Estimation of Location and Orientation for Underwater Vehicles from Range Measurements

TL;DR

This paper presents a novel two-step localization method for underwater vehicles using OWTT measurements from acoustic beacons, accounting for water condition uncertainties through polynomial optimization.

Contribution

It introduces a calibration-based approach to bound range measurement uncertainties and combines convex optimization with rigid body constraints for accurate localization.

Findings

Effective localization with uncertainty bounds demonstrated

Calibration function improves range accuracy

Experimental results validate the proposed method

Abstract

Localization is an important required task for enabling vehicle autonomy for underwater vehicles. Localization entails the determination of position of the center of mass and orientation of a vehicle from the available measurements. In this paper, we focus on localization by using One-Way Travel Time (OWTT) measurements available to a vehicle from the communication of its multiple on-board receivers with acoustic beacons, more specifically, long baseline (LBL) beacons. Range can be inferred by multiplying the OWTT with speed of sound; however, water conditions can change spatially and temporally resulting in uncertainty in range measurement. The farther a beacon is from a receiver, the larger is the uncertainty. The proposed method for localization accounts captures this uncertainty by bounding the true distance with an increasing (calibrating) function of the range measurement. Determination of this calibration function is formulated as polynomial optimization problem and is a crucial step for localization. The proposed two-step procedure for localization is as follows: based on the range measurements specific to a receiver from the beacons, a convex optimization problem is proposed to estimate the location of the receiver. The estimate is essentially a center of the set of possible locations of the receiver. In the second step, the location estimates of the vehicle are corrected using rigid body motion constraints and the orientation of the rigid body is thus determined. Numerical examples and experimental results provided at the end corroborate the procedures developed in this paper.