Euclid zardoz Again I reassert that the other issues of operating--mainly weather and grade crossings--are not taken into account. What type of response is needed for grade crossings that cannot be provided by self-driving trains?
zardoz Again I reassert that the other issues of operating--mainly weather and grade crossings--are not taken into account.
What type of response is needed for grade crossings that cannot be provided by self-driving trains?
Stopping and inspecting for damage of the rail equipment. Normally the train isn't damaged or damaged slightly. However, parts of what they struck might get hung up underneath or cause some damage. You could program the train to stop, but then it's sitting, possibly blocking that and other crossings, while waiting for someone to come check it out. There's also the part about going back to check what you hit. Mainly to see if you need an ambulence or a hearse. I suppose you could just have it automatically call for police, fire and ambulance services when an impact is registered.
I'm reminded of a JGK column in Trains about the subject. His recommendation was to install detectors at crossings. The computer could be programmed to stop the train if an object was detected. Or, and he was for this, program the train only to stop if the object was large enough to possibly damage the train, otherwise plow right through it. I read that as let the train plow through an automobile stuck on the tracks, but stop for a dump truck.
Jeff
JPS1 tree68 On single track, it would be back to the days of "fleeting." What is fleeting?
tree68 On single track, it would be back to the days of "fleeting."
What is fleeting?
Moving trains in one direction for a set time period, and then the other. Kind of like when a two lane highway is down to one lane during reconstruction. They allow so many vehicles in one direction, with or without a pilot car, and then so many vehicles in the opposite direction. The railroad does this on double/2MT lines when one track is closed or blocked. They (RR) doesn't use a pilot car
On single track lines it's one way to move multiple trains in either direction that can't fit in sidings.
jeffhergert Euclid zardoz Again I reassert that the other issues of operating--mainly weather and grade crossings--are not taken into account. What type of response is needed for grade crossings that cannot be provided by self-driving trains? Stopping and inspecting for damage of the rail equipment. Normally the train isn't damaged or damaged slightly. However, parts of what they struck might get hung up underneath or cause some damage. You could program the train to stop, but then it's sitting, possibly blocking that and other crossings, while waiting for someone to come check it out. There's also the part about going back to check what you hit. Mainly to see if you need an ambulence or a hearse. I suppose you could just have it automatically call for police, fire and ambulance services when an impact is registered. I'm reminded of a JGK column in Trains about the subject. His recommendation was to install detectors at crossings. The computer could be programmed to stop the train if an object was detected. Or, and he was for this, program the train only to stop if the object was large enough to possibly damage the train, otherwise plow right through it. I read that as let the train plow through an automobile stuck on the tracks, but stop for a dump truck. Jeff
I expect that ultimately, all of our self-driving cars will be interlocked with all railroad grade crossings to prevent any cars from fouling a grade crossing during the approach of a train.
In the meantime, Rio Tinto self-driving trains have a system that scans grade crossings and reacts if they are fouled or maybe about to be fouled with approaching vehicles.
I would expect that they and other users of self-driving trains will have them shepherded by ground crews in highway trucks to deal with any contingencies including inspecting locomotives and equipment after grade crossing collisions occur, moving trains to clear crossings, etc. I also expect that at any time, a human operator would be able to take over for the automatic self-driving system and manually run the train as though he/she were in the locomotive cab and looking out the windows.
jeffhergertThere's also the part about going back to check what you hit. Mainly to see if you need an ambulence or a hearse. I suppose you could just have it automatically call for police, fire and ambulance services when an impact is registered.
I imagine that management will not care about the victims--after all, it was their fault for getting hit; how dare they interfere with my railroad's profit margin!
Sometimes it is difficult for me to reconcile the inhumane and cold-heartedness of corporate and government alleged humans; you know--those that value profit over the lives their quest for yet more money grinds underneath. Everyone for themselves!
I could never understand why a following movement is not allowed in a DTC block without a flagman on the rear. What difference does it make if a signal tells you that a block is occupied or if your DTC authority states that it is occupied?
Mark Vinski
zardozWhich would work--sort of. If ambulance services are required, time is usually a factor. When the automated call for police, fire, coroner, or ambulance goes out, who will direct them to where needed? Who will escort them along the ROW? Do emergency services sit and wait for escort as a victim bleeds out?
