Self-Driving Trains

6105 views
147 replies
1 rating 2 rating 3 rating 4 rating 5 rating
  • Member since
    January, 2014
  • 5,577 posts
Self-Driving Trains
Posted by Euclid on Friday, September 13, 2019 9:01 PM

From the Trains Newswire:

http://trn.trains.com/news/news-wire/2019/09/13-computer-runs-train-at-test-track-nyab-announces

I don’t think anyone has yet precisely defined PSR, but it seems like a marker to the future, kind of like PTC or any form of automation.  Right in there with that general trend, I see automated running, once we figure out how to handle those broken knuckles. 

  • Member since
    May, 2003
  • From: US
  • 16,905 posts
Posted by BaltACD on Friday, September 13, 2019 10:13 PM

A test train's purpose is testing - nothing else, on a test track with a fully known geometry and without connection to 'real world' operations.   I could hook my truck up to some cars, put it in cruise control and say it was Self Driving.

  • Member since
    January, 2014
  • 5,577 posts
Posted by Euclid on Friday, September 13, 2019 10:33 PM

Maybe so, but I don't think that is the point.

  • Member since
    July, 2010
  • From: Louisiana
  • 1,820 posts
Posted by Paul of Covington on Friday, September 13, 2019 11:03 PM

   Thirty freight cars  "over a variety of terrain, starting and stopping on uphill, downhill, and flat grades."   I'm not a railroader, but this seems like they're just testing basic functions.  With just thirty cars, the terrain won't vary much over the length of the train.   How will it handle 150 cars over varying terrain with parts of the train going uphill and parts going downhill?   I think they'll need a lot of real life testing.

   By the way, I've suspected from the start that part of the reason the railroads didn't protest more over the PTC mandate was that they saw it as a possible step in the direction of automated unmanned trains.

_____________

   "A stranger is just a friend you ain't met yet."  ___ Dave Gardner

  • Member since
    January, 2002
  • From: Equestria
  • 7,030 posts
Posted by zugmann on Friday, September 13, 2019 11:14 PM

Trip optimizer and auto throttle are already being used in places.

 The opinions expressed here represent my own and not those of my employer or any other railroad, company, or person.

  • Member since
    December, 2001
  • From: Northern New York
  • 19,597 posts
Posted by tree68 on Saturday, September 14, 2019 7:07 AM

zugmann

Trip optimizer and auto throttle are already being used in places.

With PTC and near pin-point geolocation from GPS, integration with TO, etc, makes a self driving train just that much closer.

Could it be that with such a self-driving arrangement will work better with shorter trains?  If you take away the manpower issue, and run trains that only require one locomotive, you can cut some costs.

Odds are there will be problems with the 12,000 foot "land barges" that are seen on the road today, as suggested.  

LarryWhistling
Resident Microferroequinologist (at least at my house) 
Everyone goes home; Safety begins with you
My Opinion. Standard Disclaimers Apply. No Expiration Date
Come ride the rails with me!
There's one thing about humility - the moment you think you've got it, you've lost it...

  • Member since
    April, 2007
  • 2,014 posts
Posted by Convicted One on Saturday, September 14, 2019 11:22 AM

Paul of Covington
With just thirty cars, the terrain won't vary much over the length of the train.   How will it handle 150 cars over varying terrain with parts of the train going uphill and parts going downhill?

With the reduced crew costs that will accompany automation, isn't there less incentive for long trains?

Maybe the crewless "train of tomorrow" will be made up of one locomotive pulling 40 cars at 20 mph, where all meets are rolling and scheduled?

  • Member since
    December, 2001
  • From: Northern New York
  • 19,597 posts
Posted by tree68 on Saturday, September 14, 2019 12:22 PM

LarryWhistling
Resident Microferroequinologist (at least at my house) 
Everyone goes home; Safety begins with you
My Opinion. Standard Disclaimers Apply. No Expiration Date
Come ride the rails with me!
There's one thing about humility - the moment you think you've got it, you've lost it...

  • Member since
    January, 2014
  • 5,577 posts
Posted by Euclid on Saturday, September 14, 2019 1:57 PM

Larry and Convicted One,

I think you are right about self-driving trains leading the way to shorter trains.

