Now I'm no expert, but I have played Train Simulator and can therefore confidently state that they're impossible on any long route. You can maybe have an automated line within a local transit system, but for an automated train running at 250 km/h to properly interpret signalling would require digital infrastructure that just doesn't exist, I don't think.
but for an automated train running at 250 km/h to properly interpret signalling would require digital infrastructure that just doesn’t exist
You mean visually interpret signalling? If signaling is standardized a well programmed computer shouldn't have that hard of a time. And the "digital infrastructure" required otherwise could be as simple as slapping QR codes everywhere. Or you know, rail is already a well defined network, so it shouldn't be too hard to make a digital network on top of it. Whatever computer is controlling that signal could also just send it directly to the train's computer, which is what I've assumed they've been doing for a while now.
None of that is to say there shouldn't be a person there, but most of the time they shouldn't need to intervene.
I don't really believe visual processing at extreme speeds will be able to compare to human vision to the degree required anytime soon. Your other idea sounds good (I think that's basically what CBTC is), but the New York City subway, for example, has been upgrading to such a system for nearly three decades now, so I'm not optimistic about the possibility of a widespread application of that technology.
Now I'm no expert, but I have played Train Simulator and can therefore confidently state that they're impossible on any long route. You can maybe have an automated line within a local transit system, but for an automated train running at 250 km/h to properly interpret signalling would require digital infrastructure that just doesn't exist, I don't think.
You mean visually interpret signalling? If signaling is standardized a well programmed computer shouldn't have that hard of a time. And the "digital infrastructure" required otherwise could be as simple as slapping QR codes everywhere. Or you know, rail is already a well defined network, so it shouldn't be too hard to make a digital network on top of it. Whatever computer is controlling that signal could also just send it directly to the train's computer, which is what I've assumed they've been doing for a while now.
None of that is to say there shouldn't be a person there, but most of the time they shouldn't need to intervene.
I don't really believe visual processing at extreme speeds will be able to compare to human vision to the degree required anytime soon. Your other idea sounds good (I think that's basically what CBTC is), but the New York City subway, for example, has been upgrading to such a system for nearly three decades now, so I'm not optimistic about the possibility of a widespread application of that technology.