This paper presents a visual-based trajectory opti-mization framework designed to enhance the navigation capabil-ities of Autonomous Underwater Vehicles (AUVs) utilized for underwater monitoring and inspection operations. The framework leverages the optical feedback from a monocular camera to detect loop closures and refine the estimation of the robot trajectory through a pose graph optimization procedure. The solution builds upon a state-of-the-art appearance-based loop closure detection method, which has been modified to improve its robustness and efficiency in underwater scenarios. The modifications include the use of contrast-limited adaptive histogram equalization to enhance image quality, the introduction of a keyframe selection procedure to reduce computational costs, and the implementation of a motion estimation stage for computing relative rigid transformations between loop closures. The optimization strategy was tested of Fline using a dataset of real underwater images acquired during structure inspection activities at sea. In par-ticular, the implemented loop closure detection and pose graph optimization functionalities were integrated with a visual-based dead-reckoning navigation approach that utilizes a monocular visual odometry algorithm providing information concerning the linear velocity of the AUV. The experimental results demonstrate the effectiveness of the proposed system in improving navigation performance within complex and unstructured underwater environments. The trajectory optimization process significantly reduces the drift of the monocular visual-based dead-reckoning estimation, thereby enhancing the accuracy of state estimation and the 2eo-referencin2 of collected data.

This paper presents the preliminary experimental validation of a visual-acoustic-based framework for the autonomous inspection of critical underwater infrastructures using Autonomous Underwater Vehicles (AUVs). The goal is to evaluate the feasibility of a strategy that enables an AUV to navigate relative to a target while maintaining a desired configuration. The proposed target relative navigation approach employs a minimal sensor suite - comprising a frontal stereo camera, a frontal acoustic range sensor, and an inertial unit - to provide information on the relative orientation and distance of the AUV from the inspection surface. This information is used to compute control actions that allow the vehicle to dynamically adjust its pose relative to the facility, adhering to specific mission requirements. The system was validated through an at-sea experimental campaign, during which a reference AUV performed relative navigation tests in front of a port dock. During the experiments, the target relative navigation solution was integrated into the robot software architecture to provide online feedback, thus enabling the AUV to maintain the desired distance and orientation relative to the dock. The results demonstrate the effectiveness and robustness of the proposed relative navigation approach, confirming its feasibility for enhancing the autonomous capabilities of underwater robots in marine infrastructure inspection tasks. Even under deliberate perturbations introduced via joystick by an operator, the AUV successfully maintained a reference distance from the harbor quay and regulated its orientation to zero, ensuring the frontal perception payload remained aligned with the target.

This work details the conceptualization and development of an Autonomous Surface Vehicle (ASV) designed for supporting underwater navigation. The ASV, which is named MARINA (Marine Autonomous Robot for Intelligent Networking Operations), was developed at the Robotics, Automation and Autonomous Systems Laboratory of the Department of Information Engineering of the University of Pisa. First, the paper outlines the main considerations that led to the conceptualization of the prototype. Then, the main features in terms of hardware and software, which have been adopted for the realization of the desired marine robot, are summarized. Finally, results obtained during a preliminary experimental campaign in a lake environment are presented.

Autonomous Underwater Vehicles (AUVs) employed for inspection and monitoring applications require a reliable estimate of their navigation status to successfully accomplish the scheduled mission. Considering that they are typically endowed with vision systems to collect images of the surveyed environment, approaches based on Visual Odometry (VO) allow to utilise the optical payload also to perform robot navigation. Despite a large body of prior research, the topic of navigation based on stereo vision is still an open problem in the marine domain. Within this context, this paper proposes an underwater navigation framework utilising a stereo camera as a linear velocity sensor. The purpose is to assess the potentiality of a stereo vision system to indirectly measure the linear velocity of an AUV, and provide this information to the navigation framework of the vehicle. The proposed strategy consists in a visual inertial odometry solution fusing linear velocity estimates, derived from optical information, with attitude and depth measurements to retrieve the robot navigation status. Such linear velocity estimates are computed exploiting a stereo VO algorithm, relying on a 3D-to-2D approach, whose source code is made available online. The developed strategy is evaluated on real underwater datasets acquired during monitoring surveys performed by an AUV equipped with Doppler Velocity Log (DVL), bottom-looking stereo camera, inertial unit, and depth sensor. The results, obtained through a comparison with the DVL taken as benchmark, confirm the approach as feasible and exploitable in underwater inspection and monitoring contexts.

