The Enriched Sigma Point Kalman filter An adaptation of the Unscented Kalman Filter for navigation applications

摘要

The Unscented Kalman Filter (UKF) has received a lot of attention in the last few years, especially for inertial navigation applications. Indeed, it efficiently extends the non-linear capacities of the Extended Kalman Filter (EKF), and provides additional benefits, such as an absence of divergence and a black-box implementation that simplifies the maintenance of such an algorithm. Nevertheless, during the heading determination phase of an inertial navigation algorithm, the barycentre of the unscented transform used to represent the system state is not adapted because it creates an artificial position error. An alternative algorithm, using a so-called ‘Central point’ allows to benefit from the UKF capabilities while suppressing this drawback, making it possible to determine the heading of a system without any of the static constraints of a traditional inertial navigation system.