View Full Version : Really cheap INS?

15th Jul 2011, 08:22
I was thinking the other day about how cheap you could make an INS - I then realised you could probably do it with the later smartphones on the market.
I've got an HTC Desire HD and inside it's little case is -
Three-axis accelerometers
Three-axis gyros

I don't know how accurate they are or what the resolution is, but could a cheap & cheerful INS be obtained from a suitable app? (Yes I've looked, couldn't find one)
And yes GPS has made a lot of this sort of thing redundant, but it'd be interesting to see how your mobile phone works as an INS in-flight .... in flight mode of course.

15th Jul 2011, 08:56
You could - if the Desire HD actually had gyros ;) It doesn't, it deduces its orientation in space from the output of its accelerometer, wherever the gravity vector points is down :) The iPhone 4 on the other hand does have a solid state gyro system in addition to accelerometers, I suppose one could do something with that...

15th Jul 2011, 11:04
It's actually a cheap IRU that would be in demand these days, not an INS.

15th Jul 2011, 16:16
The low cost sensors are not nearly accurate enough for long range
navigation, maybe by 2 or 3 orders of magnitude in terms of noise, scale
factor and offset drift characteristics. The better sensors of the mems
or quartz type cost around 50 x as much as the consumer grade types and are
used in standalone heading reference systems from vendors like meggit
and goodrich, to name a couple.

There quite a few vendors of low cost sensor systems designed for
robotics and short range orientation and navigation applications. They
tend to use mems accelerometers and rate gyros from analog devices etc.
These have reasonable performance, but there are even lower cost mems
rate gyros and accelerometers available for use in consumer electronics
(games, mobile phones, hand held computers) and vehicle body rate
control. None of these are anywhere near as good as fibre optic
or laser ring gyros, which are used in high accuaracy ins systems.

A few examples using low cost technology:

ADXRS642 | Vibration Rejecting 250/s Yaw Rate Gyro | Gyroscopes | MEMS and Sensors | Analog Devices (http://www.analog.com/en/mems-sensors/gyroscopes/adxrs642/products/product.html)

autopilot: Do it yourself UAV (http://autopilot.sourceforge.net/index.html)

Avionics, Flight Control and Engine Control Systems by PC Flight Systems (http://www.pcflightsystems.com/?gclid=CKLjkobPg6oCFeYJtAodi1rszA)

No connection, usual disclaimers etc.

There are loads of academic papers describing projects built around low
cost sensors on the web...

15th Jul 2011, 18:04
I was thinking the other day about how cheap you could make an INSI'm curious about what started your train of thought.....
I understand you're referring to "INS" as "Inertial Navigation System".
Even a "near-perfect" INS isn't perfect, because of drift and error issues, and is usually updated from other sources, like DME fixes in the olden days, or now GPS.

syseng68k, You're right, there is a lot of low-cost sensor 'stuff' around nowadays.
Thanks for the hint, I should really start to get familiar with all that 'stuff', and see if I can come up with new ideas to use it. I'll dig into your links.

15th Jul 2011, 19:31
IRUs are expensive.About $250,000 each. Airplanes capable of oceanic or remote land mass flights almost always have three. The cost is justified for that purpose.

RNP AR requires two sources of position information for RNP of less than 0.30 or missed approach of less than 1.0. (also known as dual string)

The three IRU airplanes are good to go simply because of their oceanic/remote land mass nav capability.

The FAA, at least on paper, will accept one IRU for dual string. But, $250,000 is a stiff equippage price for a domestic aircraft to simply satisfy the entry into the RNP AR "dual string club."

An IRU with at least the accuracy of the Honeywell unit for a fraction of the price would likely find a lot of buyers. It wouldn't have to provide an attitude platform, just provide good position information for a maximum of 30 minutes in the event of loss of GPS during a dual-string RNP AR IAP.

16th Jul 2011, 00:42
Just random thoughts ..... thanks guys.

16th Jul 2011, 01:14
I believe some cheap devices depend on magnetometers to sense their attitude. They're paired with a gps so they know what the magnetic field is supposed to look like wherever "here" is but I wouldn't count on it at extreme latitudes and FL350.

For a cheap demonstrator/toy, there's a free bubble-level app for Android phones that have the appropriate hardware. I'd say it's accurate to a minor fraction of a degree. Not so good for CAT III, but (especially if the sensor is magnetic) great for noticing that your deck angle is 15 degrees when the million-dollar computers have just had a domestic spat and thrown up in your lap. Not that anyone would whip out their cellphone at that moment and start fiddling with it.

At a more professional level, this thing is about USD5000: Inertial Navigation System: INS GPS, MIDG INS/GPS, IMU Global Positioning System, unmanned vehicles, servo controller (http://www.microboticsinc.com/ins_gps.html). See also rc_uav_autopilot.html on the same site; don't know the price on that, but more than bare MIDG.

16th Jul 2011, 05:12
For a cheap demonstrator/toy, there's a free bubble-level app for Android phones that have the appropriate hardware. I'd say it's accurate to a minor fraction of a degree.

Yes I've got that app and a couple of others like it to help me do fabrication in the workshop.
This one's pretty handy at times - https://market.android.com/details?id=com.blackdot.LaserLevel&feature=search_result

Edit - This looks interesting as well - https://market.android.com/details?id=org.drfischer.android.level&feature=related_apps

16th Jul 2011, 17:39
The cost is more about the approval process, not the piece of equipment. The approval process is so long, that the electronics are hopelessly outdated before it ever comes to market.

Look at the FMS, the 'upgrade' is to 1 MB of memory, which your cell phone has more...but just try to market a new one.

We are working on the RNP redundancy requirement, because in some airports, such as in China, the only way in/out is RNP, and if you lose one, technically you are not allowed to use the RNP procedure, so you are really stuck.

16th Jul 2011, 19:28
It's a couple of years old now, but there is a paper here (http://www.cl.cam.ac.uk/%7Eojw28/ins.html) describing the use of MEMS for inertial navigation.

The basic conclusion seems to be that MEMS are about two orders of magnitude noisier than laser ring gyros etc., and that even with some heavy duty signal processing the best they managed to get was an error of 5m after a minute of being strapped stationary to a test bench.