Autonomous Trains

[John]

https://www.wsj.com/articles/how-to-move-more-goods-through-americas-clogged-infrastructure-robot-trains-11633790990?tesla=y

"Anyone who has spent Christmas morning sending a choo choo around the tree knows the primary hazards to trains are toys wandering onto the track, and taking turns at high speed—both of which are also more or less the primary hazards to trains in real life. So, you might ask, how hard can it be to make trains fully autonomous?"

[peteski]

Since trains are strictly confined to their rigth-of-way, I  would think it would be magnitudes easier than self-driving cars.

[Steveruger45]

Some shuttle trains at airports already are and have been for years.

[samusi01]

Since trains are strictly confined to their rigth-of-way, I  would think it would be magnitudes easier than self-driving cars.

Reminds me of the old story told me by a KCS engineer years ago. It's probably an apocryphal story but as told to me, the crew hit a horse. Big mess all over the front of the locomotive. They called the dispatcher to relate the equine demise and the dispatcher radioed back, "was it on the right of way?" This annoyed the crew, who responded by saying "no, we had to swerve to hit it!"

[Mike C]

 

[Steveruger45]

Reminds me of the old story told me by a KCS engineer years ago. It's probably an apocryphal story but as told to me, the crew hit a horse. Big mess all over the front of the locomotive. They called the dispatcher to relate the equine demise and the dispatcher radioed back, "was it on the right of way?" This annoyed the crew, who responded by saying "no, we had to swerve to hit it!"

😂😂😂.  I will never be able to look at the NS stallion logo quite the same now.

[Missaberoad]

Since trains are strictly confined to their rigth-of-way, I  would think it would be magnitudes easier than self-driving cars.

Maybe on a closed loop with minimal grades and road crossings...

[peteski]

Maybe on a closed loop with minimal grades and road crossings...

Regardless, still magnitudes easier than self-driving cars on standard roads.  Train can only follow its tracks.  It does not have to swerve around double parked parked cars, careless pedestrians, road construction, and a multitude of other obstacles.  Car also has to plot its own way, even in inclement weather or snow covered roads where road markings are not visible.  There are *LOTS* more last minute surprise obstacles on the road than on the tracks, plus the track ahead already cleared for the train (unlike on the automobile roads).  Just think about all the possibilities self-driving car has to deal with, vs. what is going on the train's right-of-way.

Sure, there are train dynamics to consider, but those can be calculated and dealt with.  The grade profile and the location of the tracks are fixed.  I'm not saying that it would be easy to replace a skilled train engineer, but IMO, it is still easier than perfecting a self-driving car.

[learmoia]

As an alternative to the inherent safety concerns of 1 man crews.

 I have always thought of 'Drone' style operation, a group of engineers in an air conditioned building operating out of work stations.

You would still need utility men to be able to attend to the train in case of an issue.

[Missaberoad]

Pete, I feel you are underestimating the complexity of train handling, and the amount of variation in how different trains react to identical situations...

But as you say just my opinion.

[nkalanaga]

It's been done, in a "mainline" environment, by at least two coal-hauling railroads, in Ohio and Arizona.  The Muskingum Electric Railroad on Ohio ran from 1968 to 2002, quite successfully, including grade crossings.
https://en.wikipedia.org/wiki/Muskingum_Electric_Railroad

And the Black Mesa & Lake Powell in Arizona was also technically successful, but ended up with human crews, as they could never find a way to avoid hitting herds of sheep.
https://en.wikipedia.org/wiki/Black_Mesa_and_Lake_Powell_Railroad

With good communications, it might be easier, at least in the near term, to run true remote-control trains.  A video camera on the train, and a monitor in an office, would give the engineer as good a view as from the cab, and the controls can easily be connected by radio.  After all, that's how DPUs work!  Basically, turn the prototype into a big model railroad, at least for through trains.  Much easier than trying to program the locos themselves to run the train.

For any train requiring switching, it would HAVE to have a human operator, at least until every car has remote controlled couplers and air hoses.  I can't see American railroads paying for that anytime in the next century!

One BNSF executive said, a few years ago, that they weren't looking at one-person crews for mainline trains.  In his opinion, they're fine in Europe, where trains are short, and in most places, with walking distance of a town.  Over most of BNSF's territory trains had to have two people, if they have any, for safety.  Not while the train is running, but for any emergencies.  If one person had to get off the train, then was injured, or bit by a rattlesnake, or whatever, there needed to be someone else to call for help.

For local and industrial switching, one-person crews are now common, often with remote-control locos, and have proven quite safe.  In fact, with the one person on the ground, throwing switches, uncoupling cars, etc, AND running the loco, there's less chance of a mistake than with two people trying to coordinate their activities.

[John]

I won't say "never" -- technology always seems to advance, and eventually train handling dynamics could be overcome to where you could remove that experienced crew from the cab . We are seeing it with self driving vehicles and drones delivering packages, pizza, etc.  We can drop a bomb or a pallet full of cash on a Taliban from 10,000 miles away with accuracy and command a rover on Mars to carry out activities.   Having a human engineer in the "office" running the trip might work out, but as a couple of others have said, how do you experience the feel, cues, and clues of how the train is handling by sitting sitting in the cab.  Just like flight simulators -- you can see the screen moving around, but you don't experience the G forces. 

I imagine you would start in full PTC territory, but run smaller trains to reduce the physical stresses on the train between major terminals. As others have mentioned, you would also need employees located in strategic locations along the route to work emergencies and other events as they came up. 

All that said,  while I have experience driving large naval ships weighing over 40,0000 tons, each with unique handling characteristics that I had to learn, the Navy is now experimenting with autonomous ships.  Trains will follow at some point.    The real question, at what cost?   -- now back to hogging my N scale 40 car train through the 4 loop helix

[learmoia]

[John]

Pretty cool video ..

https://www.railwayage.com/freight/rio-tinto-operates-first-driverless-freight-train/

[CBQ Fan]

Many more variables for automobiles interacting with one another than a train on a track. Not over simplifying one but stating how much more complicated the other is.  Self driving vehicles are a lot farther out than many like to suggest. The issue isn’t so much the technology, although it has many challenges to over come, it is the legal hurdles with regard to liability.  Is the driver at fault, did the system operate properly, is it a covered loss under your policy, is the manufacturer responsible.  Currently these type of claims are legal minefield. Think lawyers combined with insurance companies and auto manufacturers.  So many deep pockets and uncharted legal territory. Then throw in the mix the laws of each state being very different from one another. And then once it gets widespread enough the Fed Government will want to weigh in.