Hacker News Reader

Comment by rkangel | Hacker News Reader

> and perhaps marks the end of a long process of continuous refinement of INS technology.

Submarines are the main target for highly precise INS development over the last few decades. GPS doesn't work under water, so they rely on dead reckoning navigation over long periods of time, only getting a 'proper' fix on the rare occasions that they surface.

Also, militaries prepare for situations in which GPS isn't available. The simplest case for that being jamming - GPS is a weak signal so fairly easy to jam over an area.

View on HN

Also airplanes, satellites etc. The standard now is laser ring gyroscopes as they have no moving parts, are super sensitive, and very light in comparison to older options. Mentour pilot has an episode on how these are used in the 737.

https://en.wikipedia.org/wiki/Ring_laser_gyroscope

Specifically, Mentour Pilot talks about inertial navigation systems in https://www.youtube.com/watch?v=J89uROO8Gsc

Speaking of which, these seem like the kind of thing that would integrate well, but cell phones are still stuck with gyros/accelerometers/magnetometers that are pretty bad, especially in cities, double especially near buried power lines. The old "my heading spins rapidly in circles whenever I walk on a certain section of sidewalk" problem. Are there engineering challenges to integration or is it a spooky controlled technology conspiracy?

IIRC it's a size and power thing. You can emulate most of what a ring laser gyro does with good accelerometers and that's the better choice for most devices needing long battery life and compact size. Ring laser gyros (while small) are not that small last I checked. A quick google shows you can get them in about the size of a pocket watch. So potentially a good case addon for those that need it without the cost/power usage issues that would hit those that don't.

By "integration" I meant "turn the module into a chip" rather than "put the module in a phone" -- though your point about power could be the answer. Some lasers and optics just feel like the kind of thing that a clever team could squeeze onto a wafer, maybe with a loop in the third dimension. Or maybe just require two of them. There's probably a good answer as to why that's harder than it seems, I just don't know what it is.

It looks like there is work in that direction... but like most things it's 10years (TM) away.

https://www.nature.com/articles/s41566-018-0266-5 https://arstechnica.com/science/2018/11/tiny-laser-gyroscope...

Cool, thanks for the links!

A number of military aircraft have incorporated systems as well, and they can be combined with automated celestial fixing systems for a hybrid astro-inertial navigation system. The SR-71 Blackbird used one, the Nortronics NAS-14V2 [0], and it was pretty amazing for the time (same as everything about the SR-71 granted). Even with a much less accurate INS then this one and primitive computer/star tracker it could maintain high accuracy to enable the precise flight paths and camera activation needed for a spyplane like that to get exact images from 85k feet going Mach 3.2. At altitude an aircraft can always see the stars with the right filter even in the day, which can be used to precisely correct drift.

I'm glad this piece itself got some attention, the AIRS system to me is one of the most impressive pinnacles of electromechnical technology and fine machine crafting, feels kind of like the ultimate evolution of efforts dating back to the earliest days of watch making. But there are definitely a range of other systems from the 60s-90s still incredible in their own right but all the more so for what was made to work without all sorts of tools and modeling we take for granted now. And I think ongoing research to allow production of modern such systems at a more mass scale (perhaps a modern AINS using laser or fiber gyros and taking advantage of advancements in cheap quality digital cameras) is worthy of continued effort, just in case. Precise navigation via GPS has become central to an enormous range of modern human activity, but it is relatively fragile and easily jammed. If we can use tech to also create a reasonable fallback that seems like a valuable hedge.

----

0: https://airandspace.si.edu/webimages/collections/full/NAS-14...

The NAS-14V2 also worked on the ground - it had multiple modes (astroinertial [default], inertial, airstart and dead reckon). See 10A-17 and Figure 10A-4 in the manual. If you read the following sections it describes procedures for calibrating the system in the event of a rapid takeoff (where a fix wasn't available), for example the runway heading can be entered as a quick heading fix.

It wasn't particularly quick, mind - up to 96 minutes for the full ground procedure. They also talk about calibration in hangars, which is important because you don't want your secret spyplane sat on the apron for satellites to see for over an hour!

There are some interesting tidbits in there. The tracker automatically compensates for shockwave and thermal gradients over the sighting window, for example.

The article isn't about highly precise INS development. It's about an INS so accurate it didn't need any external references, and says such applications haven't been necessary in decades.

Submarines still use external referencing. Unlike a missile, a submarine can use the earth's magnetic field to determine orientation in at least two axis. They can also reference fixed sound sources, know their depth, and so on.

I believe highly sensitive magnetometers were used to help correlate position - namely, the Navy had done surveys and so a record of magnetometer data could be 'fit' a track to the map by a computer.

That's the point of the article. So many navigation systems can use at least some sort of external reference, a completely independent positioning system isn't necessary.

Good INS actually involves combining several sources of information to produce a result better than any of them. A Kalman filter is the normal technique for this.

You're right that magnetometry information is incorporated, but better and better INS help too. Laser Ring gyros (mentioned elsewhere) were an innovation that was first developed for submarines. I have no idea what they use now.

The AIRS has no external references beyond the initial position set by the ground station before the missile was fired.

That was why AIRS had to be so incredibly low-drift. The point of the comment in the writeup is that the DoD found that there were almost no cases where at least some sort of external reference wasn't available and thus there was no point to having such an expensive, low-drift INS.

I read that sentence as not meaning that there are no future applications of INS, but that its unlikely that any future application will need to be more accurate then the AIRS.

So its unlikely anyone will put $ into trying to improve on it, even if it continues to be used.

> What I was trying to say was that governments have continued to invest millions of dollars in improving INS in the years since the AIRS was developed. The drift you need for days/weeks/months of navigation in a submarine is much much better than what you need for a few minutes of missile flights.

The article says AIRS accuracy was not necessary back in the eighties for many applications.

It's important to note they mean drift over time. When you have at least some sort of external reference, drift is much less of a problem to compensate or correct for.

Submarines should also be able to use passive sonar and a map of the sea floor.

Two things: couldn't the sonar pings be detected if you're close enough? And does anyone actually have a detailed enough map of the sea floor to do navigation based on it?

Passive sonar doesn't use pings.

Those that do have such detailed maps, if they do, probably aren't likely to talk about them.

> Those that do have such detailed maps, if they do, probably aren't likely to talk about them.

Though I think it's a reasonable inference that the US is probably didn't have them in 2005 for the whole ocean:

https://en.wikipedia.org/wiki/USS_San_Francisco_(SSN-711)#Co...

> The seamount that San Francisco struck did not appear on the chart in use at the time of the accident, but other charts available for use indicated an area of "discolored water", an indication of the probable presence of a seamount. The Navy determined that information regarding the seamount should have been transferred to the charts in use—particularly given the relatively uncharted nature of the ocean area that was being transited—and that the failure to do so represented a breach of proper procedures.

> Nonetheless, a subsequent study by UMass Amherst indicated that the Navy's charts did not contain the latest data relevant to the crash site because the geographical area was not a priority for the Defense Mapping Agency.

Nobody said the USN had them for the whole ocean. The point is that the Navy likely has numerous methods of providing external references to a submarine's INS.

Map != position, by the way...

Reply on news.ycombinator.com