Gyroscope noisy data or mistake in cumulating deltaRadiansX/Y numbers?

Solar2D General Questions/Discussion
#1

I am playing around with the iPad’s gyroscope. Essentially I want to detect device orientation relative to when the app has been launched or a function started. I have noticed that the gyroscope data seems very noisy and biased. Have anyone noticed that or am doing something wrong?

In particular, the noise seems to come from event.xRotation inside the formula:

deltaRadiansX = event.xRotation * event.deltaTime

I then cumulate deltaRadiansX the small error builds up. At each frame I simply add a new reading for deltaRadiansX to the total calculated in the previous frame (frame = corona iteration). [import]uid: 159908 topic_id: 34895 reply_id: 334895[/import]

#2

Apparently gyroscopes (in general) give biased data, so cumulating device change in rotation data is a bad idea, unless information is filtered (and possibly cross checked with the accelerometer). Given that I do not need to drive a missile down a chimney, a handy trick is making sure that there cumulative data reverts to some fixed level (e.g. object on screen drifts back to its starting position after being displaced by shaking the device).

If anyone has some experience with filtering gyroscope data, I am always open to a better way. [import]uid: 159908 topic_id: 34895 reply_id: 138917[/import]

#3

Apparently gyroscopes (in general) give biased data, so cumulating device change in rotation data is a bad idea, unless information is filtered (and possibly cross checked with the accelerometer). Given that I do not need to drive a missile down a chimney, a handy trick is making sure that there cumulative data reverts to some fixed level (e.g. object on screen drifts back to its starting position after being displaced by shaking the device).

If anyone has some experience with filtering gyroscope data, I am always open to a better way. [import]uid: 159908 topic_id: 34895 reply_id: 138917[/import]