Train Driving and Automation

[DerekC]

The reason for that is probably that there is significantly more research and development into automatic cars. Maybe it's an easier problem to solve as all the technology needed is contained within the car and currently the best solutions we have for heavy rail is an interface between train and track that doesn't scale very well over long distances. Self driving cars are already on the roads and the financial motivation to push on with them is huge, probably far bigger than the rail network in fact. The cost of transporting freight via road would plummet overnight and vehicles could go much further in a 24 hour period as they wouldn't be slave to the dreaded tachograph.

Longer term, 50-100 years if automated cars really take off then a far smaller percentage of the population will own a car, especially in urban areas. Almost everything will become a taxi and it will reduce the need for car parking space in central areas of major towns and cities. Imagine if all the car parking space in zone 1 of london could be repurposed for something else?

I don't think automated cars are at all an easier problem - after all, there are two degrees of freedom instead of one and the environment they operate in is much more varied and chaotic. Genuinely "go anywhere" autonomous cars are further away than is generally assumed. However I accept the point that the market is much, much bigger and the investment will be in proportion. We think of rail as big and important, yet the number of cabs to be fitted with ATO across the whole GB network is a few thousand - not enough for the consumer technology big boys to even think about getting out of bed for!

I am not sure that any of us can predict how autonomous cars will be used. the suggestions about reducing private car ownership are very sensible, but will it actually happen like that?
--- old post above --- --- new post below ---

Every so often someone - usually a new member like yourself DerekC, brings up the subject of automation/driverless trains etc. They then don't like the responses being given and the whole thing descends into a 'I know best contest' . You've heard the responses from drivers and others, that isn't going to change so let's just leave at that - yes ? [emoji4]

I was trying hard not to have an "I know best" contest because I don't. What I did want is some ideas about the future role of the train driver in an automated world. However it doesn't sound as if I am going to get any because the consensus on this forum is that main line trains will be manually driven for the foreseeable future. You may all be right for a reason we haven't discussed, which is that the current main line railway structure doesn't support change - ETCS seems likely to take another 100 years at the current rate of progress, and without that ATO can't happen anyway.

 

[HLE]

Every so often someone - usually a new member like yourself DerekC, brings up the subject of automation/driverless trains etc. They then don't like the responses being given and the whole thing descends into a 'I know best contest' . You've heard the responses from drivers and others, that isn't going to change so let's just leave at that - yes ? [emoji4]

Must admit the feeling I got was that the OP has a problem with drivers in particular. The other day the OP started a thread saying ERMTS/ETCS should be implemented straightaway and now DerekC starts a thread about automatic trains.

DerekC, just because you claim to have spent 25 years commissioning ATO systems does not mean that it will be implemented anytime soon.

Full implementation of ATO and ERTMS will cost billions, probably trillions.....and for what? To potentially save a few minutes and have a bit more capacity on the network .....whilst still having to have a fully qualified driver up front to take over when things go wrong. Now is it me or....

So for a massive initial capital outlay (and where is the money coming from by the way?), where will any 'real' savings come from? Your post reads that you are 'clued up' in how the business world works these days .....yet what business would spend trillions on a project of which no real savings would be made?

 

[driver_m]

It always turns into an 'all or nothing' debate this. Isn't the obvious answer that it will be a blend of semi automated driving where we will be always needed because of the potential for failures, but with functions added incrementally to help out. The job has changed bit by bit over the years. Some of us do everything near enough manually without any aid and some of us have aids such as speed set, station call etc. The railway is seemingly incapable of doing huge changeover programmes like the ROC's replacing existing signalboxes but good at smaller ones like TPWS which like it or loathe it has made a difference.

 

[ComUtoR]

1) The technology as such isn't the problem (ETCS plus ATO should work, although the GSM-R comms is still a weak link)

Why is GSMR the weak link ? I thought that the introduction of GSMR is because it is essential for ETCS and ATO. Considering that it is the main onboard equipment and handles all the data, how and why is it the weak link ?

 

[Cherry_Picker]

I don't think automated cars are at all an easier problem - after all, there are two degrees of freedom instead of one and the environment they operate in is much more varied and chaotic. Genuinely "go anywhere" autonomous cars are further away than is generally assumed.

