Hi, hhenne - great insight & explanation thanks.
Regards On Nov 12, 3:35 am, hhenne <[email protected]> wrote: > Hi again, > > In my app, I am only interested in pitch and roll - not in yaw. > > The accelerometer can measure static pitch and roll (because of the > gravity) - but not static yaw. > Acceleration caused by movements are just unwanted noise in my app and > is filtered away by a usual digital Lowpass filter. > > The Gyro cannot measure static pitch and roll and not static yaw > either, It can only measure angular velocity - changes in pitch, roll > and yaw. > To convert angular velocity to angle, you have to use digital > integration. But, because small errors would cause the computed angles > to slowly drift away, you have to filter the slow variations away - > that is done by a digital highpass filter. You can say that is > actually to opposite working filters, but it is the correct way. > > And now the computed gyro signal cannot be used for detecting static > pitch, roll and yaw because the static part is filtered away. But then > it can be combined with the lowpass filtered accelerometer, simply > adding the two signals, to get fast and correct detection of pitch and > roll. That means if you make a sudden rotation of the device the > signal should change from one level to the other with a small rise > time and no overshoot. in signal terms the frequency response of the > transfer function is linear from DC to around 5 hz. > > When you are interested in yaw, but not pitch and roll, you have to > replace the accelerometer with the magnetic sensor. As mentioned by > gjs, it cannot help you overcome magnetic interference. But it should > be possible to get a much smoother and faster response by using > magnetic sensor and gyro in combination than by using the magnetic > sensor alone. > > I have spend a lot of time on Kalmann filters, whithout producing > results worth the efforts. What you see in my keyboard app is the > first shot of the 'complementary filter design' sketched above - and > it only took me a few hours to implement. All right - it is not > perfect yet - for instance I need to take into account the variations > in sample frequencies among the different sensors and also from device > to device and from time to time. Never the less it was a great > improvement from using the accelerometer alone. The same app runs on > devices, only having an accelerometer. So if you have got two devices, > and only one of them has a gyro, you can see and feel the > difference. > > Hardy Henneberghttp://www.maxikeys.com > > On Nov 11, 3:30 am, gjs <[email protected]> wrote: > > > > > > > > > Hi, > > > I am only suggesting you use integration ( sum & 'average' last N > > measurement over some short period of time ) for the compass/azimuth - > > however you decide to get that data - just to smooth the data, so > > these filtered reading don't jump around so much (this has got little > > or nothing to do with accuracy of the sensors). > > > Then you might use the gyroscope &/or acceleration to detect relative > > movement of the device, is it currently still or moving, to alter the > > integration you are doing with the compass/azimuth, eg to reset or > > change value of N mentioned. Kalman Filters are often mentioned here > > as a solutionhttp://en.wikipedia.org/wiki/Kalman_filter > > > If you want high accuracy you will probably need to use external high > > accuracy IMU / compass sensor devices. Even then you won't necessarily > > get around external magnetic interference. > > > Regards > > > On Nov 10, 10:17 pm, Ferro Francesco <[email protected]> > > wrote: > > > > I know what is an integration and How to do it. What I don't know it's > > > what > > > do I do with the result? How can i combine is with the compass value ? -- You received this message because you are subscribed to the Google Groups "Android Developers" group. To post to this group, send email to [email protected] To unsubscribe from this group, send email to [email protected] For more options, visit this group at http://groups.google.com/group/android-developers?hl=en