Recent developments in marine technologies allow underwater vehicles to perform survey missions for data collection in an automatic way. The scientific community is now focusing on endowing these vehicles with strong perception capabilities, aiming at full autonomy and decision-making skills. Such abilities would bring benefits to a wide range of field applications, e.g. Inspection and Maintenance (I&M) of man-made structures, port security, and marine rescue. Indeed, most of these tasks are currently carried out employing remotely operated vehicles, making the presence of humans in water necessary. Projects like Metrological Evaluation and Testing of Robots in International CompetitionS (METRICS), funded by the European Commission, are promoting research on this field by organising events such as the Robotics for Asset Maintenance and Inspection (RAMI) competition. In particular, this competition requires participants to develop perception techniques capable of identifying a set of specific targets. Within such context, this paper presents an algorithm able to detect and classify Objects of Potential Interest (OPIs) in underwater camera images. First, the proposed solution compensates for the quality degradation of underwater images by applying color enhancement and restoration procedures. Then, it exploits deep-learning techniques, as well as color and shape based methods, to recognize and correctly label the predefined OPIs. Preliminary results of the implemented neural network using restored images are provided, and a mean Average Precision (mAP) of about 92% was achieved on the dataset provided to the RAMI competition participating teams by the NATO Science and Technology Organization Centre for Maritime Research and Experimentation (STO CMRE).

This paper proposes an evaluation of the impact of the 2D feature correspondence stage on an underwater vision-based navigation solution using a monocular Visual Odometry (VO) algorithm for linear velocity estimation. In particular, this work compares three different mismatch removal methods: the Cross-Check (CC), the Lowe’s Ratio Test (RT), and the Grid-based Motion Statistics (GMS). The performance of the three methods was assessed using two datasets containing real underwater images, which were collected by an Autonomous Underwater Vehicle (AUV) during monitoring activities over two distinct marine areas exhibiting different seafloor characteristics. The comparison is conducted considering both quantity and quality of features returned by the three approaches. In addition, the influence they have on the overall VO algorithm in terms of linear velocity accuracy is taken into account, using doppler velocity log readings as a reference. The results show that the three techniques are comparable in the case of a seabed characterised by identifiable and discernible features. In contrast, when surveying a more challenging and variable scenario, the RT technique shows a greater ability than CC and GMS to filter out erroneous 2D correspondences. This ensures higher accuracy in estimating the AUV linear velocity by the monocular VO algorithm. Furthermore, in both scenarios analysed, the RT technique is also the one that leads to a lower computational cost of the entire VO algorithm, and thus a better suitability for a real-time application onboard the AUV.

Cooperation among heterogeneous marine vehicles can offer several benefits to underwater applications, such as supporting the navigation of submerged robots by leveraging the information available to a surface vehicle. Within such context, this paper describes an acoustic positioning and communication protocol developed for the localisation of multiple Autonomous Underwater Vehicles (AUVs) by means of an Autonomous Surface Vehicle (ASV) equipped with an Ultra-Short BaseLine (USBL) device. To avoid packet collisions, access to the acoustic channel is managed through a time slot division mechanism. Unlike a classical Time Division Multiple Access (TDMA) protocol, the approach considered in this work consists of a centralised scheduling handled by the ASV, in which AUVs are localised at regular time epoch but can transmit only when queried. This allows to prioritise the number of position measurements taken by the ASV and then relayed back to the submerged vehicles, with the final goal of improving the navigation accuracy of the latter. The solution is also designed to handle the latency of acoustic communication and to correctly associate each position measurement with the corresponding acquisition time. Furthermore, it does not require any a priori synchronisation between the clocks of the vehicles involved and ensures complete decoupling between their navigation algorithms. Experimental activities in very-shallow waters, involving an ASV and two target nodes, were carried out to validate the multi-AUVs positioning system and to provide a characterisation of its performance.

Periodical inspections are a fundamental operation to monitor the status of underwater structures and to assess their need for proper maintenance or repair interventions. Autonomous Underwater Vehicles (AUVs) could represent a viable option to carry out underwater inspection tasks, potentially bringing benefits in terms of safety for human operators and quality of the collected data. Aiming at developing a fully autonomous vision-based inspection strategy, this paper proposes a comparative analysis between monocular and stereo vision approaches for estimating the lateral velocity of an AUV and its orientation with respect to a target surface. The proposed analysis is performed by exploiting a dataset of real underwater images, collected during at-sea experiments in which the Zeno AUV was remotely driven to carry out a pier inspection. Specifically, the performance of the two solutions in terms of estimation of the robot lateral velocity is assessed by considering doppler velocity log measurements as benchmark. Instead, the accuracy of the estimation of the vehicle orientation with respect to the target is evaluated by taking into account both geographical information of the pier and AUV attitude observations. The comparison suggests that stereo vision provides better performance for estimating the relative orientation between the AUV and the target; on the contrary, the monocular approach produces more reliable lateral velocity estimates. The results obtained prove the suitability of the two vision-based strategies for inspection applications in a real underwater scenario, thus suggesting a possible implementation onboard the reference vehicle.

