Using quantum mechanics to track trains
Steve Foot and Steve Venables, TFL & Joe Cotter, Imperial College London
We’re going underground: no, not the 1980 hit for the Jam, but instead onto the real London underground to find out how quantum mechanics is being used to keep track of the oldest and one of the most extensive subway systems in the world. This story was first broadcast in February 2025. Chris Smith’s guides were TFL – Transport for London – engineers Steve Foot and Steve Venables, and Imperial College London physicist Joe Cotter…
Steve Foot – So the London Underground network’s really complicated and extensive. We’ve got 11 lines, there’s 272 stations, we move approximately 1.2 billion people a year, we have 400 kilometres of track, and then we’ve got five different signalling systems. So it’s a really complicated system to operate.
Chris – I’ve joked in the past, but it’s actually true. You’re moving more people under the streets of London every day than the populations of whole countries, like Sweden, for example.
Steve Foot – I am still amazed at how complicated the system is, and what we achieve each day, yes.
Chris – So tell us about the problem that, as you see it, that having better positioning of trains could help to surmount.
Steve Foot – In terms of improving the way in which we maintain the asset, and become more efficient and effective in the way that we do that, then having data is fundamental. Obviously, modern technology would allow us to use the train as a sensor, and it could run down the tracks, collect data around all those assets. And then if we’ve got an accurate positioning of where that train has collected that data from, if there’s an issue, we know exactly where to go back to collect that data. And if we keep running the trains and keep collecting the data, you can then start trending the changing condition over time. But you can only do that if you know that you’re capturing the data in the same place each time.
Chris – I was talking to someone the other day who’s working with car manufacturers. They’re collecting data collected by cars as they drive across the country’s roads. They’re kind of regarding cars now as a sensor on wheels. They know where the car is because they have GPS. In a tunnel, that presumably does not work.
Steve Foot – That’s correct. GPS, you can’t rely on it in the underground sections of our network. And therefore, an alternative form of accurate positioning is needed. The technology that signalling uses identifies the section of railway that the train’s operating on, but it doesn’t give you the absolute position. So if you want to go and identify where there’s a defect on the railway, you need the absolute position to know where to go to.
Chris – Got it. So you know a train is on a section of track between two signals, but if it has encountered a bump in the road, you don’t know where that bump is except it’s somewhere between A and B. Whereas if you had a really precise way of keeping tabs on where the train is at any moment in time, you could say that’s where the problem is. That’s where we’ve got to send the repair guys to.
Steve Foot – Yeah, so the signalling system uses a train being in a geographic section in order to keep the train safe, keep them separated. We do also have other infrastructure installed on the network, but it doesn’t give you that accuracy we desire. So using quantum sensor could give us the level of accuracy such that we could do, similar as you described with the cars, use a train as a sensor to look at lots of different assets.
Chris – Steve Foot. And this is where Dr Joe Cotter, a physicist at Imperial College London, comes in. He’s pioneering quantum sensors based on clouds of supercooled atoms, the movements of which he can read with exquisite precision and accuracy using a laser. And by adding up all the movements the atoms make, you can work out how far and how fast the thing holding them must have moved. In this case, that would be a train, but it could equally be a boat, a car or a plane. In other words, this would give us another way of keeping tabs on position, but without relying solely on mechanisms like GPS, which can sometimes fail or even be misleading.
Joe – In my labs, what we’re doing is developing a new kind of sensor that harnesses quantum mechanics to make more accurate measurement devices. One of the nice things about this approach, so inertial navigation, is that the ability to position yourself is self-contained in the vehicle that you’re travelling in. So you don’t rely on anybody outside. It’s all on that vehicle.
Chris – In basic terms then, what is inertial positioning and how does that actually work? Before we get into the quantum way of doing it, what’s the basic principle?
Joe – So inertial navigation relies on measuring the motion of the vehicle that you’re in. So in particular with the acceleration or the rotation of the platform. And then you have to do some maths to convert those measurements of inertial signals into a change in position in a map frame. So for the London Underground along the track, for example.
Chris – So if I reach in my pocket, take out my phone, it knows that it’s moved upwards, sideways, along a bit at what sort of rate, over what sort of time, and it could then work out, working backwards, doing that maths you mentioned, where it must now be relative to where it started. Is that what you mean?
Joe – That’s exactly right, yeah.
Chris – And you’re saying you want to do that with atoms, quantum level, tiny particles. Why do we need to go any further and use atoms?
Joe – By harnessing quantum mechanics, we think we can make more accurate sensors that could enable you to navigate for longer in the future. So it’s about a next generation kind of approach to this inertial navigation.
Chris – How does it work?
Joe – So for our quantum inertial sensors, we start by laser cooling a cloud of rubidium atoms to a few microkelvin in temperature, about a millionth of a degree above absolute zero.
Chris – And why do they have to be that cold?
Joe – When they’re cold like that, their quantum properties start to come to the forefront. And so to describe their motion, we need to treat them like waves. And it’s that wave-like nature that we take advantage of in our sensors.
Chris – How does that work then? We’ve got a sensor, it’s got these very cold rubidium atoms. How do they know where they’re moving? And how do you log that? How do you extract that information?
Joe – We have a laser and we release our atoms and the atoms interact with the laser. The laser is essentially just a wiggling electric field. And we use those wiggles like a ruler. And so if it’s fixed to the vehicle and our atoms are in free fall, we just measure how many of those wiggles in the laser the atom moves through in a given amount of time.
Chris – Put simply then, if the atoms are in one position and the train moves forward, the atoms are left behind a bit. So they’re going to move down the ruler a bit, your laser ruler, and you can register that. And that would correspond to an acceleration in a certain direction. Is that how it’s working?
Joe – That’s exactly right. Yeah, that’s the principle behind it. We’ve taken demonstrator systems on the tube already and we’ve had some success.
Chris – What you took, basically, I’ve seen your gear in the lab here. It’s a whole room. So how on earth did you get that on the underground?
Joe – So we packaged up one of our sensors and deployed it on a London Underground train.
Chris – Did you get any funny looks? Other passengers look a bit strangely at you?
Joe – It was a test train. There were no passengers.
Chris – Can it work in three dimensions? You’ve mentioned a sort of ruler analogy. That’s one axis. Can you do this in multiple axes so that you’ve literally got a tracking in a three-dimensional space system? Because that’s going to be needed. If we’re going to need to know where an aeroplane is or a boat or something, we’re going to need all of the degrees of movement to be tracked.
Joe – Yes, absolutely. So in the laboratory, we already do it in three dimensions. For the transportable work we’re doing with TFL, we’re focusing on just one axis for now, just to overcome these engineering challenges and environmental challenges. But no, you’re absolutely right. The next step will be to develop a new six-axis sensor for the railways.
Chris – Joe Cotter. Another London Underground engineer who was actually instrumental in getting this collaboration off the ground, is TFL’s Steve Venables. I asked him if they can get this to work, what sort of a difference it might make?
Steve Venables – This has potential to change the landscape of how we operate our railway. If the railway breaks for some reason, we’d be able to fix it quicker, cheaper, and more efficiently. We’d also be able to understand how that asset’s performing, so we can fix it before it does fail.
Chris – I also like the juxtaposition of one of the world’s biggest and oldest railway systems, lining up with some of the world’s newest cutting-edge navigation technology.
Steve Venables – Disruptive technology like this has to be looked at. We can’t continue doing the same things and expect different results. We need to start challenging how we look at things going forward, and this is just one example of where we’re looking to partner with industry, academia, to develop our own internal capabilities.