southern442
Established Member
Apologies, I was very hot and bothered when I wrote that.
I found your knolwedge of the Southern system quite accurate.
-
Our new ticketing site is now live! Using either this or the original site (both powered by TrainSplit) helps support the running of the forum with every ticket purchase! Find out more and ask any questions/give us feedback in this thread!
Driverless trains report
- Thread starter theageofthetra
- Start date
- Status
Apologies, I was very hot and bothered when I wrote that.
I found your knolwedge of the Southern system quite accurate.
Sponsor Post - registered members do not see these adverts; click here to register, or click here to log in
R
RailUK Forums
Apologies, I was very hot and bothered when I wrote that.
No problem, matter closed and over and done ands bits referring to it in my post edited out o plastic pig follower (and I can remember when someone famously got form 1'd for calling a 442 that once too often)
Last edited:
Does it mention obstacle recognition? I don't know how these trains use obstacle detection technology, but it could well just be in a dispatch context. The DLR, in it's relatively closed environment (the same as most or all of those grade 4 systems listed) zooms around happily all day with no way for an obstruction on the line to be detected automatically. Even if it weren't so, obstacle detection is a rather simpler business than obstacle recognition (which is why Network Rail are struggling with the former on level crossings - if the latter was sufficiently advanced, I'm sure they'd be considering it!).
carriageline
Established Member
- Joined
- 11 Jan 2012
- Messages
- 1,897
The sad thing is (well, it happens far too often..), the MCB-OD equipment works fine elsewhere, yet we have somehow tried to adapt it for our needs and surprise surprise, no comprende.
Driverless trains outside of the sterile DLR wil not happen nor will pilotless planes. The technology already exists to fly completely autonomously. its call a drone. However when things go wrong and the data link is down, having a MK1 computer and eyeball on the job can save the day. Hudson river for example. Computers cant replicate what happened there with a successful outcome and undoubtedly it was a very experienced human that saved the day.
You could argue that on a high speed line such as HS1 and HS2, travelling above 140mph, the driver is irrelevant in reacting to situations that a microprocessor and sensors couldn't do faster. That said, he would still be able to spot the cows or trespassers on the lineside or the OLE on the other line sagging low.
Old street tunnel incursion is another good example, spotted by the previous driver reported and IMHO saved a major incident So you might as well keep a highly skilled person sat there up front with some extra safety aids to prevent human error and help the workload but not to the point the pilot or driver dips out of the loop and therefore becomes useless in a fast developing situation And dont get me started on driverless cars
You could argue that on a high speed line such as HS1 and HS2, travelling above 140mph, the driver is irrelevant in reacting to situations that a microprocessor and sensors couldn't do faster. That said, he would still be able to spot the cows or trespassers on the lineside or the OLE on the other line sagging low.
Old street tunnel incursion is another good example, spotted by the previous driver reported and IMHO saved a major incident So you might as well keep a highly skilled person sat there up front with some extra safety aids to prevent human error and help the workload but not to the point the pilot or driver dips out of the loop and therefore becomes useless in a fast developing situation And dont get me started on driverless cars
Pertinent to this debate
http://www.telegraph.co.uk/culture/.../When-will-CGI-actors-replace-human-ones.html
http://www.telegraph.co.uk/culture/.../When-will-CGI-actors-replace-human-ones.html
Yes, this is currently at an early stage of development. It is an essential development to provide additional capacity on the busy routes, and has been, in part, brought forward due to persistent militancy amongst railway workers.
--- old post above --- --- new post below ---
That is the perception, but with shorter response times it is the human factor that becomes the weak point.
--- old post above --- --- new post below ---
I don't know what you mean by "doubt London Underground", but a closed metro system with common stock is completely different to the intricacies of the National network. It's not viable, at least in the forseeable future. And we've had loads of threads about this already.
--- old post above --- --- new post below ---
They already exist, but it's not viable for the national network.
If the report was that significant, it would be more widely available, than just one paywall site. And if it was, a Google search would find it!
