today we begin to define some PSSC for our balls project migB, and begin to think about components we need. According to our teachers, it is very hard to analyse a motion system with an IMU, that’s why i did research about it. It seems that orientation sensing is very easy, while getting velocity and position is much harder. Indeed getting velocity and position requires at least one mathematical integration. So if you get an error on acceleration, even little, and you always get a little error at best, this integration which increases the error. Therefore it is impossible to be very accurate over time. But how much do you want to be accurate ? This is very difficult to foresee which algorithm you will need to get what you want with your IMU, as soon as your project is such specific. Until now the only good one i found is the Kalman Filter, but i’m afraid that our errors could be negligible and thus Kalman Filter would become a big algorithm for nothing.
Also i installed Archlinux on the last computer available to us.