Autonomous Rail Cars

[greyhounds]

This has a whole lot of possibilities if it works, and the government doesn’t screw things up once more.

 

From the Trains newswire:

 

https://www.trains.com/trn/news-reviews/news-wire/details-emerge-of-genesee-wyoming-plans-for-first-test-of-autonomous-container-equipment/

 

First, A “Trigger Warning.”  The contents of this post will likely greatly irritate Backshop. He may even be moved to post a mean, spiteful, false personal attack on me. Again. I suggest that he approach with caution.

 

Second, I do see some issues with the concept. Such as:  signal activation, the need to accommodate 53’ containers, the need for a higher speed, and the need to couple with conventional equipment if it breaks down. I’m sure many of you can add more issues if you’re interested. I believe the issues I see can easily be dealt with. Any issues that I don’t see may be significant and hard to deal with.

 

Let’s start with my favorite underutilized rail line. The CN Iowa line. The line runs west from Chicago and enters Iowa at Dubuque. The line continues across Iowa through Waterloo to Ft. Dodge. Just west of Ft. Dodge it splits into two lines. One line goes to Council Bluffs and the other goes to Sioux City. There are branches from Waterloo to Glenville, MN, from Manchester, IA to Cedar Rapids, IA and from Wall Lake, IA to Ida Grove, IA.

 

 

The line has only two scheduled manifest trains. One per day each way between Kirk yard (Effectively Chicago) and Waterloo. The line also handles unit trains of ethanol and grain.  Everything else is served by locals. There is no intermodal service offered on CN’s Iowa line. The line obviously has excess capacity that could be used by autonomous rail cars.

 

I see numerous opportunities for business development on CN’s Iowa line, but I’ll just select one as an example. And remember, Iowa is all about the production of food. (OK, there’s that ethanol.)

 

Cedar Rapids, located at the end of a branch, produces a lot of breakfast cereal. Quaker has what’s been labeled “The World’s Largest Cereal Factory” there.  The latest volume output figure I have seen for Quaker at Cedar Rapids was 100 truckloads per workday. There is no output (the cereal) from Quaker that is shipped by rail.

 

The other cereal factory in Cedar Rapids is a recently expanded General Mills facility. It isn’t the size of the Quaker facility. But it’s big enough to warrant attention.

 

Conventional railroading just isn’t working here. Maybe autonomous cars can work. If several platoons of the cars were dispatched eastward daily, they could motor right on into the CSX intermodal terminal at Bedford Park and the containers transferred to conventional rail cars for furtherance on existing trains to eastern and southern population centers.

 

Maybe the savings will be great enough to allow CN to split the platoons near Chicago and send cereal to Milwaukee, Green Bay, etc. Maybe, just maybe, the cost savings will be enough for the railroad to be truck competitive in the Cedar Rapids-Chicago market.

 

Anybody have anything to add?

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

[daveklepper]

Yes, absolutely.

1.  The one-car or group of automous cars train is not crewed.  Automatic train oprtation is sinmply never going vto be legal and safe on any railroad with infenced and at-grade right-of way.  Complete isolation  as in rapid transit systems and within land owned by the coirpration that is the customer and/or owner of the railroad are possible, and CN, former IC, western line c definitely fails the separation requirement.

2,   CN shoukld get off its lazy butt and go after the business.  Don't need automatic cars.  Sufficirny business for a regular daily train. 

[daveklepper]

2.  Sufficient business for a daily train, anyway/

[Euclid]

daveklepper

Yes, absolutely.

1.  The one-car or group of automous cars train is not crewed.  Automatic train oprtation is sinmply never going vto be legal and safe on any railroad with infenced and at-grade right-of way.  Complete isolation  as in rapid transit systems and within land owned by the coirpration that is the customer and/or owner of the railroad are possible, and CN, former IC, western line c definitely fails the separation requirement.

2,   CN shoukld get off its lazy butt and go after the business.  Don't need automatic cars.  Sufficirny business for a regular daily train. 