Investigations in to such operations on the national network are the core of the activity. Further, it is only a news article on the activity within the industry, but a industrial report.
--- old post above --- --- new post below ---
Just like the 1960s space program, and other more recent projects.
These sort of systems have to comply with rigorous engineering standards, and already exist. They are somewhat beyond the more widely evident approaches in software development.
This is your opinion and not a fact.
A key problem is that the human can not respond fast enough at high speeds, so it makes little difference.
This happened in France in the past few weeks with human at the controls of a TGV; the operator had not noticed that he had hit a cyclist - impact detectors would have helped and would be integrated into the train management system:
https://www.youtube.com/watch?v=p34SClQtVY8
http://www.dailymail.co.uk/news/art...ist-crushed-25-MILES-accident-took-place.html
--- old post above --- --- new post below ---
The control staff do not need to based in the UK; it is feasible, as in other industries, to have backup or primary control handled elsewhere in the event of a 'problem' with UK operations.
--- old post above --- --- new post below ---
Deployment of ERTMS starts in 2016/2017. The activities are focused on a start in CP6, they do not need to wait for the entire network to be migrated.
Johnuk123
Established Member
- Joined
- 19 Mar 2012
- Messages
- 2,800
Like the driverless lorries to be trialled on British roads next year driverless trains will obviously be here sooner or later.
Like all new technology you always get scare stories, in the 70's the greens said that north sea oil drilling would be a disaster and wouldn't be worth bothering with.
Like all new technology you always get scare stories, in the 70's the greens said that north sea oil drilling would be a disaster and wouldn't be worth bothering with.
Is that based on a thorough knowledge of the driver's duties and responsibilities? At least HSTEd, as much as I don't think much of whatever goes on in this wonderful theoretical world would stand up to the realities of the railway environment, can back up his thoughts and proposals with a thorough and reasoned (even if some of the big words are made up
On those realities - I think we're some way away from being able to examine the line remotely. The driver's presence and thorough route knowledge counts for a great deal - a good number of broken rails, for example, are detected by a driver, before it deteriorates too badly, reporting a rough ride (and distinguishing that from the various bits of track that are known to always produce a bit of a kick). Defects on passing trains - would an on-board computer be able to (within a reasonable budget) detect a shifted or otherwise unsafe load? Assuming that there won't be a sudden influx of hi-tech freight stock, would a computer even be able to 'drive' a heavy freight train on a challenging route - still a task requiring a fair amount of skill, I understand.
As an aside, isolating the power supply isn't (currently - no pun intended - at least) an effective way of getting a train stopped quickly - reference to stories of electric trains coasting twenty miles or more into Euston proves that!
A good portion of the network is already mapped out in far greater detail than a human can provide through initiatives such as the NMT, data collection, instrumentation, etc.
--- old post above --- --- new post below ---
So you are going to replace all the old trains as well!
Yes, eventually, though the 455s will possibly be around into their 60s.
Can you provide a source for this, please?
The rest of your post is going over old ground and shows a lack of understanding of the realities of the operation of our railway as things stand. One day, most of what you suggest will surely be possible. Not yet though.
Studies and evaluations are under way now; implementation targeted to start in CP6.
--- old post above --- --- new post below ---
Was that the system or the data at the route cause?
Either way, sloppy system design and implementation is unacceptable.
--- old post above --- --- new post below ---
Correct, but it forms part of the platform in the proposed systems.
--- old post above --- --- new post below ---
http://www.railwaygazette.com/news/...verless-metro-train-arrives-in-hong-kong.html
--- old post above --- --- new post below ---
CNR Changchun-developed Domestic First "Driverless" Subway Train Made an Official Debut:
- 23/06/2014
http://english.jl.gov.cn/News/GeneralNews/201406/t20140623_1685822.html
--- old post above --- --- new post below ---
Heavy rail product is already on the market, with a five to seven year lead time from inception to product, the required product technology is already in the pipeline.
