Autonomous navigation for a depth measuring vehicle in an unknown naval environment

Abstract: The accuracy of depth measurements in shallow parts of the Stockholm archipelago nautical charts is relatively low at present. In order to improve it, an autonomous depth measuring craft equipped with three downward facing sonar sensors is being developed at KTH. In this report, a navigation algorithm that does not require any prior knowledge of the target environment is proposed to be implemented in the craft. It consists of a predifined sweeping pattern, a contour following that only relies on sensor data, and a way of identifying and scanning missed partitions. Performance and the impact of sensor inaccuracies in addition to environmental factors, such as wind and waves, are investigated through simulations. Results from simulations show that the proposed navigation algorithm is able to complete scans of complex environments. However, the uncertainty in sensor readings is proven to have a large inuence on performance. Finally, further improvements to the algorithm are proposed and the realism of the simulation is discussed.