The Gyro bias is estimated automatically by the filter. So there is no need to calibrate this bias away manually.

Your approach is correct but will likely fail due to inaccuracies in your “known angle”.

A better approach to estimate the bias is to hold the sensor completely steady and average the gyroscope angular velocity values over a period of time. ]]>

As said in the article, the bias is the amount the gyro has drifted. Therefore, as far as I understand, to find the gyro bias one should:

1. make the gyro rotate about a known angle (say 360°)

2. integrate the values so you get actual gyroscope output in terms of the angle

3. subtract this output from the known angle (360°).

4. divide the result by the total integration time

By using this procedure I always obtain very big and different values for the gyro bias, such as 0.78 rad/s, -0.11 rad/s, 0.5 rad/s.

My question is: am I using the correct procedure?

You compile the Android Studio project from: https://github.com/TKJElectronics/AndroidControllableLights/tree/master/ColorLightsController ]]>

Which of them are you looking for? The small 5mm diffused ones or the large 1W LED?

Both items have been purchased on eBay from China. ]]>

Thank You

]]>But Q_\theta refers to the gyroscope variance, not the accelerometer.

]]>