Two Kalman Models for Chirp-Sequence Radar Data Filtering of a Periodically Moving Hand for Conducting a Virtual Orchestra

In this paper two Kalman filter models are presented to filter a signal of a FMCW radar sensor. The measurement object is a rhythmic hand movement of a conductor. From the radar signal we get the distance and velocity. The models are intended to correct the noisy signal for later analysis of frequency and phase so that the tempo and beat of the conductor can be estimated. We evaluate and compare the models with simulated and real data from a chirp-sequence radar sensor.