Underwater robotics & exploration
My research addresses localization, state estimation and SLAM for autonomous underwater robots. Since GNSS signals do not propagate underwater, navigation relies on onboard sensors and often incomplete observations of the environment. Drift, ambiguities in sonar measurements and uncertain oceanographic conditions make it difficult to determine where a vehicle actually is.
I study navigation approaches that exploit the constraints provided by vehicle dynamics, acoustic measurements and the seabed itself. This includes terrain-based navigation, sonar-based localization and the estimation of trajectories when initial position or heading is poorly known. Set-membership methods are particularly useful when several locations remain possible and none should be discarded prematurely.
A current objective is to transfer these developments to compact micro-AUVs through the Yodac program. Polar and under-ice missions provide a demanding test case: magnetic heading may be unreliable, surfacing may be impossible, currents and tides are imperfectly known, and the robot must still find its recovery point. These constraints motivate research on reliable navigation, exploration coverage and safe return.
