Non linear estimation in sonar tracking

This paper presents an optimal recursive algorithm with a new adapted state representation for the bearings-only tracking problem. We apply nonlinear filtering techniques to derive the optimal filter. State estimation, which consists of initial position and velocity both in polar coordinate is achieved by a space discretization of the optimal filter. To apply this time-consuming approach in realistic situations, this paper presents an adaptative grid computational scheme where the integration step is controlled by the a priori probabilities. We compare this new solution to the maximum a posteriori estimator and perform Monte-Carlo simulations to compare both algorithms performance.