An approach towards online bathymetric SLAM

This study introduces a computationally efficient SLAM algorithm that can map the elevation changes in undulating terrain and simultaneously localize the vehicle´s position relative to the map. The algorithm enables autonomous navigation in unknown environments without using position fixes from external telemetry systems such as GPS and LBL. In particular, this approach can benefit a variety of underwater vehicle applications and expand the utility of autonomous underwater vehicles. The terrain-based SLAM algorithm involves severe nonlinearity due to complicated elevation changes in natural terrain, which leads to a nonlinear estimation problem. This research focused on developing a computationally efficient algorithm. Thus, the EKF algorithm incorporating depth measurements from a simple acoustic altimeter is employed to keep the computational cost at minimum. The feasibility and validity of the proposed algorithm is demonstrated through numerical simulations.