 

This objection is stated every time autonomous operation is brought up, but it is never explained in a way that justifies it.  Can you please explain, for the first time ever, what a crew will do to overcome the safety issues that you see with autonomous trains? 

 

Tell me exactly why a right of way fence is needed for autonomous trains, but not for crewed trains. 

 

This point seems like an argument put forth by people who simply oppose autonomous running because it threatens labor. 

[jeffhergert]

So do the potential customers, at both ends, have to invest in equipment to load/unload the containers off the rail vehicle? I guess they would have to have chassis available for the time the containers are off the rail and in the truck docks at their facility. Most already have yard tractors, so that's not a problem. Or will they run from intermodal terminal to intermodal terminal? That just adds more problems for handling from customer to customer. Not insurmountable, but not really convenient. 

Once, or maybe if, autonomous trucks are everywhere this will be obsolete. I'm not even sure without autonomous trucking this will take much off the highways. 

Like so many jobs or processes,  the techies see one part, usually the easiest part, to automate. They don't realize that often there's more than meets the eye.

Jeff

 

[daveklepper]

Euclid--   This is so obvious that you really surprise me with a lack of understanding.    By fence, one that does not just discoyurage trsspassers, but actually prevents tresspassing.  And of course no grade crossings.

Withouit such separation, unforeseen incidents occur regularly, and automatic operation will never totally account for them.  My estimate would be reduction of unforeseen incidents from one per-hundred trips to one per-ten-thousand trips.

We are assduming in both cases very careful train assembly, no empty well cars separating  heafvily loadeed cars, etc.

[daveklepper]

Also, self-powered railcarsc require between five and ten times the time for inspections and maintenance than trailing cars.

 

[charlie hebdo]

Euclid

 

 

daveklepper

Yes, absolutely.

1.  The one-car or group of automous cars train is not crewed.  Automatic train oprtation is sinmply never going vto be legal and safe on any railroad with infenced and at-grade right-of way.  Complete isolation  as in rapid transit systems and within land owned by the coirpration that is the customer and/or owner of the railroad are possible, and CN, former IC, western line c definitely fails the separation requirement.

2,   CN shoukld get off its lazy butt and go after the business.  Don't need automatic cars.  Sufficirny business for a regular daily train. 

 

 

 

 

This objection is stated every time autonomous operation is brought up, but it is never explained in a way that justifies it.  Can you please explain, for the first time ever, what a crew will do to overcome the safety issues that you see with autonomous trains? 

 

Tell me exactly why a right of way fence is needed for autonomous trains, but not for crewed trains. 

 

This point seems like an argument put forth by people who simply oppose autonomous running because it threatens labor. 

 

So far at least it does not appear that satisfactory, factual objections have been forthcoming.

[Backshop]

It's so cheap living rent-free in someone's head...

[Euclid]

daveklepper

Euclid--   This is so obvious that you really surprise me with a lack of understanding.    By fence, one that does not just discoyurage trsspassers, but actually prevents tresspassing.  And of course no grade crossings.

Withouit such separation, unforeseen incidents occur regularly, and automatic operation will never totally account for them.  My estimate would be reduction of unforeseen incidents from one per-hundred trips to one per-ten-thousand trips.

We are assduming in both cases very careful train assembly, no empty well cars separating  heafvily loadeed cars, etc.

 

Okay, so a fence will reduce trespassing and thus reduce accidents.  I get that.  Without a fence, and with grade crossings, people and cars will get into the fouling zone of a train and get hit.  It will make no difference whether the train is operated by a human on board or by autonomous running.  In either case, the train simply cannot stop in time. 

 

It has been deemed perfectly okay that a human engineer cannot stop the train in time because it is the responsibility of pedestrians and drivers to yield.  So if people follow the law, there is no need for a fence. 

 

Yet it is not okay that an autonomous system is unable to stop the train in time while it is the responsibility of pedestrians and driver to yield.   

 

Both the engineer and the autonomous system can see trespassers ahead and dynamite the brakes; but it takes the same amount of distance for the train to stop; and often the distance needed to stop will exceed the distance available.

