Unscented Kalman Filtering for Autonomous Underwater Navigation

The availability of a high-performance navigation state estimator is fundamental to Autonomous Underwater Vehicles (AUVs), especially when multiple vehicles are involved in autonomous tasks. The underwater environment further complicates the estimation process, reducing the number of available sensors. In this paper, a navigation filter based on the Unscented Kalman Filter (UKF) and relying on AUV onboard sensors is proposed. The performance of the presented solution is evaluated exploiting experimental data acquired by the two Typhoon AUVs, developed and built by the Department of Industrial Engineering (DIEF) of the University of Florence within the THESAURUS Tuscany Region Project for exploration and surveillance of underwater archaeological sites, during the International workshop Breaking the Surface 2014 (BtS 2014) held in Biograd na Moru (Croatia) in October 2014. An offline comparison between the estimates given by the proposed filter and those computed by a standard navigation algorithm (based on the Extended Kalman Filter, EKF) is presented. The demonstration performed at BtS 2014 constitutes a preliminary test of cooperative navigation between the two AUVs. The obtained results show that the UKF offers promising accuracy improvements with respect to the EKF; in addition, the computational load required is affordable by the typical AUV hardware. According to the achieved results, the proposed algorithm will be implemented on the Typhoon AUVs and tested online.