General Motors' Driverless Cars Announcement

GM executives Rick Wagoner and Larry Burns recently announced that GM is now heavily engaged in the development of an automatically piloted vehicle which would consign the control functions to an intelligent network. Signaling would take place wirelessly. A prototype is on display currently at the annual Consumer Electronics Show in Las Vegas

This announcement was widely reported in the business press, with many veteran automotive journalists weighing in. Not surprisingly, a lot of the reactions were skeptical, as well they should be. GM, as you'll remember, is always about to introduce fuel cell powered vehicles. It's the visionary company whose vision lies just over the horizon.

What struck me particularly about the commentators and their remarks was how uniformly ignorant they were concerning the prior history of this concept, which in truth goes back to the very dawn of electronic computing in the early fifties. Long before anyone contemplated computer graphics or computer generated spreadsheets, deep thinkers were pondering the automated control of traffic streams.

Perhaps this is because the design of missile guidance systems, a somewhat related issue, was one of the first really complex problems to engage the first generation of military programmers, who, it must be said, commanded most of the nation's computing power initially. In any case, the automated control of terrestrial vehicles was a well developed subdiscipline of transportation engineering by the late nineteen sixties.

The modeling of an urban traffic system consisting of thousands of vehicles on miles of interconnected roadway turns out to be a nontrivial mathematical problem. Indeed, modeling a small rail system with a couple of hundred automated vehicles with varying destinations is a nontrivial problem as well. No one has ever built a large automated traffic system or even a large physical model of such a system using miniature vehicles. Computer models are the sole frame reference, and so no one knows of a certainty that such a system could really work.

Maybe GM is initiating such physical modeling—obviously they have the resources to do so—but they've made no announcements to that effect. So the real extent of their commitment is unknown.

Frankly, I find the announcement puzzling in the extreme. In the case of fuel cell powered vehicles, one could readily discern a hidden agenda. Look like you're developing a zero emissions vehicle and keep producing the same old gas guzzlers. But why say you're developing automated cars? What end is being served?

I mentioned prior efforts in this regard. Most of the research and development in this area occurred in the nineteen seventies, and my own father, a civil engineering executive in the City of Los Angeles, was an expert in this field. I must say that I was not particularly interested in his researches at the time, but I explored the field thoroughly many years later.

So here I'll provide a very brief summary of work to date in automated transport and attempt to put GM's efforts into the prior context.

Transportation Moderne

Automated transport as a field has seen a multiplicity of approaches on the theoretical level. Most fit within the overall category of PRT (personal rapid transit) PRT was the brain child of one Donn Fichter, a transportation expert from Texas who wrote his first paper on the subject in 1954. Fichter envisioned a system of ultra light rail public transportation where a vehicle could be summoned through an electronic network and would quickly appear at a designated station. The rider would indicate his destination, board the vehicle, and reach that destination directly with no stops—an automated taxi system, if you will.

PRT cars were envisioned as a form of individual transport, making them quite distinct from all other rail systems. The cars would have the capacity to hold at most a few persons, or about the same capacity as conventional sedan. Those individuals would presumably be members of a single party.

A little reflection will lead one to the conclusion that what PRTs most resemble are amusement park rides or aerial cable tramways in a resort area. The approach is really quite distinct from that in any established mode of urban transport.

Given the state of computing in 1954, Fichter's notion would have almost certainly been infeasible at the time. Maybe it could have been accomplished with some kind of analog electronic network using servo systems, but I'm skeptical. But a dozen years later when mainframe computers were commonplace and relatively inexpensive, one had, at least in theory, the underlying basis for serious development efforts.

Moreover, during the late nineteen sixties, it was becoming evident in the U.S. that the automobile was an imperfect solution to the problem of commuting to work from bedroom communities to the urban cores where many businesses were located. This was particularly evident in newer cities ringed with suburbs like Los Angeles. More freeways lead to more suburbs and the congestion grew worse with every year. But then neither were such cities amenable to traditional light rail commuting solutions. Too many lines would be required to serve all of the low density suburbs, and the total cost of the system would be prohibitive.

My father, being deputy city engineer of Los Angeles at the time, and having worked for years trying to integrate the city's street system with the emerging state freeway network, was acutely aware of the nature of the problem, and used to cite the work of the historian of urbanization, Lewis Mumford, who in the nineteen twenties correctly predicted that express ways would ultimately exacerbate urban transportation problems. My father thought that the new technology of PRTs might provide an answer.