--- old post above --- --- new post below ---
When an industry evolves those with existing experience are not necessarily relevant to the future. There are plenty of examples of this. It does not mean that those in the industry are not relevant, rather that they can be reliant on the assumptions that their knowledge is wholly relevant going forward.
Of ATO on a couple of short sections of Network Rail infrastructure (no surprises there) or fully driverless trains? No plans for the latter that I'm aware of at this stage
Either way, if we can't get that right, or ERTMS (which still hasn't really proven itself), or obstacle detection on level crossings - what chance is there with something that supposedly removes the need for human intervention altogether?
Is it? A product that will be able to cope with the realities of our railway system as opposed to the sterile environment of the underground lines that have been converted to driverless operation so far?
Why is experience not important? Evolution might change the way that things are done, but it doesn't usually change the underlying issues and challenges that need to be understood. This fantasy world is clearly brilliant in theory, but the 'experts' in this thread seem unable to answer - meaningfully - some of the questions being posed by those of us who have to deal with these things every day. I'm sure we've all seen inexperienced outsiders - in any industry - come in with a fantastic idea of how things should work, but absolutely no experience of the realities - and, inevitably, their radical proposals prove completely impractical. So yes, I think that there'll always be a place for experience.
Let's assume you're right, and obstacle recognition is impractical to implement, isn't obstacle detection enough for the train captain model proposed earlier? Most of the time the train captain could have customer service duties, but when an obstacle is detected on the track, the train would notify the captain and stop just short of the obstacle. The train captain would then be able to decide how to deal with it, which may include slowly proceeding past the obstacle then returning to automated operation.
--- old post above --- --- new post below ---
I don't see why a computer couldn't be developed that could make that sort of decision, but either way it's misleading to look at specific examples. There are plenty of cases where human error has lead to an aircraft crashing, so the real question is on balance, which is more likely to make a mistake, a human or a computer?
--- old post above --- --- new post below ---
Experience is important, but that experience can be built into a machine and the useful experience becomes how you operate, maintain and upgrade that machine. How useful are hand weaving skills in modern fabric production?
So are you an expert in machine learning or artificial intelligence? I've yet to see any rail expert explain why a computer is incapable of recognising whether the track is clear or not, or indeed any problem that can't be solved with suitably sophisticated software and a member of staff on board (but not necessarily looking out of the front window at all times).
Last edited:
Yes, I agree that it might work! Others have gleefully leapt on the opportunity to point out that drivers' wages and status could be cut as a result, and that's the part that I object to - there'll still be a need for a fully competent (rules, traction and route) driver in that model. There's also the potential problem of the obstacle detection equipment regularly detecting obstacles that aren't a problem, and causing spurious brake applications as a result. After all, if it can't identify obstacles, how does it know whether an object is foul of the line or is actually well clear but directly ahead because the train is on a curve? Not impossible to overcome, I'm sure, but also not as simple as it might seem.
I can't disagree with that. The difference is that you're suggesting building the existing experience into the machine, rather than an entirely theory-based design that completely ignores the realities of the job!
I'm not, and I don't claim that computers will never be able to do what you suggest. I do think that it's a long way off though, and that there'll be a need for a fully competent driver, even if not restricted to the cab in normal operation, for a long time even beyond that.
I agree that the first implementations of ATO on the mainline would need human supervision.
In that case, the ideal way to run it would be keeping the driver and keeping them in the cab but also giving them the safety-related duties of the guard, particularly with respect to self-dispatching when the train handling the driving aspect would allow the driver/captain/guard to observe the platform-train interface safety with their hand on an emergency stop button. All customer-facing duties (ticket inspections, enquiries etc) would then be handled by dedicated customer service staff on inter-regional and inter-city trains, or by remote intercom to the operator's call centre on commuter routes.
This would be much simpler than you think, I can think of two methods which would be quite useful.
Cruise missiles were navigating themselves using terrain contour matching back in the late 1960s. Recycle that concept (which western military services have now left behind in favour of GPS) and you'd instantly have a useful system for recognising what obstacles are in the train's path. The fact that the rails are steel and therefore the best possible radar reflectors will be a great help!
