Autonomous Rail Transportation

If you want to know how far we are from self-driving trains, just listen to http://www.railroadradio.net/ for a while.

  • BNSF has a computerized dispatch system, "Movement Planner," that doesn't seem to understand local yard characteristics and requirements.
  • Last week, two sections of a transcontinental passenger train were kept from reaching the depot in Spokane for Four Hours when a BNSF freight broke down in the junction entering town. Apparently nobody thought to bring a unit from the shop and yard the sucker out of the way to work on. Did a mechanic come out? Can't say, went to bed, but the first half hour some tech on a radio was telling the crew "switch this knob, push that button" without noticeable effect.

Right now, the mechanicals ain't there and the electronics ain't there. Maybe someday.

The only railroads in the world comparable to U.S. railroads are in Russia, China and India for distance, time and tonnage. Other comparisons are pointless. Not to say that other rail systems aren't valuable, of course they are. The German and Japanese systems are inspiring. But Germany is the size of Wisconsin + Minnesota and Japan is the size of California. So, no comparison.

:B~)
 
I don't know Jon, it seems to me that you pick incomplete information to justify broad predictions. As you report, there is a shortage of truck drivers in the U.S., but you overlook that there are some 3 million professional truck drivers still driving trucks. The current spike in fuel prices is causing quite a bit of disruption in that industry, but long haul trucking isn't going away.

Linking health effects to jobs has nothing to do with whether or not those jobs will continue. How many office workers are obese? Are those jobs threatened too? Jobs disappear because of changing economic conditions. Coal mining has traditionally been quite unhealthy, but if there is a demand for coal and coal mining wages to be earned, there are people who will take those jobs. Yet the demand for coal is decreasing for reasons that have nothing to do with the health of coal miners and the industry is generally losing jobs over the long term.

You propose that the railroad industry could invest more to improve reliability. Yet most North American railroads are currently embracing PSR (Precision Scheduled Railroading?) to cut costs and deliver more profits to stockholders at the expense of reduced service to the smaller railroad customers. Don't get me wrong, I've read that BNSF has been spending millions to increase the capacity of their southern transcontinental railroad route from Los Angeles to Oklahoma because the demand to haul tonnage using that route can justify the expense of the improvements. Other railroads are investing in projects that improve operation with the best cost vs. benefit ratios as well.

The current coupler and air brake system used in North American freight trains is antiquated. Better systems have been proposed. But no one wants to junk all of the equipment using the current system to replace it with a newer system. The cost is too prohibitive despite the benefits of the proposed systems.

I don't think that anyone here has stated that railroads will never switch to robotic trains. Given the current circumstances, it doesn't look likely to me in the near term.
 
Slightly OT, but here in southern Idaho we have both an interstate freeway and a main UP haul line. I would LOVE to see more of those truck trailers put on the train and off the freeway. We seem to have more than our share of accidents with trucks between Twin Falls and Mountain Home.
 
If you want to know how far we are from self-driving trains, just listen to http://www.railroadradio.net/ for a while.

  • BNSF has a computerized dispatch system, "Movement Planner," that doesn't seem to understand local yard characteristics and requirements.
  • Last week, two sections of a transcontinental passenger train were kept from reaching the depot in Spokane for Four Hours when a BNSF freight broke down in the junction entering town. Apparently nobody thought to bring a unit from the shop and yard the sucker out of the way to work on. Did a mechanic come out? Can't say, went to bed, but the first half hour some tech on a radio was telling the crew "switch this knob, push that button" without noticeable effect.

Right now, the mechanicals ain't there and the electronics ain't there. Maybe someday.

The only railroads in the world comparable to U.S. railroads are in Russia, China and India for distance, time and tonnage. Other comparisons are pointless. Not to say that other rail systems aren't valuable, of course they are. The German and Japanese systems are inspiring. But Germany is the size of Wisconsin + Minnesota and Japan is the size of California. So, no comparison.

