Windowless driver cabs?

[TwistedMentat]

A discussion on an Aviation forum I'm on about windowless airliners got me thinking. Would there be any benefit to a windowless driving cab?

The thinking here is that with modern and near future cameras and augmented reality technology we could move train cabs to 'better' areas or modify the design of the nose.

For example, on a diesel locomotive you could move the driving cab away from the nose. Improving crash survivability. You could even build a single control station that rotates depending on the direction of travel.

And on a passenger train you could redesign the nose to be stronger and/or lighter.

Now of course the camera and display technology would have to be properly high def and smooth, something like a high end PC gaming rig. But you would also have the ability to make use of low light and infrared cameras for night running and augmented reality tech to give drivers extra help in identifying things on the line like highlighting signals.

And on EMUs with restricted visibility due to gangway doors the cameras and monitors could help actually improve their sightlines.

And before the idea is thrown out as a kneejerk reaction. It is worth pointing out that London City Airport is about to commission a remote control tower that uses a large number of cameras and other augmented reality tech to have an airport control tower without the physical crewed control tower. So some safety first industries are already looking into and starting to use similar tech.

And as a continuing thing could we move towards windowless passenger cabins? Using screens and cameras to display the outside. Giving manufacturers the ability to build lighter and stronger train cars as they can avoid the huge holes needed for windows.

We're talking 20+ years from now stuff, not suggesting the next train should have all this stuff.

 

[takno]

And before the idea is thrown out as a kneejerk reaction. It is worth pointing out that London City Airport is about to commission a remote control tower that uses a large number of cameras and other augmented reality tech to have an airport control tower without the physical crewed control tower. So some safety first industries are already looking into and starting to use similar tech.

And as a continuing thing could we move towards windowless passenger cabins? Using screens and cameras to display the outside. Giving manufacturers the ability to build lighter and stronger train cars as they can avoid the huge holes needed for windows.

We're talking 20+ years from now stuff, not suggesting the next train should have all this stuff.

A camera-based remote control ATC tower seems more like moving your signal boxes to PSBs and ROCs than it does anything to do with windowless cabs/cockpits, so in that sense your other safety-first industry is about 60 years behind the times on this.

As to the use of screens in passenger cabins, you seem to be missing the point of windows entirely. I want natural light, not a pretty picture. If I was just going to go round looking at everything on a screen I might as well stay home and not travel at all. I have no doubt your giant claustrophic metal coffins would have great structural integrity, I just question whether they would have any passengers on them

 

[Bletchleyite]

We're talking 20+ years from now stuff, not suggesting the next train should have all this stuff.

20+ years from now it all needs to be ATO and on-board staff should be helping passengers, not being a human automaton that's by then inferior to an actual one.

 

[Whattraintoday]

20+ years from now it all needs to be ATO and on-board staff should be helping passengers, not being a human automaton that's by then inferior to an actual one.

Absolutely. Also, by then AI / Machine Learning will be advanced enough to correctly interpret the inputs from the sensors that are scanning ahead (as well as the feeds from static sensors on corners, near level crossings and other higher-risk areas)

Before anyone says that passengers won't stand for it, by then it'll be the norm in all industries. People could even be worried about travelling in a vehicle controlled by a human, with all their failings and foibles.

 

[WelshBluebird]

Be careful what you wish for. More technology that is relied upon to operate a service = more opportunity for things to go wrong.
Do you really want services cancelled because "the in cab camera systems have failed"?
Also, by the time such technology could probably work reliability, we'd more than likely be looking more at automated trains where the member of staff is more of a supervisor than a driver anyway.

And before the idea is thrown out as a kneejerk reaction. It is worth pointing out that London City Airport is about to commission a remote control tower that uses a large number of cameras and other augmented reality tech to have an airport control tower without the physical crewed control tower. So some safety first industries are already looking into and starting to use similar tech.

Regarding London City, there are some additional things built into the system that I don't think could be replicated in your idea. For one, the data between the sites takes 3 totally separate routes to ensure that redundancy in the event of a failure of one or two. And for two, they have multiple redundant cameras that would take over from a camera failing. How would you replicate that on a train? What would happen in the event of an incident where, for whatever reason, the cameras / data connection are interfered with (say by an accident or vandalism).

And as a continuing thing could we move towards windowless passenger cabins? Using screens and cameras to display the outside. Giving manufacturers the ability to build lighter and stronger train cars as they can avoid the huge holes needed for windows.

Look at the hissy fits some people have about small windows (pendolinos) or windows not lined up with seats (multiple examples) or some seats not having a window (IET's). That just isn't going to happen.

