The innovation here appear to be the chamber. (caveat: I still need to read the paper). A few thoughts follow.
They are going for a UHV***, probably <1e-10 mbar, but want to do it with a passive pumping design. In an ordinary chamber to get there you'd do a pump down, then bake it to remove the water. Then use an ion pump plus occasional firing of a titanium sublimation pump to remove the hydrogen that leaks in (ion pumps don't work so well for hydrogen at low pressures). They have some expensive chamber that doesn't suffer from these leaks.
Still a lot of other work to do though. I see a table full of optics to make the MOT*. Also see a pair of anti-Helmholtz coils for the magnetic field. They could maybe replace those coils (which probably drive a few amps) with permanent magnets (assuming they don't need to switch the fields on and off which I imagine they don't because it can't imagine a pulsed design that wouldn't risk missing data)
Passive getters work for a good long time with low leak rates. These you can just buy**. To be honest I'm not sure why you need such a fancy chamber. I'd guess a good NEG pump, a small chamber, and good seals would make you fine for a long time, but I'd need to do the math to be sure.
Permanent magnets are typically a no-go, because these sensors employ matter-wave interferometry, i.e. interference between atoms. Those unfortunately do not only react to gravity, but also to magnet fields, so these need to be suppressed as far as possible, especially gradients.
And yes, this will absolutely be a pulsed design.
I think, the breakthrough here is that they maintain a decent vacuum even without NEGs, which you can of course buy off the shelf, but they still tend to be quite bulky.
Also, they state a vacuum of about 10^{-7} torr. Far from UHV.
> Quantum sensors for PNT use atomic accelerometers or gyroscopes
There is one fundamental problem with using any sensors for global positioning: a stationary frame of reference is indistinguishable from the one moving with a constant speed on a straight (actually geodesic [2]) line (inertial frame of reference, general principle of relativity [1]). So accelerometers and gyros indeed can give you accurate orientation and acceleration/deceleration data, but they will inevitably accumulate uncertainty for long eventless movements, so use of GPS will be needed to occasionally calibrate the true position. The obvious options is to augment this data with readings of wheels rotational, airspeed sensors, radars, etc, then such systems may last even longer without cross-calibration with GPS.
True, but what you describe is related to the measurement and sampling accuracy. Until such sensors hit the Heisenberg uncertainty levels of precision on the readout/sensing, such uncertainties may be decreased further, there is no physical constrain on that (more sensors may help with that also). While it is in principle physically impossible to even identify the fact of movement linearly with constant speed without introducing external information (radar, GPS, wheels rotation, airspeed, magnetic and gravitational field change, Doppler shift in radio waves, etc). This is the relativity principle in action, which says that all the inertial frames of reference are equivalent.
But then it also usually impresses me with how people think companies haven’t thought of and tried the things they thought of in reading the article and thinking about it for a few minutes.
These forums have an interesting kind of collective intelligence. People commenting here rarely (if ever) have full knowledge of the specifics of the particular instance of the topic at hand, but often have very deep knowledge of the associated theory.
In my experience, if you take any 10 people and form a team or company to make some product, they won't have anywhere near the breath or depth of knowledge that people here could bring to bear. What they have is time and persistence, but I guarantee you that they're "flailing about" and wasting time on many wrong or suboptimal solutions. In most cases, someone here could set them on the right track within minutes.
It's so strange to me that the Internet enables this, but we use this superpower for essentially nothing more than casual chit-chat. We show off our knowledge, where we happen to have a particularly impactful trick up our sleeve, win some meaningless points, and feel smug about ourselves for an afternoon. I do this all the time. It's fun.
But what would happen if we did this properly? Like Uber? Made it into an app where startups, research teams, and software developer houses could use the collective knowledge and advice of hundreds of thousands of people instead of just the dozens on their staff? Just click here to get expert advice on your question, $55 guaranteed price? Would you click?
Maybe it is time that we all plugged into the group consciousness and started contributing to the collective... /s
These are code words I almost never use. I'm no Gary Larson [0].
But someday scientists will discover that they were plum fools for dropping the advancing momentum in vacuum science as badly as they did when solid-state electronics needed to be quickly developed & adopted.
>Made it into an app where startups, research teams, and software developer houses could use the collective knowledge and advice of hundreds of thousands of people instead of just the dozens on their staff?
Maybe it would be needed that those (startup, research teams, software developer houses) learn before to listen to their customers/users.
