Design of a robust vision-based sensor of position and rate for the guidance of autonomous underwater vehicles

This paper is concerned with the design of a vision-based algorithm for on-line estimation of position and rate control errors in the guidance of autonomous underwater vehicles for path tracking of underwater lines. The algorithm uses techniques of pattern recognition with different degrees of morphological operations. Ad-hoc experiments with a sub aquatic vehicle in a test tank show the features of our approach under strong conditions of light perturbations and cloudy water.