The second method would essentially be a rehashing of the road surface reflectors which allow a driver to detect a kangaroo crossing the road at night by just seeing that the retro-reflecting light got interrupted. Reflectors attached to electrification masts, station platforms, lineside fences, other trains and so on would allow a UV emitter to determine whether any detected objects are nearer to the train than objects with proper reflectors on them.
Carefull or they will tell you that you cannot use obsolete technolgy from the past in the future next. Best just to stick fingers in ears and go la la la it cant be done and be happy.
carriageline
Established Member
- Joined
- 11 Jan 2012
- Messages
- 1,897
Crying out loud. Surprise surprise with this forum, the same suspects spouting the same rubbish.
No one has said it can't be done chip, we all know the technology is capable of moving and stopping a train. In 50 or so years when they want ATO who knows what technology will hold. I'm guessing it will be a different world, If we look to 50 years in the past, and look at the computers we have now, the internet, smartphones. I don't think anyone (well not the main posters in this thread anyway) are seriously doubting the technology.
What I doubt (and so does Tom by the look of it) is how useless it will be not having any staff on the train (a bloke in a control center watching a bank of screens? Lol) for a whole myriad of reasons.
I'm not sure what you do chip, so don't mean to be patronising (I'm sure you work on the railway though, aswell as your other half) but people outside the industry don't realise what we deal with day in day out, and how useful it is to have a member of staff who's willing to get their hands dirty.
One example I will use, and I don't think even HSTEd will have an answer for. With these adverse weather conditions which are apparently going to get more frequent. We all know trees often come down and block the line all over the place. Of course, a train running ATO software will stop in time for that, but what next? I have known drivers to jump down and saw the tree down, or help shift it.
If no one was on board the train, the MOM/pway/S&T would no doubt be busy dealing with other incidents in the area. You could end up with a lot of trains stood, with the ground staff running around like a blue ass fly trying to chop it all up.
Sent from my iPhone using Tapatalk
No one has said it can't be done chip, we all know the technology is capable of moving and stopping a train. In 50 or so years when they want ATO who knows what technology will hold. I'm guessing it will be a different world, If we look to 50 years in the past, and look at the computers we have now, the internet, smartphones. I don't think anyone (well not the main posters in this thread anyway) are seriously doubting the technology.
What I doubt (and so does Tom by the look of it) is how useless it will be not having any staff on the train (a bloke in a control center watching a bank of screens? Lol) for a whole myriad of reasons.
I'm not sure what you do chip, so don't mean to be patronising (I'm sure you work on the railway though, aswell as your other half) but people outside the industry don't realise what we deal with day in day out, and how useful it is to have a member of staff who's willing to get their hands dirty.
One example I will use, and I don't think even HSTEd will have an answer for. With these adverse weather conditions which are apparently going to get more frequent. We all know trees often come down and block the line all over the place. Of course, a train running ATO software will stop in time for that, but what next? I have known drivers to jump down and saw the tree down, or help shift it.
If no one was on board the train, the MOM/pway/S&T would no doubt be busy dealing with other incidents in the area. You could end up with a lot of trains stood, with the ground staff running around like a blue ass fly trying to chop it all up.
Sent from my iPhone using Tapatalk
There have been lots of people on this thread who have said it cant be done for this that or the other reasons - go back and read them.
Ive never said they shouldnt have anyone on board either as that would be rather daft to do and would never get passed out either.
What I have done is given decent enough reasons about object detection and how it could be implemented and all these points have either not been answered or have been told that nope, it wont work, a human will be better and is the only option.
But thanks for being coindescending and telling me I am spouting the same old rubbish even though I have proposed solutions to the issues. Bet you feel dead happy dont you you smug *******.
carriageline
Established Member
- Joined
- 11 Jan 2012
- Messages
- 1,897
Sorry chip, the surprise surprise bit wasn't to you, more to the other posters (literally, spouting the same crap!) I have missed since I was last on here! (I did try to make that point by doing another paragraph!), again sorry!
