Abstract
Recently, research concerning the navigation of autonomous surface vehicles (ASVs) has been increasing. However, a large-scale implementation of these vessels is still held back by several challenges such as multi-object tracking. Attaining accurate object detection plays a big role in achieving successful tracking. This article presents the development of a detection model with an image-based Con-volutional Neural Network trained through transfer learning, a deep learning tech-nique. To train, test, and validate the detector module, data were collected with the SENSE ASV by sailing through two nearby ports, Leixoes and Viana do Castelo, and recording video frames through its on-board cameras, along with a Light De-tection And Ranging, GPS, and Inertial Measurement Unit data. Images were ex-tracted from the collected data, composing a manually annotated dataset with nine classes of different vessels, along with data from other open-source maritime datasets. The developed model achieved a class mAP@[.5 .95] (mean average precision) of 89.5% and a clear improvement in boat detection compared to a multi-purposed state-of-the-art detector, YOLO-v4, with a 22.9% and 44.3% increase in the mAP with an Intersection over Union threshold of 50% and the mAP@[.5 .95], respectively. It was integrated in a detection and tracking system, being able to continuously detect nearby vessels and provide sufficient informa-tion for simple navigation tasks.

Abstract
Underwater autonomous manipulation is the capability of a mobile robot to perform intervention tasks that require physical contact with unstructured environments without continuous human supervision. Being difficult to assess the behaviour of existing motion planner algorithms, this research proposes a new planner evaluation metric to identify well-behaved planners for specialized tasks of inspection and monitoring of man-made underwater structures. This metric is named NEMU and combines three different performance indicators: effectiveness, safety and adaptability. NEMU deals with the randomization of sampling-based motion planners. Moreover, this article presents a benchmark of multiple planners applied to a 6 DoF manipulator operating underwater. Results conducted in real scenarios show that different planners are better suited for different tasks. Experiments demonstrate that the NEMU metric can be used to distinguish the performance of planners for particular movement conditions. Moreover, it identifies the most promising planner for collision-free motion planning, being a valuable contribution for the inspection of maritime structures, as well as for the manipulation procedures of autonomous underwater vehicles during close range operations.

Abstract
Underwater environments pose unique challenges to optical systems due to physical phenomena that induce severe data degradation. Current imaging sensors rarely address these effects comprehensively, resulting in the need to integrate complementary information sources. This article presents a multimodal data fusion approach to combine information from diverse sensing modalities into a single dense and accurate tridimensional representation. The proposed fusiNg tExture with apparent motion information for underwater Scene recOnstruction (NESO) encoder-decoder network leverages motion perception principles to extract relative depth cues, fusing them with textured information through an early fusion strategy. Evaluated on the FLSea-Stereo dataset, NESO outperforms state-of-the-art methods by 58.7%. Dense depth maps are achieved using multi-stage skip connections with attention mechanisms that ensure propagation of key features across network levels. This representation is further enhanced by incorporating sparse but millimeter-precise depth measurements from active imaging techniques. A regression-based algorithm maps depth displacements between these heterogeneous point clouds, using the estimated curves to refine the dense NESO prediction. This approach achieves relative errors as low as 0.41% when reconstructing submerged anode structures, accounting for metric improvements of up to 0.1124 m relative to the initial measurements. Validation at the ATLANTIS Coastal Testbed demonstrates the effectiveness of this multimodal fusion approach in obtaining robust tri-dimensional representations in real underwater conditions.

Abstract
With the expansion of offshore infrastructure, the necessity for efficient Operation and Maintenance (O&M) procedures intensifies. This article introduces DADDI, a multimodal dataset obtained from a real offshore floating structure, aimed at facilitating comprehensive inspections and 3D model creation. Leveraging Unmanned Aerial Vehicles (UAVs) equipped with advanced sensors, DADDI provides synchronized data, including visual images, thermal images, point clouds, GNSS, IMU, and odometry data. The dataset, gathered during a campaign at the ATLANTIS Coastal Testbed, offers over 2500 samples of each data type, along with intrinsic and extrinsic sensor calibrations. DADDI serves as a vital resource for the development and evaluation of algorithms, models, and technologies tailored to the inspection, monitoring, and maintenance of complex maritime structures. © 2024 IEEE.

Abstract
This work presents a bio-inspired Autonomous Underwater Vehicle (AUV) concept called Raya that enables high manoeuvrability required for close-range inspection and intervention tasks, while fostering endurance for long-range operations by enabling efficient navigation. The AUV has an estimated terminal velocity of 0.82 m/s in an optimal environment, and a capacity to acquire visual data and sonar measurements in all directions. Raya was designed with the potential to incorporate an electric manipulator arm of 6 degrees of freedom (DoF) for free-floating underwater intervention. Smart and biologically inspired principles applied to morphology and a strategic thruster configuration assure that Raya is capable of manoeuvring in all 6 DoFs even when equipped with a manipulator with a 5 kg payload. Extensive experiments were conducted using simulation tools and real-life environments to validate Raya's requirements and functionalities. The stresses and displacements of the rigid bodies were analysed using finite element analysis (FEA), and an estimation of the terminal forward velocity was achieved using a dynamic model. To assess the accuracy of the perception system, a reconstruction task took place in an indoor pool, resulting in a 3D reconstruction with average length, width, and depth errors below 1. 5%. The deployment of Raya in the ATLANTIS Coastal Testbed and Porto de Leix & otilde;es allowed the validation of the propulsion system and the gathering of valuable 2D and 3D data, thus proving the suitability of the vehicle for operation and maintenance (O&M) activities of underwater structures.

Abstract
Accurate dynamics modelling of Unmanned Under-water Vehicles (UUV s) is critical for optimizing mission planning, minimizing collision risks, and ensuring the successful execution of tasks in diverse underwater environments. This paper presents a structured approach to estimating the hydrodynamic coeffi-cients of UUV s. Initially, it follows a detailed methodology for estimating hydrodynamic coefficients using simple geometries, a sphere and a spheroid, using the Computational Fluid Dy-namics (CFD) software OpenFoam, and comparing the results to analytical solutions, enabling the validation of the simulations approach. Following this, the paper provides an in-depth analysis of the damping and added mass coefficients for the Raya UUV, offering valuable insights into its hydrodynamic behaviour. © 2024 IEEE.