Grant: Discovery Grants Program - Individual

All Canadians are impacted by, and benefit from, research into tools that: contribute insight on Canada's under-ice oceans and their connection to changing climate; safe and secure navigable northern waters, and long duration devices to monitor, patrol and track underwater activity for Canadian Arctic sovereignty and defence. Autonomous underwater vehicles (AUV) are tools tasked to work in all three. The limiting capability in all three is the AUV ability to navigate and localize over long ranges unassisted by external sensors. This does not exist yet to the extent that it should. The outcomes of the proposed research program contribute important milestones towards this. It provides a transformational capability to address these topics and supports the ocean sector in an unprecedented way. For AUVs to fully realize their value by operating at more than arms-length from a ship or operator, they must navigate and localize long distances underwater and under-ice without assistance from a ship or external-to-the-AUV positioning sensor which are not options under-ice. The external aiding is necessary as the positional error of inertial systems grown unbounded with distance travelled. With simultaneous localization and mapping (SLAM) the AUV will navigate an unknown area using inertial sensors which could be externally aided. During the mission, the AUV builds the map and concurrently localizes itself within it so its survey sensor measurements are geo-referenced. With AUVs, inertial navigation is limited by the range of the external positional aiding and cannot operate far from the seabed. This solution restricts the AUV operational range possible. Beyond the state-of-the art SLAM with inertial navigation aided by external sensors, my research program studies featureless (abstract features) SLAM with inertial navigation aided by geophysical or terrain navigation instead of external sensors as it better suits the under-ice environment. Similar to fusing GPS or acoustic position aiding sensors with inertial navigation, terrain-aided (bathymetric) navigation (TAN) features in a priori maps are used to bound the position error growth of the inertial sensors. Unlike underwater inertial or acoustic navigation methods, TAN methods are not limited in operational range or altitude. TAN is a potentially powerful solution towards long range under-ice AUV navigation and localization. To work towards on-board on-line SLAM with bathymetric TAN, the SLAM method favored for the proposed research is pose-graph.