He wasn't alone. Both the Johnson and Nixon administrations allocated quite a bit of money to PRT research, and Raytheon actually got a grant to develop a system. A number of pilot networks were launched in various parts of the world, including, most famously a tramway servicing the state university in Morgantown, Virginia. The import of these experiments is still steeped in controversy—the whole topic of PRT is a polarizing one in transportation circles—but the outcomes of individual experiments were generally bad. The Raytheon project stalled amidst massive cost overrides and the pilots, such as they were, consisted of rudimentary people mover systems rather than full blown PRTs. And many of those were problematic as well, leading skeptics to suppose that full scale urban systems could never work.

At this point one might ask, what was the attitude of the established players in the transportation industry, the manufacturers of automobiles on the one hand and of light rail systems on the other?

This is where it gets really interesting and where the GM announcement begins to make a certain sense.

Manufacturers of light rail equipment have always opposed PRT. They see it—correctly, in my opinion—as a threat, while at the same time most view such systems as unsaleable and infeasible. Very few light rail manufacturers have made any attempts to develop PRT systems in spite of the fact that the latter are in essence light rail systems.

Now one would think that the auto makers would be even more implacably opposed to PRT, and perhaps they are on some level, but their response in the sixties was one of pre-emption rather than forthright opposition. Rather than decry the whole concept they came up with something rather similar known as the dual mode transport system, which is really what GM is talking about today.

Dual Mode

In a dual mode transport system the individual vehicles in the network are privately owned and can be operated as ordinary motorcars. But they can also be remote controlled by network intelligence, relieving the passenger from the obligation of piloting them.

Whereas most straight PRT systems bear a family resemblance to one another, dual mode systems take many forms. In some cases the vehicle will run exclusively on ordinary automotive thoroughfares, and this appears to be GM's model, while in other cases the vehicle will have the additional capability of running on some sort of guideway, and, in many cases, of drawing power from that guide way—in other words, behaving like a vehicle in a light rail system. Yet another approach is the pallet or ferry system where the individual automobile is automatically loaded onto a sort of flat car traveling over a guide way, and automatically disembarks at its destination. This is perhaps not a dual mode system in the strictest sense.

Dual mode systems have also been proposed which are not highly automated. A number of vehicles, for instance, have been designed that can utilize railroad tracks as well as city streets. Most such vehicles would be utilized in either freight transport or in moving masses of commuters.

Any dual mode system would pose many problems. Selling a sufficient number of such vehicles to justify the construction of a guide way would be a challenge since the benefits would not immediately be available to the first purchasers. Determining if individual vehicles were in good working order and barring the entry of malfunctioning units onto the network would present further difficulties. And the fact that the system is open rather than closed would increase its complexity tremendously. Finally, developing interoperability standards that would enable such a system to function with diverse brands of vehicles would be a monumental undertaking.

Moreover, a dual mode vehicle would almost have to be a hybrid-electric system or a pure electrical vehicle. A system designed for the use of internal combustion engines on guide way would provide fewer benefits because the energy efficiency of the system would be on a par with ordinary automotive transport. One really wants a system that is all electric because its energy efficiency will be incomparably greater.

Which brings us back to the question of why GM is promoting a system which apparently will be based upon ordinary internal combustion vehicles.

Interestingly, Ford was the principal proponent of such systems in the past. Their primary appeal, so far as the auto manufacturers were concerned, was that they would permit the company to develop something perceptibly new that could yet incorporate almost every component utilized in an ordinary automobile. The novelty would be in the control system which would be a bolt on addition, not the radical redesign of some pre-existing part or system.

So what would be the benefits of such an approach for the driver, other than hands free navigation and operation, which many of us might not want in the first place?

In theory, a transportation system with centralized control and monitoring would be less subject to bottlenecks and congestion than one where each vehicle was left to its own devices. The system controller could select the most expeditious routing for each vehicle and could drastically reduce the number of collisions and stoppages, provided, that is, that all or most of the vehicles were under centralized control. If only a handful of vehicles were automated and the rest were individually directed, the full benefits of the scheme would be unrealized.

The possibility of such a system being hacked is of course terrifying, and the fact that the proposed GM network utilizes wireless connections and transponders in every car makes it inherently vulnerable.

The Larger Context

How serious is GM? They're predicting systems ten years off which could be possible. Ten years is enough time to plan and build such systems if all goes smoothly.

But in the meantime the first true PRT networks are likely to come on line. One is currently in construction in Dubai and another in China, and recently a surge of startup technology companies has appeared in this field. The first real commercial system is likely to appear somewhere in the world in the next decade. That first system will probably determine if any others are built. If it works satisfactorily and proves reasonably cost effective, PRT could be the transportation system of the 21st century. If it fails, investment is likely to depart from the sector and the concept may not be revisited for decades.