Everyone of course is different and has different experiences, but in the last few years (personally) I found an increasing "detachment" between the people that write the software and their users.
Maybe it us just me, but it seems like users (even paying ones in case of commercial tools) are considered more a nuisance than a resource.
To some extent, that's just StackOverflow (and the other StackExchange sites). Not quite as commercially focused as you imagine, but at least the code parts definitely are used like that.
For more individualized problem solving, this just amounts to consultancies, no?
I think engineers tend to have a borderline-compulsive need to understand why the simple/obvious solution didn’t work. Sure sometimes the comments are along the lines “duh, just do X” but I think they’re usually “why not do X?” asked earnestly. It’s hard to tell on the internet but I think most engineers ask questions like that not to second-guess decisions but to get a better understanding of complicated systems.
We're talking about quantum mechanics and vacuum chambers. If
you can spec such a system... congratulations, you're an expert. Kathy down the street doesn't even know this technology exists.
On patents...if you're in a sufficiently unexplored space, simple ideas become eminently patentable. So -- given the context -- yeah, pretty much any thought made that is aware of these systems is likely patentable.
Profitable to patent? That's a different question.
The OP seems to know something, but I wouldn't go as far as taking the comment above as coming from an "expert"... the fact they feel confident to make recommendations about improvements on the chamber without even reading the paper actually tells me they are just armchair expert.
I am an expert. I have a PhD in experimental atomic physics. My work thesis work was on laser cooled atoms and ions. I have multiple publications in this topic area^. I started and run a company that is making the first commercial instruments that rely on laser-cooled atoms.^^
There is nothing wrong with making few conversational comments without being aware of the details in a particular paper. This is how scientists talk to each other BTW. If I was at a conference with the authors I'd say something similar to the OP above, then they'd agree or disagree.
I don't know about that. At least for myself, when I say that I've read a paper I mean that I've read it thoroughly from end-to-end and thought deeply about it, not unlike what I do when I'm doing a peer review. (I.e., "grokking" the paper.)
There are way too many papers to do that for every one that comes out in my field, though, so I've gotten pretty good at skimming a paper quickly and making a snap assessment. I might say I'm aware of or acquainted with said paper, but I'd never say I'd really _read_ it.
So charitably speaking, I'd interpret the top-level commenter (avsteele) as meaning that they've only skimmed the paper and these are off-the-cuff comments, rather than that they haven't looked at the paper at all.
Does passive pumping mean a chamber full of baked zeolite or something that's then chilled to adsorb all the gas? I guess I should read the paper too, but I don't have a copy.
I think CRTs were mostly manufactured using conventional high-vacuum pumps with moving parts or liquids, not sorption pumps. They didn't normally include a pump in the cathode-ray tube itself; the pump stayed at the factory. And I think you're mistaken that the CRTs ever included any zeolite. Perhaps you're confusing zeolites with getters, which react irreversibly with any stray molecules that manage to get into the vacuum?
DARPA has been funding work on better gyros and accelerometers for about 10 years now. MEMS gyros keep getting better. The good ones are down to a few degrees per hour.[1] That IMU/GPS is $3,386.09. Not down to the phone/vacuum cleaner price point yet.
This is nowhere near as good as the device described in the parent article, but it's good enough for a lot of applications. If you can get a GPS fix every few minutes, it's good enough for many nav applications.
(When we did a DARPA Grand Challenge vehicle in 2005, we had a MEMS gyro with over 10 deg/minute drift. That was too much for the mapping to work properly. We needed < 2 deg/minute for the laser scans to align. Better IMUs were back-ordered due to the Iraq war, and attempting to work around the problem with filters and a magnetic compass were not quite good enough.)
"The whole spectrum" for GPS FHSS being a relatively tiny frequency range. So still trivial to jam.
There are physical restrictions on transmitter aerials which make really wide FHSS impossibly difficult. Especially from a satellite. So GPS still transmits on the same standard 5 frequencies, with some special codings.
The antijamming features work a little differently. If you know a random code sequence you can use various digital noise reduction techniques to improve the SNR and help punch a signal through noise.
But if you blanket blast the transmitter frequencies and their associated components with noise, you're still going to have an effective jammer. It will just have a smaller effective range.
In the military context, violence is an option. Powerful radio transmitters are vulnerable to anti-radiation missiles: https://en.wikipedia.org/wiki/Anti-radiation_missile In a free fire zone a GPS jammer would have a lifespan measured in minutes. A hypothetical conflict with a near-peer would involve a lot of jam and counter-jam drones destroying each other, until both sides run out of equipment.
