Is route knowledge an outmoded concept?

[AlterEgo]

In-cab signalling only gives you SIGNALLING.
It does not give you safe access points, places of limited clearance, places subject to adhesion problems, etc.

Is there any reason why this wouldn't be technologically possible?

 

[AlexNL]

In-cab signalling only gives you SIGNALLING.
It does not give you safe access points, places of limited clearance, places subject to adhesion problems, etc.

German drivers have a system called EBuLa which provides them with information about the infrastructure which lies ahead, such as speeds and platforms. A system like that could be enhanced to show all the information which drivers need to operate a train, and which can be updated in near real-time.

 

[BRblue]

Not once has anyone mentioned trespassers... and trust me we get loads of those.
If the signaller needs to tell the driver or vice versa route knowledge is essential.
Cars can go around obstacles/persons on the road and you don't find any at 30,000ft but you do on the railway and that is one very good reason to retain route knowledge.

 

[whoosh]

Yes once in cab signalling happens I can't see why route knowledge would be required, it can just calculate the exact breaking curves for the driver and apply them.

Here's a special stop order for New Barnet. Can you accept it with your eight car train?

What restrictions are there regarding platforms at Kings Cross?

You are travelling towards London on the Up Fast taking an empty train to Hornsey Depot, whereabouts will you need a speed reduction on your in-cab signalling? Because if you don't get one then you'll continue on the Up Fast and the Signaller will be wrong routing you.


There's more to route knowledge than you think.

 

[DynamicSpirit]

Yes but trains dont have those systems so at present there is no option other than route knowledge. You can't drive slower on a train as its foggy as it holds everyone else up, and of course lets say there is fog at Uckfield, the train comes through East Croydon 10 minutes late, its now taking a slot that belongs to something else, so holding them up.

Hope this doesn't go too off-topic, but reading that made me wonder: If it's really foggy - for example, too foggy to see where you are - how does route knowledge help you? Are you just expected to deduce where you are because you've been driving for X minutes at Y mph since the last station, or is there something else? And what happens if it's too foggy to see the signals ahead until you're very close to them?

 

[mrmartin]

Is there any reason why this wouldn't be technologically possible?

Of course it is, how come some tube lines can do this? And the DLR can also do this. I've never seen a DLR miss the platform because the rail adhesion wasn't right, despite taking it most days for the past 2-3years.

In cab signalling also gives you very accurate positioning of the train and can easily give target/upcoming speed profiles.

 

[gsnedders]

Of course it is, how come some tube lines can do this? And the DLR can also do this. I've never seen a DLR miss the platform because the rail adhesion wasn't right, despite taking it most days for the past 2-3years.

In cab signalling also gives you very accurate positioning of the train and can easily give target/upcoming speed profiles.

The DLR does have platform overruns because of low adhesion at times, but I don't know how the overrun rate compares with something similar manually driven which is arguably the more interesting question.

 

[Bromley boy]

Hope this doesn't go too off-topic, but reading that made me wonder: If it's really foggy - for example, too foggy to see where you are - how does route knowledge help you? Are you just expected to deduce where you are because you've been driving for X minutes at Y mph since the last station, or is there something else? And what happens if it's too foggy to see the signals ahead until you're very close to them?

You then resort to sounds, and other visual cues such as line curvature. For example, at one of my locations, my braking point from 75mph is the third overbidge after exiting a long tunnel.

This works well whatever the weather conditions. In bright sunlight I can see the bridges. In thick fog, at night, I can hear the bridges passing over me, I can see the line 5 feet ahead of the cab, I can recall that the line begins to curve to the right shortly before the third bridge.

Signals come into it but if you're running on greens you continue at line speed because, barring anything out of course, the worst the next signal will be is a double yellow, with signal spacing commensurate with line speed and braking distance. An out-of-sequence signal aspect means you drop the lot regardless of weather conditions or anything else you can (or can't) see. Once you're at a stand you contact the signaller for further explanation/instructions.

In answer to the general thread, given the Victorian infrastructure in place on much of the UK's railways it's still far cheaper to train and employ human drivers and equip them with the requisite route knowledge to drive over it than it is to reinvent the wheel, at great cost, by changing the infrastructure.

In terms of safety, the existing aincient infrastructure is evidently working extremely well, given we have the safest railway in Europe.

