Estimation of Angular Velocity and Acceleration from Shaft-Encoder Measurements

This paper suggests a Kalman-filter approach to the estimation of angular velocity and acceleration from (quantized) shaft-encoder measurements. Finite-difference estimates deteriorate as sampling rates are increased. For small sampling periods, we show that the filtering problem is the dual of the cheap control problem, and we jus tify the use of all-integrator models. We investigate Kalman filtering with constant sampling rate, and also with measurements triggered by encoder pulses. Simulation and experimental results are given.