I expect the railroads to latch onto the self-driving train concept with a vigor that makes their sudden zest for PSR pale by comparison.  For one thing PSR is more marketing hype than actual substance.  That is why nobody can define it.  But there is real substance in self-driving trains and all the automation that will also make operations safer.  Now you’re talking real substance, and it will finally be able to deliver the precision and timing that PSR now only implies as a wish.  It will move PSR from a slogan to a reality.

The monster trains now in vogue are justified by moving more tonnage with fewer operators on board.  This was the fundamental advantage of trains going back to their inception.  Dieselization maximized that advantage by providing M.U., allowing additional locomotives without requiring additional crews to operate them.  But still, all the power needed to be on the head end, and so the drawbar strength was the limit to this M.U. enabled maximum train length. 

Then came D.P.U. which shattered the M.U. limitation of drawbar strength.  Now, with D.P.U, the only limitation to train length under the control of one engineer is control signal strength.  Yet I expect that increasing the D.P.U. signal capability will be a piece of cake for a society that plans to colonize Mars. 

Nevertheless, monster trains are ponderous to handle during any contingencies such as breakdowns, and the delay they cause at grade crossings is raising the hackles of the public.   So just because D.P.U. makes even longer trains possible, I think the industry has ridden that horse as far as it can go.  It is almost a last gasp in reaching the highest productivity by exploiting the long train potential that was the fundamental to the train concept when first introduced. Even in the 1960s, this trend of longer trains made possible by M.U. was starting to be questioned because although it offered the minimum manpower for the tonnage moved, it tended to interfere with the quick delivery that customers wanted.  The longer the train, the more operational factors that are dependent on each other. 

So now it is time for the short, nimble trains running fast on close headway, all made possible by PTC automation, and maybe even ECP brakes, depending on how this all shakes out.  Short trains may even be semi-permanently coupled. And then what makes these short trains ultimately possible is that they are self-driving.  No need to get the maximum tonnage under the control of one engineer; just eliminate the engineer instead.  Then instead of moving tonnage in long trains with distributed power, the industry will be moving tonnage in distributed trains. 

It will change from power from being distributed in the trains, to short trains being distributed on the railroad.   

  • Member since
    April, 2007
  • 2,014 posts
Posted by Convicted One on Saturday, September 14, 2019 2:24 PM

Euclid
It will change from power from being distributed in the trains, to short trains being distributed on the railroad.   

I don't think that the trains would necessarily have to be "fast". I think the speed they would be run at would be optimized to coordinate meets.

I picture a system running at a 20 mph constant with passing sidings  located with sufficient frequency where  few, if any, trains truly had to stop in route for meets.

  • Member since
    March, 2003
  • From: Central Iowa
  • 4,934 posts
Posted by jeffhergert on Saturday, September 14, 2019 4:58 PM

zugmann

Trip optimizer and auto throttle are already being used in places.

 

The EMS technicians that come out in the field from time to time have told us they have been working on the auto control of the air brakes.  That it wasn't far off.

I may see one person crews, but I don't think I'll ever see completely crewless trains.  (I don't expect to see totally driverless trucks,either.)  It will certainly change the job of locomotive engineer, more to an observer ready to take control when the system fails. At least on rotes equipped for such operation.

I work with these EMS auto control systems.  They work almost flawlessly at times, some types of trains better than others, and they've caused train separations and even a minor derailment.  Sure, technology can and will get better as time goes on.  Still, one must remember that these devices trying to replace imperfect humans is designed, built, and (most importantly) maintained by imperfect humans.  That maintaining part will be the important factor.  It probably won't be maintained as well as it should be.  

Jeff     

  • Member since
    September, 2002
  • From: ShelbyTwp., Michigan
  • 412 posts
Posted by SD60MAC9500 on Saturday, September 14, 2019 9:52 PM

The fully autonomous train of the future is going to pass right by locomotive hauled trains for self powered platforms for a true DP setup.. They can couple decouple on the fly with fully automated couplers.. Yet I wouldn’t expect em all to be short.. I can still see a train of 50-60 5 pack nextgen self powered maxi stack IV’s carrying hundreds of ISO boxes from port to fully automated IM terminals.. The current ICE power source replaced with a simpler battery/capacitor layout, or Maglev, and the complete elimination of, crossings, and carload freight will have to happen before trains become truly autonomus. Trucks will eventually just be boxes on wheels with large batttery or capacitior packs. I assume this will lead to a truly integrated road-rail network. Of course these items are 4-5 decades away, or more... 

