摘要：AbstractThe navigational drift for Autonomous Underwater Vehicles (AUVs) operating in open ocean can be bounded by regular surfacing. However, this is not an option when operating under ice. To operate effectively under ice requires an on-board navigation solution that does not rely on external infrastructure. Moreover, some under-ice missions require long-endurance capabilities, extending the operating time of the AUVs from hours to days, or even weeks and months. This paper proposes a particle filter based terrain-aided navigation algorithm specifically designed to be implementable in real-time on the low-powered Autosub Long Range 1500 (ALR1500) vehicle to perform long-range missions, namely crossing the Artic Ocean. The filter performance is analysed using numerical simulations with respect to various key factors, e.g. of the sea-floor morphology, bathymetric update rate, map noise, etc. Despite very noisy on-board measurements, the simulation results demonstrate that the filter is able to keep the estimation error within the mission requirements, whereas estimates using dead-reckoning techniques experience unbounded error growth. We conclude that terrain-aided navigation has the potential to prolong underwater missions to a range of thousands of kilometres, provided the vehicle crosses areas with sufficient terrain variability and the model includes adequate representation of environmental conditions and motion disturbances.
