amounts of leisure time. One predictable result is greatly increased travel. The resulting traffic situation could be chaotic, unless some changes are instituted beforehand. 

It is obvious that the traffic problems cannot be solved simply by building more and larger highways, for the for costs are too high, both in dollars and in the amount of land. Many alternative solutions have been suggested: high-speed surface rail transportation; a high-speed, electrically powered, air-cushioned surface transportation system… However, in the opinion of the writers, a majority of the of the public will not be satisfied with only city-to-city transit or even neighborhood-to-neighborhood transit via some form of public transportation. One needs only to witness the common use of private automobiles where such transit already exists. The role of a personal transportation unit is certainly justified by the mobility, privacy, and freedom afforded the occupants. It seems certain that this freedom, which dictates the spatial pattern of their lives, will not be relinquished.

I don’t know about adequate incomes or increased amounts of leisure time leading to greatly increased travel, but what’s definitely true is that more and more people are commuting longer and longer distances to get to work. The authors were certainly correct that the more time we spend in our vehicles, the more important autonomy becomes. At the same time, individual car ownership and usage is starting to get replaced by services that are more decentralized, but autonomy will enable that as well. It’s funny to see that mention of a “high-speed, electrically powered, air-cushioned surface transportation system”; it sounds like they were foretelling the Hyperloop.

In this light, one satisfactory solution would be the automation of individual vehicles. This approach has been examined by a number of researchers, for in addition to the retention of the individual transportation unit, it appears that considerable improvement in highway capacity and safety as well as a considerable reduction in driver effort can be achieved. However, there is an extremely large number of possible systems for achieving this goal—the writers have counted 1296—and great care must be exercised so that an optimum or near optimum one is chosen.

The approach described in this article involves the concept of a dual-mode system, whereby the vehicle (which must be specially equipped) is manually controlled on nonautomated roads and automatically controlled on automated ones.

The system that the authors suggest, which is typical for automated driving system ideas of that era, relies heavily on infrastructure built into the highway directly. The well-defined and highly structured environments of highways are where today’s autonomous cars both perform the best and are the most valuable. The authors of the 1969 article weren’t all that worried about automating other types of roads, because they figured that it simply wouldn’t be worth the hassle and expense. In the near term, this is where many autonomy projects are finding success as well.

However, highway autonomy that’s based on the highway itself is much harder to expand, and the total infrastructure overhaul needed to implement it in the first place wouldn’t be cheap. Even in 1969 dollars, it sounds expensive:

It is expected that with the introduction and extended use of microcircuits, it will be possible to install all necessary equipment in the vehicle for several hundred dollars. The total investment in computers and highway-based sensors would probably average anywhere from $20,000 to $200,000 per lane mile (about $12,000 to $120,000 per lane kilometer), depending on the form of the chosen system and future technological advances.

One can expect two principal returns from such an investment: greatly increased lane capacity at high speeds and a reduction in the number of highway accidents. Estimates of the former range up to 800 percent and would depend, of course, on the chosen system design. The expectation of fewer accidents arises from the fact that an electronic system can provide a shorter reaction time and greater consistency than a driver can. 

It’s interesting that Fenton and Olson sound like they’re more focused on increasing capacity than on safety, which is the opposite of how most vehicle autonomy projects are presented right now. This is likely because the efficiency benefits require some critical mass of autonomous vehicles all working together. It’s much easier to achieve this with expensive automated highways and cheap automated vehicles, rather than ‘dumb’ highways that require each vehicle to have its own automation system, which is what we’ve got going now.

As for the vehicles themselves, the “near future” might provide something better than internal combustion, the authors hope:

It is probable that vehicles would be powered by the internal combustion engine; however, a number of other prime movers— the DC motor, the gas turbine, the steam engine, and the linear induction motor— may be available and practical in the near future. The eventual choice will probably be dictated by such factors as air pollution and the continuing availability of cheap fossil fuel.

One attractive possibility involves the use of electrically powered cars, which would be self-powered via batteries on non-automated roads and externally powered on automated ones. Here, power would be supplied through a pickup probe protruding from the vehicle, and control could be obtained by simply controlling the power flow.

