Visual sensing for autonomous underwater exploration and intervention tasks. (1st January 2015)

• Record Type: Journal Article
• Authors: Bonin-Font, Francisco; Oliver, Gabriel; Wirth, Stephan; Massot, Miquel; Lluis Negre, Pep; Beltran, Joan-Pau
• Is Part Of: Ocean engineering. Volume 93(2015)
• Page Start: 25
• Page End: 44
• Publication Date: 2015-01-01
• Subjects: Underwater intervention and exploration -- Vision-based navigation -- Visual object detection and tracking; Ocean engineering -- Periodicals; Ocean engineering; Periodicals; 620.4162
• DOI: 10.1016/j.oceaneng.2014.11.005
• Languages: English
• ISSNs: 0029-8018

Abstract

Abstract: Underwater activities, such as surveying or interventions, carried out by autonomous robots, can benefit greatly from using a vision system. Optics based systems provide information at a spatial and temporal resolution higher than their acoustic counterparts. At present, they are the best option when high precision maneuvering and manipulation is needed, if there is good visibility. This paper presents a new system designed to provide visual information in submarine tasks such as navigation, surveying, mapping and intervention. The main advantages of our system, called Fugu-f (Fugu flexible), are its robustness in both the mechanical structure and the software components, its flexibility, since it is installed as an external module and is adaptable to different vehicles and missions, and its capacity to operate in real-time. Experiments of surveying and object manipulation carried out in real conditions in the context of the TRIDENT project show the suitability of the system and its scientific and industrial potential applications. Abstract : Highlights: Success in underwater intervention depends on a fine object recognition & tracking. Vision has revealed to be a highly effective sensor for underwater vehicles. Traditionally, visual sensors are rigidly integrated in the vehicle structure. We propose a new versatile and external visual module for AUVs for up to 250 m depth. We highlight the methods for object detection & tracking, proven to work in the sea.