Not as trivial as jamming detection and countermeasures, so simple that it has found its way into base model commercial components. If you are in the air: use directional antennas - why are you listening to the ground? If you are on the ground: you get an easy 30deg arc pointing to the hostile transmitter... put a building or terrain between it and you.
I thought so for a long time, but then learned about spread-spectrum and frequency-hopping (with cryptographically randomized pattern) radio communication techniques which are employed in military-grade communications, and suddenly reliably jamming GPS without jamming literally everything becomes very hard.
I'm literally experiencing GPS jamming when I'm driving my car near president palace. Like one second I'm moving on road and next second I'm jumping to random places few hundred meters away until I passed long enough. Navigator sometimes goes crazy because of that. So at least civilian GPS is possible to jam.
The US doesnt have a presidential palace, so this must be in a foreign country. There isnt really anything the US can do for localized jamming in a foreign country.
TLDR; Sandia researches dead reckoning with a new kind of atomic accelerometer and gyroscope.
It has nothing to do with avocados, aside from someone comparing the device’s size to that of an avocado, and that was probably the only thing the journalist understood so they went with it…
Indeed. Also, the old system apparently "needs thousands of volts of electricity," which is too much for airplanes, despite them heating food with microwave ovens which typically use > 10 kV to drive the klystrons that generate the microwaves.
To be fair the "thousands of volts of electricity" nonsense is in their own press release not just this fluffy article.
It's really hard to get the people writing for a lay audience to actually understand the difference between "Simpler for lay people to understand" and "Simply wrong". This is sort-of fine if you're making a Hollywood movie, but it makes serious research look pretty stupid. Unfortunately the rare person who understands and knows how to communicate this stuff to lay people is probably more valuable as a lecturer or in some management role than writing press releases :/
How is this an improvement over other INS like LRG?
INS mechanisms suffer from the same problem: that they only detect linear or rotational acceleration. To reach position, double integration must be performed, which rapidly accumulates error (drift) because velocity and position both have to be maintained to calculate the next position.
While I heard that the effect of double integration is bad, attempting it as a student with consumer MEMS devices was quite dramatic. Really underscores the importance of Kalman filters and updates from sensors with fewer derivatives.
I used to work on AUVs with inertial navigation, and a Kalman filter accepting input from GPS at the surface and a DVL or similar underwater, as well as the inertial unit, was pretty much standard equipment. The device in the article has to have miraculously low error in order to eliminate the need for GPS.
even worse, your (single) integrated gyro estimate must then be fed back into your double integrated accel filter to subtract gravity and further compound errors :)
I would think camera-based navigation would be fairly successful at this point. We can easily store large amounts of image and terrain data, and deep learning based image recognition systems are quite powerful. I would think that by limiting your image search space based on your likely position would make the problem even easier to solve.
Terrain based navigation is certainly nothing new(1950s?), so there is probably a large pool of research into it, but I wonder what's been done with recent pattern recognition advances?
There’s still the issue of compute power required for a lot of the CV algorithms you’d want to use. There’s also the question of what data you are using; Satellite imagery is much harder to match with a ground perspective and ground imagery(e.g street view) is very tedious to collect and keep updated.
In areas with unique features (dense urban) localizing to within ~50ft has been demonstrated pretty robustly but more accurate methods of localization are a lot more variable and depend on the datasets, cameras (monocular vs stereo), compute, etc.
> I would think camera-based navigation would be fairly successful at this point. We can easily store large amounts of image and terrain data, and deep learning based image recognition systems are quite powerful.
Aside from the fact that the amount of data required for that would be orders of magnitudes larger than you can reasonably store, that wouldn't work at all in areas with little terrain features like deserts or the open sea.
Don't look down. Look up. That's been a navigation standard for years. Funnily enough replaced by GPS. https://airandspace.si.edu/collection-objects/astro-tracker-...
That doesn't really work in a sub though.
Also compasses (especially of the 3D variety) can be a great help.
It’s certainly conceivable that you could store a compressed representation of the image features. Especially if you’re operating in a more limited environment (e.g. a large metro area), it would be very feasible to only store extracted features from the image data that you could use for matching.
Just like dead reckoning works today, you start from a known point at known attitude and integrate up accelerations and rotations to keep an up to date position.
Whenever you can get a GPS signal you use that to reduce the drift.