 

[Bletchleyite]

Absolutely. Also, by then AI / Machine Learning will be advanced enough to correctly interpret the inputs from the sensors that are scanning ahead (as well as the feeds from static sensors on corners, near level crossings and other higher-risk areas)

Before anyone says that passengers won't stand for it, by then it'll be the norm in all industries. People could even be worried about travelling in a vehicle controlled by a human, with all their failings and foibles.

I agree strongly. Most transport accidents are caused by human error. Once automation is up to it, remove the human and the accidents go away. Hence my view that guard-only operation is the future (on the DLR model - the train releases doors automatically, the guard dispatches) and DOO is a blip.

Get the humans dealing with human factors - that's what they do best.

 

[furnessvale]

A discussion on an Aviation forum I'm on about windowless airliners got me thinking. Would there be any benefit to a windowless driving cab?

The thinking here is that with modern and near future cameras and augmented reality technology we could move train cabs to 'better' areas or modify the design of the nose.

For example, on a diesel locomotive you could move the driving cab away from the nose. Improving crash survivability. You could even build a single control station that rotates depending on the direction of travel.

Why would it need to rotate if the driver was viewing cameras?

It wouldn't even need to be on the train and trains could be driverless like the Rio Tinto ore trains in Australia.

 

[Dave1987]

In a word..... no.

 

[D6975]

Why would it need to rotate if the driver was viewing cameras?

Driving backwards?
Just like the SPV for those of a certain age.
It might be wise to rotate it though as it may lead to motion sickness.

 

[furnessvale]

Driving backwards?
Just like the SPV for those of a certain age.
It might be wise to rotate it though as it may lead to motion sickness.

True, but if he sits in the signalbox next to the signaller there will be no motion to be sick of!

Better still, give the driver a super duper i phone and he can drive his train from home, won't even need to get out of bed to do it.

 

[Dave1987]

I agree strongly. Most transport accidents are caused by human error. Once automation is up to it, remove the human and the accidents go away. Hence my view that guard-only operation is the future (on the DLR model - the train releases doors automatically, the guard dispatches) and DOO is a blip.

Get the humans dealing with human factors - that's what they do best.

Until the sensors are fed erroneous data or the computer system is hacked. I’m really not bothered about automation at all as it will kill off jobs for the whole of society and our Government refuses to do anything about it and almost denies its existence but I still maintain you don’t really understand what you are talking about, with regards to trains. But hey ho you are entitled to your own opinion.

 

[D6975]

Not having the driver on the train is surely a bad move as a driver can often rectify a minor fault on a train, something he couldn't do if sitting many miles away. OK in a suburban area where a fitter could be rapidly sent out to attend, but if you're at Ribblehead or Harrison's sidings or Drumochter then definitely not desirable.

 

[Whattraintoday]

Until the sensors are fed erroneous data or the computer system is hacked. I’m really not bothered about automation at all as it will kill off jobs for the whole of society and our Government refuses to do anything about it and almost denies its existence but I still maintain you don’t really understand what you are talking about, with regards to trains. But hey ho you are entitled to your own opinion.

Preventing them being fed erroneous data (which is the main malicious reason why you would hack the system) can be dealt with through decent cyber security. No more of an issue than any other part of the UK's critical infrastructure - as it is, the ROCs could be hacked today and carnage wrought right now. We don't see it happening because we're not at war with a nation-state adversary and terrorists (as yet) don't have good enough cyber warfare capability.

As to "you don’t really understand what you are talking about, with regards to trains" - if you're going to be peremptorily dismissive, we'd appreciate you explaining exactly what is fundamentally wrong with driverless mainline trains (for that's what we're ultimately talking about here) in 20+ yrs time

Right now - God no!
In 20-30 years, with sufficient investment* - probably.


Your opinions on the morality / social acceptability of automation generally are your own and no worse than anyone else's.


*I know, I know!

 

[TRAX]

Isn’t the Class 385 cab basically windowless ?

 

[GB]

...ah, those magical "senors" again.

 

[Dave1987]

Preventing them being fed erroneous data (which is the main malicious reason why you would hack the system) can be dealt with through decent cyber security. No more of an issue than any other part of the UK's critical infrastructure - as it is, the ROCs could be hacked today and carnage wrought right now. We don't see it happening because we're not at war with a nation-state adversary and terrorists (as yet) don't have good enough cyber warfare capability.

