The AutoNaut USV is a novel wave-propelled vessel with a very low noise profile. Assuch, it is uniquely suited to passive acoustic monitoring, which can be carried out usinga towed hydrophone array. However, the method of propulsion inherently presents somechallenges to employing bearing estimation techniques by potentially exacerbating theuncertainty in array shape. The aim of this research is to investigate and demonstratethe capabilities of such a vessel, to quantify its performance and to evaluate the stepsrequired to fully realise its potential.A recursive Bayesian array shape estimation method was developed to combine thedata from all available sensors on the array. This was tested and verified with simulateddata and then used to estimate the motion of an array in operation during the UnmannedWarrior ’16 trial. This trial data was used to demonstrate bearing estimationto a sound source, with an analysis of the performance increase from compensating forthe perturbed shape. The outcomes of this were then used to investigate the optimalpositioning of non-acoustic sensors on an array.The main result was the successful demonstration of bearing estimation to a soundsource from experimental data. The results of the array shape estimation processsuggest that the array is tilted and periodically perturbed into a bowing shape with anamplitude of less than 0.1m. Compensating for this results in up to a 5.8dB increasein the output spectrum from the MUSIC bearing estimation algorithm.While a perturbed array shape results in a slight drop in performance, a conclusionof this work is that the tilt on the array and the resulting detection of a multipatharrival presents a further challenge in interpreting the results. Nonetheless the researchpresented here represents a successful first step to enabling the AutoNaut’s fullcapabilities.