There has been a lot of talk of late about self-driving (or autonomous) road vehicles, and about how current passenger train riders will desert trains for autonomous vehicles in 14-odd years.

Much of the talk arises because high-profile, high-tech companies have gotten into the autonomous vehicle business. And carmakers, of course, take advantage of the hype by demonstrating how far along they are in development. The public then longs for the Mercedes or the Chevy that will get them to their jobs while they read the morning paper or finish up that report that's due at the morning meeting.

Self-driving trains, however, are entirely another matter.

The long, slow march toward driverless trains was precipitated by the invention of train brakes. Operated from the engine cab, they no longer required people on several cars of a train to operate the brakes by hand upon the engineer's signal.

Thus, the size of an operating train crew on freight trains has diminished from more than 10 people to, in some cases, one. The size of crews on passenger trains — from Amtrak to regional transit — has diminished in like kind, with the only bottom limiting factor on heavy passenger trains being the amount of customer service the railroad wants to provide.

We're talking averages here, because there are prominent examples of crewless trains out there. My research indicates that no fewer than 17 rapid transit systems operate at the highest level of automation in the United States. As defined by an international body, this means that no attendant is present for driving, loading or unloading passengers. Internationally, Paris, Sao Paulo and Kuala Lumpur are heavily invested in this kind of rapid transit train, each having more than one line in operation.

Heavy rail switching has been done by remote control for years, but this is always with the driver standing on the ground with his train in full view — something like operating a model train layout. This is not truly autonomous operation.

The last-person-standing hurdle is something of a matter of confidence. After a fatal crash in 2009 under computer control, Washington Metro went back to manual operation. It returned gradually to computer operation after five years, but with a driver in the cab. To remove the driver completely must signify the railroad's ultimate confidence in the computer hardware, software and peripherals that do the driving.

Positive Train Control (PTC) has been presented to the general public as a safety measure. As envisioned by Congress and the railroads, it should reliably and completely stop several kinds of railroad accidents, all of which have the potential for heavy loss of life.

The most prominent to be prevented are missed signals resulting in collision, excessive speed resulting in derailment, collisions without missed signals (either head-on or rear-end), end-of-track derailments (such as recently in New Jersey), wrong track-wrong train, and asleep or incapacitated engineer. With the proper lineside peripherals, PTC can also prevent accidents due to landslide, bridge out or wash out, inoperative grade crossing signals and gates, and even some trespassing.

What the general public does not ordinarily hear, however, is that PTC can increase railroad capacity by tightening schedules and decreasing time between trains — and, in its full deployment, run trains without any crew at all. As I said, this needs the ultimate confidence that nothing can or will go wrong, or the willingness to take the risk because of savings in crew costs.

So why, then, do the railroads appear to be resisting PTC?

"Many of the top people (at American railroads) are perhaps not technical people. That makes them more cautious," Steven Ditmeyer, a former director of the research and development office of the FRA, recently told railway-technology.com. "Also, technology is expensive to implement. It has potential major benefits, but some are not fully comfortable with it."

That's it in a nutshell: The comfort level is still low, and the cost is high — and getting higher.

Another reason to resist is political show. Railroads are just not happy with the government telling them what they can and cannot implement. As much as any other business, government regulations that are capricious or difficult to implement go right to the railroads' bottom line. Even if the eventual rewards of implementation are significant, the railroad would rather it be seen as their idea.

Cultural inertia is yet another reason railroads seem to resist PTC. Everyone working on the project knows it will eventually result in a loss of jobs for operating personnel. Railroaders, as a group, do not like kicking members out of the fraternity.

So when will we see full implementation the full potential of PTC — autonomous trains?

Ditmeyer says 10 years. Knowing what kind of societal pressure can be exerted on a corporation that makes a single misstep — for railroads, that would be the first accident involving a driverless train — not to mention the resources that would have to be poured into claims and redevelopment of a failed system, I'd say much, much longer.

But this is going to be true of autonomous road vehicles as well. So while you're waiting for that first ride on a driverless Amtrak train, don't buy stocks in driverless Uber cars. We've still got some development and confidence-building to do.