Navigation using 3D features from side-scan sonar data for a deep-sea AUV

Simultaneous navigation and mapping (SLAM) of an autonomous underwater vehicle (AUV) based on side-scan sonar data has significant peculiarities that make standard SLAM techniques inapplicable. In particular, recognition of already visited places (loop closure) which is an important tool improving the navigation accuracy cannot be done with the raw sonar data. This and other navigation tasks can be more conveniently performed based on the 3D seafloor shape (elevation map). In this report we first present an extension of a well-known sonar data reconstruction method to account for nontrivial AUV motion, and then discuss various algorithms useful in the context of SLAM processing elevation data.