This paper is the second in a series of two papers documenting the research activities in the area of vision-based navigation for unmanned underwater vehicles. The earlier paper, published in Part B7 of IMarEST Proceedings, covered tracking of pipelines and cables and moving targets such as fish. This paper covers the topics of motion estimation and station-keeping and then goes on to explain how this can be used in mosaic-based positioning. The most significant of the previous efforts at each are examined, including an overview of the image processing techniques used, the various motion estimation algorithms employed and the different styles of approach. Methods for reducing the error in each technique are discussed. Both monocular and binocular vision is examined and the outstanding challenges still to be solved are presented. Finally, some conclusions from both papers are offered.
展开▼