In summary, it ain't broke, so nobody wants to pay to fix it!

 

[dctraindriver]

Hope this doesn't go too off-topic, but reading that made me wonder: If it's really foggy - for example, too foggy to see where you are - how does route knowledge help you? Are you just expected to deduce where you are because you've been driving for X minutes at Y mph since the last station, or is there something else? And what happens if it's too foggy to see the signals ahead until you're very close to them?

You just know. All those journeys you learn and know where you are and to be honest I love it when it's foggy. Keeps you on your toes and shows how good your knowledge is.

When you feel a set of points through your arse and the way the train moves you know instinctively where you are.

 

[theironroad]

You just know. All those journeys you learn and know where you are and to be honest I love it when it's foggy. Keeps you on your toes and shows how good your knowledge is.

When you feel a set of points through your arse and the way the train moves you know instinctively where you are.

Driving in thick fog is always going to be the best test of an individual's route knowledge. It really does focus the mind.

 

[MisterT]

With ERTMS, route knowledge could become obsolete (or: there is only a minimum amount of route knowledge required).
Unfortunately, not entirely, because the last ERTMS Baseline 3 standard doesn't account for stations (even though it was suggested, but they wanted to 'keep the planning area clean').

When I look at the current Dutch signalling system, we already need a limited amount of route knowlegde only due to the speed based signalling system. As the Dutch signalling system is the designed to give every driver 'a fair chance' to drive its train safely, it is completely possible to drive without any route knowledge. The only thing, as always, are the exceptions, and that is exactly what we need to know in the Netherlands for route knowledge.

 

[najaB]

You just know. All those journeys you learn and know where you are and to be honest I love it when it's foggy. Keeps you on your toes and shows how good your knowledge is.

When you feel a set of points through your arse and the way the train moves you know instinctively where you are.

Driving in thick fog is always going to be the best test of an individual's route knowledge. It really does focus the mind.

That's getting back to my original point: is this the best we can do in 2017 where safety is concerned (and capacity)? A system that keeps you on you toes? What happens when your knowledge isn't as good as it should be? And drivers should be focused, but I'd rather that instead of lower-order brain functions (memory) which can be automated, they were focused on the higher-order reasoning functions that can't yet be.

 

[AlterEgo]

Driving in thick fog is always going to be the best test of an individual's route knowledge. It really does focus the mind.

In some other transport industries though, you don't need to know the layout of the infrastructure in fog to be able to conduct yourself through it safely.

I'm interested in why it might not be technologically possible to have this on a display in the cab.

 

[dctraindriver]

That's getting back to my original point: is this the best we can do in 2017 where safety is concerned (and capacity)? A system that keeps you on you toes? What happens when your knowledge isn't as good as it should be? And drivers should be focused, but I'd rather that instead of lower-order brain functions (memory) which can be automated, they were focused on the higher-order reasoning functions that can't yet be.

It's 2017, it works just fine. Autonomous isn't there. If a Tesla can kill its driver by smashing into another vehicle it shows an element of risk be it human or digital factor.

Redesign the Victorian rail system and who knows maybe it'll work.

And I want to be kept on my toes. It's what I'm paid well for.

 

[glbotu]

I mean, to everyone's "a computer can fail", computers fail less than people (and that includes everything up to and including the systems upon which the computer is built - power etc). We intrinsically don't trust computers, because we have an inherent belief that human traits are somehow better (re-inforced by millenia of that being true, the human brain is a pretty great computer), but if I asked 10000 train drivers to perfectly recall a route, versus 10000 computers, I can guarantee the computer would recall them better, because fundamentally speaking, that's the computer's sole function. The computer driver could easily take a few extra readings and figure out what the adhesion is like and how to behave accordingly, because that is, essentially what the squishy driver is doing.

The squishy driver has other important functions, like reproduction, drinking beer and what's for tea, which the computer driver doesn't have to worry about. All the computer has to worry about is where am I, what are the conditions and how much traction/brake should I apply. The squishy driver should absolutely still be there for dealing with trespass issues, with problem solving and with interfacing with the public when things go wrong.

Now you can argue about slippery slopes and "if we don't need the driver to drive the train, then why do we need them at all", but really that's an argument about a whole other squishy populace, management and politicians, which I'd argue should also be replaced with computers, much to everyone's joy............

