Hacker News Reader

Comment by SoftTalker | Hacker News Reader

The goal is totally autonomous train operations.

That doesn't seem likely considering their terrible safety record with people at the scene who should be empowered to detect and stop it.

Rail has a very good safety record compared to other forms of freight, considering the quantity of freight that moves.

I doubt it. From what I understand, on top of squeezing the workers, railroad companies have been underinvesting in the infrastructure. I don't see how they would accept to spend billions in R&D (to my knowledge there is no fully automated train yet) and track/signals upgrades.

> (to my knowledge there is no fully automated train yet)

Several metro lines are fully automated and driverless--Paris Metro Line 1 has been fully automated for 11 years now.

I should have been more specific. I am aware of some automatic metro systems but in my opinion the challenges they overcame are very different from the challenges freight or even long distance passenger trains would have to overcome.

Fully automated freight rail is not possible with the infrastructure and technology we currently have unless the rail companies accept insane risk.

For example, if one of the locomotive engines stops working and the train must stop, you need to keep the brake lines pressurized. For some archaic reason they fail open instead of closed, so you have to keep the engines running at all times. Otherwise, you need to send a guy out to manually apply the emergency brakes. This was part of the cause of the Lac-Mégantic disaster.

So it's not impossible, but there's zero chance that NS or CSX would ever invest $MM to upgrade the brakes on all their trains to let them run automated. But the real crux of it is that they'd be responsible for any disaster, and their go-to strategy is to just blame whoever happened to be in the cab that night.

https://en.wikipedia.org/wiki/Lac-M%C3%A9gantic_rail_disaste...

> For some archaic reason they fail open instead of closed, so you have to keep the engines running at all times

This sounds incredibly stupid. Do you happen to know the archaic reason?

I think most of it is legacy. Trains have been around for a long time, and changing something on tens of thousands of locomotives and millions of train cars would be very expensive and complicated. Most systems now have a reserve tank on each car that holds a certain pressure, and the braking signal is sent by reducing the pressure from the main unit. But it's still not nearly as safe as automotive air breaks that fail closed, since eventually all the air will leak out of the tanks.

https://en.wikipedia.org/wiki/Railway_air_brake

There are tons of automated metro lines, and the EU standard for rail signalling, ETCS, is capable of Automatic Train Control and has heavy doses in automation in it already, especially used on high speed lines (where the driver can't really do much anyways due to the speeds involved)

However, without full grade separation, this is extremely risky because humans are stupid. It's also very expensive, especially if retrofitting on existing track without interrupting traffic.

Trains have been getting longer in recent years, increasing the amount of time people have to wait at train crossings. Seems likely that people will respond by taking more chances at beating the train and some percentage will lose their bet that they'll get across in time.

Literally impossible in the terrain they operate on

Bold claim

Being ignorant makes many realities seem like fanciful thinking.

Which is only slightly less hard than full self-driving vehicles. The traffic/rail conditions may be simpler by comparison, but the added complexity of the vehicles in question more than makes up for it.

Reply on news.ycombinator.com