Title :
Estimation of iceberg motion for mapping by AUVs
Author :
Kimball, Peter ; Rock, Stephen
Author_Institution :
Dept. of Aeronaut. & Astronaut., Stanford Univ., Stanford, CA, USA
Abstract :
Iceberg-relative navigation for Autonomous Underwater Vehicles (AUVs) will enable a new mode of data collection for studies of free-floating icebergs. Compared to current data collection methods, AUVs offer substantially expanded coverage area and continuous sampling. However, because icebergs translate and rotate through inertial space, standard vehicle navigation methods which rely on inertial sensors are unable to provide iceberg-relative position estimates. One means of obtaining iceberg-relative position estimates is to correlate incoming sonar measurements of range to the iceberg surface with an a-priori three-dimensional iceberg map. This requires the generation of a map. The key challenge in creating such an a-priori map is accounting for the translation and rotation of the iceberg over the duration of map data collection. Previous iceberg mapping work by the authors accounts for iceberg motion by assuming a constant velocity model. It uses loop closure (small overlapping areas of sonar data at the beginning and end of a circumnavigation) in post-processing to identify the value of the iceberg velocity. The result is a self-consistent iceberg map suitable for iceberg-relative vehicle navigation. However, the map can be somewhat warped relative to the true shape of the iceberg due to any innacuracies in the constant velocity assumption. This paper presents an extended iceberg mapping method which uses a more complex motion model, capable of more accurately representing the motion of the iceberg and therefore of providing a more accurate iceberg map. To identify the parameters in this model, the extended method incorporates Doppler sonar measurements which give information about the velocity of the iceberg throughout the circumnavigation. Because the Doppler velocity measurements depend on the trajectory of the iceberg and also on their locations on the iceberg surface, the extended method estimates simultaneously the parameters in the iceberg motion model and- - the locations of the recorded sonar soundings in an iceberg-fixed reference frame. The paper details the multi-objective least-squares estimation computation used in the iceberg motion and shape estimation. Included is a basis spline implementation for modeling and parameterizing the potentially complex iceberg trajectory. The paper presents results from simulated data to support the explanation of the method and demonstrate that the iceberg shape it recovers is less warped (relative to truth) than that generated under the assumptions used in previous work.
Keywords :
Doppler measurement; least squares approximations; motion estimation; remotely operated vehicles; sea ice; sonar; underwater vehicles; AUV; Doppler sonar measurements; autonomous underwater vehicles; data collection; iceberg motion estimation; iceberg-relative navigation; mapping; multi-objective least-squares estimation; sonar data; Mathematical model; Position measurement; Space vehicles; Spline; Trajectory; Velocity measurement;
Conference_Titel :
Autonomous Underwater Vehicles (AUV), 2010 IEEE/OES
Conference_Location :
Monterey, CA
Print_ISBN :
978-1-61284-980-5
DOI :
10.1109/AUV.2010.5779664