Preliminary Results in Decentralized Estimation for Single-Beacon Acoustic Underwater Navigation

We report a decentralized, vehicle-based extended information filter designed to enable single-beacon acoustic navigation of multiple underwater vehicles. In single-beacon navigation, ranges from a single reference beacon to a moving underwater vehicle are used in addition to the vehicle’s inertial navigation sensors to perform absolute (as opposed to relative) localization and navigation. For this implementation we assume a moving reference beacon from which range measurements are calculated using asynchronous acoustic data broadcasts that also contain information about the reference beacon’s position and recent sensor measurements. We assume that other than these acoustic data transmissions the vehicle has no knowledge of the beacon’s position or sensor measurements. This implementation allows the simultaneous navigation of multiple vehicles within acoustic range of the reference beacon. Within this framework we show that the decentralized information filter yields identical results to a centralized extended Kalman filter at the instant of each range measurement; in addition we show that between range measurements the results from the two filters differ only by linearization errors. We compare the state estimation results of the decentralized information filter to that of a centralized extended Kalman filter using a simulated data set.