Bearings-only target localization for unmanned underwater vehicles
Donald P. Eickstedt · The Journal of the Acoustical Society of America · 2011
This paper reports on target localization for autonomous underwater vehicles (AUVs) with acoustic bearing sensors where the bearing sensor provides detection and beamforming capabilities and reports target bearings with the associated measurement uncertainties. Estimating the state of a moving target using a bearing sensor on a moving platform is a difficult problem due to both the nonlinear nature of the measurement equations with respect to the unknown target parameters but also the observability conditions for the unknown parameters which require the observer to maneuver. This paper will report on a maximum likelihood (ML) method using Levenberg–Marquardt (LM) optimization for estimating the target state parameters. A known issue with ML solutions to this problem is that the algorithm may have difficulty reaching the global minimum unless the initial solution guess is close to the true solution. This paper will report on a genetic algorithm approach to making the initial solution guess. The performance of this approach, obtained from Matlab simulation, is presented. The implementation of the algorithms in the MOOS architecture is described using a very fast Lapack implementation for the LM parameter estimation to support the goal of optimal AUV maneuvering to improve the target state estimate.