 

[Llanigraham]

It does not give you safe access points, places of limited clearance, places subject to adhesion problems, etc.

But a combination of GPS, balises (for where GPS reception leaves something to be desired) and specialist maps does.

And there is no-where to display that information.

 

[MichaelAMW]

Here's a special stop order for New Barnet. Can you accept it with your eight car train?

What restrictions are there regarding platforms at Kings Cross?

You are travelling towards London on the Up Fast taking an empty train to Hornsey Depot, whereabouts will you need a speed reduction on your in-cab signalling? Because if you don't get one then you'll continue on the Up Fast and the Signaller will be wrong routing you.


There's more to route knowledge than you think.

They don't strike me as good examples as they are all just pieces of information out of the sectional appendix. In this brave new world of less route knowledge all of those things would become the responsibility of the IT systems or the signaller. If he/she forgets to turn you into the depot at Hornsey then that's (now) their problem, even if currently it might be yours.

 

[Johncleesefan]

That's getting back to my original point: is this the best we can do in 2017 where safety is concerned (and capacity)? A system that keeps you on you toes? What happens when your knowledge isn't as good as it should be? And drivers should be focused, but I'd rather that instead of lower-order brain functions (memory) which can be automated, they were focused on the higher-order reasoning functions that can't yet be.

.... And this is one of the reasons train drivers are paid the way they are, and why it is a tough job to get into, and also why you are expected to maintain competence in your knowledge.

Yes its 2017, and management have made damn sure we get pop quizzed about the lot

 

[Johncleesefan]

They don't strike me as good examples as they are all just pieces of information out of the sectional appendix. In this brave new world of less route knowledge all of those things would become the responsibility of the IT systems or the signaller. If he/she forgets to turn you into the depot at Hornsey then that's (now) their problem, even if currently it might be yours.

Sorry I'm not buying that, put the knowledge in the head of the person driving to make the decisions about wrong routing. Not some geek in control.

So you get wrong routed, now you have to make a wrong direction movement to return to the junction. Something else the driver probably knows nothing about since a computer seems to be spoon feeding him. Does this sound safer to you?

 

[DynamicSpirit]

You then resort to sounds, and other visual cues such as line curvature. For example, at one of my locations, my braking point from 75mph is the third overbidge after exiting a long tunnel.

This works well whatever the weather conditions. In bright sunlight I can see the bridges. In thick fog, at night, I can hear the bridges passing over me, I can see the line 5 feet ahead of the cab, I can recall that the line begins to curve to the right shortly before the third bridge.

Thanks, that's interesting.

 

[driver_m]

Simplest answer to that is that a computer won't really know what's an unsuitable location should a major problem occur. One thing we are tested on, with our simulator, is to be given a fire scenario. A computer would likely just stop. But in our scenario, if that's running through the tunnel heading into New Street, then we know thanks to our route knowledge that the tunnel is very much an unsuitable location. Even ETRMS couldn't make that decision. We can.

 

[HSTEd]

There's nothing like a Mk1 brain & eyeballs for route knowledge. DAS & C-DAS require electricity

The trains control systems require electricity.
If the train loses power it will eventually even lose the air brake, at which point the train si totally out of control.

and GPS (which the yanks can switch off) or a GSM-R signal to correlate all of the information, any part of that chain goes down, DAS/C-DAS stops working,

Tell the driver to not move the train if any of that kit isn't working?
Additionally the US Government effectively cannot disable GPS any more - not without causing ridiculous numbers of aviation accidents in the US itself, it has already cashed the monetary saving from decommisioning most of its fixed navigation aids in favour of GPS.

Simplest answer to that is that a computer won't really know what's an unsuitable location should a major problem occur. One thing we are tested on, with our simulator, is to be given a fire scenario. A computer would likely just stop. But in our scenario, if that's running through the tunnel heading into New Street, then we know thanks to our route knowledge that the tunnel is very much an unsuitable location. Even ETRMS couldn't make that decision. We can.

You could simply have someone mark locations as "unsuitable for fire emergency stop" or similar in the giant database we are probably proposing be loaded into the train's computer system.

So you get wrong routed, now you have to make a wrong direction movement to return to the junction. Something else the driver probably knows nothing about since a computer seems to be spoon feeding him. Does this sound safer to you?

