A distributed localization hierarchy for an AUV swarm

Abstract

Localization of teams of autonomous underwater vehicles (AUVs) still remains as a challenge in large-scale ocean currents. In this study, moving references, drifting under the influence of the ocean background flow, were employed in order to improve the cooperative localization (CL) performance of an AUV swarm in harsh ocean flows. More capable AUVs (dubbed as mother AUVs) with less localization error were utilized as moving references for improving localization error of less capable AUVs (called daughter AUVs). Limitations of a previously proposed modified extended Kalman filter (MEKF) were identified. A particle filter (PF) based algorithm was proposed to address those issues. The performance of the PF algorithm was compared with the MEKF algorithm in several simulated examples including CL in an N-vortex background flow field. Both algorithms can effectively avoid the diverging behavior of localization error in pure CL. The PF algorithm is more robust in choosing the better localized AUV. With a large number of particles, the PF algorithm outperforms the MEKF algorithm at the expense of computational efforts.