Terrain referenced navigation using sand dunes and sand ripples in presence of tides

We investigated using terrain referenced navigation in shallow water in the presence of unknown tide, with an autonomous underwater vehicle equipped with a high resolution bathymetric sensor. The mission area had both variations, with large sand dunes, and flatness with small sand ripples. In order to deal with the unknown substantial tide and shallow water, we tested a non-linear estimator for horizontal position, tide level and sound speed errors, a four dimensional Rao-Blackwellized point mass filter. The algorithm received an initial position estimate from the inertial navigation system (INS), computed the probability density function for the position and environmental errors, tested for convergence, and provided a Gaussian approximation that could be used by the INS. The results showed that the algorithm converged consistently in both areas, more often in the sand dune area, but surprisingly with comparable to higher accuracy in the sand ripple area. This demonstrated that terrain referenced navigation worked as expected, on sand dunes, in shallow water with substantial tide. The flatness is sometimes only a matter of scale, and this technique should be considered used everywhere when equipped with high resolution sensors.