 

So the overall results are the same for a manually operated train as they are with an autonomous train. 

 

Therefore, why is it okay to run trains with human engineers without a fence, but not okay to run autonomous trains without a fence?

 

A fence will indeed save lives, but it will do so equally for either system of operation.  So if we must have a fence for autonomous operation, why shouldn’t we also have a fence for manual operation?

[Overmod]

Those who have been following the tech will be amused, but not particularly surprised, to see that the moron idea of 'autonomous bogies' using the container as a stressed member has been 'reconsidered' so there is now a perimeter frame with corner support and presumably twistlocks.  It seems they are using a physical probe-type sensor to govern the platooning, probably to go through several further rounds of 'reconsideration' before they get something rational.

Strangely they remain entirely ignorant of the possibility that, now that they have a structural frame and a ginormous traction battery/cap system, they can implement the emergency magnetic track-brake system that Erik and I were discussing in a couple of the interminable Bucky emergency-brake-desirability timeless-topic threads.  Program the platoon to fire emergency from 'last car' to 'lead' a few milliseconds apart so that any with 'impaired' brake implementation come into smooth contact and can be decelerated with appropriate modulation.  And no flat wheels.

As with the Weems electric railroad for M&E in the 1890s, the elephant in the room is physical security for the autonomous container loads.  Where there is none for crewed stack trains, imagine the fun with conservatively-programmed and more than a little 'dumb' AI models...

[tree68]

Euclid

This objection is stated every time autonomous operation is brought up, but it is never explained in a way that justifies it.  Can you please explain, for the first time ever, what a crew will do to overcome the safety issues that you see with autonomous trains?    Tell me exactly why a right of way fence is needed for autonomous trains, but not for crewed trains.    This point seems like an argument put forth by people who simply oppose autonomous running because it threatens labor. 

Do a search on "autonomous automobiles" and pay attention to the accident reports.  

Yes, they are sharing the road with everyone else, but not always successfully, and it isn't usually the "other guy's" fault.

Such operation on a single track line will require the ability to conduct meets.  Individual autonomous vehicles will end up in sidings, waiting for opposing traffic.

Out on the plains, you're talking cattle country, with all that entails.  

The autonomous railcar idea has merit, but we aren't there yet, nor will we be for a while.

[BaltACD]

Lets say we have a line for autonomous operations.  Two hundred miles long with 20 customers at each and and 20 customers spaced at locations along the 200 miles.

Each of these 60 customers ship and recieve from all the other customers in carload lots.

How does autonomous schedule, originate, move and deliver these cars?  What support facilities will be required to make this a self sustaining operation?

Transportation is simple when dealing with single events at a time.  When dealing with a multiplicity of events at the same time it begins to become infinately more complex.

[Backshop]

All I will say (I wouldn't want to disappoint Greyhounds) is that it would take more than the revenue from the one route to cover all the R&D costs associated with having autonomous rail cars.

[Overmod]

Actually, most of the "R&D cost" of these things doesn't matter.  There is over $37 million of VC money already in Parallel Systems (probably more now) and ALL the heavy lifting for Level 4 autonomy will be developed by the automobile/truck/aircraft groups with far more market potential than any railroad application.

[Euclid]

tree68

 

 

Euclid

This objection is stated every time autonomous operation is brought up, but it is never explained in a way that justifies it.  Can you please explain, for the first time ever, what a crew will do to overcome the safety issues that you see with autonomous trains?    Tell me exactly why a right of way fence is needed for autonomous trains, but not for crewed trains.    This point seems like an argument put forth by people who simply oppose autonomous running because it threatens labor. 

 

 

Do a search on "autonomous automobiles" and pay attention to the accident reports.  

Yes, they are sharing the road with everyone else, but not always successfully, and it isn't usually the "other guy's" fault.

Such operation on a single track line will require the ability to conduct meets.  Individual autonomous vehicles will end up in sidings, waiting for opposing traffic.

