Towards comparison of Kalman filter methods for localisation in underwater environments

Abstract

Kalman Filters are utilised for filtering and estimation in a large set of application. Here, this methodology is utilised for trajectory estimation of an underwater robot. In this work, three Kalman Filter methods are proposed for trajectory estimation. There are: Extended Kalman Filter (EKF), Unscented Kalman Filter (UKF) and Central Difference Kalman Filter (CDKF). Simulation results are presented and discussed, where UKF and CDKF presented better performance than EKF with data that were collected from our dataset. However, UKF had a slightly smaller execution time than CDKF with almost the same error.