Comment by swiftcoder
1 day ago
> Trains are still not able to assess properly where they are, even using satellites and inside train sensors. Of course there is research on that, but it's still an unsolved problem
Throwing up our hands about how hard a problem this is seems like a ridiculous position to stake. If we can design self-driving cars, that operate safely in far less controlled environments, we can figure out the linear offset of a large vehicle traveling along a physical rail running along a fixed pre-determined path.
> If we can design self-driving cars, that operate safely in far less controlled environments
Self-driving cars aren't supposed to drive quicker than they can look ahead, which makes the problem way easier because you don't have to figure out how to detect obstacles ahead of time and can simply rely on vehicle-local information to stop if necessary. And because cars can brake way more rapidly than trains, it means you can still drive reasonably quickly even within that limitation.
Plus in an emergency, cars have steering as an additional degree of freedom to attempt getting out of any nasty situation that may have developed.
(And of course the socially expected safety standards for cars are quite a bit lower than for public transport…)
Trains on the other hand have braking distances on the order of a kilometre or so, and high-speed trains at full speed several kilometres, so trying to get by on local data only is hopeless and the usual strategy is instead having central signalling system that can accurately track trains and coordinate their movements.
I said nothing about "throwing up out hands", this is a current research topic. But you can hardly blame the workers for not using a system, the researchers haven't invented yet.
> If we can design self-driving cars
We can not design self-driving cars to the safety standard that is required for railways. They are way less safe than the current half-manual system for railways. What you suggest is using less automatic safety systems than what we currently do.
Why is it that these fatal train accidents have happened multiple times in Spain, but never in Japan, even though Japan has been doing it high speed rail longer, and at 10x the passenger volume than Spain? Do they have some alien safety technology unavailable to Spain?
The were several other train accidents, where it was found out, that the drivers have made races, who can stay above the speed limit longer. But I think that was already in the 90s. There is/was definitely something at fault with the safety culture.
But there was a recent incident in Germany as well (https://en.wikipedia.org/wiki/Burgrain_train_derailment), which is somewhat similar to the earlier accident in Spain this year. Targeting high value projects, while not paying for proper maintenance is not something that only happens in a single country.
10 replies →
indeed. Japan is a compliance and shame based society. that alone is the foundation of a very strong safety culture, something that southern europe is famous for lacking.
2 replies →
Has there ever been a case where two self driving cars collide with each other? I hate to be part of the peanut gallery, but as a totally naive person when it comes to railway safety research it really does seem amazing train collisions can’t be easily prevented with gps, lasers, sonar or various other trivial ways. Even the crash in question seems like it might have been easily prevented by having a second human in the loop. Maybe I and others just suspect the difficulty isn’t so much technological as it is financial.
As far as I am aware, there has been no non-supervised train in a general train network. (They exist in closed systems.) However whether the train is self-driving is really not the issue here. In current railway safety systems (even from the 80s) a crash is about the same probability as random bitflips. However reality is too messy, so that the availability is not satisfactory enough to not use a human fallback, and then you are back to the assisted human error rate.
Current operation of a train is basically self-driving and supervising driver. So if you suggest "just use self-driving", you make the system less safe, not more, because it means just getting rid of the human supervisor. This is not currently done, because it would be too insecure. It might be done in the future, but rather because less safety is going to be accepted in favor for getting rid of the costly human.
A self-driving train will happily cause a crash, if the target is "drive with high-speed in that segment currently occupied by another train". The issue is that that becomes a target, not that there is a human in the loop.
2 replies →
> Has there ever been a case where two self driving cars collide with each other?
I don't think that's relevant to trains. Cars can avoid a collision much more easily than trains can. Cars can swerve, and if they are at a distance at which line of sight tells them they will hit an obstacle, they are moving slow enough to stop before hitting that obstacle. Trains have to stay on tracks, and they need up to 4 kilometers to stop moving.
The constraints are very different, so the solutions and tradeoffs will be different too.
> We can not design self-driving cars to the safety standard that is required for railways.
That is completely false. PTC does that, and we've had it for several decades. There are several self-driving cars operating in the world right now. I have never heard of a PTC system causing accidents; those occur in places where the system isn't required (like entering Grand Central).
You keep pretending that avoiding train collisions is too magical to achieve with current tech, but train traffic is easy to limit (but slightly harder to optimize), and systems exist to do that, because rail is unique amongst all forms of travel: if you drop speed to zero for any trains at risk of collision, they are safe. Planes will plummet; most cars are human-operated and you can get rear-ended, etc.
> That is completely false. PTC does that, and we've had it for several decades.
I think you are misunderstanding the sentence you quoted.
It says something about the safety of self-driving cars. That is they don’t meet the standards required for railways. In other words the expected safety for a self-driving CAR is lower than the expected safety of a RAILWAY. (Caps for emphasis, sorry about the shouting.)
Your comment reads as if the quote claims that we cannot make “self-driving trains”. It does not say that.
> You keep pretending that avoiding train collisions is too magical to achieve with current tech
I don’t think that they do that. Definietly not in the comment you responded to.
> That is completely false. PTC does that, and we've had it for several decades.
And they somehow work with radar, lidar and GPS from inside the train? No they work with classical block based systems and on signalboxes. So all the safety systems are outside of a train.
> You keep pretending that avoiding train collisions is too magical to achieve with current tech
Huh?? Railway interlocking exists for more than a century. It's very safe and I didn't wrote anything contradicting that.