I disagree. Cars are so much more responsive than trains, they are lighter, grippier and can be controlled on line of sight which fits in well with current sensor technology. Google have completed about 1.5 million miles of autonomous driving on public roads in the USA. Just last month the university of Nevada successfully used their system to negotiate a 1500 mile road trip in Mexico, Ford have built a fake town in Michigan to test their system. Tesla have said they have a system which works. Volvo have said they a system that works. GM are working on one. We are probably no more than five years away from self driving cars being on the open market. Trains will take much longer to happen.

 

[Radedamer]

I'm sure the technology for driverless trains will benefit from what is learnt by Google and the car manufacturers. Obviously not all of that will be applicable and there will be additional problems to be overcome. I doubt if adhesion will be one though, as this is surely akin to antilock brakes? But the application of the technology will depend on economic factors. Whereas driverless cars will become a commercial success if they appeal to the general population as a reduced stress, easier alternative to driving, the take up of self-driving trains will be more akin to that of self-driving buses and lorries: is the initial cost justified by the savings in wages, increased driving hours, reduced fuel and maintenance from more moderated driving style, etc? While these are all applicable to road vehicles, with rail there is probably less to be gained in terms of fuel and maintenance or even increased hours, so that just leaves the wage bill. In the more distant future it might be possible that ultra-high-speed passenger services could be run at speeds which the human hand and eye just can't cope with, but this would obviously be dependent on engine, signalling and even rail technology.
--- old post above --- --- new post below ---

I disagree. Cars are so much more responsive than trains, they are lighter, grippier and can be controlled on line of sight which fits in well with current sensor technology. Google have completed about 1.5 million miles of autonomous driving on public roads in the USA. Just last month the university of Nevada successfully used their system to negotiate a 1500 mile road trip in Mexico, Ford have built a fake town in Michigan to test their system. Tesla have said they have a system which works. Volvo have said they a system that works. GM are working on one. We are probably no more than five years away from self driving cars being on the open market. Trains will take much longer to happen.

Slight disagreement. One of the advantages of robocars over meatware cars is that they do not depend solely on line of sight.

 

[Cherry_Picker]

No, but GPS aside the sensors on an autonomous car (LIDAR, RADAR and camera) are all 'reactionary' sensors which work within distances comparable to line of sight and light vehicles connected to a rough road surface by sticky tyres meshes well with that technology. Autonomous cars (and lorries) will happen soon.

Trains will happen also, like I said in my first post on this thread, it's inevitable. I just happen to think it could be decades before existing lines are fitted with trackside equipment needed to make it work and rolling stock modified to work too. I'm sure all new trains going forward will be 'future proofed' to take automation into consideration but it just isn't going to happen on anywhere near as an accelerated scale as road vehicles. The economics of it just don't work as well.

 

[underbank]

Surely rather than using tech for auto/driverless trains, it should be used to remove/replace/improve the signalling system. It's a little antiquated to rely on very restrictive lights/semaphores to keep trains apart (even if supported by the modern warning and auto brake systems).

There's no reason why two trains can't run far closer together. If they are "linked" electronically via wifi/satellite or whatever, they could travel far closer, the following train automatically slowing at the same pace at the front train, thus effectively stopping/slowing together as if they were one long joined train.

That would free up a lot of extra capacity which is one of the major issues on the network.

The old fashioned idea of signals being used to keep adequate separation should have long been consigned to the rubbish heap of history.

 

[notadriver]

Surely rather than using tech for auto/driverless trains, it should be used to remove/replace/improve the signalling system. It's a little antiquated to rely on very restrictive lights/semaphores to keep trains apart (even if supported by the modern warning and auto brake systems).

There's no reason why two trains can't run far closer together. If they are "linked" electronically via wifi/satellite or whatever, they could travel far closer, the following train automatically slowing at the same pace at the front train, thus effectively stopping/slowing together as if they were one long joined train.

That would free up a lot of extra capacity which is one of the major issues on the network.

The old fashioned idea of signals being used to keep adequate separation should have long been consigned to the rubbish heap of history.

It's called ERTMS and it's already being used in parts of Europe. It requires massive investment.

 

[Cherry_Picker]

It's being used on the Cambrian line in Wales too.

 

[47802]

No, but GPS aside the sensors on an autonomous car (LIDAR, RADAR and camera) are all 'reactionary' sensors which work within distances comparable to line of sight and light vehicles connected to a rough road surface by sticky tyres meshes well with that technology. Autonomous cars (and lorries) will happen soon.

