Intelligent automatic interpretation of active marine sonar

Abstract

This dissertation explores the problems raised by the design and
\nconstruction of a real-time sonar interpreter operating in a three dimensional
\nmarine context, and then focusses on two major research
\nissues inherent in sonar interpretation: the treatment of observer
\nand object motion, and the efficient exploitation of the specularity
\nof acoustic reflection. The theoretical results derived in these
\nareas have been tested where appropriate by computer simulation.
\nIn the context of mobile marine robotics, the registration of sensory
\ndata obtained from differing viewpoints is of paramount importance.
\nSmall marine vehicles of the type considered here do not
\ncarry sophisticated navigational equipment, and cannot be held stationary
\nin the water for any length of time.
\nThe viewpoint registration problem is defined and analysed in
\nterms of the new problem of motion resolution: the task of resolving
\nthe apparent motion of objects into that part due to the movement of
\nthe observer and that due to the objects' proper motion. Two solutions
\nto this under constrained problem are presented. The first
\npresupposes that the observer orientation is known ~ priori so that
\nonly the translational observer motion must be determined. It is
\napplicable to two and three-dimensional situations. The second solution
\ndetermines both the translational and the rotational motion of
\nthe observer, but is restricted to a two-dimensional situation. Both
\nsolutions are based on target
\nextensively tested in two
\ntracking techniques, and have
\ndimensions by computer simulation.
\nbeen
\nThe
\nnecessary extensions to deal with full three-dimensional motion are
\nalso discussed.
\nThe second major research issue addressed in this thesis is the
\nefficient use of specularity. Specular echoes have a high intrinsic
\ninformation content because of the alignment conditions necessary for
\ntheir generation. In the marine acoustic context they provide a significant
\nproportion of the information available from an acoustic
\nranger. I suggest a new method that uses directly the information
\npresent in specular reflections and the history of the vehicle motion
\nto classify the specular echo sources and infer the local structure
\nof the objects bearing them. The method builds on the output of a
\nmotion resolution system. Six distinct types of specular echo source
\nare described and three properties useful for their discrimination
\nare discussed. A suitable inference system for the analysis and
\nclassification of specular echo sources is also proposed.