Designing embedded fish sensor for underwater robot

Abstract

Commercial fish finders have already been known and widely used in so many real applications. However, sometimes these kinds of instrument are not suitable for underwater robotic or submarine robot applications. The size and the interfacing features are not designed to meet the requirements of underwater robot application purposes. This system relies on an Artificial Neural Network implemented on an embedded microcontroller. By using a proximity ping sensor widely used in mobile robot with some dedicated signal preconditioning and processing of extracted features with the proposed algorithm, a fish detection and classification system has been realized. The proposed system gives satisfactory achievement with respective maximum values of 100% for detection and 94% for classification. Also the existence and the type of fish can be known and the behavior of group can also be revealed by statistically interpretations such as hovering passion and sparse swimming mode.