Out on the plains, you're talking cattle country, with all that entails.  

The autonomous railcar idea has merit, but we aren't there yet, nor will we be for a while.

 

I do not necessarily disagree with your points, but they have nothing to do with what you quoted from me, which was in response to Dave Klepper and the always cited “problem” of autonomous trains requiring completely fenced track corridors. 

 

In any case, I do not believe autonomous cars or trucks running on public highways will ever be in full operation unless the government builds separate infrastructure for them so they don’t mix with non-autonomous vehicles on a common roadway. 

 

That may be plausible for commercial trucks because there is an attractive business model with trucking. 

 

In any case, I do expect it to soon begin developing for railroads because their natural guideway of tracks eliminates the most difficult aspect of autonomous running; which is found with cars and trucks needing to be manually guided on plain roadways. 

 

But even with that , railroad advantage, its logistical methods and extensive unpredicted variables, yard terminal operations, monster loose car trains, and all of the operational contingencies will have to be streamlined into ideal, predicable operating patterns.   Otherwise, as it is, that may be a showstopper even for applying autonomous operation to railroads.   But, once all that is sorted out, applying a form of autonomous train operation similar to Rio Tinto may be possible. 

 

But autonomous private automobiles make no sense to me.  And for automated trucks, that vision may well emerge as “trucking” on rail, which brings it closer to the autonomous railcar.   With the single rail line Greyhounds mentioned, it would be interesting to see if we could pencil out a practical system of autonomous running. 

[tree68]

Euclid

I do not necessarily disagree with your points, but they have nothing to do with what you quoted from me, which was in response to Dave Klepper and the always cited “problem” of autonomous trains requiring completely fenced track corridors. 

My point is that autonomous autos do not always recognize problems.  I have little doubt that the same will be true of autonomous railcars.  A 100+ ton railcar (or several) zipping down the line all by itself doesn't need any challenges like trespassers, animals, objects thrown on the tracks, etc.  

[Backshop]

Overmod

Actually, most of the "R&D cost" of these things doesn't matter.  There is over $37 million of VC money already in Parallel Systems (probably more now) and ALL the heavy lifting for Level 4 autonomy will be developed by the automobile/truck/aircraft groups with far more market potential than any railroad application.

 

I'm not talking about the software.  Even just engineering self-propelled railcars with onboard motive power wouldn't be cheap, especially for such a limited use.

[zugmann]

Backshop

I'm not talking about the software.  Even just engineering self-propelled railcars with onboard motive power wouldn't be cheap, especially for such a limited use.

If we're doing self-propelled why not just have the containers go directly to the customer?  Why bother with all this in between stuff?

 

Self-propelled flatcars are like the RedBox of the transportation world. 

[BaltACD]

Self propelled autonomous movement.

Remote controlled movement with genuine locomotives still needs to be perfected.

https://www.progressiverailroading.com/federal_legislation_regulation/news/NTSB-issues-preliminary-report-of-fatal-accident-in-CSX-rail-yard--70333 

[NittanyLion]

Euclid

In any case, I do not believe autonomous cars or trucks running on public highways will ever be in full operation unless the government builds separate infrastructure for them so they don’t mix with non-autonomous vehicles on a common roadway. 

I wouldn't go so far as to say never.  I'm cooler on automation than a lot of people, but we're already pretty close to being able to do autonomous cruise on limited access highways.  The main obstacle won't be (or isn't) the capabilities of the vehicles. Instead it'll be the rate of vehicle turnover in the national fleet.  At some point, every vehicle sold will be capable of autonomous operation and it will take 15 to 20 years after that point before the fleet is entirely capable.  We're looking at a 45 to 50 year horizon there.  That's far off, but not never.

[SD60MAC9500]

.