Rahhhhhhhhh!!!!
  • Member since
    June, 2009
  • From: Dallas, TX
  • 4,228 posts
Posted by CMStPnP on Sunday, September 15, 2019 4:33 AM

Based on my knowledge of what IT and computers can currently handle I would have to agree with the rail union perspective that we are nowhere near crewless frieght trains on the national rail network.   It's going to take decades  of training AI to get to the point where it is reliable enough to use in place of the cab based operating crew without causing or creating collisions or derailments.   It's not going to happen anytime soon.

  • Member since
    January, 2014
  • 5,577 posts
Posted by Euclid on Sunday, September 15, 2019 7:45 AM

CMStPnP
...I would have to agree with the rail union perspective that we are nowhere near crewless frieght trains on the national rail network.

Can you cite a source for that rail union perspective?

  • Member since
    September, 2003
  • 8,428 posts
Posted by Overmod on Sunday, September 15, 2019 9:34 AM

Euclid
Can you cite a source for that rail union perspective?

You can start by perusing the responses mentioned here..   Then a quick search on some of the names of the responding organizations should quicly flesh out their current positions. 

Note that in some respects this is a very fast-moving technology and in others almost petherically retarded.  It requires more than the usual care to assess what is and isn't practical, or more importantly what is and isn't required for a particular intended use of the technology.

(Personally I don't think fully autonomous trains of any particular size will ever be wholly practical in the North American political and legal climate, even were there to be some version of a Price-Anderson act or some Amtrak-like cap on full liability for a given incident.  That isn't important in and of itself, but I have been following the history of autonomous control from my early childhood, when the New York subways got it to work on the Grand Central shuttle for a while, and I continue to study autonomous vehicle tech intensively, and would like to see at least a successful version of one-man operation with autonomous-tech support.)

  • Member since
    September, 2017
  • 1,920 posts
Posted by charlie hebdo on Sunday, September 15, 2019 10:14 AM

CMStPnP

Based on my knowledge of what IT and computers can currently handle I would have to agree with the rail union perspective that we are nowhere near crewless frieght trains on the national rail network.   It's going to take decades  of training AI to get to the point where it is reliable enough to use in place of the cab based operating crew without causing or creating collisions or derailments.   It's not going to happen anytime soon.

 

You mean more accidents and derailments than we have now with humans running the show? 

  • Member since
    May, 2003
  • From: US
  • 16,905 posts
Posted by BaltACD on Sunday, September 15, 2019 10:41 AM

Euclid
 
CMStPnP
...I would have to agree with the rail union perspective that we are nowhere near crewless frieght trains on the national rail network. 

Can you cite a source for that rail union perspective?

Can you cite your source for your perspective?

  • Member since
    January, 2014
  • 5,577 posts
Posted by Euclid on Sunday, September 15, 2019 11:40 AM

BaltACD
 
Euclid
 
CMStPnP
...I would have to agree with the rail union perspective that we are nowhere near crewless frieght trains on the national rail network. 

Can you cite a source for that rail union perspective?

 

Can you cite your source for your perspective?

 

My own deductions are my source, and what I have said based on them is only my opinion.  I am not challenging what CMStPnP said about the rail unions' perspective.  I am only curious to know exactly what the rail unions said.  I would expect them to be vehemently opposed to self-driving trains for whatever reason they can think of.  But for them to say it won't happen soon seems odd to me.

That almost sounds like people who predict something won't happen because they don't want it to happen. 

  • Member since
    May, 2003
  • From: US
  • 16,905 posts
Posted by BaltACD on Sunday, September 15, 2019 12:09 PM

Euclid
 
BaltACD 
Euclid 
CMStPnP
...I would have to agree with the rail union perspective that we are nowhere near crewless frieght trains on the national rail network. 

Can you cite a source for that rail union perspective? 

Can you cite your source for your perspective? 

My own deductions are my source, and what I have said based on them is only my opinion.  I am not challenging what CMStPnP said about the rail unions' perspective.  I am only curious to know exactly what the rail unions said.  I would expect them to be vehemently opposed to self-driving trains for whatever reason they can think of.  But for them to say it won't happen soon seems odd to me. 

