Experimental Filter Comparison of an Acoustic Positioning System for Unmanned Underwater Navigation

Abstract

The demand for Unmanned Underwater Vehicles (UUVs) for various operations has increased. Automating these operations requires good knowledge of the location of the UUV. This study investigates the filter lag in the commercial low-cost Underwater GPS Explorer Kit, a short baseline (SBL) from Water Linked. Furthermore, different filters are designed in an attempt to minimize the filter lag. Minimizing the lag is beneficial for real-time navigation, as delays have a negative impact on stability. Three different filters are designed; a model-free simple moving average (SMA) filter, a Kalman filter (KF) based on a constant velocity motion model, and lastly, an unscented Kalman filter (UKF) based on the non-linear model derived from the robot's dynamics. A large filter lag is found in the original filter; however, it can be concluded that a model-based filter with knowledge of the system dynamics can be used for minimizing the filter lag.