Isn’t the track itself, already wired and somewhat ‘smart’?
There is something called positive train control that can automatically slow a speeding train but it’s not present at the site of today’s accident.
I was wondering the same thing. What do train engineers do, exactly, that a computer couldn’t do much better? Seems like the most monotonous job ever.
It’s not present anywhere in the US, as I understand it, except a stretch of new track in California. Even where it is installed, the automatic-slowing feature remains somewhat experimental, from what I read.
It’s a really good question, isn’t it. Some of the challenges for self-driving cars are:
[ul]
[li]Roads are of variable size and quality, vehicles are not physically constrained to use any particular part of them[/li][li]The vector and intent of other vehicles is unpredictable[/li][li]Things that are not vehicles (i.e. humans) can wander freely into the road[/li][/ul]
If you described roads to an alien and posed those three challenges for making the vehicles driverless, the alien would invent railways.
The technology to run a train fully automatically (driverless) exists and operates all over the world including the USA (ex., people movers at airports), but is currently limited to short distance tracks or metro lines in urban areas. It’s more difficult, i.e., costly and complex, to implement Automatic Train Operation (ATO) in a commuter train or long-distance line because of the countless number of variances that may occur, such as cars, animals, debris, or even people on the tracks. The number of signals and sensors necessary to insure that the train stops and goes when it’s safe to do so, and retrofitting them to legacy tracks is very cost prohibitive. It’s easier to construct brand new train lines from the ground up, such as the new linear train in Japan that is scheduled to start service in 2027, to be unmanned ATO.
The article Dewey Finn linked to refers to Automatic Train Protection (ATP) and this is a safety installation, only to slow or stop a train when it should. It has nothing to do with making a train go and you will still need a conductor.
Interestingly, the same sort of accident where a train exceeded the speed limit of a curve occurred in Japan 10 years ago. The train derailed and wrapped itself around an apartment building that was situated right at the apex of the curve. The accident killed over 100 and injured over 500 passengers and we just observed its 10-year memorial two weeks ago. The findings were the exact same as the article; had Japan Rail implemented ATP faster, (they were in the middle of doing this), the accident could have been prevented. Very sad indeed to lose lives because of human error.
Same thing with this one in Spain. Spain train crash: Video footage emerges showing moment train derails and kills 80, as police interview driver Francisco Jose Garzon Amo | The Independent | The Independent
How come self-driving trains aren’t already the norm?
You need to have a person, not a company, to blame when things go bad.
That way, when things go bad, the person is reprimanded or fired, not the company.
Doesn’t BART run automatically? When I rode it some 30+ years ago, the operator’s job was to close the doors when the passengers were aboard. As soon as (s)he closed the doors, the train moved.
There was a case in DC a few years ago where there was a backlog and trains were held at stations. One operator stepped onto the platform to have a smoke. When his back was turned, the train left the station and stopped at the next two stations without opening the doors. At the 3rd station, a passenger had the foresight to hit the emergency button.
Even without ATO, can’t they have radar at key points to signal the operator that (s)he is going too fast? With today’s technology, it sounds like a piece of cake.
I don’t understand how a human could be better equipped to handle those things. Trains are, to my understanding, incapable of stopping within any sort of short distance; by the time the engineer in the train sees an obstacle on the track, it’s far too late to stop the train. So how does having an engineer in the train do the job better than a computer, when both are equally incapable of stopping the train before colliding with an unexpected and unknown obstacle in the path?
I didn’t mean that humans could operate trains better than computers. I was pointing out why ATO was easier to implement on metros and dedicated tracks. You don’t have cars and nature and other random things to deal with in an underground tunnel or on an elevated track at an airport. For a commuter train travelling long distances, you do. Trying to get sensors and signals to account for all these things get very expensive and it’s cheaper to use the current system and add safety features like ATP, instead.
The entire Montreal Metro runs automatically and has since it was opened nearly 50 years ago. The Times Square-Grand Central shuttle is much older and has always, AFAIK, operated automatically. But underground operation with no switching is much easier to automate.
Partly to be a responsible and accountable (and ‘accuse-able’) person. People like to have a person they can point to instead of a undefined hard to comprehend or understand chain of responsibility.
I suppose I phrased it poorly. What I meant was, the only relevant sensors a human brings to the table (as far as I know) are their eyes, and by the time the obstacle is close enough to be detected by human eyesight, the train is already too close to the obstacle to stop if it hasn’t already been slowing down.
Is there a particular aspect to this that I am missing, or am I wrong in some of my understanding?
Not true. Per this article, it’s present from Washington to north of Baltimore and from New Haven to Boston.
Seems to me that the safest system is where the trains run automatically with a human on stand-by if something goes wrong.
If I’m the human and I’m just on standby, it’s going to be difficult not to fall asleep or at least be goofing off on my phone. Just standing there but not actually running the thing is gotta be really boring.
The 42nd Street Shuttle only operated automatically for a brief time in the 1960s. The Grand Central fire destroyed the automation equipment and the line was rebuilt without it.
Currently the only NYC subway line with full automation is the Canarsie (L) line. The Flushing (7) line will be next, with automated operation expected to begin in early 2016. Like the Canarsie line, doing this required a multi-year, multi-million dollar project to outfit the line with the necessary transponders and equipment, upgrade all the wayside signalling apparatus, and purchasing and/or retrofitting cars to use the new system.
If you look at the map, you’ll note that the Flushing and Canarsie lines have something in common: they’re both self-contained and do not join or branch from any of the other lines in the system. The complexity of routing in the rest of the NYC subway system, with multiple types of rolling stock sharing the same tracks and inter-system routing between the BMT and IND makes installation of ATO extremely complex. Still, after Flushing they’re going to give Queens Blvd a shot. Unlike the other services which are completely automated, automating just the Queens Blvd line means that you will still need an operator once each train makes it across the tunnel to the Fifth Ave, Broadway, or Eighth Ave trunks.
Detroit has the People Mover, but it runs on an elevated track, around a single loop in the downtown area.
Still, other people are making an excellent point. Trains are much simpler to control than cars, by at least a couple of orders of magnitude. If one of the primary purposes of the engineer is to be someone to throw under the [del]bus[/del] train when something goes kablooey, then perhaps we should just automate the process, and accept that if someone wanders onto the tracks at the wrong moment, they’re fucked. I mean, applying the brakes is mostly a symbolic gesture anyway, since trains weight many many tons, and the laws of physics will be obeyed.
I’m an automation engineer who programs intelligent devices to automate switches and breakers on the power grid. My name is on those design drawings along with my professional engineer license number. If shit hits the fan and people are killed, they blame me, whether I was on site or not. As far as I can tell, that’s the whole reason professional engineer licenses exist.
I don’t think “we need a person to blame” is the reason we still have locomotive engineers. I think “it’s cheaper that way” is certainly reason enough.