[SD60MAC9500]

I think all the AV Rail concepts miss the point.. Not only does this not handle 53' containers, and the incompatibility with existing equipment..You'd need to automate the physical plant as well.. It has horrible utility.. What market are they trying to serve? You don't use containers for domestic short haul, containers require expensive multi-million dollar Mi-Jacks... Even side loaders while cheaper are still a cost.. TOFC is the way to go for short haul. In fact autonomous loading/unloading at a Iron Highway type facility would be more ideal. Then you could handle any type of equipment C/C (Container/Chassis), Flatbed, Bulkmatic, DV (Dry Van) Reefer, etc..Oh yeah you can't double-stack these on most routes as they exceed plate H..

The head of strategy at Parallel Systems and myself got into a little back and forth over PM's on Linkedin due to my former comments..

[daveklepper]

Euclid, suppose the tresspasser is a small child?  Suppose the specific clothing or whatever fools the asutomatic train's sensors?  Suppose the human engineer sees that an accident is in the making, even if not definite, just a possibility that the automatic  sensors would never recognize?

And you did not address the unproductive time for maintenance and uinspections.

 

[daveklepper]

I was four years old when I pulled a little red wagom with yellow-cream lettering (Third Avenue Transit colors) just about everywhere I went.  Possibly I was pretending to be a train.  We were crossing the West 85th Steet and Central Park West intersection exiting from the park, no traffic light at the time, one installed about 1945-6, and the driver of a Buick was impatiant and ran over my little red wagon, after he saw the heads  of mom, dad, and me clearing his path. 

[Euclid]

daveklepper

Euclid, suppose the tresspasser is a small child?  Suppose the specific clothing or whatever fools the asutomatic train's sensors?  Suppose the human engineer sees that an accident is in the making, even if not definite, just a possibility that the automatic  sensors would never recognize?

And you did not address the unproductive time for maintenance and uinspections.

 

 

Well let’s just fence the right of way for the manually operated trains that we have today.  It would make everything safer, so why not?  Also, suppose the human engineer fails to see a small child, whereas an autonomous system sensor would have seen the child. 

 

Railroad management seems to be enthused about autonomous operation.  And just like Elon Musk promoting autonomous vehicles as a means of increasing safety, railroad management is promoting autonomous trains as a means to make railroad operations safer.  In either case, the improved safety comes from removing human error. 

[Erik_Mag]

I would think that the autonomous railcar problem would be much simpler than the autonomous truck problem. As OM mentioned, braking for an autonomous railcar should be significantly better than braking action on freight trains, so the cars can move at reasonable speeds and still be able to stop for obstructions. In addition, the increased traffic on the line may discourage trespassers - though the Brightline experience unfortunately may suggest otherwise.

Here in California, the state legislature passed legislation to prohibit driverless trucks on the road, but the legislation was vetoed by the governor (governor and majority of the state legislature are members of the same party, so this wasn't a partisan spat). This would indicate that there isn't insurmountable opposition to autonomous railcars.

[zugmann]

Erik_Mag

As OM mentioned, braking for an autonomous railcar should be significantly better than braking action on freight trains,

I don't know.  Single car means you don't have much wheel-rail surface contact.  Small trains/locals can be a pain to run because of this.  

I don't see the cost savings in having to build maintain 100s of 'locomotives' ( and the neccesary track upgrades for same) to move the same freight as one.  And how do they work in tracks with restricted speed is the required operation?  I guess hav ea guy at the gate with a remote to run them?  Eats into savings... but I'm not a retired MBA so what do I know. 

[ChuckCobleigh]

zugmann

... but I'm not a retired MBA so what do I know. 

Maybe stay at a Holday Inn Express, then?

[tree68]

zugmann

I don't know.  Single car means you don't have much wheel-rail surface contact.  Small trains/locals can be a pain to run because of this.

I'll second that.  It's the stuff of flat wheels...

   

zugmann

I don't see the cost savings in having to build maintain 100s of 'locomotives' ( and the neccesary track upgrades for same) to move the same freight as one.  And how do they work in tracks with restricted speed is the required operation?  I guess have a guy at the gate with a remote to run them?  Eats into savings... but I'm not a retired MBA so what do I know. 

Details, shmetails...