This paper presents a novel and practical convolutional neural network architecture to implement semantic segmentation for side scan sonar (SSS) image. As a widely used sensor for marine survey, SSS provides higher-resolution images of the seafloor and underwater target. However, for a large number of background pixels in SSS image, the imbalance classification remains an issue. What is more, the SSS images contain undesirable speckle noise and intensity inhomogeneity. We define and detail a network and training strategy that tackle these three important issues for SSS images segmentation. Our proposed method performs image-to-image prediction by leveraging fully convolutional neural networks and deeply-supervised nets. The architecture consists of an encoder network to capture context, a corresponding decoder network to restore full input-size resolution feature maps from low-resolution ones for pixel-wise classification and a single stream deep neural network with multiple side-outputs to optimize edge segmentation. We performed prediction time of our network on our dataset, implemented on a NVIDIA Jetson AGX Xavier, and compared it to other similar semantic segmentation networks. The experimental results show that the presented method for SSS image segmentation brings obvious advantages, and is applicable for real-time processing tasks.

This brief further explores the problem of finite-time synchronization of delayed recurrent neural networks with the mismatched parameters and neuron activation functions. An improved sliding mode control approach is presented for addressing the finite-time synchronization problem. First, by employing the drive-response concept and the synchronization error of drive-response systems, a novel integral sliding mode surface is constructed such that the synchronization error can converge to zero in finite time along the constructed integral sliding mode surface. Second, a suitable sliding mode controller is designed by relying on Lyapunov stability theory such that all system state trajectories can be driven onto the predefined sliding mode surface in finite time. Moreover, it is found that the presented control approach can be conveniently verified and does not need to solve any linear matrix inequality (LMI) to guarantee the finite-time synchronization of delayed recurrent neural networks. Finally, three numerical examples are exploited to demonstrate the effectiveness of the presented control approach.

This paper addresses an autonomous navigation method for the autonomous underwater vehicle (AUV) C-Ranger applying information-filter-based simultaneous localization and mapping (SLAM), and its sea trial experiments in Tuandao Bay (Shangdong Province, P.R. China). Weak links in the information matrix in an extended information filter (EIF) can be pruned to achieve an efficient approach-sparse EIF algorithm (SEIF-SLAM). All the basic update formulae can be implemented in constant time irrespective of the size of the map; hence the computational complexity is significantly reduced. The mechanical scanning imaging sonar is chosen as the active sensing device for the underwater vehicle, and a compensation method based on feedback of the AUV pose is presented to overcome distortion of the acoustic images due to the vehicle motion. In order to verify the feasibility of the navigation methods proposed for the C-Ranger, a sea trial was conducted in Tuandao Bay. Experimental results and analysis show that the proposed navigation approach based on SEIF-SLAM improves the accuracy of the navigation compared with conventional method; moreover the algorithm has a low computational cost when compared with EKF-SLAM.

Navigation technology is one of the most important challenges in the applications of autonomous underwater vehicles (AUVs) which navigate in the complex undersea environment. The ability of localizing a robot and accurately mapping its surroundings simultaneously, namely the simultaneous localization and mapping (SLAM) problem, is a key prerequisite of truly autonomous robots. In this paper, a modified-FastSLAM algorithm is proposed and used in the navigation for our C-Ranger research platform, an open-frame AUV. A mechanical scanning imaging sonar is chosen as the active sensor for the AUV. The modified-FastSLAM implements the update relying on the on-board sensors of C-Ranger. On the other hand, the algorithm employs the data association which combines the single particle maximum likelihood method with modified negative evidence method, and uses the rank-based resampling to overcome the particle depletion problem. In order to verify the feasibility of the proposed methods, both simulation experiments and sea trials for C-Ranger are conducted. The experimental results show the modified-FastSLAM employed for the navigation of the C-Ranger AUV is much more effective and accurate compared with the traditional methods.

This paper proposes a prediction strategy for the hydrodynamic performance of bionic fish. The major challenges are meshing and building prediction model. The NACA0012 airfoil is used to replace the fish driven by the body and/or caudal fin (BCF), and a two-dimensional swimming geometric model is constructed. The geometric model is divided into hybrid meshes using the overset mesh method. The classical traveling wave model is studied using Matlab, and an improved self-propelled motion model is established. The hydrodynamic performance of the self-propelled model is numerically simulated based on computational fluid dynamics (CFD). In this strategy, the geometric model and the self-propelled motion model are integrated with user defined functions (UDF). The influences of the parameters such as inflow velocity, frequency, wavelength, and head fluctuation amplitude on the motion performance are studied. The results show that when inflow velocity is uniform, the self-propelled motion will eventually reach a quasi-steady state. According to the numerical simulation results, a hydrodynamic performance prediction model is established based on multilayer perceptron (MLP). The model is used to predict the performance of the optimized traveling wave parameters, and the error is within 3 $\mathrm{\%}$. The accuracy and generalization ability of the MLP prediction model is verified.