Whilst I agree people have said the technology isn't possible, they haven't exactly been the best posts apart from (as you say it) the lalala it won't happen posts, which frankly I have ignored. The serious posters in this thread have accepted that, but arguing a different point.
I know you have said there would be a member of staff on board (or at least should!), I was just reiterating the point Tom was trying to make about lack of staff, and the point I have been trying to make from the start.
Sent from my iPhone using Tapatalk
Whilst I agree people have said the technology isn't possible, they haven't exactly been the best posts apart from (as you say it) the lalala it won't happen posts, which frankly I have ignored. The serious posters in this thread have accepted that, but arguing a different point.
I know you have said there would be a member of staff on board (or at least should!), I was just reiterating the point Tom was trying to make about lack of staff, and the point I have been trying to make from the start.
Sent from my iPhone using Tapatalk
I won't lead this down the road of the DOO debate, but the other aspects sound fair. That said, if the driver's got to be there anyway, I'm sure that there's a lot to be said for him retaining driving duties (unless capacity constraints really do require ATO, as per Thameslink), supervised by the computer, and thus being more alert, and ready to respond in an emergency, than if the computer's doing all the work.
I'm sure that neither would be impossible, but I'm not convinced that it'd be as easy as it sounds! How would it cope with diverging routes, for example? Cruise missiles have the advantage of being able to navigate themselves, rather than following a path that they can't change - in time, perhaps the technology will be able to relate the layout that it sees ahead of it to the route set by the signalling system. In the meantime, how does it differentiate between an obstacle on a diverging route that it's going to take through a complex junction and between a train approaching safely from that diverging route when it's going the other way? What would it do about a large bull within the boundary fence, well clear of the running line at the moment but posing a potentially significant danger to trains (see Polmont) that needs to be reported quickly (worthy of the red button and/or emergency protection) in accordance with the Rule Book?
I have no doubt that all of the above, and more, could be overcome with (lots of) time and advances in technology. What's the point, really, though if you still need a fully competent (and fully paid) driver?
dannypye9999
Member
- Joined
- 28 Jun 2013
- Messages
- 392
I don't fancy travelling on a 140mph train with a computer driving it, especially in this day and age of computer viruses, Trojans and hackers.
The airline world (which has a safety culture far superior to the rail world) would suggest you're wrong, and that it would be better for the driver to be relaxed and monitoring things than stressfully occupied with the fatiguing actions of manually driving.
A train captain who performs the guard's duties leaving a station while the train handles the driving (including safeworking, applying power efficiently, monitoring for wheelslip many times a second and so on) and then settles down to monitoring the automatic operation will be far more rested and alert than a driver who has their full attention occupied even before a contingency gets thrown into the mix.
Same way it does for a curve. Knowing with precision the path it is scheduled to take, and checking it is clear of obstructions. Hot spots for obstructions could have fixed object detection units placed there until the cause of the obstructions is eradicated.
A millimetre wave radar (off the shelf tech) would even be able to check that point blades are in the correct position from the moment it gets into line of sight for the radar even if that's well over a kilometre away (and therefore still within stopping distance) and instantly take action if it is not as it should be. A human driver can't do that until it's too late.
We've dealt with curves and junctions, so what's next? "Ooh, what about a bridge, bet you hadn't thought of that one!"
Because "we pay them to do the job" has been totally discredited as a safety strategy, many times over.
Automatic systems would do the driving duties better. For example, a number of train drivers themselves have posted on this forum that they have at times had differing ideas on the best way to apply power efficiently, while an automatic system would be programmed to do it right and to recalculate the energy use strategy many times every second to meet the schedule at the next stop. More efficient energy use inevitably leads to lower wear and tear, and therefore reduced maintenance costs.
The driving competency of a train captain would not need to be the same as a "proper" driver. The rule knowledge required would be appropriately lower as they would only need the knowledge appropriate to incident responses and to handling low-speed limping home to the next station, and the driving skills required would be similarly simplified.
