Distributed implementation of Kalman object tracker with discrete asynchronous measurements

Abstract

Typical tracking systems employ tracking filters in order to consecutively estimate the kinematics of a moving object, based on some limited observations of its position and/or velocity which are contaminated by different types of uncertainties. The Kalman filter is the optimal filter for some simple cases. However, in practical scenarios, designing more complicated filters or making modifications to the Kalman filter may be necessary, in order to provide the tracking process with adequate reliability and accuracy. An appropriate design should consider the discipline under which measurements are gathered and the type of disturbances to which the measurement process is exposed. This study investigates the case of quantised, unreliable, asynchronous velocity and position measurements, and proposes an effective design which realises a distributed Kalman filter with two- or three-dimensional state vectors. It is shown via simulations that the proposed design not only smooths the output and combats the effects of quantisation but also keeps on tracking when the measurements are lost for a quite long duration of time. © The Institution of Engineering and Technology 2018.