Trains will happen also, like I said in my first post on this thread, it's inevitable. I just happen to think it could be decades before existing lines are fitted with trackside equipment needed to make it work and rolling stock modified to work too. I'm sure all new trains going forward will be 'future proofed' to take automation into consideration but it just isn't going to happen on anywhere near as an accelerated scale as road vehicles. The economics of it just don't work as well.

Well I think you are right to large degree Train wise, Road wise how long it will be before driverless road vehicles are readily available, especially one that work down UK narrow country roads, I think remains to be seen some say 5 years some say quite a bit longer. But when it does happen I expect it will happen quickly and as its already been pointed out by some experts driverless vehicles will work better if they are all driverless, and not getting cut up by boy racer so for those who still want to drive like me it might be tough luck, and from an employment viewpoint its going to have vastly more impact than any potential loss of Train Drivers jobs.

 

[D1009]

It's called ERTMS and it's already being used in parts of Europe. It requires massive investment.

ERTMS is happening already in the UK. What is not happening is moving block (level 3) on the main line, anywhere in Europe if I'm not mistaken.

 

[Dave1987]

I think driver_m post is the most accurate. In the future I think technology will be used to assist drivers and reduce the workload and fatigue of driving trains much like automation does in the airline industry with pilots.

DerickC it's not that people on this forum are totally anti automation and refuse to accept it, it's just this has been discussed to death on previous threads in some fairly heated discussions and the same answers always come up. You mention how you believe ATO will make train diving more efficient and time savings but I genuinely don't agree with you there. You are discussing the future role of a train driver as if it's a definite that ATO will be fully rolled out everywhere within the next few decades which is probably not the case. ERTMS will be rolled out as it provides up to 30% capacity improvements on the current infrastructure as it is but that still needs a driver. And ERTMS is extremely expensive and is not proven to be more reliable or as safe as TCB.

 

[muz379]

Anyhow, this thread doesn't seem to be going the way I hoped (i.e. a discussion about the future role of the main line driver). One last time - are there some thoughts about that out there?

Ive not previously commented by my frank and hoenst thoughts are that the mainline railway will go through a period of DOO with a secondary member of staff on board in many places carrying out revenue and customer service duties , then after that we will go through an obvious transition period of with the roll out of ATO systems and then when all of the issues have been ironed out I personally think you will end up with only one member of "traincrew" who spends most of their time dedicated to customer service duties but who is able to take control of the train in the case of a catastrophic system failure .Even if that just means being trained in how to bring it to a stand safely

Im in my late 20's now and I honestly dont think that this is going to happen before I retire from the railway . We have parts of the network with semaphores , track circuits and token block sections that are not even going to be updated in the next 15 years . We have sections where electrification is a very distant hope . I dont even think we are going to see a widespread roll out of DOO for another 10 years .

 

[thenorthern]

I think we are a long way off mainline driverless trains as many things such as onboard faults, speed restrictions, obstructions on the line and derailments would need to be factored in.

Considering some of the signaling in used in this country dates from before the 1930s (I think) Network Rail wouldn't want to be bothered with the cost of installing the equipment for driverless trains.

 

[DerekC]

It always turns into an 'all or nothing' debate this. Isn't the obvious answer that it will be a blend of semi automated driving where we will be always needed because of the potential for failures, but with functions added incrementally to help out. The job has changed bit by bit over the years. Some of us do everything near enough manually without any aid and some of us have aids such as speed set, station call etc. The railway is seemingly incapable of doing huge changeover programmes like the ROC's replacing existing signalboxes but good at smaller ones like TPWS which like it or loathe it has made a difference.

I think that's right operationally - what makes sense is a kind of auto-pilot for trains, so that the driver is still in control and can intervene when needed (that's what Thameslink is getting). The problem is that the way railway safety rules work, it has to sit over a full blown ATP system - TPWS isn't considered good enough - so that has to be ERTMS, which is a big technology change for the railway.

 

[notadriver]

I find it interesting that on the newest high speed lines in the world, manual driving rather than ATO is the rule despite it being easy to implement.

 

[DerekC]

Why is GSMR the weak link ? I thought that the introduction of GSMR is because it is essential for ETCS and ATO. Considering that it is the main onboard equipment and handles all the data, how and why is it the weak link ?

