Technology and Equipment for Underwater Robots

As more and more attention is paid to marine resources and furthering marine scientific research, underwater robots, offering unique advantages and great potential, are playing an indispensable role in many respects, including in deep-sea exploration, marine ecological monitoring, salvage and rescue, and underwater engineering operations. This Special Issue focuses on the research and development of underwater robot technology and equipment. The included papers cover fields such as underwater target recognition, path tracking, task operations, wireless charging, and deep-sea driving. To enhance the depth of the discussion, we also cite further papers in the same fields for comparison and draw some interesting conclusions.

Target recognition technology serves as the “sharp eyes” of underwater robots, enabling them to accurately identify various targets and providing crucial information for subsequent scientific research, resource exploration, and safety assurance. Compared with optical images, sonar images have a longer detection range and stronger penetration ability, and they play an important role in underwater target detection and identification as well as in navigation and obstacle avoidance [1]. However, the resolution of sonar images is much lower than that of optical images. Moreover, these images are easily affected by suspended particles and bubbles in the water, further degrading image quality. Therefore, it is critical to process underwater sonar images in order to improve the efficiency and accuracy of target recognition.

The authors of [2] propose an underwater target-tracking method based on particle filtering. It involves the use of Zernike motion feature matching and a first-order autoregressive model to optimize the particle state transition, enhancing the robustness of underwater target recognition.

In [3], the authors employed two pre-trained CNN architectures trained via transfer learning. Using data with different imbalance ratios and basic data augmentation, their method combines multiple weak classification results into an ensemble average, solving the problem in which human-shaped objects in side-scan sonar images were automatically detected by the Police Robot for Inspection and Mapping of Underwater Evidence (PRIME).

The authors of [4] propose a new underwater target detection model, introducing the SRFD module, RFCAConv, and Dysample in the yolov8 network to address the problems of information loss and feature fusion in side-scan sonar images for underwater target detection.

In [5], the authors propose a new sonar image feature extraction method (FMSE) and combine it with the support vector machine algorithm to address the difficulty of feature extraction and the low recognition accuracy in underwater acoustic target recognition pertaining to divers.

Underwater robots’ path-tracking ability allows them to navigate precisely in unpredictable water currents and on complex terrains, effectively ensuring that they can efficiently execute various tasks along the preset trajectory. Different from the terrestrial environment, the dynamic changes of water currents and the unknown interferences caused by complex obstacles undoubtedly pose great difficulties in the path-tracking control of underwater robots [6].

The authors of [7] considered roll angle during the motion of an unmanned sailboat, built corresponding kinematic and dynamic models, and employed a sail–rudder collaborative control method based on model predictive control (MPC). Their method solves the problem wherein the traditional sail–rudder control systems of unmanned sailboats are prone to excessive yaw and roll angles, and it also improves the path-tracking ability of unmanned sailboats.

Based on the adaptive sliding mode strategy, the authors of [8] sought to address the path-tracking control and multi-degree-of-freedom coordinated crawling control of an underwater hexapod robot by designing a path-tracking control system that includes a line-of-sight (LOS) guidance system, a super-twisting sliding mode controller, and a fuzzy mapping system, improving the tracking accuracy, motion stability, and coordination of the robot.

The authors of [9] constructed an integrated path-planning and tracking control framework using the proximal policy optimization (PPO) algorithm. Their approach solves a path-planning problem, namely, maximizing the energy harvested, and a path-tracking problem, that is, minimizing the tracking error and avoiding collisions with the turbine, for energy-harvesting autonomous underwater vehicles operating in marine currents in a stochastic ocean environment.

In [10], the authors address the cooperative path-planning problem in which air–sea heterogeneous unmanned vehicles must complete an underwater target search-and-tracking mission. They did so by using task allocation algorithms based on different initial positions to solve problems in the search phase and an improved particle swarm optimization (PSO) algorithm to address a path-planning problem concerning obstacles in the tracking phase, improving search efficiency and allowing the shortest tracking path to be found.

The underwater environment is complex and changeable. Factors such as water pressure, water temperature, and water quality are unstable. Furthermore, the speed and direction of currents are irregular, and they may carry debris. These complexities and interferences make it difficult for underwater robots to accurately position themselves, operate stably, and accurately identify targets during intervention operations, greatly increasing the difficulty of control and associated operational risks [11].

