Because it can't. There are always cumulative errors in INS. There is no way around this without external references, but then it's no longer inertial navigation and it's inertial-assisted navigation.
An underground navigation system based on triangulation of UWB cells would be a better solution than some nonstarter project the size of a refrigerator that requires liquid nitrogen.
There is no liquid nitrogen involved here. The instrument from the article is actually rather big, current generations of quantum IMUs are roughly half this size with lots of room for miniaturization.
One big advantage of these atom interferometers is that they actually don't need to be recalibrated because the reference is the wavelength of the lasers which can be controlled with extreme precision.
A big disadvantage is however the limited repetition rate, which is on the order of only 1 Hz at the moment. Currently, combinations with "classical" IMUs seem most promising, and there is lots of interest in these devices for applications in planes, cars and spacecraft.
Actually, in a train context, inertial is pretty damn high accuracy because you already have the world's best odometer showing (A) how many times your known size wheel has turned on your known track; and (B) you can test many axels simultaneously to ensure no slippage (I guess you wouldn't test the wheels as they move faster creating a higher frequency sampling requirement for ~no benefit whereas a dedicated axel feature could live within an environmental enclosure protected from dust and grime); (C) you are constantly doing exactly the same trip-segments over and over again.
So for a train with an offline positioning requirement, I'd suggest that an odometer based solution is close to ideal.
Re odometer - don't the wheels of a train slide a little bit? I know the train is big pile of heavy iron, but still. My naive thinking is that when braking or during a rush start some extra distance can be covered.
If you are averaging axels across multiple carriages slippage is unlikely to amount to much. Moreover, because you are traveling known segments, you can reset your position at each station or known intersection/detectable segment terminus, so you're going to have zero accrued drift at that point. It's a perfect deployment scenario for an odometer based solution. If you want to be higher confidence, sensor-fusion with an IMU, laser TOF/LIDAR, camera, ambient light sensor, radio signals, or MEMS microphone can verify position. Et voila! - no need for GNSS.
Generally trains do everything they can to avoid slipping. They have anti-lock breaks and traction control just like your car, and stuff your car doesn’t have, like “sanders” which pour sand onto the track just in front of the wheels.
Regardless of material, dynamic friction is always lower than static friction. So for maximum acceleration and breaking it’s important to ensure you wheels stay in “rolling” mode of interaction, and don’t slip.
Is that necessarily true of this quantum thing? I know nothing about it except this article, theoretically if it kept track of exact Plank lengths or something, then there would be no errors to accumulate, right? Lots of the things that seem intuitively true break down in weird ways when dealing with quantum effects.
TL;DR-TL;DR: says the opposite of your implied claim, "atom-gyros are set to outperform light-based gyros"
TL;DR: this is a StackExchange question with 1 answer, noting it is indeterminate if a quantum gyroscope would be more accurate than a laser-atom-based one.
It looks like you rushed through and missed that in this context, TFA is describing an atom gyro.
That leaves conversation at a point where either A) we assume the scientist interviewed knows what they're doing, or B) following your unstated lead, assume they're a crackpot and the whole article is irrelevant because they're untrustworthy, and thus in an ideal world, there's 0 comments on the article.
All accelerometers tell you is the direction of the acceleration vector (ie how speed is changing and in which direction). You still have to add the individual vectors to derive where you actually are.
And if you don't sample fast enough and your acceleration has frequency components at frequency comparable to your sampling, the acceleration you measure may not reflect where you actually are (ref Nyquist sampling theorem)
Imagine sampling at 1hz, and you just happen to have a bump every 1 sec (eg your wheel happens to have a flat spot and is turning at 1Hz), followed almost instantly later by a bump in the opposite direction. Your sampling only sees (say) the +ve components, misses the -ve and accrues a bunch of error.
If you can sample fast enough, you can minimize this sort of error, but you can't really make it go away.
Oh, btw, if you make it work well enough you're considered munitions for export control purposes, so limits the number of countries you can sell to. Same reason civilian GPS units stop working somewhere around 1200mph
These are all excellent elucidations of classic mechanical principles making it hard. I'm not sure they're enough to make me say the scientist/institution in the article a priori has it wrong, especially because it's not a one-off dude just messing around.
An underground navigation system based on triangulation of UWB cells would be a better solution than some nonstarter project the size of a refrigerator that requires liquid nitrogen.