Antoni Martorell-Torres

This article focuses on the coordination and communication aspects of Marine Multi-Robot Systems (MMRS). A new MMRS coordination approach is proposed where a heterogeneous group of marine vehicles must explore an unknown area and communicate the gathered information to a central agent. This article proposes a task allocation method to coordinate the vehicles and decide when and how to communicate. Different allocation optimization algorithms are proposed in order determine which strategy produces optimal results.

This work introduces a comprehensive hardware and software solution for an underwater Stereo Vision System. Rigorous robustness and resistance studies were conducted using finite element analysis, accompanied by a meticulous selection of materials and design considerations. Multiple tests were executed to determine optimal camera configuration parameters, ensuring the acquisition of high-quality images and accurate 3D information. A Robot Operating System-based software for fish detection and classification was also integrated into the system, enabling the creation of 3D point clouds. The entire system was successfully tested in a real marine environment, and the results demonstrated the robustness of the hardware design and the potential of the software for fish biomass estimation applications. These findings confirm that the proposed solution is effective and can be practically applied in underwater environments.

Marine invasive macroalgae alter the environment in which they settle, changing food chains, generating structural variations and, sometimes, displacing native species. Halimeda incrassata (H.i.) is a tropical seaweed that settles mostly on sandy substrates and that has expanded almost sevenfold in the Balearics from 2011 to nowadays. Measuring its coverage is crucial to estimate its expansion rate and plan effective eradication actions. In the last years, the Marine Robotics team of the Systems, Robotics and Vision group (University of the Balearic Islands) has collaborated actively with biologists in the observation of sensible marine habitats using an Autonomous Underwater Vehicle to get data and image processing techniques to infer information of biologic interest. Several marine areas of Cabrera National Park were surveyed by our robot for H.i. video recording. Images of videos were used to form photo-mosaics, mark all H.i. shoots and calculate automatically its coverage with software. Time, extension and depth of data collection campaigns increase, offering measurements more accurate, and with higher temporal and spatial resolutions than those obtained with traditional techniques based on divers and quadrat frames.

In this paper, we present the design, development and a practical use of an Autonomous Surface Vehicle (ASV) as a modular and flexible platform for a large variety of marine tasks including the coordination strategies with other marine robots. This work tackles the integration of an open-source Robot-Operating-System (ROS)-based control architecture that provides the ASV with a wide variety of navigation behaviors. These new ASV capabilities can be used to acquire useful data from the environment to survey, map, and characterize marine habitats. In addition, the ASV is used as a radio frequency relay point between an Autonomous Underwater Vehicle (AUV) and the ground station as well as to enhance the Acoustic Communication Link (ACL) with the AUV. In order to improve the quality of the ACL, a new Marine Multirobot System (MMRS) coordination strategy has been developed that aims to keep both vehicles close to each other. The entire system has been successfully designed, implemented, and tested in real marine environment robotic tasks. The experimental tests show satisfactory results both in ROS-based navigation architecture and the MMRS coordination strategy resulting in a significant improvement of the quality of the ACL.

Visual Loop Detection (VLD) is a core component of any Visual Simultaneous Localization and Mapping (SLAM) system, and its goal is to determine if the robot has returned to a previously visited region by comparing images obtained at different time steps. This paper presents a new approach to visual Graph-SLAM for underwater robots that goes one step forward the current techniques. The proposal, which centers its attention on designing a robust VLD algorithm aimed at reducing the amount of false loops that enter into the pose graph optimizer, operates in three steps. In the first step, an easily trainable Neural Network performs a fast selection of image pairs that are likely to close loops. The second step carefully confirms or rejects these candidate loops by means of a robust image matcher. During the third step, all the loops accepted in the second step are subject to a geometric consistency verification process, being rejected those that do not fit with it. The accepted loops are then used to feed a Graph-SLAM algorithm. The advantages of this approach are twofold. First, the robustness in front of wrong loop detection. Second, the computational efficiency since each step operates only on the loops accepted in the previous one. This makes online usage of this VLD algorithm possible. Results of experiments with semi-synthetic data and real data obtained with an autonomous robot in several marine resorts of the Balearic Islands, support the validity and suitability of the approach to be applied in further field campaigns.

During the past few decades, the need to intervene in underwater scenarios has grown due to the increasing necessity to perform tasks like underwater infrastructure inspection and maintenance or archaeology and geology exploration. In the last few years, the usage of Autonomous Underwater Vehicles (AUVs) has eased the workload and risks of such interventions. To automate these tasks, the AUVs have to gather the information of their surroundings, interpret it and make decisions based on it. The two main perception modalities used at close range are laser and video. In this paper, we propose the usage of a deep neural network to recognise pipes and valves in multiple underwater scenarios, using 3D RGB point cloud information provided by a stereo camera. We generate a diverse and rich dataset for the network training and testing, assessing the effect of a broad selection of hyperparameters and values. Results show F1-scores of up to 97.2% for a test set containing images with similar characteristics to the training set and up to 89.3% for a secondary test set containing images taken at different environments and with distinct characteristics from the training set. This work demonstrates the validity and robust training of the PointNet neural in underwater scenarios and its applicability for AUV intervention tasks.