Interval methods based AUV localization in the context of an acoustic network with experimental results

This article presents an application of interval methods for the dynamic localization of an autonomous underwater vehicles (AUV) using acoustic measurements of range and/or bearing to one or more underwater or surface nodes. A node can be either a vehicle or a buoy equipped with an acoustic modem and positioning aid. The position and speed vector of each node is communicated acoustically underwater to all of the other nodes. Popular methods used in solving localization problems use a probabilistic approach (e.g. Kalman filtering or Particle filtering). While such methods consider the probabilistic distribution of variables, the interval methods consider sets of possible values of a variable. The interval methods are deterministic in nature and have certain advantages over probabilistic methods such as native handling of non linear equations or solving problems of high dimension, guaranteed result under the right assumptions, and handling inconsistent data. The interval localization method is tested on a dataset acquired during an experiment on the Mediterranean sea.