AUV Dynamic Obstacle Avoidance Method Based
on Improved PPO Algorithm
ABSTRACTD：esigning a reasonable obstacle avoidance method for AUV 3D path planning is difﬁcult, and existing obstacle avoidance methods have certain drawbacks. For example, they are only applicable to 2D planar applications and cannot effectively handle dynamic obstacles. To address these problems, we design an obstacle collision prediction model (CPM). Based on the results of the simulation of obstacles’ inertial motion, the safety of the AUV navigation is evaluated to improve the model’s sensitivity to dynamic obstacles. Then, we enhance the learning ability of the sequence sample data by combining it with a long short-term memory (LSTM) network, thus improving the training efﬁciency and effect of the algorithm. The trained proximal policy optimization (PPO) network can output reasonable actions in order to control the AUV to avoid obstacles, forming an AUV 3D dynamic obstacle avoidance strategy based on the CPM-LSTM-PPO algorithm. The simulation results show that the proposed algorithm has good generalization in uncertain environments. Moreover, it achieves dynamic AUV obstacle avoidance in different three-dimensional unknown environments, providing theoretical and technical support for real path planning.
RSI-Net: Two-Stream Deep Neural Network for Remote Sensing Images-Based Semantic Segmentation
ABSTRACT: For semantic segmentation of remote sensing images (RSI), trade-off between representation power and location accuracy is quite important. How to get the trade-off effectively is an open question, where current approaches of utilizing very deep models result in complex models with large memory consumption. In contrast to previous work that utilizes dilated convolutions or deep models, we propose a novel two-stream deep neural network for semantic segmentation of RSI (RSI-Net) to obtain improved performance through modeling and propagating spatial contextual structure effectively and a decoding scheme with image-level and graph-level combination. The ﬁrst component explicitly models correlations between adjacent land covers and conduct ﬂexible convolution on arbitrarily irregular image regions by using graph convolutional network, while densely connected atrous convolution network (DenseAtrousCNet) with multi-scale atrous convolution can expand the receptive ﬁelds and obtain image global information. Extensive experiments are implemented on the Vaihingen, Potsdam and Gaofen RSI datasets, where the comparison results demonstrate the superior performance of RSI-Net in terms of overall accuracy (91.83%, 93.31% and 93.67% on three datasets, respectively), F1 score (90.3%, 91.49% and 89.35% on three datasets, respectively) and kappa coefﬁcient (89.46%, 90.46% and 90.37% on three datasets, respectively) when compared with six state-ofthe-art RSI semantic segmentation methods.
The Impact of Different Mapping Function Models and Meteorological Parameter Calculation Methods on the Calculation Results of Single-Frequency Precise Point Positioning with Increased Tropospheric Gradient
ABSTRACT: Tropospheric delay is one of the main errors in precise point positioning (PPP). )e inaccuracy of the tropospheric delay model will inevitably lead to a decrease in PPP accuracy.)erefore, the inﬂuence of the tropospheric gradient on the positioning accuracy should be considered in the processing of tropospheric delay. At the same time, the eﬀects of diﬀerent mapping function models and meteorological parameter calculation methods on the tropospheric delay accuracy of single-frequency PPP (SF PPP) are analyzed. Twelve MGEX stations, which are evenly distributed in the world, are used in this article. Taking into account the seasonal variation of the tropospheric delay, the observation times adopted are 2016, 2017, and 2018 for diﬀerent seasons (winter, day of year (DOY): 22–28; spring, DOY: 92–98; summer, DOY: 199–205; and autumn, DOY: 275–281). )en, according to diﬀerent mapping function models and meteorological parameter calculation methods, a total of 7056 tests and 9072 tests are performed, respectively. )e following results were obtained after comparative analysis. (1) When the same method is used for calculating meteorological parameters, the percentage with improved tropospheric delay repeatability calculated by the Hopﬁeld mapping function model (MFM3) is the highest, reaching more than 70%, and by Vienna Mapping Functions 3 (VMF3, grid resolution is 1°), the mapping function model (MFM8) is the lowest, less than 67.5%. )e percentage with improved position repeatability is highest in the north (N) direction and lowest in the up (U) direction. (2) Using the same mapping function model, the correction of the tropospheric gradient model has a greater impact on calculating the repeatability percentage of the tropospheric delay and the position. Compared with standard atmospheric parameters, other calculation methods of meteorological parameters have little eﬀect on the percentage increase of the tropospheric delay value and the positioning result after adding the tropospheric gradient model. It shows that diﬀerent calculation methods of meteorological parameters have little eﬀect on the calculation of tropospheric delay and position, diﬀerent mapping function models have a large eﬀect on the calculation of tropospheric delay and position, and the tropospheric gradient model has the greatest inﬂuence on the calculation of tropospheric delay and position.
