Focusing on coordination and consensus, this paper, addresses a control problem for a group of unmanned underwater vehicles (UUVs) by tracking a maneuvering target with varying velocity and time delays. To realize coordinated target tracking, consensus control of multiple UUVs requires neighboring UUV state information, detected target state information, and estimated target acceleration information. At least one UUV is assumed to be capable of obtaining information about the target, and the communication topology graph of the vehicle is assumed to be undirected connected. We consider the convergence analysis of this multi-UUV system for two conditions: information interaction between the target and an UUV with time delays and information interaction between the target and an UUV without time delays. Sufficient conditions for the uniform ultimate boundedness of the tracking errors and estimation errors are obtained by utilizing a Lyapnov-Krasovskii functional. Two simulations are presented to demonstrate the effectiveness of the proposed method.

To solve the unavailability of a traditional strapdown inertial navigation system (SINS) for unmanned underwater vehicles (UUVs) in the polar region, a polar grid navigation algorithm for UUVs is proposed in this paper. Precise navigation is the basis for UUVs to complete missions. The rapid convergence of Earth meridians and the serious polar environment make it difficult to establish the true heading of the UUV at a particular instant. Traditional SINS and traditional representation of position are not suitable in the polar region. Due to the restrictions of the complex underwater conditions in the polar region, a SINS based on the grid frame with the assistance of the OCTANS and the Doppler velocity log (DVL) is chosen for a UUV navigating in the polar region. Data fusion of the integrated navigation system is realized by a modified fuzzy adaptive Kalman filter (MFAKF). By neglecting the negative terms, and using T-S fuzzy logic in the adaptive regulation of the noise covariance, the proposed filter algorithm can improve navigation accuracy. Simulation and experimental results demonstrate that the polar grid navigation algorithm can effectively navigate a UUV sailing in the polar region.

More and more underwater robots are deployed to investigate marine biodiversity autonomously, and tools are needed by underwater robots to discover and acknowledge marine life. This paper has proposed a convolutional neural network-based method for intelligent fish detection and recognition with a dataset used for training and testing generated and augmented from an open-source Fish Database regarding 6 different types. Firstly, to improve image quality, a hybrid image enhancement algorithm is used to preprocess underwater images with a weighted fusion strategy of multiple traditional methodologies and comparisons have been made to prove the effectiveness according to various indexes. Secondly, to increase detection and recognition accuracy, different attention modules are integrated into the YOLOv5m network structure and the convolutional block attention module(CBAM) has outperformed other modules in recall rate and mAP while maintaining the capability of real-time processing. Lastly, to meet real-time requirements, lightweight adjustments have been made to CBAM-YOLOv5m with the GSConv module and C3Ghost module and a nearly 25% reduction in network parameters and a 20% reduction in computational consumption are obtained. Besides, the lightweight network has realized better accuracy than YOLOv5m. In conclusion, the method proposed in this paper is effective in real-time fish detection and recognition with practical application prospects.

A novel network for enhancement of underwater optical images is proposed in this paper. It contains a Reinforcement Fusion Module for Haar wavelet (RFM-Haar) images. Fusion is performed to enrich available underlying information for better enhancement. As this network transform Haar Images into Fusion Images, it is called HIFI-Net. The experimental results based on Mean Square Error (0.1952, 0.1512 and 0.4683), Peak Signal to Noise Ratio (25.2241, 26.3342 and 21.4253) and Structure Similarity Index Measure (0.8265, 0.8819 and 0.8012) on three public datasets (EUVP, UFO-110 and UIEB) demonstrate that the proposed HIFI-Net performs better than some existing state-of-the-art methods.