Hi fellow forum members,
I have been playing with rate gyros for some time now and have been experiencing poor accuracy due to spurious readings from the rate gyro, potentially driven by vibration and other environmental considerations. The difficulty has been to differentiate between the true value and the noise from normal operation.
Whilst researching the problem on the Web, I came across a number of articles relating to Kalman filtering. Kalman filtering is a really clever mathematical algorithm that allows you to extract a true value from the noise.
Whilst there are many articles available on the web describing Kalman filtering, I found this article to be a good start http://bilgin.esme.org/BitsBytes/Kalman ... mmies.aspx. It provides a very simple explanation and mathematical example of how a Kalman filter works.
Attached to this post is a pic example of a Kalman filter. This example has been based on the mathematical model contained within the article I have added 10 data points to the simulation in order to make the example run. If you refer the values on the LCD to the table in the above document you will find that they track, meaning that I have got the maths right.
Appreciate your feedback on this topic – I know there may be some interest out there and would appreciate any input to help further refinement. I also have a dsPic version if anyone is interested .
All the best,
Pete