As to "you don’t really understand what you are talking about, with regards to trains" - if you're going to be peremptorily dismissive, we'd appreciate you explaining exactly what is fundamentally wrong with driverless mainline trains (for that's what we're ultimately talking about here) in 20+ yrs time

Right now - God no!
In 20-30 years, with sufficient investment* - probably.


Your opinions on the morality / social acceptability of automation generally are your own and no worse than anyone else's.


*I know, I know!

Erroneous data doesn’t just come from being hacked. Sorry I’m just the organic bit that sits at the front of the train what do I know eh?? I know that for many many years designers have come across real world problems that never existed in the bubble of their cosy office on their computer screen so when their fantastic product is released in the real world it fails spectacularly. Seen it many times. You may consider me dismissive, I consider myself a believe it when I see kind of person.

 

[underbank]

Also, by the time such technology could probably work reliability, we'd more than likely be looking more at automated trains where the member of staff is more of a supervisor than a driver anyway.

More likely the "driver" would be sat in a control centre somewhere miles away, using the same kind of technology where the air force "pilots" control their drones from a different country!

 

[Whattraintoday]

Erroneous data doesn’t just come from being hacked. Sorry I’m just the organic bit that sits at the front of the train what do I know eh?? I know that for many many years designers have come across real world problems that never existed in the bubble of their cosy office on their computer screen so when their fantastic product is released in the real world it fails spectacularly. Seen it many times. You may consider me dismissive, I consider myself a believe it when I see kind of person.

And I wholeheartedly agree that it needs to deliver greater safety and reliability than a human driver and be thoroughly tested and proven in that area. The driving force to replace drivers may not be strong enough to be worth the investment.
Any AI would need to be able to determine what sensor inputs are incorrect - just as a human driver can. Signalling and in-cab systems can tell you one thing, your eyes another. Which do you go on? That’s where your training and experience kick in.
AI gets trained too. At the moment it’s hilariously bad at it. In 20+ yrs time it’ll probably be viable.

 

[Dave1987]

And I wholeheartedly agree that it needs to deliver greater safety and reliability than a human driver and be thoroughly tested and proven in that area. The driving force to replace drivers may not be strong enough to be worth the investment.
Any AI would need to be able to determine what sensor inputs are incorrect - just as a human driver can. Signalling and in-cab systems can tell you one thing, your eyes another. Which do you go on? That’s where your training and experience kick in.
AI gets trained too. At the moment it’s hilariously bad at it. In 20+ yrs time it’ll probably be viable.

We will agree to disagree I will believe it when I see it. I firmly believe I will retire at the pointy bit of the train and I have a good 30 years until I will retire. If I’m am wrong I’m sure there will be a long line of people at the dole queue ahead of me

 

[Whattraintoday]

We will agree to disagree I will believe it when I see it. I firmly believe I will retire at the pointy bit of the train and I have a good 30 years until I will retire. If I’m am wrong I’m sure there will be a long line of people at the dole queue ahead of me

Indeed. To be honest, I’d quite like to be wrong too about automation.
My field (cyber security) is firmly in the firing line for it, way before they go for the guys at the front of trains.
Returning to the original point of the thread (as we’ve deviated somewhat) - the benefits, effort & political courage to make a cab windowless are much the same as remote driving, which are in turn much the same as full automation.
So they probably wouldn’t bother with those intermediate steps.

 

[bastien]

Driving backwards?
Just like the SPV for those of a certain age.
It might be wise to rotate it though as it may lead to motion sickness.

Haha. Beat me to it!

 

[bastien]

We will agree to disagree I will believe it when I see it. I firmly believe I will retire at the pointy bit of the train and I have a good 30 years until I will retire. If I’m am wrong I’m sure there will be a long line of people at the dole queue ahead of me

I think you're right, and that 'machine learning' will soon be found out to be a lot of snake oil.

 

[WelshBluebird]

Whilst I do think real AI and systems able to say drive mainline trains safely without a driver sat in the front cab are further away than some people like to suggest, I do think we shouldn't be too complacent about how technology progresses.

20 years ago most people people didn't have internet access at home, and those of us who did dealt with dial up speeds where you could barely listen to (really bad quality) audio. Now the majority of country have very high speed connections that allow the streaming of pretty damn good quality live video, even on a mobile device when not at home! Back then, mobiles looks like this for! - https://en.wikipedia.org/wiki/Nokia_5110. Amazon UK had only just launched, Google was only just being founded, most people didn't own a DVD player and relied on VHS etc etc. 20 years is a long time!

We have no idea what the state of play will look like in 20 years time.

