Visually-augmented acoustic subsea navigation

The use of underwater vehicles is increasing fast. Their use covers a wide range of applications, including inspection of structures, seabed mapping, and environmental monitoring. Since GNSS/GPS does not work underwater, one fundamental issue in an underwater environment, is that navigation and constantly knowing your whereabouts can be challenging and expensive. This is also a key to autonomous operation of underwater vehicles. This project aims to explore ways of integrating optical sensors into traditional acoustic navigation systems. The integration of complementary sensors can help optimize the navigation of underwater vehicles. The project will be conducted through theoretical research combined with experimental verification. The latter reflects the uncertainty and lack of knowledge in current models. The research will also address the limits in harsh environments, which has importance when considering which applications will benefit the most from integrating optical sensors. The integration of optical sensors into traditional acoustic navigation systems is a promising technique in various applications such as mapping and vehicle navigation. It allows for compensating the drawbacks of the low resolution of acoustic sensors and the limitations of optical sensors in bad visibility conditions. The research will investigate how the complementary sensors can be optimized, which will help improve the accuracy of underwater navigation systems.

The project’s primary objective is to develop an enhanced near-seafloor underwater navigation system by complementing traditional acoustic navigation systems with visual observations and processing. The secondary objective is is to establish whether the capabilities will meet requirements for potential commercial applications of this technology, and in particular for autonomous navigation for new, low-cost vehicles. This addresses the needs of expanding offshore wind energy as well as the existing needs of the offshore oil & gas industry. This will involve researching optimal camera technology, visual odometry capabilities, integration with acoustic-inertial navigation, methods for optimization of sensors and potential for SLAM functionality and application of machine learning towards such goals. The project will extensively combine theoretical research with experimental verification reflecting the present uncertainty in models and lack of knowledge in this area. The research will address the potential and also limitations of visual capabilities in various conditions and the relevance for applications in harsh environments. A fundamental parameter of a useful application is the ability to identify visually features of sufficient quality. This is in contrast to the subsea visual (or acoustic) identification of features of infrastructure. Feature recognition of the seafloor is a substantial challenge and largely unknown for most environments.