OnStar already gives us the location of the vehicle, that the airbags have deployed, and if there has been contact with the occupants of the vehicle. With PTC GPS, the location of the head end should be readily available.
Around here it's single track territory with few trains. A call to the emergency number (which dispatch has) would likely have train movement status available to responders even before they reached the area of the scene.
This is dark territory - the train in question will have an EC-1 (or whatever CN calls them, coming soon) for exclusive occupancy.
Larry 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...
mvlandsw I could never understand why a following movement is not allowed in a DTC block without a flagman on the rear. What difference does it make if a signal tells you that a block is occupied or if your DTC authority states that it is occupied? Mark Vinski
DTC was used in DARK territory. NO Signals to protect the rear end of the train. With the person needed to protect the rear of the train thousands of feet from the rear end of the train when it stops - the ABILITY to provide flag protection is SERIOUSLY COMPROMISED.
DTC blocks were 10 or more miles in length. The only way a train could handle such a Occupied Block would be at Restricted Speed for the entirety of the block.
When operating with signal indication - signal spacing is on the order of 2 to 3 miles between signals. Getting a Restricted Proceed signal indication identifies that the rear of the preceeding train will be encountered PRIOR to the next signal.
Operating at Restricted Speed is mentally taxing for train and engine crews - searching for all the conditions that restricted speed requires stopping within 1/2 the range of vision of - there are more conditions to be protected against than just a train ahead.
Never too old to have a happy childhood!
Unlike the concept of self-driving road vehicles, I do not see any practical impediments to self-driving trains. The major impediment to self-driving road vehicles is that they need to be largely defensive. They need to be able to sense any sort of threat for collision all along their route. This sensing must not only be focused on the foul zone of the road right of way, but also must be focused on a broad area outside of their right of way. They must search outside of the right of way in order to sense any type of threat developing there which might encroach on their right of way as they approach.
However, trains have their own dedicated, private right of way. Also, their large mass and robust construction make them less vulnerable to damage or derailment from encroaching vehicles, machinery, or other wayside hazards. And unlike self-driving vehicles, self-driving trains cannot react or stop fast enough to avoid encroachments, whereas self-driving vehicles can. So, detecting threats outside of the railroad right of way is less necessary than vehicles detecting threats beyond the roadway.
Euclid, your comment is correct assuming excellent maintenanc of all rolling stock, track, and control equipment, absolue control of freight loading procedures, complete grade separtion with zero grade crossings, completely fenced RoW with some form of electrical or electronic protection against invasion of the RoW.
Today, automatic train operqtion is a reality. At this moment there are "people movers" at airports that meet these conditions and are not manned. There are also industrial freight railroads in remote areas. The PATCO rapid-transit line (legally an interurban!), Philadelphia - Camden - Lindewald, is largely automated, still with an single operator for security, and both the L and 7 subway lines in New York City are close in degree of automation under normal operting conditions. The 7 is a bit more complex with both express and local service, with some train identification and switching thus involved. Eventurally, the entire subway system will be opreated similiarly.
Perhaps in the far furure the North American freight network will meet the conditions outlined. In the far future.
daveklepperEuclid, your comment is correct assuming excellent maintenanc of all rolling stock, track, and control equipment, absolue control of freight loading procedures, complete grade separtion with zero grade crossings, completely fenced RoW with some form of electrical or electronic protection against invasion of the RoW.
Why do you assume that these extra protections are necessary with self-driving trains, but not necessary with manually opperated trains? Trains cannot stop for these obstructions no matter whether they are self-driving or manually operated by an engineer in the cab.
EuclidWhy do you assume that these extra protections are necessary with self-driving trains, but not necessary with manually opperated trains? Trains cannot stop for these obstructions no matter whether they are self-driving or manually operated by an engineer in the cab.
A lot depends on the peripheral vision of the automated system. A live engineer might catch the movement out of the corner of his eye of someone approaching a crossing at a high rate of speed and attempt to react (subject to speed and weight variables).
A live engineer can react if the obstruction is no longer an obstruction. The logic stream for this could get interesting for the "electronic engineer."
