Ask HN: Why aren't trains fully automated?

9 points by legostormtroopr ↗ HN
Prompted by this news article: https://www.brisbanetimes.com.au/politics/queensland/just-17-5-per-cent-of-required-new-qr-train-drivers-ready-to-roll-20170901-p4yvog.html

Why are train drivers necessary in an increasingly automated world?

Trains run on fixed tracks, at fixed speeds, have fixed stops and have right of way (mostly because they are very difficult to stop quickly).

If there is an obstruction on the tracks, even if a human driver is in charge, a train is nearly impossible to stop.

So what are the challenges that prevent the cockpit of a train being replaced by a n automated driver?

17 comments

[ 3.0 ms ] story [ 62.6 ms ] thread
I'm guessing there's a union involved somewhere, and automation would threaten jobs.
This and also the fact that implementing automation would cost too much in the short-term than employing human drivers, so nobody wants to do it.
It depends on the area - not sure which region you're talking about (I guess Australia) but for example in the UK rail operators are private companies and there are multiple of them - they build maintain their rolling stock independent of each other, so the challenge would also be to convince/force them all to agree on a standard network/communication protocol for their trains.
Southern Rail is having a famously hard time trying to convince its union to accept a reduction in the number of staff per train.
Are drivers a significant cost to running a train? Just like when optimizing code, you have to profile first and go for the low hanging fruit. I'd be willing to bet that the rolling stock and infrastructure are where the low hanging fruit is.
Not sure how accurate this is but a colleague told me that some trains on the London underground are fully automated, but they still have staff standing in the driver compartment to put passengers at ease.
Some trains are fully automated, and machines have been able to handle general operation of a train better than a human for a while. But train drivers have some advantages when it comes to making decisions on shutting doors, handling emergencies, and making out announcements and their cost is spread over a lot of passengers so eliminating them has never really been a priority.
There are fully automated trains. However full automation requires not just a capable "train cockpit" but also a capable signal system (with supporting auxiliary systems). Introducing the latter is, as far as I can tell, the hard/expensive part.

As a note, not all trains have fixed speeds or fixed stops. As an example some signal systems may allow a train to enter a section where another train is already present if the driver drives slow enough to be able to stop before colliding.

In Paris, "ligne 14" of the subway has been built fully automated, it opened in 1998 (~20yrs ago now!). "ligne 1" has also been transformed from manual to automated in 2012. "ligne 4" is in progress of being transformed.

This is possible. Easier for subway because the tracks are fully protected from any external impact: trees, animals, cars, humans, ... There has been a problem once with some stuff that fell onto the track from works on a roof. Quite rare in the subway though compared to trains.

Also, in a subway, the distance would be shorter for employees of the company to come and assist passengers: it might take a few minutes between stations, where it could take dozens of minutes, even hours, on train tracks.

Removing driver is not the goal, at least for now. Look at ETCS operation modes: https://en.wikipedia.org/wiki/European_Train_Control_System#...

By design it support exceptional situations where full automation is not possible. Also infrastructure will need decades for upgrade. Currently even routes used by high speed trains aren't 100% covered and there is low or none financial incentive to upgrade old, regional lines.

You still want someone on the train for emergency repairs, so why not let that person drive it?
I was on Amtrak a few years back and we made an unscheduled stop out in the middle of nowhere. A few minutes later, one of the conductors got on the intercom to let us know that the train had run over a shopping cart, and that they were going to have to dig the mangled remains out from underneath the engine.

I went back to my game and we were under way soon enough, but can you imagine if an engineer had to drive out to meet the train and put it back into operation? We would have waited hours just for them to drive hundreds of miles out to where the train was. I imagine it's less of a problem on a little subway system.

There was a brief discussion about this the other day after talking about autonomous cars here https://news.ycombinator.com/item?id=15116967

The biggest key take away I read was the problem with "unions" and the following.

> If we had autonomous planes it wouldn't significantly reduce cost or increase usage...

Cars have 2 or 3 passengers per drivers while trains often have dozens. Cost per person ratio is different.