But will there seriously ever be trains with NO ONE up the front? Things fail period. Be it signalling equipment, train equipment, and someone will have to be there to step in and drive it manually.
The only way a failure of the atuomatic train control system can cause a dangerous fault is if it fails by some mechanism that prevents any of the watchdog systems detecting it (or you have to assume that the watchdogs all fail which is so unlikely its difficult to even calculate).
It is even harder to contrive a mechanism that could produce such massive cascade failures but would not disable the trains 'manual' controls - which on many trains are far from actually being manual these days.
And remember that in the worst case scenario an electric train, which for the majority of lines is the future, can be stopped remotely by the simple expedient of tripping the power supply breaker.
And drivers are terribly expensive to train and pay.
I'm jot completely sure what the future holds for technology, but having one person watch X amount of screens with trains on them doesn't sound like the way forward.
All the person watching the screens has to do it look for anything completely out of the ordinary, having dozens of screens watching the train run to the eaxctly programmed schedule would be rather pointless.
It is more to do with using cameras to react to out-of-ordinary situations such as a train losing communication with the control centre.
At that point, having your eyes in a central control centre may in fact be beneficial as it avoids the 'fog of war' effect that conerns staff in the field.
Plus what happens if things like examination of the line needs to be carried out?
Examination of the line?
If in response to some trigger incident rather than simply prophylactic then the man in the control centre can switch to the appropriate feed from a train in the appropriate position and watch it in real time or even record it since all feeds could be practically buffered for a significant amount of time.
And for prophylactic monitoring a large fraction of the capability can be gained by simple leaving a screen on a randomly selected train whenever it is not in use for something else.
What about the things drivers ring up us signallers now to report (ie trespassers, smoke from a location cabinet, something wrong with a passing train), I can't see an automatic train will deal with them.
Trespassers can be detected through proper deployment of intrusion alarms of various types, additionally the increasing isolation of the railway from the general population (the security measures I see trackside are far in excess of historical norms).
Smoke from a location cabinet could be handled by the simple expedient of fitting heat detectors to trackside cabinets - and remember that the amount of trackside wiring is going to be significantly reduced by ETCS L2 (which will eliminate the signal wiring) and eventually by ETCS L3/R (which will eliminate track circuits/axle counters).
In the long run all that will be left will be the pointwork controllers.
And something wrong with the passing train?
What exactly?
Computerised diagnostics aboard the train would be able to detect a wide variety of faults that would have been difficult to diagnose historically - it is hard to think of faults that would escape notice by the computers but that would be a serious matter.
Too much to go wrong in my eyes, I think the future will just be similar to what we have on the tube lines. Otherwise, are you going to have no one on certain trains at all? Or are they going to bring guards back and make them GOO
As I understand it Line 14 of the Paris Metro (the METEOR) operates with noone on the trains at all.
That is the future, especially on our increasingly underground metro operations (where PEDs will be standard in the future).
And pattern recognition obstacle detection is drastically better than it used to be, the Google Car proves that.