You'd also get improvements to customer service. With the operational tasks of the guard shifted to the train captain, customer service personnel could focus on their jobs completely instead of being hampered by multitasking.
- Current DOO trains would see an increase in safety as the train captain would be able to devote full attention to the guard's duties at stations while the train drives itself
- Trains where the guard has to handle multitaksing between operational and customer-facing roles would see a revenue-neutral increase in customer service as the conductor would be just a conductor and not a conductor+guard.
- Trains where a guard currently performs only operational tasks would see a revenue-neutral introduction of customer service by that guard being replaced by a conductor, or costs reduced as the guard gets phased out with their operational tasks transferred to the train captain.
As demonstrated by a risk assesment process that results in planes with 300 people on board being permitted by the avaiation authorities and some some plane operating companies to fly over active warzones where the locals are in the habit of regularly shooting down aircraft in order to avoid a 15 minute delay....
Back to rail a lot of debate seems to revolve around the need for route knowledge. AIUI In the USA speed signalling is used and drivers do not need to know the route.
Last edited:
I don't really disagree with the majority of your post, but I think it's clear that we're a very long way from much of it becoming a reality! I don't know if some of the signalling technology in use elsewhere in the world differs, but nothing - current or, as far as I'm aware, proposed - in this country tells the train which route it's being signalled over, only (in the case of ATP and ERTMS?) how far it's allowed to go and the speed 'profile'. Please correct me if I'm wrong!
I do disagree that, even with full ATO, the driver's status (and pay) will be reduced. As with many others, they only really start to earn their money when things go wrong and they have to delve into the 95% of their vast knowledge that's not required in routine operations. Low speed or otherwise (bearing in mind how much is currently done under caution!), it's still got great potential to go seriously wrong if the rules are followed - and it's not necessarily just about limping a broken train home, as there's plenty of failure conditions on the infrastructure side to consider. Route knowledge will hopefully continue to be an important part of a driver's competency - I, for one, certainly don't want to be on the receiving end of an emergency call where the driver can't meaningfully identify his location. Unless I'm very much mistaken, the ATO train operators on London Underground are on similar (or the same?) pay and conditions as their colleagues on non-ATO lines?
But the system to run it does not have to be connected to the outside worl which would negate all those issues. Of course there would have to be measures in place to ensure that whatever system that was used to talk to the train if not done the DLR way was infiltrated the same way as lineside cable equipment then they could either shut it down or have a go slow failsafe.
And to get back to the 'object detection' issue that so many people seem to have a problem with, they should look up the Leica geomapping cameras. Whilst these are only static cameras in use at the minute they are amazing at what they can capture from up to nearly 1000mtrs and there is no reason that linking them to current technology would not give an effective real time object detection and identification process to then let the computer assess and the apply an action.
Again, not impossible in the longer term, but not a realistic prospect in the near future either! Even if the current technology was up to the job, it'd take a lot of work and time to prove that it was safe. I still maintain that it'd take a fair amount of further development to get it even to that stage though, for the specific railway application where it's trying to establish that the path that it must follow is clear as opposed to other applications where it can choose its own path to negotiate the obstacles that it identifies. It'd still need someone competent to drive it manually too!
Everything takes time to prove that something is safe, I do believe someone earlier has already said that they are developing such a system so it is being done here and now whether you like it or not. The rate of technological advancement is alarming and I wouldnt bet against a system being in place within 10 years.
You are still hung up about obstacle detection too yet other posters have already said about other systems which were very basic in mapping the the ground to fly and they were back in the 80s - do you really not think that that, coupled with the Leica stuff I mentioned earlier and also coupled with the tech that is on the NRM - which as you knwo can inform the controller of an issue in real time - that this stuff is possible now.
And yes, the person on board would need to be able to shunt it to either the next signal or station should something happen like an obstruction but in reality nowadays just how often are trains disrupted during the day because of an obstruction on the track?
- Status