In order to allow the accurate control of soft robots in the presence of contact constraints, the authors of [12] took contact constraints into account, established a spatial dynamics model of soft robots, and improved the operational compliance of these robots. The method they developed can provide new ideas for the control of intervention operations carried out by underwater soft robots.

The authors of [13] used the Adaptive Neuro-Fuzzy Inference System (ANFIS) to predict the heave motion of a Tether Management System (TMS) for the docking of work-class remotely operated vehicles (ROVs) in order to address the issue of ROVs needing to rely on the pilot’s experience to estimate TMS heave motion and mitigate the difficulties posed by limited power and high drag forces that limit ROVs’ ability to achieve autonomous docking.

The authors of [14] designed a multi-claw shipwreck-salvaging system with a semi-active heave compensation (SAHC) function and used a multivariate long short-term memory (LSTM) neural network to predict the motion of a barge in order to address unknown interference from sea waves affecting the system and the system’s own delay, thus improving the stability of shipwreck salvage.

The authors of [15] proposed a human autonomy teaming (HAT) framework for ROV teleoperation by integrating autonomy algorithms, virtual reality (VR), and sensory augmentation methods and conducted a human subject experiment to compare different control methods, aiming to reduce the learning and operational burdens on ROV operators and enhance our understanding of ROV work statuses.

Wireless power transfer (WPT) technology, a new underwater power transmission method, offers advantages such as safety, reliability, convenience, and concealment. However, compared with aerial wireless power transfer systems, its transmission power and efficiency are still relatively low [16,17].

The authors of [18] proposed a novel magnetic coupling structure with a solenoid transmitting coil and dual combined planar receiving coils for an AUV wireless power transfer (WPT) system and analyzed and optimized it theoretically and experimentally to address power transmission efficiency reductions caused by misalignment in underwater WPT systems.

In [19], the authors analyze the effects of the hulls of autonomous underwater vehicles (AUVs) on underwater capacitive wireless power transfer (UCWPT) and underwater inductive wireless power transfer (UIWPT) systems through simulations and experimental work and compare the different performances of the two systems to provide a reference for the design of underwater wireless power transfer systems for AUVs.

The authors of [20] focused on long-range underwater wireless power transfer systems. They adopted the LCC-S-S compensation circuit and designed a portable omnidirectional magnetic resonant extender with two orthogonal and decoupled coils, addressing the large output decay and fluctuation caused by the flow of water and enlarging the resonance range and improving the efficiency and stabilization of the output.

During the underwater wireless charging process, the charging coils can easily be misaligned in the horizontal and vertical directions, making it difficult to maintain constant current or voltage output. Therefore, the authors of [21] designed a new detuned inductive power transfer (IPT) system structure that improves misalignment tolerance in underwater wireless charging.

The deep sea remains a vast and inaccessible space, presenting enormous challenges for exploration due to its extreme pressure, temperature, and darkness [22].

The authors of [23] proposed a light robot joint created using thermally activated paraffin suitable for deep-sea environments and analyzed the paraffin phase change process and the performance of the joint. Unlike the traditional driving structures, which cannot operate in the deep sea for a long time, this joint allows suitably long deep-sea operation.

Inspired by the structure of the deep-sea snailfish, the authors of [24] designed a dielectric elastomer material for the flapping fins of a robot by integrating electronic devices into a silicone matrix to reduce interfacial shear stress. Additionally, a cableless soft robot was developed, solving the problem wherein rigid containers and pressure compensation systems need to be used due to high pressure in deep-sea environments.

The authors of [25] aimed to make miniature underwater robots that can move quickly and agilely in a high-pressure environment, features the existing versions of such robots lack. They employed a design in which piezoelectric bi-jet actuators are directly embedded in a spherical shell. By optimizing the structural parameters and working frequency of the actuators, multi-degree-of-freedom, agile movement of the robot without bulky pressure-resistant devices was achieved.

This article introduces the relevant research on underwater technology and equipment pertaining to robots, covering technical fields such as underwater acoustic target recognition, path tracking, intervention operation control, energy supply, and deep-sea driving. It provides technical support and innovative ideas for the development of underwater robots in marine exploration and operations. Regarding the research on underwater robot technology and equipment, in addition to the scope covered in this Special Issue, the technologies related to underwater robots also include multi-robot cooperation, underwater communication, autonomous positioning and navigation technology, etc. We expect more scholars to make more contributions in the field of underwater robot technology and equipment to help humans explore the ocean more comprehensively and efficiently and tap its seemingly endless potential and value.