Evaluation of High Precision Localization Approach for a Fleet of Unmanned Deep Ocean Vehicles

This paper addresses an unconventional approach for localizing underwater robots through sensor fusion and inverted Long Baseline (LBL) method. Initially, the paper describes the practical difficulties faced in developing such a system and then provides a solution for the same. Underwater localization finds its application in precise bathymetry scanning and underwater navigation. Apart from the normal LBL localization, an inverted moving LBL approach is introduced by means of Ultra-Short Baseline (USBL) modems. For the development of localization algorithm, a fleet of underwater robots and autonomous surface vehicles equipped with USBL modems are utilized. These modems, equipped with an atomic clock are able to localize its position with respect to another USBL modem. Measurements from the USBL modem which is fitted in the underwater robot are then fed to a tested outlier rejection algorithm. Filtered values are then given to a localization algorithm which is built upon time of arrival (TOA) measurements and depth sensor readings. The output of the localization algorithm is combined with Inertial Navigation System (INS) readings by an unscented Kalman Filter. By this method, long-term accuracy of USBL and short-term accuracy of INS are well utilized. Analysis of various errors caused by the system has been addressed in the second half of the paper. Finally, the paper finishes with a few test results from the simulation. Simulation results prove the localization approach and the degree of error caused by false readings. The simulation results had an error of 9m in localization with a 100ms difference in TOA without moving LBL approach. This paper addresses a method to overcome such an error.