"Failing"? Can you name a company with profit margins better than Google? Here's one: Union Pacific.

http://fortune.com/2015/06/04/union-pacific-railroad/

The problem with trains is the last mile: you obviously can't drive them to every store etc that needs a delivery. But the process of loading and unloading containers from ships to trains (and trucks) at ports is getting more and more automated as well.

I mean the market forces are failing us, not that train companies are.

Then of course a transportation company with 20% profit margin is very likely the result of market failure...

Throw the fraction of the money used for promoting and developing self-driving and platooning, coupled with the cost of building roads to public transport, trains, tracks and ships and you will have a much more efficient albeit less cool system.

Tangential, but letting (semi) self-driven convoys on the road is quite irresponsible, imagine the headlines if they were to get in an accident.

Is it irresponsible if the self-driving truck is believed to be as safe or safer than a human driver? Humans get into accidents quite frequently.

If they did it for the sake of this challenge then it is an added risk imo, so yes (the real alternative is no human operated traffic at all). Some people believe a controlled car hack on a highway is safe, or at least not substantially more dangerous than everyday highway business, some car just break down on the commute.

The last mile is where automated trucks are least likely to work, though. They have to deal with local streets, pedistrians, cyclists, poor roads, unmarked roads, all the things that are most difficult for automated vehicles.

Fortunately, autonomous trucking doesn't have to solve many of the tricky last mile problems before it becomes immediately economically advantageous. You could imagine local depots just outside of the city center where platoons pull into to switch from autonomous to human drivers.

Truckers would only have to work inside a local area rather than deal with the logistics of long haul routes and drive challenging surface streets rather than the soul numbing monotony of long interstates.

Yes, trains exist.

What the mentioned platooning achieves though, is rapid re-grouping once a truck joins or leaves the platoon. Imagine that you save 10-20% of fuel within a platoon (as pointed out in the article), then precise scheduling entering the highway at the right time makes suddenly sense.

Trains can't do that with the same flexibility.

Fuel consumption is about the last thing you want to bring to the fore when comparing trucks to trains. You could take all the diesel in a truck, burn it in a oil power plant and use the generated electricity to power a train. You would go much further.

I only compared the ability of those truck platoons to re-group dynamically to those of trains.

Trains are much more heavily constrained by grade, radius and stopping distance than trucks are. You can't build a railway to every door. And they're far more expensive. Their main advantage is in axle weight.