A Kalman filter algorithm has been developed to estimate angular velocity and acceleration from encoder measurements. Asynchronous observations are used, namely the time instants when encoder pulses are emitted. In addition, in order to help low-velocity performance when few pulses are generated, the fact that the encoder output has not changed is also fed to the estimator, by clock-driven sampling to limit the duration of an interval without samples. The algorithm has been implemented in real time, at a 1KHz sampling frequency. It will also be desirable to incorporate tachometer measurements into the estimator.
A. Helmy, P. Canet, P.R. Bélanger