That almost sounds like people who predict something won't happen because they don't want it to happen. 

Your source is seriouly suspect and has no basis in real world experience in the subject area.

  • Member since
    February, 2003
  • From: Guelph, Ontario
  • 3,804 posts
Posted by Ulrich on Sunday, September 15, 2019 1:06 PM

More likely, some trains (such as closed loop mining operations) will be completely autonomous while others will require a crew of two or more (switching a busy industrial branch perhaps). 

It should be noted that we’re all in the same boat.. regardless of how you earn your living, technology will transform your job in a big way over the next decade. Or maybe not.. my crystal ball has been wrong before. I go for my early morning walks thinking about 5G, Blockchain, the Internet of Things.. yet the garbage is still picked up via a man in a truck as it was 50 years ago. Sometimes indeed the more things change the more they stay the same.

 

  • Member since
    January, 2014
  • 5,577 posts
Posted by Euclid on Sunday, September 15, 2019 1:41 PM

I would think that technology would be less of an impediment to self-driving trains than to highway vehicles.  Trains already have their own mechanical guidance system and PTC takes a big step toward the technology to control self-diving trains.

This biggest impediment to self-driving trains will be the labor unions and their interest in railroad employee job security.  They will argue that self-driving trains are unsafe and endanger the public.  But at the same time, self-driving cars are being promoted by the claim that they will make driving much safer.  That same view could be applied to self-driving trains. 

So with self-driving cars, trucks, and trains, replacing an imperfect human operator with an infallible machine will be the promotion thrust.  Technology will be the only limitation and it will apply mostly to road vehicles, and hardly at all to trains. 

I think the railroad industry will enthusiastically welcome self-driving trains because it will save money, and they are already getting half the concept foisted on them in the form of PTC.  Self-driving trains can also be blended into human operation train traffic, so there is no need for a national system conversion as is the case with ECP brakes, for instance.  

  • Member since
    September, 2017
  • 1,920 posts
Posted by charlie hebdo on Sunday, September 15, 2019 2:41 PM

BaltACD

 

 
Euclid
 
BaltACD 
Euclid 
CMStPnP
...I would have to agree with the rail union perspective that we are nowhere near crewless frieght trains on the national rail network. 

Can you cite a source for that rail union perspective? 

Can you cite your source for your perspective? 

My own deductions are my source, and what I have said based on them is only my opinion.  I am not challenging what CMStPnP said about the rail unions' perspective.  I am only curious to know exactly what the rail unions said.  I would expect them to be vehemently opposed to self-driving trains for whatever reason they can think of.  But for them to say it won't happen soon seems odd to me. 

That almost sounds like people who predict something won't happen because they don't want it to happen. 

 

Your source is seriouly suspect and has no basis in real world experience in the subject area.

 

More likelty the unions have PR spinmeisters who figured that saying it won't happen because of [flaws, IT limitations, etc.] plays better than saying because we don't want to lose dues.

  • Member since
    June, 2009
  • From: Dallas, TX
  • 4,228 posts
Posted by CMStPnP on Sunday, September 15, 2019 6:28 PM

Euclid
Can you cite a source for that rail union perspective?

Its in TRAINS NEWSWIRE, why do folks keep asking me for a source without looking via this website themselves...........Oy!

 

http://trn.trains.com/news/news-wire/2019/09/13-union-reacts-to-computer-run-train-test-thats-not-the-real-world

 

 

  • Member since
    June, 2009
  • From: Dallas, TX
  • 4,228 posts
Posted by CMStPnP on Sunday, September 15, 2019 6:31 PM

charlie hebdo
You mean more accidents and derailments than we have now with humans running the show? 

Well psychologically you have to ask yourself which type of accident is more excusible by society.    Human error or automated computer error.   I would definitely argue the former as society seems to think computers should be less error prone than humans, especially if it is safety related.

  • Member since
    January, 2014
  • 5,577 posts
Posted by Euclid on Sunday, September 15, 2019 7:18 PM

Regarding the union argument that the recent test is not real world conditions:  That’s right, it’s not.  But one test will lead to another, and eventually, the tests will be with real world conditions.  Testing always begins with just limited examples of the intended full application.  Test data is gathered and analyzed.  Design is improved, and then retested.  Eventually it is deemed ready to be put to work and still the testing and revisions continue.  So, I don’t think the unions’ argument that self-driving trains are not possible will prevent them from happening. 

