RU-SLAM: A Robust Deep-Learning Visual Simultaneous Localization and Mapping (SLAM) System for Weakly Textured Underwater Environments.

Accurate and robust simultaneous localization and mapping (SLAM) systems are crucial for autonomous underwater vehicles (AUVs) to perform missions in unknown environments. However, directly applying deep learning-based SLAM methods to underwater environments poses challenges due to weak textures, image degradation, and the inability to accurately annotate keypoints. In this paper, a robust deep-learning visual SLAM system is proposed. First, a feature generator named UWNet is designed to address weak texture and image degradation problems and extract more accurate keypoint features and their descriptors. Further, the idea of knowledge distillation is introduced based on an improved underwater imaging physical model to train the network in a self-supervised manner. Finally, UWNet is integrated into the ORB-SLAM3 to replace the traditional feature extractor. The extracted local and global features are respectively utilized in the feature tracking and closed-loop detection modules. Experimental results on public datasets and self-collected pool datasets verify that the proposed system maintains high accuracy and robustness in complex scenarios.

Adaptive Navigation Algorithm with Deep Learning for Autonomous Underwater Vehicle.

Precise navigation is essential for autonomous underwater vehicles (AUVs). The measurement deviation of the navigation sensors, especially the microelectromechanical systems (MEMS) sensors, is a crucial factor that affects the localization accuracy. Deep learning is a novel method to solve this problem. However, the calculation cycle and robustness of the deep learning method may be insufficient in practical application. This paper proposes an adaptive navigation algorithm with deep learning to address these questions and realize accurate navigation. Firstly, this algorithm uses deep learning to generate low-frequency position information to correct the error accumulation of the navigation system. Secondly, the χ2 rule is selected to judge if the Doppler velocity log (DVL) measurement fails, which could avoid interference from DVL outliers. Thirdly, the adaptive filter, based on the variational Bayesian (VB) method, is employed to estimate the navigation information simultaneous with the measurement covariance, improving navigation accuracy even more. The experimental results, based on AUV field data, show that the proposed algorithm could realize robust navigation performance and significantly improve position accuracy.

Coronavirus Mask Protection Algorithm: A New Bio-inspired Optimization Algorithm and Its Applications.

Nowadays, meta-heuristic algorithms are attracting widespread interest in solving high-dimensional nonlinear optimization problems. In this paper, a COVID-19 prevention-inspired bionic optimization algorithm, named Coronavirus Mask Protection Algorithm (CMPA), is proposed based on the virus transmission of COVID-19. The main inspiration for the CMPA originated from human self-protection behavior against COVID-19. In CMPA, the process of infection and immunity consists of three phases, including the infection stage, diffusion stage, and immune stage. Notably, wearing masks correctly and safe social distancing are two essential factors for humans to protect themselves, which are similar to the exploration and exploitation in optimization algorithms. This study simulates the self-protection behavior mathematically and offers an optimization algorithm. The performance of the proposed CMPA is evaluated and compared to other state-of-the-art metaheuristic optimizers using benchmark functions, CEC2020 suite problems, and three truss design problems. The statistical results demonstrate that the CMPA is more competitive among these state-of-the-art algorithms. Further, the CMPA is performed to identify the parameters of the main girder of a gantry crane. Results show that the mass and deflection of the main girder can be improved by 16.44% and 7.49%, respectively.

A High Precision Modeling Technology of Material Surface Microtopography and Its Influence on Interface Mechanical Properties.

In order to accurately and effectively obtain the contact performance of the mating surface under the material surface topography characteristics, a numerical simulation method of rough surface based on the real topography characteristics and a multi-scale hierarchical algorithm of contact performance is studied in this paper. Firstly, the surface topography information of materials processed by different methods was obtained and characterized by a measuring equipment; Secondly, a non-Gaussian model considering kurtosis and skewness was established by Johnson transform based on Gaussian theory, and a rough surface digital simulation method based on real surface topography was formed; Thirdly, a multi-scale hierarchical algorithm is given to calculate the contact performance of different mating surfaces; Finally, taking the aeroengine rotor as the object, the non-Gaussian simulation method was used to simulate the mating surfaces with different topographies, and the multi-scale hierarchical algorithm was used to calculate the contact performance of different mating surfaces. Analysis results showed that the normal contact stiffness and elastic-plastic contact area between the mating surfaces of assembly 1 and assembly 2 are quite different, which further verifies the feasibility of the method. The contents of this paper allow to perform the fast and effective calculation of the mechanical properties of the mating surface, and provide a certain analysis basis for improving the surface microtopography characteristics of materials and the product performance.

Numerical Simulation and Accuracy Verification of Surface Morphology of Metal Materials Based on Fractal Theory.

This paper presents a numerical simulation method to determine the surface morphology characteristics of metallic materials. First, a surface profiler (NV5000 5022s) was used to measure the surface, and the morphology data thereof were characterized. Second, fractal theory was used to simulate the surface profile for different fractal dimensions D and scale coefficients G, and statistical analyses of different surface morphologies were carried out. Finally, the fractal dimension D of the simulated morphology and the actual morphology were compared. The analysis showed that the error of fractal dimension D between the two morphologies was less than 10%; meanwhile, the comparison values of the characterization parameters of the simulated morphology and the actual morphology were approximately equal, and the errors were below 6%. Therefore, the current method used to evaluate the surface morphologies of parts processed by the grinding/milling method can be replaced by the simulated method using the corresponding parameters. This method makes it possible to theorize about the surface morphologies of machined parts, and provides a theoretical basis and reference value for the surface morphology design of materials, with the potential to improve the assembly quality of products.