Loading? See the "why not to stand near the tracks" thread. Unless there is a near constant means to verify that loads are within a permissible template, things could go south in a moment.
How would self driving trains handle fallen trees like those on these videos?
https://www.youtube.com/watch?v=u7QCnv0Qpfg
https://www.youtube.com/watch?v=V1xeMZo6lBg
1. Obviously, the RoW would be wide enoigh and planting controlled to prevent falling trees from ever being a problem.
2, By extra protection, the fencing must be such to prevent access or have access detected. Seeing people invading RoWs is a daily. even hourly, occurance on most railroads. Horn blowing works almost all the time. Wh will see and blow the horn on an automated train without a operator?
Electroliner 1935How would self driving trains handle fallen trees like those on these videos?
Most of the first video involved conifers sans large limbs - not hard to bust through.
The tree in the Amtrak video looked to be a little more substantial.
We've had instances where parts of the tree get under the plow/pilot, where they could do damage.
And as I rode our line yesterday I saw some freshly cut trees that would have done substantial damage. We're talking deciduous 8"-12" in diameter, and sticking out about handrail height over the tracks. MOW found them before our trains, but that's why we carry chain saws on our trains.
Here in the Adirondack forest we can only cut them back so they don't interfere with our operation. You'll see plenty of examples in plain view, many of which are, as I said, substantial.
Clearing the ROW enough to prevent such occurances may be problematic, f'rinstance if there are 100' trees just outside the usual 100' ROW...
Electroliner 1935 How would self driving trains handle fallen trees like those on these videos? https://www.youtube.com/watch?v=u7QCnv0Qpfg https://www.youtube.com/watch?v=V1xeMZo6lBg
It would detect something as large as a fallen tree and stop the train. If it were a big enough tree to cause a derailment, and it was it was not visible until the train was too close to stop, it would derail the train; just as would be the case with a manually operated train.
If it were a serries of snow laden trees bent down into interference with the train, the system could be overridden. But generally, I don't see smashing through bent over trees as a practice utilized by an modern, automated railroad. They would run a special machine specially built to clear the trees.
daveklepper 1. Obviously, the RoW would be wide enoigh and planting controlled to prevent falling trees from ever being a problem.
That's funny. They don't do that now because it would cost money. I doubt they'll care in the future. The ROW is getting so overgrown in places you can't see wayside signals until you're on top of them. (Of course with PTC or an automated train, there eventually won't be wayside signals.) They only trim or remove trees when they have to. Usually either after complaints are turned in or a storm drops trees on top of the tracks or too close to the ROW. (I heard a train over the radio say they hit a hanging limb and it may have done some damage to the cab. My conductor said his last trip they reported the limb after striking it. He said hit his sun shade on the side of the cab.)
They haven't done much meaningful vegatation control since they stopped maintaining communication and signal pole lines. I actually think they like that the ROW has grown over because it's a barrier for sound and site between the railroad and the NIMBY element.
The biggest thing an automomous train will have to deal with when going over a tree is when it snags an air hose or cut (uncoupling) lever. It's going to be hard for it to recouple itself.
Jeff, are you saying that there will not be a robot at one end of each car which will be programmed to take care of any possible bad situation (pulled drawbar excepted)? Perhaps there will also be a knuckle in the robot's compartment?
It seems rhat the proponents of non-manned trains are not familiar with the possible difficulties entailed in getting over the road.
Johnny
jeffhergert daveklepper 1. Obviously, the RoW would be wide enoigh and planting controlled to prevent falling trees from ever being a problem. That's funny. They don't do that now because it would cost money. I doubt they'll care in the future. The ROW is getting so overgrown in places you can't see wayside signals until you're on top of them. (Of course with PTC or an automated train, there eventually won't be wayside signals.) They only trim or remove trees when they have to. Usually either after complaints are turned in or a storm drops trees on top of the tracks or too close to the ROW. (I heard a train over the radio say they hit a hanging limb and it may have done some damage to the cab. My conductor said his last trip they reported the limb after striking it. He said hit his sun shade on the side of the cab.) They haven't done much meaningful vegatation control since they stopped maintaining communication and signal pole lines. I actually think they like that the ROW has grown over because it's a barrier for sound and site between the railroad and the NIMBY element. The biggest thing an automomous train will have to deal with when going over a tree is when it snags an air hose or cut (uncoupling) lever. It's going to be hard for it to recouple itself. Jeff
PSR doesn't allow for trivial vegation clearance. Especially clearing vegitation on property adjacent to the right of way and can fall blocking the right of way. I have my doubts if weed eradication activities are taking place these days either.