Electrically powered vehicles that harvest electricity from highways and use batteries to get around elsewhere are currently being tested in Sweden.

Another important contemporary issue that the 1969 article mulls over: Should a human play any role at all in operating the autonomous vehicle?

An important question that is frequently raised is the advisability of allowing the driver to override the system. If he were able to do so, there would be a large measure of randomness in the system, which would be undesirable from a standpoint of both safety and system efficiency.

This remains one of the more difficult problems with consumer vehicle autonomy: unless you can promise 100-percent capability for your system, a human “driver” has to be involved. But how do you make sure that the human is alert and able to take over when necessary? Tesla reminds drivers to keep their hands on the wheel, and will begin to gradually slow the car if you persistently ignore it. Google, on the other hand, doesn’t trust humans even that far, which is why it’s working on autonomous cars without steering wheels.

The authors of the 1969 article have a slightly different perspective, but their concern about the “undesirable randomness” of human drivers foreshadows the day (it’s getting closer, we promise) when the most dangerous car on the road will be the one driven manually by a human.

There is also a need for communication links to a central station so that there will be a complete “picture” of the traffic state at all times.

Combining vehicle autonomy with systems that allow vehicles to communicate with each other (and with a central coordination system) will allow for all kinds of exciting things, like flow optimization, cooperative route planning, and even intersections without traffic lights.

There seems little question that vehicle automation is technologically feasible; however, a tremendous amount of effort in both research and development will be required before a satisfactory automatic system is in operation. This effort must involve not only vehicle-control studies, but also an intensive investigation of the present driver-vehicle complex, since the knowledge gained will be necessary for the proper specification and introduction of the control system components. Further, the need exists for intensive overall system studies so that optimum strategies can be chosen for headway spacing control, merging and lane changing, and the interfacing of automated highways with other modes of future transportation. 

The authors’ conclusion is as true now as it was in 1969. Since then, the focus has changed somewhat, with highway autonomy seen as just a step towards full autonomy—and not necessarily a step that can be taken independently. Technology has improved to the point where it’s now feasible to pack all of the sensors and computers needed for autonomy into vehicles themselves, rather than having to rely on external infrastructure. This makes the transition to autonomous vehicles more straightforward, partly because it’s something that can be motivated by the market rather than by the government.

At the same time, though, I certainly appreciate the vision that the authors had for expensive automated highways that the average driver could take advantage of with a relatively inexpensive car. Here in Washington, D.C., for example, adding autonomy infrastructure to the beltway alone would vastly improve the commuting experience for an enormous number of people on a daily basis—that is, if existing cars could be cheaply retrofitted with the technology. This isn’t the trajectory we’re on anymore, but it’s interesting to look back and think about what would have happened if we’d taken a different road.

Mark Fields, the chief executive of Ford Motor Company, said his company would sell completely self-driving cars by about 2025, after first providing them via ride-hailing service, in 2021.

Such cars would have “no steering wheel, no brake pedal,” he said. “Essentially a driver is not going to be required.”

At first these robocars will cost more than conventional cars, he admitted, but the ride-hailing application will make up for that by saving the salary of a professional driver. Later, the rising scale of production will lower the sticker price enough to justify offering the robocars for sale. Ford can make money either way.

“Now vehicle miles traveled are just as important as the number of vehicles sold,” Fields said.

As robocars proliferate and cities impose congestion fees and other measures to limit traffic, total car sales may well drop. “But you can also argue that autonomous vehicles will be running continuously and will rack up more miles—and that that will mean more replacement.”

Ford has begun framing itself as a mobility company rather than a mere car company, and it has emphasized the point recently by announcing ventures to provide cities with electric-bicycle services and shuttle services. Asked about recent drops in the company’s share prices—a sign that investors aren’t happy with a program that can only bear fruit a decade hence—Fields said his company wasn’t managed for the short run alone.

He quoted Wayne Gretzky, the famed Canadian hockey player: “You’ve got to skate to where the puck is going to go.”

News from the past and future of information http://infostory.wordpress.com/