:B~)

The AI system works when things work and fails miserably when there's the slightest glitch. We see this with our own simple OOPS programming of our AI drivers as we string together commands, marker, and scripts to instruct them to move from one location to another. An incorrect switch, marker, or instruction will leave the AI scratching their virtual heads and throwing up virtual arms when an error in our logic is encountered. With the larger virtual dispatcher, this system most likely works fine with smaller transit systems and other situations where things are less prone to go wrong or are more controlled, but in a large system with lots of interconnections, something is bound to break somewhere and then things fall down pretty fast.

With a regular train, I'm sure it would be easy to pull the lead loco and put in a replacement, but my assumption here is the AI system is connected to some encoder-system on the locomotive. This encoding-system then has some interconnect with the dispatching and signaling system so with a new disconnected loco is brought in, it'll run "blind" from the dispatcher because the dispatching office is located 1,000s of miles from Spokane with no one in that location having ever seen the tracks, signals, switches, or anything related to the yard in Spokane. In the meantime, the dead encoder is confusing the AI dispatcher into thinking that dead train is still occupying the space similar to tripping an ATLS crossing trigger then sitting across it while switching instead of moving completely past the trigger.
 
All this can be computerized and unmanned.
Not right now it can't

There is no reason under the sun for any freight railway employee these days to not come home each and every work night to sleep in his own bed.
Obviously there is.

A freight train could be sent from California to New York totally unmanned.
No it can not, unless your unicorn AI railroad system exists.

How well could Dave, Geoff or Bald drive a physical train?
Terribly - a complete disaster that nobody is willing to risk.

Auto trains could have radars just like Toyota cars with safety sense. They can have GPS. They can have cameras, heat detectors and motion detectors to detect trouble at crossings and ahead on the tracks. The AI computers would be tied in with railroad signaling.
A good bit of this is already on trains as I pointed out before. it still doesn't make full automation practical.

There is no reason to put humans inside long-haul freight trains anymore. Sitting on board a train for long hours and days away from home is not a healthy lifestyle just as OTR truck driving is not healthy. Manned freight vehicle operation on road or rails can be limited to short haul.
Says who and why? Many professions that are critical to our society are dangerous or unhealthy. While technology helps in this regard, it doesn't remove the human element altogether. There is no means by which to accomplish this as of yet. Even the best systems available today to map every rail change and adjust locomotive power will draw complaints from engine drivers that could do the job 10 times better and with half the hassle. Same with train routing on a signal system.

The American trucking industry is also short about 50,000 drivers nationally because nobody wants to be cooped up in a rig away from home.
Incorrect. Debatable. Presumptuous.

I still think AI is now feasible enough to send unmanned freight trains long haul as between two yards hundreds or thousands of miles apart.
You see - this is an incorrect and wild assumption.

In this day of unmanned spaceships to Mars, there is no excuse for unreliable rolling stock on lowly earthbound railways.
Trains travel on the planet surface amongst humans and animals, cities and structures and is a far difference challenge than pointing a device through open space.

If locos and rail cars tend to crap out a lot, maybe then Toyota should build them.
Idk if this was a joke, but... what?

I beg to disagree. I am a visionary. This is how I see tomorrow today.
Well sure, we all want matter-anitmatter reactors and space elevators and trips to the moon, but saying that doesn't make you visionary. In practical terms, you must deal with what you have right now if you are to make a statement like "All this can be computerized and unmanned." Perhaps it should say "could be computerized and unmanned one day" but even then you have to face the fact that you are hardly the first or only person to think of this.

I predict that AI freight hauling by rail will become commonplace in America before I die of old age. I'm sure interested in following its progress. It's not a wild assumption at all. We all know AI rail transportation is already happening somewhere in this world today.
That may be, I don't know how old you are but I hope its pretty young. You have a lot to learn.
 
Landslides, washouts, flooding, snow, fog, hackers, drunk people, sducidal people, break downs, sabotage, power outages, malfunctions. Yet we believe AI can or will handle all this. What are we doing with all the people we propose to eliminate?
 