Deggesty Jeff, are you saying that there will not be a robot at one end of each car which will be programmed to take care of any possible bad situation (pulled drawbar excepted)? Perhaps there will also be a knuckle in the robot's compartment? It seems rhat the proponents of non-manned trains are not familiar with the possible difficulties entailed in getting over the road.
Self-driving trains will scan the right of way for possible hazards, but that requirement will not be nearly as stringent it will be with self-driving highway vehicles.
Here is an article about Rio Tinto driverless trains. The first video gets into some of the operational details. I am sure that U.S. driverless train technology will include the trains being shepherded by ground personnel in highway trucks to take care of any manual chores such as pulling pins, coupling air hoses, replacing broken knuckles, inspecting trains, etc. They won’t be actually following a specific train, but will be stationed at points in a network and be ready to assist any train that needs it.
https://www.riotinto.com/ourcommitment/spotlight-18130_25692.aspx
BaltACD I have my doubts if weed eradication activities are taking place these days either.
Watched a sprayer on the Deshler railcam this past summer.
Most lines in heavy growth areas get Roundup ever 3 years or so. Also a tree cutter comes along and wacks down the trees within about 20 feet of the track every few years.
On another point, most train accidents are due to crew fatigue (up goes the din of anger!). The computer won't go to sleep and accident rates might shrink.
petitnjMost lines in heavy growth areas get Roundup ever 3 years or so. Also a tree cutter comes along and wacks down the trees within about 20 feet of the track every few years. On another point, most train accidents are due to crew fatigue (up goes the din of anger!). The computer won't go to sleep and accident rates might shrink.
Or Microsoft takes revenge and a CTRL-ALT-Delete reboot requires a technician to be sent to the middle of nowhere at Zero dark thirty. ie. the computer went to sleep.
Euclid Deggesty Jeff, are you saying that there will not be a robot at one end of each car which will be programmed to take care of any possible bad situation (pulled drawbar excepted)? Perhaps there will also be a knuckle in the robot's compartment? It seems rhat the proponents of non-manned trains are not familiar with the possible difficulties entailed in getting over the road. Self-driving trains will scan the right of way for possible hazards, but that requirement will not be nearly as stringent it will be with self-driving highway vehicles. Here is an article about Rio Tinto driverless trains. The first video gets into some of the operational details. I am sure that U.S. driverless train technology will include the trains being shepherded by ground personnel in highway trucks to take care of any manual chores such as pulling pins, coupling air hoses, replacing broken knuckles, inspecting trains, etc. They won’t be actually following a specific train, but will be stationed at points in a network and be ready to assist any train that needs it. https://www.riotinto.com/ourcommitment/spotlight-18130_25692.aspx
I'm sure with one or no person crew trains they'll have a utility man covering a specific zone. The failed contract I mentioned earlier called for a "master conductor" to do that. The thing is there will be times when they won't have enough, others when they have too many. The bean counters will only see the times they have too many and want to cut positions and lengthen the territories they cover.
While I have no doubt automated trains could be run here, I don't think you'll see truly autonomous trains. I do think you'll see one person trains eventually. Rio Tinto has a couple of things going for it that railroads in this country, for the most part, don't. The trains are unit trains, either loaded or empty. Those are the easiest to automate. While LEADER and Trip Optimizer can run all kind of trains, the mixed manifests are the ones it seems to have the most trouble with, especially the large ones. (To be fair, it's the same for human engineers, too.)
