First they autonomize the cars....

and I didn't speak up because I love trains.

Here's the thing, though. All the buzz is about self-driving cars but you know that the trucking industry will go there first just to get rid of the Teamsters. How far behind will self-driving locos be? Trains lose 90% of the variables programming cars and trucks will require.

A long time after we are all dead...

They could have self driving trains tomorrow if they wanted. But it seems costs and politics will keep that from happening for hopefully many years.

They're already doing it in Australia:

http://www.smh.com.au/business/mining-a ... 3z28v.html" onclick="window.open(this.href);return false;

Just wait till this guy learns that they've had driverless metro trains operating all over Europe for years now.

There are a whole lot of things that make operating U.S.-sized freight trains by computers more difficult than passenger trains. There are, however, just three technologies necessary (things other than cost, regulatory and political forces) to remove the engineer from the freight equation. I won't say on here what they are, but one has existed for over 50 years (relatively minimal cost), the second has existed for over 15 years (slightly larger cost) and the third is having it's foundation implemented right now (huge initial cost but second-level development costs much lower). Given a few more major human-factor operational mishaps, particularly if there happens a severe rash of them in a short time frame, I give it twenty years at most.....

The New York City subway system's 42nd Street Shuttle (Times Square to Grand Central Terminal, no intermediate stops) is effectively an isolated line. (It has a track connection to the rest of the system, for use in maintenance-related switches of equipment, but most of the time this connection is covered by a movable barrier with a pedestrian bridge on it.) In the 1950s it was proposed to convert it to driverless operation: the equipment was set up and tested. Reverted to human drivers after (at most-- I don't remember the details) a few days. I believe after a fire that was thought to be the result of sabotage. FWIW.

On a factory tour I did this past fall for my work they presented this to us. This is was accomplished back in 2013.

https://www.youtube.com/watch?v=CKqJccK_EkM

Let that sink in for a bit.

Lastly I'll say maybe self driving trains won't be a thing for a little bit, but I can certainly see a situation where one engineer will be driving multiple trains from a central location kind of like a drone. This sort of thing would be pioneered of course by the big class 1's at first who have the financial power and motivation to bring it into operation.

One of the proposed features of self-driving cars is virtual highway trains. The idea is that if each vehicle knows when the one in front is slowing down it need not maintain such a large distance between vehicles. They can slow down and speed up as a unit. What allows this is direct vehicle to vehicle communication. Presumably the pace will be set by the one in the lead.

What can't this be done by actual trains?

CLamb--
This -- a "virtual" coupling of self-propelled railway vehicles without a material coupler -- has been tried in a railway context, albeit a rubber-tired railway like some Paris Metro lines. A fair bit of development effort was put into it as part of an effort to develope a fully automated mass transit system in France, under the name "Aramis." In the event it was decided that the precise speed control needed to run independent cars as a "virtual" train was too difficult for then-existing technology, and so that part, at least, of the project was abandoned.