Two Vision-Based Algorithms for Image-Properties Extraction in the Path Tracking of Underwater-Vehicle Navigation

MARIO A. JORDÁN; CARLOS E. BERGER; JORGE L. BUSTAMANTE; SANTIAGO HANSEN

Ciudad Autónoma de Buenos Aires

Congreso; XXII Congreso Argentino de Control Automático; 2010

Asociación Argentina de Control Automático

In this paper two vision-based algorithms for on-line estimation of position control errors in a guidance system for path tracking of patterned lines are developed. The algorithms use techniques of pattern recognition, however with a high degree of morphological simplifications. Particularly, one of the approaches employs operations of thresholding and filtering while the second one applies brightness corrections and contrast improvements. The resulting features in the image-property abstraction are quite different. Generally, both algorithms run in real time providing the controller position as well as orientation of a patterned scene during the path following. Ad-hoc experiments with a subaquatic vehicle in an test tank show the performance of the visual sensor under strong conditions of light perturbation and cloudy water.