Cheap systems can do this for seconds, 10k$ buys you 15minutes (MEMS), 100k an hour (laser gyro), everything above is under export restrictions and NDAs so noone will tell you what the state of the art is.
Looks like with these frozen gasses one can build even better sensors, which will make no difference for civilian use because it will be even more expensive than existing systems and also because we wouldn't be allowed to use it.
The whole premise that this might be something in everyday use one day seems completely flawed. Even at the 1cm cube of physics it’s huge compared to GPS and it uses incredibly expensive materials.
If you’re a nuclear submarine or stealth bomber it makes loads of sense. For anything else I don’t really get the point.
If you're anything to do with military, I bet you _really_ badly want this over GPS. GPS can be spoofed or blocked over an area. So every drone, every group of soldiers, every truck, every boat really needs one.
Even just as a person, having a nav system that doesn't need to sync before it can be used would be nice, and no possible issues with buildings, trees, hills, etc.
I recon that’s what said about solid state gyros and accelerometers, vibrating structure gyros were initially developed for the military now they are like a $1 a pop…
If this works I can’t see a reason why in 10-15 year you won’t be able to fit it on a single chip.
There's a fairly obvious application in pretty much any military theatre. GPS is vulnerable to spoofing, having a second source of truth lets you know you're even being attacked. Comparing an IDS to a GPS at least gives you a canary.
> how will it determine its position without a reference
It won't. It needs a starting position from an external source, and without periodic updates from some external source, errors will build up in this system just as with any inertial navigation system. As I understand it, this system is just supposed to build up errors much more slowly than existing inertial navigation systems, so it can go longer between updates from an external source before its error becomes too large to be acceptable.
I'm also confused by this, does it get set a reference point once? If so what is the accuracy degradation, is it fixed over time or acceleration?
If it's updated periodically what does that system look like? Is the advantage that it can manage on its own "offline" for a usable period, unlike GPS which needs an active connection?
And I agree that the answers to my questions are probably not public, this is likely going to be very useful for things that go boom, in addition to a lot of other civilian applications.
This reminds me of a spacecraft navigation system that employed an almost perfect sphere, that floated inside the centre of chamber that employed a detection system that corrected the spacecrafts direction to keep the sphere in the centre of the chamber. On the drawings it was referred to as the space crafts "consciousness"
Unfortunately i cannot find any reference to it anywhere.
> Using a starting point as a fixed reference and calibrated to the patterns of Earth’s magnetic field, data from a quantum navigation system will give us all the information we need to accurately determine where we are.
Why is the Earth's magnetic field important? And isn't that just as easy to "spoof" (if not more) than a GPS signal?
Magnetic field strength falls off rapidly - a spoofer would be extremely large and heavy to reach even a very short range. It’s not going to work against aircraft.
The 10kv power supply for an ion pump could be quite small, as the load current is proportional to leakage. The real innovation seems to be offering a small package without the voltage induced by the ion pump possibly effective the rest of the environment.
The actual paper itself would make it clear what the real advance is, but it's paywalled. 8(
Url changed from https://asiatimes.com/2021/11/atomic-avocado-could-change-th..., which seems to be copying from other sources and mentioning them, but not linking to them. That's a shoddy practice and from a quick look at other submissions like they do it a lot, so I've banned the site.
If anyone knows of a better article on this, we can change the URL again.
They are going for a UHV***, probably <1e-10 mbar, but want to do it with a passive pumping design. In an ordinary chamber to get there you'd do a pump down, then bake it to remove the water. Then use an ion pump plus occasional firing of a titanium sublimation pump to remove the hydrogen that leaks in (ion pumps don't work so well for hydrogen at low pressures). They have some expensive chamber that doesn't suffer from these leaks.
Still a lot of other work to do though. I see a table full of optics to make the MOT*. Also see a pair of anti-Helmholtz coils for the magnetic field. They could maybe replace those coils (which probably drive a few amps) with permanent magnets (assuming they don't need to switch the fields on and off which I imagine they don't because it can't imagine a pulsed design that wouldn't risk missing data)
Passive getters work for a good long time with low leak rates. These you can just buy**. To be honest I'm not sure why you need such a fancy chamber. I'd guess a good NEG pump, a small chamber, and good seals would make you fine for a long time, but I'd need to do the math to be sure.
*https://en.wikipedia.org/wiki/Magneto-optical_trap
**https://www.saesgetters.com/products-functions/products/neg-...
***https://en.wikipedia.org/wiki/Ultra-high_vacuum