Hydrodynamic loads on subsea robots

Free floating subsea robots are often subjected to complex hydrodynamic loads due to waves/currents and wakes from nearby structures. This makes control extremely difficult. Good hydrodynamics data might counteract this. This paper presents hydrodynamics data for two configurations undergoing various combinations of steady translation and approximately steady rotation. The two configurations are: a spherical body with a single fixed link arm and a two link arm without a body. For both configurations neural networks were found to provide a reasonable fit to the data. We believe that such fits could be used to improve control.