UKF-Based Inverted Ultra-Short Baseline SLAM with Current Compensation

Abstract

This paper presents an inverted ultra-short baseline (iUSBL) simultaneous localization and mapping (SLAM) based on Unscented Kalman filter (UKF) with current compensation. The complexity of the underwater environment prevents using a Global Positioning System (GPS) or visible-light cameras. Therefore, we rely on an acoustic positioning system to achieve autonomous underwater vehicle (AUV) navigation relative to the beacons. Our system uses an AUV equipped with a time-synchronized iUSBL array to measure the range and angle between the vehicle and acoustic beacons via one-way travel-time. Then, we propose UKF-based online SLAM to filter these acoustic measurements, and add the compensation for inherent underwater current. Compared to traditional SLAM applications usually involving many landmarks, this iUSBL acoustic system requires only sparse beacons to estimate the vehicle pose. Consequently, it does not cause UKF to have an excessive computational burden. Numerical results confirm the validity of the proposed method. In particular, the proposed algorithm prevents cumulative errors from proliferating quite well when using double beacons.