These trains are running in the middle of nowhere I doubt if it would happen in an area with a population
 
I don't know. I might eventually drop my worries about RR technology. Perhaps, the RR industry will have to pay more money in payroll to have an incentive for human freight drivers to take long cross country trips. It might actually still cost more for RR's to adopt and maintain AI technology than to pay humans even handsomely to man trains in the lonesome and boring long runs away from home. Men on board trains or no men on board trains, the American RR's (the automobile industry and other enterprises) will always take the cheapest route that's both profitable and that the law will allow. Much of emerging technology is economics driven. It's also driven by ever-changing government regulations as in quests to save the environment and improve public safety. Automobile design is such a mess due to the EPA.

Some afterthoughts as curiosity is still prevalent: the big question is WHY auto trains, to begin with?

What is the distinct advantage to the railway company in Australia for employing an Auto Haul freight train from a mine to a coast without a human crew? We must always attack the premise. Why are Auto trains even needed anywhere? Are they just experimenting with technology for now? Is the mining company a test bed for auto train R&D?

https://www.youtube.com/watch?v=8A-GAg69rAM


Let's take this thread in a new direction just for a barrel of fun:

Let's now list all the pros and cons of autonomous trains:


PROS

-no human train crews will get hurt should the train wreck being that the auto train is unmanned by its very nature


CONS

-human job opportunities are potentially lost since no human railroad workers are needed on board autonomous trains under way


Fellow Trainzers, please add to my pros and cons list if you may. This should be interesting. Both manned trains and unmanned trains will have advantages and disadvantages.
 
Last edited:
Some afterthoughts as curiosity is still prevalent: the big question is WHY auto trains, to begin with?

What is the distinct advantage to the railway company in Australia for employing an Auto Haul freight train from a mine to a coast without a human crew? We must always attack the premise. Why are Auto trains even needed anywhere? Are they just experimenting with technology for now? Is the mining company a test bed for auto train R&D?

Quite simply, the benefit outweighs the disavantages, due to the unique situation that is the Pilibara iron ore mining railways - they're basically set up already as less a railway and more a glorified conveyor belt, with set train lengths and services running day in day out without end. Add in that the lines run through what is virtually inhospitable desert from the isolated mine sites to the ports on the coast, in a state physically larger than France but where over two thirds of the population of 2.2 million reside in the capital city only thousands of kilometres away, and you can start to see that it's not a simple matter of having a staff of drivers and mechanics nearby on call 24/7 - most have to be flown up and reside on site for a period of time ('Fly In/Fly Out' or FIFO as it's known), at significant cost to the mining companies. So autonomous technology starts to look attractive for such a situation.

Also it needs to be said the Rio Tinto trains are not truely 'autonomous' anyway - they're more accurately 'remote controlled', still with a human driver, only the driver is thousands of kilometres away in a nice comfortable air-conditioned office in Perth, watching the track ahead on a video feed and controlling the train with a joystick, rather akin to a game console controller (and thus it's much like playing a simulator like Trainz). The trains have cameras mounted on the front and in the cab which gives the remote drivers a full view they would have had in person. This is very costly to set up however, and at the moment of the various mining companies operating here only Rio Tinto has taken the Remote Control plunge (they've also installed remote control in the mines with the dumpers and diggers also remotedly controlled from the same location in Perth).

This situation is not suitable for all railways though, and certainly not long distance/transcontential trains which have varying schedules, routes or locomotive replacements, and thus creates potential problems that requires human oversight. It's the whole 'horses for courses' thing. And as others have said, full autonomous operation of all railways is probably something that will not be happenning any time soon, if at all.
 
Last edited:
Ok, it's now simple to understand: in short, right now, auto trains work well for special situations in some parts of the planet but are not yet feasible for every rail transportation situation worldwide.

This video suggests that American railroads may have to eventually go autonomous to some degree to compete with the trucking industry should it go seriously autonomous:

https://www.youtube.com/watch?v=Q79BHfxfaSI


One thing is certain: people and things will still need to be moved in the future as well as in the present, it will be interesting to see how the transportation industry will go about it as time marches on
 
Last edited:
Back
Top