Don't trains run on a fixed path, meaning we can use more traditional "positioning systems" like, umm, math? Or placing passive sensors or paint a big number on the wall?
back
2 comments
Maths isn't a sensor technology. Wheels slip, air pressure fluctuates, rotary counters misread, errors accumulate, so better raw data could get you a more precise estimate of your current position. Various position technologies are already in use (e.g. "balise") but they have their limits if you're after cm-level accuracy.
That said, I'm not buying that quantum INS is required for this over more established techniques like visual-inertial odometry. However, the UK seems to have a whole bunch of active quantum INS research projects, so I wonder if the tube stuff is just a weak public cover for the military/civil emergency applications that make the technology actually interesting.
> so I wonder if the tube stuff is just a weak public cover for the military/civil emergency applications that make the technology actually interesting.
The project is quite explicitly a defence project. IanVisits is just a transport blog, so the article focus on the transport aspect of this project. But reporting else where makes it quite clear this a defence project.
https://www.theguardian.com/science/article/2024/jun/15/lond...
The only reasons for the Tube being involved is as a test bed for the technology. It’s cheaper to put this box on a train and test it, than it is to put it on a boat or a submarine. It’s also handy that the researchers involved are based in London, so the commute to their test bed is nice and short.
An experimental cryogenic device doesn't sound very good in terms of reliability. A train could have a few free rubber-coated wheels dedicated for precise odometry, there could be cameras with optical flow/reference markers on the tunnel walls every so often, virtually anything seems to be better than sinking millions in quantum devices.
Maybe they used trains as an example because that's the only application for which the accelerometer technique is currently sufficiently accurate.
The worlds longest linear encoder?