The other thing is they (RT) operate in mostly sparsely populated areas. Not much interaction with the public. While parts of the US west are that way, others are not. It may not, for the forseeable future, be 'politically' possible for a fully autonomous train. (When UP started using RCO in Des Moines, a TV news interviewer said she was told that while RCO could be used on main tracks outside of yards, it wasn't politically 'feasible' yet.) We shall see.
https://www.freightwaves.com/news/rio-tintos-autonomous-trains-cant-work-in-north-america-yet
The last I knew RT wasn't fully automated yet, but they are working towards that goal. I would like to know, from an objective source - not the salesman, not the company - how things are really working out.
What is the solution of the rare but deadly vehicle running in to train ? Remember the Cal Z accident a few years ago. Automation might not even know that such collision happened even if a vehicle ran into the second loco. Leaking fuel tank might spread diesel for miles.
EDIT
Worse still ---------- Vehicle strikes a volitile tank car that starts spilling onto tracks and is on fire with fire spreading to other cars. The expression " fire train coming " leads to new heights of a movie disaster.
blue streak 1What is the solution of the rare but deadly vehicle running into train?
Closest thing to an 'answer' is the kind of barrier that I described earlier. This of course does not address the 'people who turn "by mistake" onto the railroad from the road, or collisions where trucks run downhill and wreck track lining or collide with consists, etc. There are ways to minimize these but they involve either outlandish line-separation fencing/instrumentation cost or placing active elements in the loading gage of operation -- if those elements stick or drag they put trains in danger themselves.
Note that the crossing camera system would give very prompt and reasonably accurate notice of any 'late hit' collisions, as well as identifying any out-of-gage anomalies in the loads, if properly designed and 'sensor-fused'. The problems involved with generating 'blips' for identifying and tracking these events do not have the problem encountered with, say, Aegis tracking radar; an icon representing multiple detected 'anomalies' can be detected, sent to local law enforcement, and passed 'intact' to railroad police and other agencies as necessary in a short period of time without meaningful impairment of represented information.
I would note that the cost of instrumenting even key high-speed crossings with cameras may be comparable to that of equipping key train consists with ECP brakes... the issue though is that 'doing less' may be politically inadvisable, if not unthinkable, and there are no effective lower-cost alternatives that I have seen so far, with little lack of lookin'.
blue streak 1Worse still ---------- Vehicle strikes a volitile tank car that starts spilling onto tracks and is on fire with fire spreading to other cars. The expression " fire train coming " leads to new heights of a movie disaster.
Well, in defense on that one - with today's "land barges" it could happen to a manned train as well. Trains of 11,000 feet are not uncommon, and one or two curves, a hill, or any other such view block would make such an incident near the rear of the train virtually invisible to the crew.
Odds are their first indication would be a brake hose melting and dumping the air.
I’ve given a lot of thought to this over the years and what I believe really out to be done in the railroad industry is the development of self-driving CARS.
It is the loose-car business, especially boxcars, that has really lost out over the last 40-50 years and loose car traffic is continuing to decline.
Railroads have an extremely tough time trying to compete with trucks on the pick-up and delivery at shippers’ docks. Therefore, to bring this traffic back, could there be any way to automate it?
I do not believe that this would be technologically impossible but there would be some logistical obstacles to overcome.
How could the cars be powered? They would only need to be self-powered in order to get them from the shipper’s dock to the yard. After that a locomotive could take over.
Driverless freight cars would also necessitate a completely fenced off right of way free of grade crossings and that would be expensive.
But if the railroads could pull this off, it could well pay off in the long run.
Something to at least think about, though.
Regards,
Fred M. Cain
I can’t believe they’re even thinking about this. I have a tire pressure sensor in my car that has basically always been wrong. The mechanics at the dealer say sure we’ll re-set it but ignore it because those things don‘t work. So we want trains and autos dependent on all kinds of sensors? Crazy.
Lithonia OperatorSo we want trains and autos dependent on all kinds of sensors? Crazy.
Boeing did.
_____________
"A stranger's just a friend you ain't met yet." --- Dave Gardner
Paul of Covington Lithonia Operator So we want trains and autos dependent on all kinds of sensors? Crazy. Boeing did.
Lithonia Operator So we want trains and autos dependent on all kinds of sensors? Crazy.
Now THERE’S an endorsement!
Our community is FREE to join. To participate you must either login or register for an account.