The problem is first that GSM-R is old technology and the support will get more and more expensive and second that it is "circuit switched" so each train using ERTMS has to have a dedicated channel to the control centre, plus a channel for each voice call which is being made at the same time. There just aren't enough channels to cope with a busy piece of main line. The answer is simple in terms of technology - change over to a modern "packet switched" system like 3G or 4G, but the railway has just spent loads of money on GSM-R. And ERTMS needs to be modified too.
--- old post above --- --- new post below ---

Ive not previously commented by my frank and hoenst thoughts are that the mainline railway will go through a period of DOO with a secondary member of staff on board in many places carrying out revenue and customer service duties , then after that we will go through an obvious transition period of with the roll out of ATO systems and then when all of the issues have been ironed out I personally think you will end up with only one member of "traincrew" who spends most of their time dedicated to customer service duties but who is able to take control of the train in the case of a catastrophic system failure .Even if that just means being trained in how to bring it to a stand safely

Im in my late 20's now and I honestly dont think that this is going to happen before I retire from the railway . We have parts of the network with semaphores , track circuits and token block sections that are not even going to be updated in the next 15 years . We have sections where electrification is a very distant hope . I dont even think we are going to see a widespread roll out of DOO for another 10 years .

Some good thoughts - the idea of a "train steward" who is mainly customer service but can drive when needed is just the way systems like DLR work. but I can't see it being realistic in a main line commuter situation until failures hardly ever happen, which is a very long way away. By the way, systems like ERTMS are very good at bringing a train to a stand safely - it's getting it going again which is the problem!
--- old post above --- --- new post below ---

I find it interesting that on the newest high speed lines in the world, manual driving rather than ATO is the rule despite it being easy to implement.

Note that HS2 is planning for ATO. I wonder why the difference?

 

[muz379]

I think we are a long way off mainline driverless trains as many things such as onboard faults, speed restrictions, obstructions on the line and derailments would need to be factored in.

Considering some of the signaling in used in this country dates from before the 1930s (I think) Network Rail wouldn't want to be bothered with the cost of installing the equipment for driverless trains.

Faults are really one of the big things standing in the way .

Emergency or temporary speed restrictions would pose no challenge because these can be controlled remotely , if anything it would speed things up when an emergency speed restrictions is needed because there would be no need for someone to go and put an ESR indicator up anyway , meaning the ESR could be implemented with immediate effect .

obstructions would also not be a big deal because you can have technology that detects these much like technology that automatically applies brakes in a car that senses an impending impact . The only thing that would present a major challenge is looking for trespassers , something signalers frequently ask drivers to do , given the fluid nature of the surroundings of the line this would not be easy for technology to do . But maybe an increased roll out of CCTV could be a work around for this meaning a signaler can examine the line himself remotely .

 

[DerekC]

obstructions would also not be a big deal because you can have technology that detects these much like technology that automatically applies brakes in a car that senses an impending impact . The only thing that would present a major challenge is looking for trespassers , something signalers frequently ask drivers to do , given the fluid nature of the surroundings of the line this would not be easy for technology to do . But maybe an increased roll out of CCTV could be a work around for this meaning a signaler can examine the line himself remotely .

The problem with obstruction detection on the front of the train is that train braking distances exceed by a long way the distance you can sense ahead - the devices for cars work on line of sight (as somebody else said). On the other hand the same is true for drivers, so maybe we would be no worse off than now. High speed image analysis coupled to forward facing CCTV may be capable of detecting trespassers. It could certainly detect track workers in PPE. And if the train had high speed data communications with the control centre it could upload anything suspicious for review off-line.

 

[hulabaloo]

I'm completely in favour of ATO as it will mean the drivers can dedicate their time and expertise to the more important issue of having the air-conditioning running at the right temperature!

 

[Cherry_Picker]

I'm completely in favour of ATO as it will mean the drivers can dedicate their time and expertise to the more important issue of having the air-conditioning running at the right temperature!

That will always be impossible. Even if you think it's too hot, the person sitting two seats over will insist it's too cold. Can't please all of the people all of the time.

 

[carriageline]

The problem with obstruction detection on the front of the train is that train braking distances exceed by a long way the distance you can sense ahead - the devices for cars work on line of sight (as somebody else said). On the other hand the same is true for drivers, so maybe we would be no worse off than now. High speed image analysis coupled to forward facing CCTV may be capable of detecting trespassers. It could certainly detect track workers in PPE. And if the train had high speed data communications with the control centre it could upload anything suspicious for review off-line.

