search

hanley6 / IMUSensorModels

This repository contains a C++ class for IMU sensor modeling.
70 stars 29 forks source link

Interpretation of the Allan Variance and the different values #3

Closed anfebit closed 7 years ago

anfebit commented 7 years ago

Greetings Hanley,

I wanted to create a new thread for several questions and concerns that I have. I hope you can help me.

  1. You stated that here that

noise = Angle Random Walk + Bias Instability + Rate Random Walk.

However, let's assume that I'm finding the different values for an accelerometer data. Where the units are going to be g/sqrt(s),g/s and g/s^2 for Random Walk, Bias Instability and Rate Random Walk respectively. How are these values added with different units?.

  1. In the following image. I have a concave section and the beginning. I can fit most of the other curves, but I was trying to understand the reason of this concavity. Is there any other type of noise that I'm not aware? What happens when I get curve such like this one? Would this part just discarded for the the noise analysis?

  2. If one of the lines doesn't fit well, how can this be interpreted? Would it mean that particular parameter is not dominant in the noise analysis?

Thank you for your help and response beforehand, any help will be appreciated it.

anfebit commented 7 years ago

Hi Hanley, I was able to solve the problem. I got a better understanding of this by reading this paper. https://pdfs.semanticscholar.org/76ef/7150505bbcedf445ae458f2f9002f0d9fa2b.pdf

The are able to add the noise actually because they have the same value. You can close this question. Thank you for your help.