The only argument that the unions have is that self-driving trains will compromise safety.  But that is going come right up against the counterargument that self-driving trains will be safer because they will eliminate human error.  In this sense, self-driving trains are just like PTC, and nobody is arguing that PTC will compromise safety. 

  • Member since
    February, 2003
  • From: Guelph, Ontario
  • 3,804 posts
Posted by Ulrich on Sunday, September 15, 2019 7:24 PM

It really boils down to economics. Labor represents a relatively small portion (compared to most other industries) of overall spend. Why fix what ain't broken? Two people on a ten thousand foot long train sounds pretty economical to me. Sure.. cut that back to one where feasible.. but we're pointlessly bumping up against the law of diminishing returns when limited cerebral resources would be better deployed elsewhere (such as increasing that all important top line through greater sales).. 

  • Member since
    June, 2009
  • From: Dallas, TX
  • 4,228 posts
Posted by CMStPnP on Sunday, September 15, 2019 8:01 PM

Euclid
So, I don’t think the unions’ argument that self-driving trains are not possible will prevent them from happening. 

The way I interpreted the remarks is that the Union was saying the technology is not ready today at this moment, I don't think they said it would never be ready.....at least I did not read that in their comments. 

 

  • Member since
    March, 2003
  • From: Central Iowa
  • 4,934 posts
Posted by jeffhergert on Sunday, September 15, 2019 8:07 PM

   

I work with PTC every time I go to work on a train.  EMS almost every time, much of our fleet is equipped with integrated PTC/EMS.     

I'm sure the railroad bean counters would employe autonomous systems today, proven or not.  All they see is (what they think will be) cost savings.  The reality of it is that while the technology is good, it's not there yet.  There are a lot of failures of both PTC and EMS.  Sometimes it is a minor issue, sometimes a major issue where trains have to run with out using them.  (My train today, the PTC wouldn't pass a departure test.  That is, it wouldn't make a penalty brake application during the test.  I don't know how it left it's originating terminal in that condition.  Even though the law doesn't require a PTC equipped leader yet, our rules have progressed to that point on my territory.  Someone's using a very broad definition of 'enroute failure'.  I was able to log into the EMS -Trip Optimizer- but it never became active.)  

The EMS auto throttle works fairly well much of the time.  LEADER has improved, but from what I hear it doesn't save as much fuel as was originally promised.  Trip Optimizer has been reprogrammed to increase fuel savings.  Much of the time it now runs 8 to 10 mph under speed now.  (There's one spot at the bottom between two hills on my run where almost everytime on certain types of train it will bottom out about 10 under, start up the next hill and go into dynamic braking until about half way up.)  It used to run a lot better but wasn't saving fuel doing so.  They won't let the very hot Z trains use EMS.  And despite what the salemen say, EMS has caused train separations.  

It's because of what I see every day I don't think I'll see truly autonomous operation.  Someone will be on hand, if not to take manual operation, at least to try to reboot the computer.  Yes, they may have utility people running around for situations like that.  But you know they won't have enough when multiple trains have issues.  We have already lost in many places, a utility mechanical department trouble shooter who could make minor repairs beyond what the crew could do.  All for the sake of saving money.

Jeff       

  • Member since
    August, 2005
  • From: At the Crossroads of the West
  • 10,205 posts
Posted by Deggesty on Sunday, September 15, 2019 8:20 PM

Jeff, do you have any idea as to how much disemploying (yes, another way of saying "firing") the utility mechancal department trouble shooter has cost over the cost of continuing to use his services?

Johnny

  • Member since
    July, 2014
  • 19 posts
Posted by Ajsik on Sunday, September 15, 2019 8:29 PM

Ulrich

yet the garbage is still picked up via a man in a truck as it was 50 years ago.

 

 

It was probably two guys. Fifty percent improvement in fifty years.

Join our Community!

Our community is FREE to join. To participate you must either login or register for an account.

Search the Community

Newsletter Sign-Up

By signing up you may also receive occasional reader surveys and special offers from Trains magazine.Please view our privacy policy