Like you've said, a true driverless train is very far off, and for the foreseeable future we will have what Thameslink are getting, with driver 'monitored' ATO. I'm personally quite interested in ATO/ERTMS, and seeing the way the train drives, especially on restrictive aspects is very interesting, as the train knows EXACTLY where to stop for the red, and so will brake late and on time. ERTMS can also be used to split signal section, such as the Core will be getting in a couple of years.

People say it only makes a small difference, but I've sat watching a 2 minute delay soon turn into a 3000 delay, there's no such thing as a small difference. Getting trains stopped and moving again is where the difference will be, trains following each other now on yellows is where delays explode.

 

[SPADTrap]

That will always be impossible. Even if you think it's too hot, the person sitting two seats over will insist it's too cold. Can't please all of the people all of the time.

Usually just me in my cab anyway

 

[Dave1987]

DerekC it seems you are trying to convince people that widespread ATO is just around the corner. Like has been said previously the costs involved are astronomical for very little gain. Some parts of our railway network are fairly up to date but other parts are still using the same technology invented in the early part of the 20th century. NR said originally that it wanted all signalling to be controlled from Regional operation centres by 2017. That plan has been pushed back so many times the NR signallers I know doubt it will ever happen! Considering that the current major projects like the GW, TP and MML electrification are all way way over budget and taking far far longer than originally forecast I really don't see ATO even being considered until at least the second half of this century.

 

[FordFocus]

Quite an interesting article on ERTMS here: http://www.railengineer.uk/2015/08/28/ertms-a-reality-check

Although the topic is on ATO but seen as ERTMS has been mentioned.... I'm unsure of the benefits of ERTMS to the GWML. The lineside signalling will still be there for 378s, freight, HSTs, HEX and other rolling stock. The only advantage I can see is that the ATP system will be upgraded from the old BR 90s system. Is there an increase of linespeed from 125mph to 140mph for the IEP stock?

Seems to be a waste of money overlaying level 2 if your having the same signalling sections.

ERTMS on the ECML is different as it's removed signals to save on maintenance costs and the majority of rolling stock having the in cab equipment fitted with a lot of units been brand new.

GSM-R requires an update too as it's still under 2G technology, see the above article.


Back to ATO. It has issues on the LUL as previously said, it can't cope with low adhesion and if the wheels lock up, the train is more than likely to get 'lost' because it can't calculate speed and distance and throw the brakes in. I can't see ATO been required on the mainline with the exception of a small number of routes in London because of the very limited benefit of retrofitted infrastructure and trains for it at a huge cost. If you put ATO out on the mainline, any additional capacity can be achieved with additional signals, shortening sections but upgraded from 3 aspect to 4 aspect.

 

[notadriver]

In Paris on the RER A and I think the B line too, rather than use ATO they've used SACEM which is basically an ATP system that allows proceed on sight driving at low speeds.

The RER A is claimed to be the busiest commuter line in the world outside Japan.

 

[miami]

I wonder how the economics of automated hgv driving will affect the railway - presumably freight will suffer.

For capacity purposes why not terminate high speed trains on the edge and use the permanent way to slow automated high frequency coaches. Just 1 coach every 10 seconds on a dedicated route otherwise earmarked for hs2 would allow 10000 passengers an hour in ridiculous comfort and space, and run happily for 10 miles into the city in just 10 minutes, maybe less, and would be far cheaper.

For the suburban lines, said coaches (or something smaller) could offer a PRT style service, turn up and go with a 2 minute wait, going direct to your destination. End to end journey times far less than with trains even if the peak speed is lower.

For timescales I'm thinking 2066. Sure we'll all be retired or dead, but my youngest will be in his 50s, how will he commute (assuming commuting is still going, which is unlikely)

 

[notadriver]

There has been talk of converting railway to roadway before. Do you mean conventionally driven coaches ?

 

[najaB]

For capacity purposes why not terminate high speed trains on the edge and use the permanent way to slow automated high frequency coaches. Just 1 coach every 10 seconds on a dedicated route otherwise earmarked for hs2 would allow 10000 passengers an hour in ridiculous comfort and space, and run happily for 10 miles into the city in just 10 minutes, maybe less, and would be far cheaper.

Two things to consider... Firstly, all you've done is shift the capacity issues slightly further out from the city centre, any issues with conflicting movements, etc. will still exist. Second, you have removed one of the key benefits of rail travelling directly from city centre to city centre. If you're going to have to take a bus out from the city centre, why not take a bus to the airport?