The periodical hull inspection represents a necessary task to ensure the maintenance of a vessel since it allows to counteract decay, check for structural damages, and fight the biofouling phenomenon affecting the navigation efficiency. Typically, this task is executed by divers, resulting in a dangerous job for the human operator, or by Remotely Operated Vehicles, driven by highly trained users. Aiming at automating the task and increasing its operational safety, this work proposes a strategy to perform the ship hull inspection using an Autonomous Underwater Vehicle (AUV), equipped with a stereo camera and a proximity sensor, without a prior knowledge of the target shape. At first, the images from the stereo vision system allow to estimate the lateral velocity of the vehicle and its orientation with respect to the hull surface. Then, the proximity measurement, properly projected along the normal axis to the surface of the target, provides a measure of the distance of the AUV from the surveyed structure. Lastly, the robot control system exploits these estimates to perform the mission with a constant lateral velocity, maintaining both a predefined safety distance from the target and the optical axis of the camera orthogonal to the examined surface. The proposed approach has been tested in a simulated environment, performing the investigation of a simplified model of ship hull. The results suggest the feasibility of the strategy: during the simulations, the AUV completes the mission with a full autonomy, safely, obtaining a 3D reconstruction of the surveyed structure.

Oceans preservation and protection have become increasingly relevant topics to tackle climate change. To this end, Autonomous Underwater Vehicles (AUVs) provide a useful means to carry out inspection and monitoring operations in full autonomy. A particular scenario in which AUVs are crucial involves the detection and mapping of underwater gas leaks, whether these are due to damaged offshore structures or naturally released from the seafloor. In this context, the proposed work investigates the effects of gas seeps on the navigation performance of AUVs. Indeed, the navigation of underwater vehicles mostly relies on acoustic sensors, as Doppler Velocity Log (DVL), which can be negatively affected by the presence of gas bubbles. The paper explores two solutions, based on two different acoustic sensors working at distinct frequencies: a DVL sensor and an Ultra-Short BaseLine (USBL) device. Both strategies have been implemented and tested during at-sea experiments, where gas leaks have been artificially reproduced. Results showed that both methods suffer from the presence of gas bubbles, causing erroneous DVL measurements and lost of USBL connectivity, respectively.

Autonomous Underwater Vehicles (AUVs) performing visual surveys aimed at the preservation of marine environments are equipped with optical sensors for image acquisition. In addition, an altitude sensor is usually installed on-board to control the distance from the seabed and avoid possible collisions. Within this context, this work proposes a navigation strategy for underwater monitoring scenarios, which fuses a single bottom-looking camera and altitude information for linear velocity estimation. This allows to exploit the payload already required by monitoring activities also for navigation purposes, thus reducing the number of sensors onboard the AUV. The linear velocity is provided by a monocular Visual Odometry (VO) technique that switches between homography and epipolar models for motion estimation and leverages altitude measurements to overcome the scale ambiguity issue. The navigation framework relies on an Extended Kalman Filter (EKF) that combines visual-based linear velocity with attitude and depth measurements for trajectory estimation. The proposed strategy has been tested on real data acquired by using Zeno AUV, equipped with bottom-looking camera, DVL, Attitude and Heading Reference System (AHRS), and depth sensor. The performance has been assessed comparing the estimated linear velocities with the DVL readings, and the VO-based estimated trajectory with that provided by a DVL-based dead-reckoning approach, yielding to a maximum absolute error of 2.16m for a reference trajectory of 166m. Given the promising results, this strategy could represent an affordable solution for underwater navigation where visibility conditions allow the use of optical sensors.

In the robotics field, cooperative approaches involving robot swarms represent a challenging task, in order to guarantee robots safety and accomplish mission goals. This problem becomes more severe when missions involve under-water vehicles, due to electromagnetic waves attenuation in underwater environment which results in several limitations on communication and navigation. In this paper, we describe a stepping stone to an acoustic-based cooperative navigation strategy between an Autonomous Surface Vehicle (ASV) and an Autonomous Underwater Vehicle (AUV). The goal is to provide the ASV with tracking capabilities in order to assist the AUV during the mission and enhance acoustic communication. The proposed tracking approach exploits within an Extended Kalman Filter (EKF) the measurements from the Ultra Short BaseLine (USBL) device installed on-board the ASV. The estimated AUV position is then used by the ASV to implement a pursuit strategy of the underwater vehicle, which consists of moving forward while pointing towards the underwater vehicle. The developed system has been tested in a real marine scenario to assess its behavior and the quality of tracking and pursuit performance. The experimental results show that the tracking algorithm allows the ASV to correctly track the AUV trajectories, while the strategy adopted to follow the AUV effectively allows to decrease distance between the two vehicles.

Recent technological developments have paved the way to the employment of Autonomous Underwater Vehicles (AUVs) for monitoring and exploration activities of marine environments. Traditionally, in information gathering scenarios for monitoring purposes, AUVs follow predefined paths that are not efficient in terms of information content and energy consumption. Informative Path Planning (IPP) represents a valid alternative, defining the path that maximises the gathered information. This work proposes a Genetic Path Planner (GPP), which consists in an IPP strategy based on a Genetic Algorithm, with the aim of generating a path that simultaneously maximises the information gathered and the coverage of the inspected area. The proposed approach has been tested offline for monitoring and inspection applications of Posidonia Oceanica (PO) in three different geographical areas. The a priori knowledge about the presence of PO, in probabilistic terms, has been modelled utilising a Gaussian Process (GP), trained on real marine data. The GP estimate has then been exploited to retrieve an information content of each position in the areas of interest. A comparison with other two IPP approaches has been carried out to assess the performance of the proposed algorithm.