Abstract
A dense and accurate disparity map is relevant for a large number of applications, ranging from autonomous driving to robotic grasping. Recent developments in machine learning techniques enable us to bypass sensor limitations, such as low resolution, by using deep regression models to complete otherwise sparse representations of the 3D space. This article proposes two main approaches that use a single RGB image and sparse depth information gathered from a variety of sensors/techniques (stereo, LiDAR and Light Stripe Ranging (LSR)): a Convolutional Neural Network (CNN) and a cascade architecture, that aims to improve the results of the first. Ablation studies were conducted to infer the impact of these depth cues on the performance of each model. The models trained with LiDAR sparse information are the most reliable, achieving an average Root Mean Squared Error (RMSE) of 11.8 cm on our own Inhouse dataset; while the LSR proved to be too sparse of an input to compute accurate predictions on its own. © Springer Nature Switzerland AG 2020.

Abstract
The use of robotic solutions in tasks such as the inspection and monitorization of offshore wind farms aims to, not only mitigate the involved risks, but also to reduce the costs of operating and maintaining these structures. Performing a complete inspection of the platforms in useful time is crucial. Therefore, multiple agents can prove to be a cost-effective solution. This work proposes a trajectory planning algorithm, based on the Ant Colony metaheuristic, capable of optimizing the number of Autonomous Surface Vehicles (ASVs) to be used, and their corresponding route. Experiments conducted on a simulated environment, representative of the real scenario, proves this approach to be successful in planning a trajectory that is able to select the appropriate number of agents and the trajectory of each agent that avoids collisions and at the same time guarantees the full observation of the offshore structures.

Abstract
The offshore wind power industry is an emerging and exponentially growing sector, which calls to a necessity for a cyclical monitoring and inspection to ensure the safety and efficiency of the wind farm facilities. Thus, the multiple domains of the environment must be reconstructed, namely the emersed (aerial) and immersed (underwater) domains, to depict as much as possible the offshore structures from the wind turbines to the cable arrays. This work proposes the use of an Autonomous Surface Vehicle (ASV) to map both environments simultaneously producing a multi-domain map through the fusion of navigational sensors, GPS and IMU, to localize the vehicle and aid the registration process for the perception sensors, 3D Lidar and Multibeam echosounder sonar. The performed experiments demonstrate the ability of the multi-domain mapping architecture to provide an accurate reconstruction of both scenarios into a single representation using the odometry system as the initial seed to further improve the map with data filtering and registration processes.

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
This paper presents an algorithm for mapping monopiles from Offshore Wind Farms (OWF). The ASV (Autonomous Surface Vehicle) surveys the environment, detects and localizes monopiles using situational awareness system based on LiDAR, GPS and IMU (Inertial Measurement Unit) data. The position of the monopile is obtained based on the relative localization between the extrapolated center of the structure that was detected and the ASV. A positive detection of a monopile is referenced to a global positioning frame based on the GPS. Results in a simulator environment demonstrate the ability of this situational awareness system to identify monopiles with a precision of 0.005 m, which is relevant for detecting structural disalignments over time that might be caused by the appearance of scour in the structure's foundation.

Abstract
The large-scale deployment of smart home technologies will unlock the flexibility of prosumers, which in turn will be transformed into electricity market services by aggregators. This paper proposes a new network-constrained bidding optimization strategy to coordinate the participation of aggregators of prosumers in the day-ahead energy and secondary reserve markets. This bidding optimization strategy consists of a decentralized approach based on the alternating direction method of multipliers, where aggregators negotiate with the distribution system operator to obtain network-constrained energy and secondary reserve bids. For a case study of 2 aggregators and 1 distribution system operator, the results show that the network-constrained bidding strategy computes cost-effective and network-feasible energy and secondary reserve bids, as opposed to a network-free bidding strategy. In addition, the network-constrained bidding strategy preserves the independent roles of aggregators and the distribution system operator, and the data privacy of all agents.