 

[takno]

Whilst I do think real AI and systems able to say drive mainline trains safely without a driver sat in the front cab are further away than some people like to suggest, I do think we shouldn't be too complacent about how technology progresses.

20 years ago most people people didn't have internet access at home, and those of us who did dealt with dial up speeds where you could barely listen to (really bad quality) audio. Now the majority of country have very high speed connections that allow the streaming of pretty damn good quality live video, even on a mobile device when not at home! Back then, mobiles looks like this for! - https://en.wikipedia.org/wiki/Nokia_5110. Amazon UK had only just launched, Google was only just being founded, most people didn't own a DVD player and relied on VHS etc etc. 20 years is a long time!

We have no idea what the state of play will look like in 20 years time.

20 years ago we were experimenting with ETCS level 2, and the rollout of cheap functional Ansaldo signalling on the main line. Looking back it seemed such an optimistic time, when we could look forward to the time 20 years from then when tilting trains would be running all over a network transformed, where we were no longer slaves to the old manual levers on poles. DLR and the Underground were leading the way, showing us how trains would all be a simple push-button affair in the future. The more optimistic amongst us dreamed of self-driving cars and lorries running in automated convoys that were surely just around the corner.

Instead, I've got a fancy new phone which, much like my old phone, I can use to summon a minicab. In a stroke of brilliance though, instead of just doing that by phoning a number from my contacts, there's this dodgy app that lets me pay some Americans to order it for me.

 

[GB]

The more optimistic amongst us dreamed of self-driving cars and lorries running in automated convoys that were surely just around the corner.

Why on earth would you dream of such things?

 

[takno]

Why on earth would you dream of such things?

It's like sheep - helps me nod off

 

[James James]

Until the sensors are fed erroneous data or the computer system is hacked. I’m really not bothered about automation at all as it will kill off jobs for the whole of society and our Government refuses to do anything about it and almost denies its existence but I still maintain you don’t really understand what you are talking about, with regards to trains. But hey ho you are entitled to your own opinion.

This attitude is what leads to economies becoming inefficient, and ultimately causing more unemployment in the long run. It's poisonous. People have to understand that some jobs will go away, the savings in one industry then lead to new jobs in other industries.

Trains put stagecoach drivers out of business at one point. E.g. a single train could've put 100 stagecoach drivers out of business (not counting those to take care of horses/maintenance/etc). It created new jobs in train operation and construction industries, more importantly it created new opportunities for business due to faster travel / easier goods transport / etc.

We no longer lament the loss of horse-drawn carriages and stagecoaches. Likewise, we won't lament the loss of train drivers jobs (just like not many in the real world lament the use of DOO). It's not quite the paradigm change that the introductions of trains saw, and it's going to take quite some time (e.g. it's easier/more cost effective on high frequency lines in a city, less so on infrequent rural lines) - but it will happen.

Anyhow, on the security topic: there's no reason trains should be hackable, first off because trains don't need to be connected to the internet. For now, everyone is connecting everything (including nuclear power stations) to the internet, but standards will evolve the more cracking goes on.

 

[route:oxford]

Look at the hissy fits some people have about small windows (pendolinos) or windows not lined up with seats (multiple examples) or some seats not having a window (IET's). That just isn't going to happen.

Easy enough to do fake windows. They are very realistic if you've ever taken the "Hogwart's Express" at Universal Studios...

From about 3:30 in.

 

[underbank]

The more optimistic amongst us dreamed of self-driving cars and lorries running in automated convoys that were surely just around the corner.

I think remote operated trains are closer than driverless cars but still probably a couple of decades away. It amazes me how people think that driverless cars are just a few years away but driverless trains will never happen! I just can't see widespread driverless cars all over the country this century - maybe in selected limited/controlled areas, but not everywhere.

 

[James James]

I think remote operated trains are closer than driverless cars but still probably a couple of decades away. It amazes me how people think that driverless cars are just a few years away but driverless trains will never happen! I just can't see widespread driverless cars all over the country this century.

Hate to break it to you, but driverless cars are actually extremely close:
https://arstechnica.com/cars/2018/0...es-of-testing-putting-it-far-ahead-of-rivals/

American roads are certainly easier to drive on, and busy cities are going to take much longer (harder to deal with the pedestrians/cyclists/etc), but actual driverless cars are very close to coming into operation across the pond.

// Edit: a more significant article: Lyft will actually begin operating self-driving cars in commercial service in the coming months:
https://www.cnbc.com/2017/11/07/way...-with-its-fleet-of-self-driving-minivans.html