A Navigation Algorithm for Under-the-Ice Robotic Operations

The operational conditions encountered in under-the-ice environments are extreme and demanding, thus preventing human-centred exploration. The study of these environments is dependent on the development of robust and flexible navigation strategies to guide autonomous vehicles in these barely known ecosystems. This work proposes an integration of different underwater navigational strategies with a view to addressing a Simultaneous Localization and Mapping scenario. The navigational algorithm combines Dead Reckoning and Scan Matching techniques, complemented with Single Beacon navigation. In addition, the latter strategy includes a new method to identify and reject erroneous range measurements, which only depends on the collected range information. The navigation system is rooted in Kalman filtering theory, and relies on bathymetric information from the underwater ice cap and proprioceptive measurements. Through the implementation of a vehicle's simulator, the simulations highlight Single Beacon navigation as a fundamental strategy to reduce the error in the vehicle's position estimation, even in the presence of outliers in range measurements. The complete navigational strategy shows that the existence of correct terrain matches can improve vehicle's position estimates, motivating the search for new sensor configurations capable of optimising Scan Matching techniques.