Vessel maintenance requires periodic visual inspection of the hull in order to detect typical defective situations of steel structures such as, among others, coating breakdown and corrosion. These inspections are typically performed by well-trained surveyors at great cost because of the need for providing access means (e.g., scaffolding and/or cherry pickers) that allow the inspector to be at arm’s reach from the structure under inspection. This paper describes a defect detection approach comprising a micro-aerial vehicle which is used to collect images from the surfaces under inspection, particularly focusing on remote areas where the surveyor has no visual access, and a coating breakdown/corrosion detector based on a three-layer feed-forward artificial neural network. As it is discussed in the paper, the success of the inspection process depends not only on the defect detection software but also on a number of assistance functions provided by the control architecture of the aerial platform, whose aim is to improve picture quality. Both aspects of the work are described along the different sections of the paper, as well as the classification performance attained.

No abstract

Vessels constitute one of the most cost effective ways of transporting goods around the world. Despite the efforts, maritime accidents still occur, with catastrophic consequences. For this reason, vessels are submitted to periodical inspections for the early detection of cracks and corrosion. These inspections are nowadays carried out at a great cost. In order to contribute to make ship inspections safer and more cost-efficiently, this paper presents a novel framework to turn a Micro-Aerial Vehicle (MAV) into a flying camera that can virtually teleport the human surveyor through the different structures of the vessel hull. The system architecture has been developed to be reconfigurable so that it can fit different sensor suites able to supply a proper state estimation, being at the same time compatible with the payload capacity of the aerial platform and the operational conditions. The control software has been designed following the Supervised Autonomy paradigm, so that it is in charge of safety related issues such as collision avoidance, while the surveyor, within the main control loop, is supposed to supply motion commands while he/she is concentrated on the inspection at hand. In this paper, we report on an extensive evaluation of the platform capabilities and usability, both under laboratory conditions and on board a real vessel, during a field inspection campaign.

Hand-crafted point descriptors have been traditionally used for visual loop closure detection. However, in low-textured environments, it is usually difficult to find enough point features and, hence, the performance of such algorithms degrade. Under this context, this paper proposes a loop closure detection method that combines lines and learned points to work, particularly, in scenarios where hand-crafted points fail. To index previous images, we adopt separate incremental binary Bag-of-Words (BoW) schemes for points and lines. Moreover, we adopt a binarization procedure for features’ descriptors to benefit from the advantages of learned features into a binary BoW model. Furthermore, image candidates from each BoW instance are merged using a novel query-adaptive late fusion approach. Finally, a spatial verification stage, which integrates appearance and geometry perspectives, allows us to enhance the global performance of the method. Our approach is validated using several public datasets, outperforming other state-of-the-art solutions in most cases, especially in low-textured scenarios.

SLAM approaches rely on loop closure strategies to avoid and/or correct the inconsistencies in the resulting map. These inconsistencies are mainly caused by the effect of sensor noise in odometry sources. For the case of visual SLAM, loop detection typically rely on the repetitive detection and matching of texture-based keypoints. Weakly textured environments, however, can lead to scenes lacking these kind of points and, hence, poor-performing loop detectors. An alternative for these environments is the use of geometrical cues such as line segments, which are frequently present within human-made, structured environments. Under this context, in this work, we introduce a novel appearance-based loop closure detection method that integrates lines and points to enhance performance in these scenarios. For this purpose, we build an incremental Bag-of-Binary-Words scheme for each visual cue to retrieve previously seen images from the two complementary perspectives. Furthermore, we rely on a late fusion strategy to combine the image candidates resulting for both visual vocabularies. An effective mechanism to group similar images close in time is applied next to reduce the effort of the image candidate search. Finally, we propose a novel scheme to validate geometrically the loop candidates, integrating lines into the procedure. The proposed approach compares favourably with other state-of-the-art methods for several datasets.

No abstract

No abstract

No abstract