Proper way to use Allan Variance

[BrunoOtavio98]

Hi, im trying to use your Kalman Filter architecture in my navegation system and i am new in this field. How should i proceed to get the accelerometer and gyroscope data to pass to allan_imu and get the errors parameters? Now i tried to leave them stationary for 20 minutes MatLab Ref, thats correct? Or they should be in movement?

If a run in this way, i get this error:

error: dtime(1): out of bound 0 (dimensions are 1x0)
error: called from
allan_overlap at line 534 column 5
allan_imu at line 181 column 30

Also in allan_overlap i get this two lines:

allan_overlap: regular data (0 freq data points @ 130 Hz)
allan_overlap: phase data generated from fractional frequency data (N=1).

[ravindraji4]

Hi Bruno,

Firstly, the data should be collected with IMU at rest and this should be inputted to calculate the different Allan Variance Values.

Secondly, I tried the inbuilt Matlab example and it works.

It would be easier. If, Could you please upload the data. So that we can find the problem.
Just writing the error/encountered problem, it is hard to say/explain the potential solution!

Best regards,
Ravindra

[BrunoOtavio98]

Hi, Im re-uploading the dataset, i changed the ned velocity with one thats not dependent of imu heading. Together, one "ground truth" displacement with RTK receiver.
IMUBNO085data.txt
GPSdata 2.txt
gpsRef.txt

[BrunoOtavio98]

And this is IMU static data to allan variance.
https://drive.google.com/file/d/1cmiveOUbyou3R2IGo7KDycCSwq3M7sYH/view?usp=sharing

[rodralez]

Hi Bruno,

Just give me a couple of days to process your new data.

Best regards.

[rodralez]

Hi, Im re-uploading the dataset, i changed the ned velocity with one thats not dependent of imu heading. Together, one "ground truth" displacement with RTK receiver.
IMUBNO085data.txt
GPSdata 2.txt
gpsRef.txt

Hello Bruno,

I attach just a template to help you start processing your IMU and GNSS data by using NaveGo.

I think your data is not correctly sampled, or in other words, the sampling time of both sensors seems to be unstable. Check the output of test_time_quality().

I hope this could help you to use NaveGo!

Best wishes.

bruno_navego.zip

[BrunoOtavio98]

Hello @rodralez, first of all, thank you for your support. Ok, i will see the code.

[rodralez]

Hi, im trying to use your Kalman Filter architecture in my navegation system and i am new in this field. How should i proceed to get the accelerometer and gyroscope data to pass to allan_imu and get the errors parameters? Now i tried to leave them stationary for 20 minutes MatLab Ref, thats correct? Or they should be in movement?

Hello Bruno,

It is convenient to have between 2 and 12 hours of static data for an Allan variance analysis.

More info:

IEEE-SA Standards Board. IEEE Standard Specification Format
Guide and Test Procedure for Single-Axis Interferometric Fiber Optic
Gyros. ISBN 1-55937-961-8. September 1997.

Naser El-Sheimy et at. Analysis and Modeling of Inertial Sensors
Using Allan Variance. IEEE TRANSACTIONS ON INSTRUMENTATION AND
MEASUREMENT, VOL. 57, NO. 1, JANUARY 2008.

Oliver J. Woodman. An introduction to inertial navigation. Technical
Report. ISSN 1476-2986. University of Cambridge, Computer Laboratory.
August 2007.

Best regards.

[BrunoOtavio98]

Thank you for your answer @rodralez i will read the papers.

[rodralez]

By the way, I am reading the article about Allan variance by MathWorks that you provided above is excellent. It is a concise and clear explanation of the the Allan variance.

Best regards.