The March 2017 Issue of E&T carries several articles about driverless cars but why haven't we got driverless mainline trains?
The technical 'problem' should be far simpler to solve than for a road vehicle. The position on the 'road' can be predicted and determined easily with precision. There is essentially no collision problem to solve, that has been done with the existing signalling system.
There is no need for communication with the train, no need for additional infrastructure. All that is needed is to observe and act on the existing fixed signals.
Of course such a basic system can be improved upon to produce a 'super driver' capable of reacting to unplanned obstructions, greasy rails etc.
The human driver is perhaps the last link to be made 'fail-safe' in the railway safety regime. Our efforts to 'improve' the driver-train interface have probably added new problems. Regular signal spacings, standard aspects and driver alerts must surely increase the boredom and inattention factor. An example of this was the Shap Roll-back in August 2010 where a driver correctly observed adverse signals, came to a stop, then allowed the train to roll-back, acknowledging the retreating adverse signals on the way, until the train exceeded 50 mph. Presumably he was half asleep?
I suspect the real 'problem' is a social one, it will be a tragedy if we can't solve that one.
but as it isn't (afaik) published by Network Rail or Bombardier I'd better not go much further. And I may be an order of magnitude or two out (I wrote it from memory), but it's of around that order.