Hacker News Reader

Comment by cperciva | Hacker News Reader

Never mind one-person crews; why don't we have zero-person crews? In most cases the crew are just responding to signals anyway; trains accelerate slowly enough that it's very rare that train crew can see a problem in time to do anything about it anyway.

View on HN

How do you find the Skytrain in Vancouver? As an occasional visitor I've found it reliable and friendly, but it is also a fully-isolated track (right?). I'm not sure how well current automation would cope with the semi-open train yards and crossing points that we have in Seattle's Interbay or SODO districts.

Skytrain is awesome. Tracks are isolated for some definition of "isolated"; skytrain doesn't have to interoperate with any other rail traffic (unless you count maintenance vehicles) but there isn't any wall separating the track from the station platform like e.g. the Las Vegas airport has (it has been discussed as a safety measure, but skytrains don't presently park accurately enough to line up the doors) and there are many places where you can hop a fence and get onto the tracks (resulting in trains being stopped and police showing up to escort you away).

The DLR in London is another example of how zero-person trains may work.

That's still a significant difference from running traditional trains on unprotected track. Skytrain has no grade crossings.

Here's why:

https://en.wikipedia.org/wiki/Lac-Mégantic_rail_disaster

RIP to all 47 victims.

That was an interesting read, thanks for posting that.

However, in the write-up the TSB cites several sources of human error and a _lack_ of automation such as rail circuits as key factors.

The article notably specifically cites oil as an hazardous & risky thing to carry. Let us not find fault with the whole idea for but one aspect which we ought not implement (carrying oil).

> Let us not find fault with the whole idea for but one aspect which we ought not implement (carrying oil).

We can't find fault with that; oil's got to move somehow. It can move by truck, by rail, by ship, by air, or by pipeline.[1] How much worse is rail?

[1] Technically, I guess we could also pour it into the ocean somewhere and hope it naturally floats to somewhere more useful.

Add mecsred notes, this disaster illustrates the risks of rail, especially with hazardous cargo (which could have been carried far more safely by pipeline!) but it also illustrates an over reliance on humans. A zero-human-drivers railway would have (a) had automatically locking trains rather than relying on humans to hit the breaks, and (b) more remote monitoring to detect if anything went wrong.

A zero-human-drivers railway sounds extremely vulnerable to "the train is on fire, the fire brigade shut off the generator, which it turns out the brakes depended on" which is the essence of what happened there.

That's absolutely a problem if a train designed to have a driver doesn't have one (which is what happened here). If trains were designed to not need drivers, this wouldn't have happened, because (a) enough mechanical brakes would have been applied automatically, (b) the engine fire would have been detected and reported automatically, and (c) if the brakes failed there would have been an immediate alarm when the train started moving.

Does anyone have any up-to-date info regarding fully autonomous freight trains? Self-driving cars and semi-trucks seem to grab all the headlines, but someone's got to be working on automating freight trains. It seems like a much easier problem to solve. Or is it?

The incentive is also much lower, given the ratio between driver comp and freight volume.

Reply on news.ycombinator.com