Why would he know nothing?
Since he know longer has to expend many hours memorising every feature of every route he works - he can spend all his training hours practicing for exactly that sort of unusual situation.

 

[InOban]

And why shouldn't a computer be programmed with data on unsuitable locations?

And of course the train will always need someone on board who can drive it in an emergency:max speed, say, 10 mph.

But, fog - that's why we have AWS/TPWS.

Speed limits - ditto. The problem with human drivers is that their mind wanders, they start thinking about something else, and suddenly they've forgotten to slow down. (Morpeth, anyone?) And as many of the skills a driver used to need - getting the best out of a steam engine, driving an unbraked freight train - become redundant, they are more and more likely to lose concentration. (This is a well-known Safety risk in many occupations).

The core section of Thameslink will be driven under ATO because it is considered that human drivers would not be able to deliver the 24tph reliably.

 

[driver_m]

You could simply have someone mark locations as "unsuitable for fire emergency stop" or similar in the giant database we are probably proposing be loaded into the train's computer system.

This sounds like replacing a perfectly good car with a Bugatti Veyron and then only ever using it to go to the shop that's a quarter of a mile up the road. Doable, but a waste of money and effort.

 

[Johncleesefan]

The trains control systems require electricity.
If the train loses power it will eventually even lose the air brake, at which point the train si totally out of control.


Tell the driver to not move the train if any of that kit isn't working?
Additionally the US Government effectively cannot disable GPS any more - not without causing ridiculous numbers of aviation accidents in the US itself, it has already cashed the monetary saving from decommisioning most of its fixed navigation aids in favour of GPS.


You could simply have someone mark locations as "unsuitable for fire emergency stop" or similar in the giant database we are probably proposing be loaded into the train's computer system.


Why would he know nothing?
Since he know longer has to expend many hours memorising every feature of every route he works - he can spend all his training hours practicing for exactly that sort of unusual situation.

The thing is though is that route knowledge, local instructions for sure and rules all crossover to a degree, if you take their route knowledge away then implementing a rule can be difficult, and when a local instruction supercedes the rule book, would a driver be expected to know that or would control tell them that too. I'm sorry but I wouldn't feel a confident driver without my knowledge, that is our job

 

[driver_m]

I can just imagine a computerised train slipping and then tying itself in knots over whether it's a suitable place to stop. Imagine that happening dropping into Lime St. Computer then suddenly has the bright idea of trying to get onto actual Lime St because its more suitable. Cue calls of 'dont be stupid, that wouldn't happen' 'oh..er......damn.....might need to think this one through a bit more....'



No doubt I'll be labelled a luddite train driver by some on here for coming out with my point of view, but we do have a few Professor Frinks (off the Simpsons) on here who have some fantasy ideas that don't belong in the real world.

 

[dctraindriver]

Well the nerds on here have made the decision that us drivers aren't needed. That's cool. Can you have a word with the Govt. to make the necessary changes to implement all what you've suggested. Hopefully they'll have this done in the next few weeks for a few quid and I can have a lie in ??????

 

[cjmillsnun]

I fully expect to get flamed for this one, so here goes...

In 2017 is it right and correct that we depend on human memory - in the form of route knowledge - as a primary safety system?

It used to be the case that pilots flew by eye and had to have route knowledge as they flew from landmark to landmark. This placed severe operational limitations on the airlines and so visual navigation was replaced by ground-based navigational aids and charts, and now even physical charts are disappearing as they are replaced by electronic flight bags.

This means that the modern airline pilot's minds are focused on the things that they do best - actually flying the plane, rather than having to be full of facts, and they aren't limited to only flying routes that they have a mental map of.

In contrast, on the railway we still depend on route knowledge as the first and most important layer of the protections that keep the railway safe. And it does do a good job of keeping the railway safe, this isn't a criticism of the men and women who do a good job of keeping the system working. It's also not an attempt to do drivers out of a job - there's already [thread=139956]a thread[/thread] about driverless trains and I agree that we're a few (many?) years away from that reality.

But there *is* an operational downside to depending on route knowledge - how many times do we hear versions of "Can't use that diversionary route, they don't have route knowledge"? Or trains have to get cancelled because a driver who signs the planned route isn't available for one reason or another.

