State of the art in stable accelerometers?

You can do short term corrections with accelerometers, but GPS will give much better absolute readings. You need both for INS, and they are fairly low cost anyway.

This is what Boeing was using in their airliners a couple years ago:

It was in the news because they were charged with violating export technology restrictions by selling them (in their airliners) without a license.

No breakthroughs in MEMS accelerometers.

You can fly around the world and get to within a mile of your target without GPS, but the gyros and accelerometers will cost more than the rest of the plane.

The price of the gyros and accelerometers goes down by orders of magnitude if you use GPS (or Loran, or star-sight or whatever) aided navigation.

The state of the art is GPS aided inertial nav systems, where the GPS provides the long term stability that the inertial measurements lack, and the inertial measurements provide the short-term noise smoothing that the GPS lacks.

I assume you mean pure inertial navigation, though? Why do you need it, in a world with GPS, Loran, &c?

I hope that you have some independent means on finding a an airfield and land safely, in case these external systems are suddenly shut down during your flight.

Paul

Tim Wescott wrote

The trigger for this study is a Eurocontrol (the organisation which, for the most part, rules IFR airspace in Europe) proposal to mandate a backup for GPS.

Heavy stuff already has INS but the light stuff is GPS only (if you are talking PRNAV accuracy, VORs don't really count).

Eurocontrol have repeatedly said they will mandate INS or Loran, and that "new silicon chips" will enable miniaturised INS systems.

I looked into this 3 years ago and found absolutely nothing, and I still think they are talking nonsense.

In any case, the USA is happy with GPS for the most precision stuff (using WAAS for vertical guidance applications) so none of the big avionics manufacturers will make the stuff anyway.... [looking for an emoticon of a head banging against a wall] but that has never stopped the Europeans from making some ruling which is impractical, or in this case excludes general aviation from certain classes of airspace (PRNAV) which is THE real concern here.

I am attending a Eurocontrol conference later this year and getting myself clued up on the state of the art, and if there is nothing I will tell them, because sure as hell nobody else is going to. This bizzare proposal has been in the making inside Eurocontrol for about a year.

Jim Stewart wrote

This appears suitable for the applications listed, which are basically attitude control, not navigation. There are many devices on the market used for that, for artificial horizons etc. The spec says

Short Term Bias Stability(100 sec at const. temp)< or =0.01°/sec, typical *

How does one translate that into a positional error?

In any case, a spec over 100 seconds is not useful - a 100 second GPS outage is a very short time.

Tim Wescott wrote

Is this on the market?

Oh, sure. Just Google "AHRS" (attitude and heading reference system). The Wikipedia article lists about a dozen manufacturers.

A GPS-aided Kalman filter can readily learn and compensate for systematic errors in the inertial sensors, such as bias and scale factor. The thing that kills you over the long term (during a GPS outage) is the "random walk" in your system state that is the result of integrating the sensor noise.

Any system that can place an aircraft to the accuracy that you're looking for can also place a missile, and therefore will be subject to tight military and/or export restrictions (ITAR in the USA).

-- Dave Tweed

David Tweed wrote

OK, I know that one, e.g. Garmin G1000, Avidyne, etc.

But none of these have any navigation capability. They are just artificial horizons, basically.

Agree re export controls on inertial nav technology. This would be a problem. They obviously get over it in airliners though, otherwise Boeing could not sell any planes. But those are FOGs anyway.

Presently I cannot see any sub-FOG technology which is any good for navigation. The regulators here (Europe) claim this can be done now, but for an INS which can work from DME/DME fixes one would need say

Does anyone have a formula for working out the likely accuracy, from the stability/noise specs of accelerometers?

I think Crossbow technologies has something like this for light aircraft

-- dig up their website (searching on "crossbow" and "IMU" or "INS" should get you there) and take a look.

They may even have, or be willing to cough up when asked, a chart of error vs. time after GPS blinks out.

You can get a rough (and considerably underestimated) idea of the angular error vs. time of a gyro by taking the noise (it'll be expressed in degrees/root-hr) and multiplying it by the square root of time (in appropriate units). This calculation only applies if you have a GPS-aided INS that's had time for the Kalman filter to acquire before the GPS blinks out. It'll be a disheartening number for any affordable gyro.

You can get a similar idea for velocity (not position!) error vs. time of an accelerometer the same way, then you can roughly integrate that velocity error by multiplying it by the time again. Once again you'll be disheartened.

What's the future of LORAN, in your mind? It should work like gangbusters if there's coverage, and the same environment that makes GPS receivers cheap can make a LORAN receiver cheap, too. It works as an aid to INS solutions, to boot.

-- snip --

Just because a device is a Defense Article doesn't mean it's all that hot. Per State Department rules a Defense Article is anything at all that's specifically designed to go into a Defense Article. It's sort of like bureaucratic VD.

So if I'm working on some weapons or sensor system that is export controlled and I make a drawing of a cruddy DB-9 to DB-9 cable for that system, then even though I haven't made anything more special than what you can get at the local Radio Shack, I've made a gen-e-u-ine Defense Article, and (technically) I, and my client's export control officer, can go to jail if I post a drawing of it here.

It's silly, but it does relieve the State Department of the need to understand what the heck they're regulating.

As mentioned, those gyros are very nice for what they do. They're light, they're rugged (IIRC Systron has some that can be shot out of cannons and still work), they work at fairly high rates, they're easy to apply, etc.

But they're not good enough for INS use without a lot of help from an external reference like GPS (or LORAN).

Tim Wescott wrote

I found them at xbow.com and this

seems to be their 'navigation' product. The spec (if I read it right) is 0.1 degree per second drift, which is insufficient for GPS-lost navigation by a factor of 100 at least.

Then go somewhere that sell the real DE9 connectors and confuse the idiots.

Not much change in the last decade except lower power, smaller and 3 axis packages. But there are still 2 basic designs, the common piezioresistive type available from many manufacturers, which has relatively high noise levels, and the much harder to fabricate capacitive accelerometers (harder due to a much deeper etch). AFIK the capacitive type are only manufactured by Kistler and not cheap, but they have more than an order of magnitude lower noise levels than piezioresistive types of the same range.

And if it's too cheap you can probably look at what the weapon industry is doing:

Terrain reference navigation!

For the extreme case, see this month's issue of Scientific American, the article on "The Big Bump". It discusses a super accurate platform used in a space probe designed to detect the presence of "frame dragging".