Abstract
In recent years, research concerning the operation of Autonomous Surface Vehicles (ASVs) has seen an upward trend, although the full-scale application of this type of vehicles still encounters diverse limitations. In particular, the docking and undocking processes of an ASV are tasks that currently require human intervention. Aiming to take one step further towards enabling a vessel to dock autonomously, this article presents a Deep Learning approach to detect a docking structure in the environment surrounding the vessel. The work also included the acquisition of a dataset composed of LiDAR scans and RGB images, along with IMU and GPS information, obtained in simulation. The developed network achieved an accuracy of 95.99%, being robust to several degrees of Gaussian noise, with an average accuracy of 9334% and a deviation of 5.46% for the worst case.

Abstract
Understanding the surrounding 3D scene is of the utmost importance for many robotic applications. The rapid evolution of machine learning techniques has enabled impressive results when depth is extracted from a single image. High-latency networks are required to achieve these performances, rendering them unusable for time-constrained applications. This article introduces a lightweight Convolutional Neural Network (CNN) for depth estimation, NEON, designed for balancing both accuracy and inference times. Instead of solely focusing on visual features, the proposed methodology exploits the Motion-Parallax effect to combine the apparent motion of pixels with texture. This research demonstrates that motion perception provides crucial insight about the magnitude of movement for each pixel, which also encodes cues about depth since large displacements usually occur when objects are closer to the imaging sensor. NEON's performance is compared to relevant networks in terms of Root Mean Squared Error (RMSE), the percentage of correctly predicted pixels (delta(1)) and inference times, using the KITTI dataset. Experiments prove that NEON is significantly more efficient than the current top ranked network, estimating predictions 12 times faster; while achieving an average RMSE of 3.118 m and a delta(1) of 94.5%. Ablation studies demonstrate the relevance of tailoring the network to use motion perception principles in estimating depth from image sequences, considering that the effectiveness and quality of the estimated depth map is similar to more computational demanding state-of-the-art networks. Therefore, this research proposes a network that can be integrated in robotic applications, where computational resources and processing-times are important constraints, enabling tasks such as obstacle avoidance, object recognition and robotic grasping.

Abstract
Structures associated with offshore wind energy production require an arduous and cyclical inspection and maintenance (O&M) procedure. Moreover, the harsh challenges introduced by sub-sea phenomena hamper visibility, considerably affecting underwater missions. The lack of quality 3D information within these environments hinders the applicability of autonomous solutions in close-range navigation, fault inspection and intervention tasks since these have a very poor perception of the surrounding space. Deep learning techniques are widely used to solve these challenges in aerial scenarios. The developments in this subject are limited regarding underwater environments due to the lack of publicly disseminated underwater information. This article presents a new underwater dataset: NEREON, containing both 2D and 3D data gathered within real underwater environments at the ATLANTIS Coastal Test Centre. This dataset is adequate for monocular depth estimation tasks, which can provide useful information during O&M missions. With this in mind, a benchmark comparing different deep learning approaches in the literature was conducted and presented along with the NEREON dataset.

Abstract
Exploiting stronger winds at offshore farms leads to a cyclical need for maintenance due to the harsh maritime conditions. While autonomous vehicles are the prone solution for O&M procedures, sub-sea phenomena induce severe data degradation that hinders the vessel's 3D perception. This article demonstrates a hybrid underwater imaging system that is capable of retrieving tri-dimensional information: dense and textured Photogrammetric Stereo (PS) point clouds and multiple accurate sets of points through Light Stripe Ranging (LSR), that are combined into a single dense and accurate representation. Two novel fusion algorithms are introduced in this manuscript. A Joint Masked Regression (JMR) methodology propagates sparse LSR information towards the PS point cloud, exploiting homogeneous regions around each beam projection. Regression curves then correlate depth readings from both inputs to correct the stereo-based information. On the other hand, the learning-based solution (RHEA) follows an early-fusion approach where features are conjointly learned from a coupled representation of both 3D inputs. A synthetic-to-real training scheme is employed to bypass domain-adaptation stages, enabling direct deployment in underwater contexts. Evaluation is conducted through extensive trials in simulation, controlled underwater environments, and within a real application at the ATLANTIS Coastal Testbed. Both methods estimate improved output point clouds, with RHEA achieving an average RMSE of 0.0097 m -a 52.45% improvement when compared to the PS input. Performance with real underwater information proves that RHEA is robust in dealing with degraded input information; JMR is more affected by missing information, excelling when the LSR data provides a complete representation of the scenario, and struggling otherwise.

Abstract
Underwater scenarios pose additional challenges to perception systems, as the collected imagery from sensors often suffers from limitations that hinder its practical usability. One crucial domain that relies on accurate underwater visibility assessment is underwater pipeline inspection. Manual assessment is impractical and time-consuming, emphasizing the need for automated algorithms. In this study, we focus on developing learning-based approaches to evaluate visibility in underwater environments. We explore various neural network architectures and evaluate them on data collected within real subsea scenarios. Notably, the ResNet18 model outperforms others, achieving a testing accuracy of 93.5% in visibility evaluation. In terms of inference time, the fastest model is MobileNetV3 Small, estimating a prediction within 42.45 ms. These findings represent significant progress in enabling unmanned marine operations and contribute to the advancement of autonomous underwater surveillance systems.

Abstract
Marine growth impacts the stability and integrity of offshore structures, while simultaneously preventing inspection procedures. In consequence, companies need to employ specialists that manually assess each impacted part of the structure. Due to harsh sub-sea environments, acquiring large quantities of quality underwater data becomes difficult. To mitigate these challenges a new data augmentation algorithm is proposed that generates new images by performing localized crops on regions of interest from the original data, expanding the total size of the dataset approximately 6 times. This research also proposes a learning-based algorithm capable of automatically delineating marine growth in underwater images, achieving up to 0.389 IoU and 0.508 Dice Loss. Advances in this area contribute for reducing the manual labour necessary to schedule maintenance operations in man-made submerged structures, while increasing the reliability and automation of the process.