So my question is this: we accept that it's perfectly safe for a driver to be conducted along a route that (s)he doesn't sign. Why does that conductor have to be a human being? While train positioning hasn't yet reached the point where it can reliably be used to control the train directly (one of the issues with driverless trains), I'm sure that it's accurate enough to be used as input to a virtual route conductor that could provide the same information as a human conductor - e.g. distance to signals, recommended speeds, braking points.

A system such as this would be invaluable in times of disruption, but could also be useful for regular day-to-day use as well as it would give the driver a virtual secondman if it was a near-realtime system that was updated with the WONs, ESRs, etc.

I'm interested in your thoughts?

(I've got my flame-retardant underwear on! )

Airline pilots rarely fly the plane except during takeoff and landing. The plane flies itself with the pilots monitoring systems.

 

[Roast Veg]

I can see route knowledge ceasing to be a requirement (or having a lower standard) over certain sections for ECS or light engine moves in the future, but not for passenger or freight services. There's a reasonable call for this, as in certain places a diversionary route around a blockade would be very useful for getting stock and drivers in and out of the day's diagram - a relatively common time when route knowledge causes a short area of disruption to be magnified. Another example might be transferring stock into/out of/between depots sooner by an unusual path that might not be available otherwise.

As long as the start and end location fall under the existing route knowledge of the driver, more live information could substitute gaps in it along the way. Appropriate risk mitigations can be made, such as reduced speed or no changing of ends en route (though I don't know what would happen in an emergency situation there).

 

[glbotu]

This sounds like replacing a perfectly good car with a Bugatti Veyron and then only ever using it to go to the shop that's a quarter of a mile up the road. Doable, but a waste of money and effort.

Not really. Storage space is cheap. I currently have the entire Sectional Appendix on my machine. It's about 300MB in total size. I own about 4TB of storage space over a few hard drives (that's a factor of 13333x as much storage space as is required), whose total cost is about £300. Compared to the cost of a single train (£4-5 million these days), £300 is pittance.

The real cost would be in the computer system to actually make the decisions and the sensors used to get the information (for adhesion etc).

I can just imagine a computerised train slipping and then tying itself in knots over whether it's a suitable place to stop. Imagine that happening dropping into Lime St. Computer then suddenly has the bright idea of trying to get onto actual Lime St because its more suitable. Cue calls of 'dont be stupid, that wouldn't happen' 'oh..er......damn.....might need to think this one through a bit more....'

I mean, that wouldn't happen unless someone programmed it to do that. I mean, that's not how trains work either...... :P Computers don't have personalities, they don't just decide to do something. When a computer does something unexpected, that's usually because it has been programmed badly (and therefore doesn't expect you to use it the way you have used it). Now that's often a risk when looking at non-safety-critical software (like most people find themselves using every day), because the testing just isn't as rigorous. But if you're creating a train control system that puts people's lives at risk, you make damn well sure it works (because if it doesn't the lawyers will make you poor). Deciding if its a suitable stopping location (to use your example), is a simple integer comparison. The train would know how big it is and know the length of the platform and go "yes, I can stop here". It's also less likely to do it than a driver who's been driving a 4 car all week getting an 8 car trying to stop at a 4 car platform, because the computer instantly knows how long the train is and keeps that data in memory.

The point where computers fall down and people are better is when the volume of input is too great. Adhesion, not a problem - some sensors could pick up things like atmospheric pressure and air humidity and apply an appropriate adhesion model. Trespassers, problem - the computer has to constantly scan for stuff it doesn't expect, which means constantly comparing data to known data etc etc - that's hard. Our brains are great at this, because that's a huge part of how animals function.

If you want to make a case for keeping train drivers in a job, making arguments based on stuff that isn't true doesn't help anyone. If something simple like comparing bits of data or figuring out how to get somewhere is your argument, computers are great at that, way, way better than people. Using sensors is a thing computers are good at too, just ask a fighter pilot who's flown a Typhoon. They have stacks of sensors that both report to the pilot and are used to modify the pilot's inputs so that the plane stays in the air. But how does a computer respond to a traction supply failure, or a component defect, or a trespasser, that's where you earn your keep as a person, and until we hit the singularity, that's where you're going to keep doing it.