A method of establishing an instantaneous water level model for tide correction
ABSTRACT:Tide correction is important in ship-board bathymetric data. Currently, most tide correction algorithms use a space-time interpolation method to regain the composite sea surface morphology. These interpolation algorithms are mostly based on geometric interpolation. However, when the tide stations are insuﬃcient in the survey region, the spatial pattern of tide change may not comply with the geometric trend. Some of the algorithms use tide simulation to obtain a space-time astronomical tide model for tide correction in tide-stationinsuﬃcient regions, because the instantaneous water level changes are mainly caused by astronomical tides in normal conditions. However, in some cases, the instantaneous water level eﬀect of the short-period stochastic meteorological factors results in short-term water level anomalies, which can be diﬃcult to simulate using a tide simulation method. Thus, in this paper, an instantaneous water level model for tide correction in tide-stationinsuﬃcient regions is proposed. The model includes the simulation of astronomical tide and deviation-tidal components. We ﬁrst simulate the astronomical tidal model using a two-dimensional MIKE21 Flow Model. Then, we propose a deviation correction method to mitigate the deviation-tidal components. Using the revised instantaneous water level model, we present the instantaneous tide correction (ITC) algorithm. Then, we compare the ITC algorithm with the two commonly used algorithms of Discrete Tidal Zoning and TCARI. The results show that the ITC algorithm is superior to the common algorithms in terms of accuracy and applicability with respect to a tide-station-insuﬃcient survey region. Our conclusion is that ITC algorithm is a feasible tide correction algorithm when tide stations are not suﬃciently measured.
An Effective Method for Submarine Buried Pipeline Detection via Multi-Sensor Data Fusion
ABSTRACT: Submarine pipelines are important resource delivery devices between land and ocean. For safety reasons, pipelines are often embedded beneath the seabed at a certain depth, to reduce the risk of direct damage to the pipeline. In the past, various kinds of detection equipment have been used for pipeline inspection, to ensure the normal operation of pipelines in practical applications. Acoustic detection technology is the dominant method to monitor buried submarine pipelines. Extracting and integrating the information in acoustic images, such as the route and burial depth, can help to monitor the status of a pipeline. However, most of the existing methods are based on limited parameters, and they cannot be used to precisely detect and locate a submarine pipeline under complex conditions. In this study, a multi-sensor surveying system was used, which integrates a sub-bottom proﬁler (SBP) and the Shipborne Over- and Under-Water Integrated Mobile Mapping System (SiOUMMS) on the same ship. The data acquired in this system include acoustic proﬁle images and the over- and under-water topography of the pipeline route area. We also designed a position deviation correction method to improve the accuracy of the pipeline detection positioning, i.e., pipeline positioning correction in the real-time kinematic (RTK) positioning data and pipeline horizontal route correction in the integrated data. Compared with the uncorrected pipeline detection positioning result, the reliability of the pipeline inspection result is greatly improved, and the effectiveness and merit of the proposed method are clearly demonstrated. Finally, we conducted a buried pipeline safety assessment for the installation of newly designed wharf piles at Mawan Port of Shenzhen, China, where the results showed that one of the ﬁrst rows of wharf piles would collide with the sewage pipeline.
An investigation into real-time GPS/GLONASS single-frequency precise point positioning and its atmospheric mitigation strategies
ABSTRACT: One of the challenges in the innovative application of global navigation satellite systems (GNSS) lies in real-time single-frequency precise point positioning (RT-SFPPP). The well-known problems associated with SFPPP are its slow convergence and lower positioning accuracy due to the effects of various errors inherited by the GNSS positioning, including the atmospheric error. In order to mitigate the two above-mentioned problems, several scenarios for the reduction of the atmospheric delays in RT-SFPPP, including the ionospheric constraint and tropospheric constraint, were investigated in this research. The performance of the RT-SFPPP utilizing the above-mentioned methods was evaluated using one-week observations from multi-GNSS experiment stations in a simulated kinematic mode. Results showed that the convergence time of the RT-SFPPP was signiffcantly shortened when the slant ionospheric delays derived from the real-time ionospheric products provided by Centre National d’E´tudes Spatiales were applied as the pseudo-observations. Results also showed that the convergence time of GPS + GLONASS RT-SFPPP can be reduced by modeling the frequency-dependent part of the GLONASS receiver uncalibrated code delay as a quadratic polynomial function of GLONASS frequency number. The a priori zenith wet delays (ZWDs) derived from forecast Vienna Mapping Functions 3 (VMF3_FC) were compared to the reference ZWDs from globally distributed radiosonde stations. The mean root mean square error of the a priori ZWDs resulting from VMF3_FC at 381 radiosonde stations in 2020 was 1.53 cm compared to the radiosonde-based ZWDs. When the ZWDs derived from VMF3_FC were used as pseudo-observations in the position estimation system, the convergence time for the vertical positioning reaching a 0.3 m accuracy was considerably shortened.
Dam Structure Deformation Monitoring by GB-InSAR Approach
ABSTRACT: Ground-based synthetic aperture radar (GB-SAR) has been proved to be one of the cutting-edge techniques for the timely detection of slope failures in both natural and engineered slopes. This paper focuses on the structure deformation monitoring on the dams using GB-SAR data. Temporal sequence data was collected by ground SAR equipment from 29 July to 1 August for the Geheyan dam and the SAR images with high quality were selected through the exhaustive spatial-temporal coherence analysis based on permanent scatterer (PS) theory in this paper. A practical solution for dam structure deformation extraction after the atmospheric effect reduction is proposed in depth. The deformation of the dam spillway gates is greater than that of the dam body monitored by this GB-SAR campaign, and with the increase of the water level in the reservoir area, the displacement increases along the direction of water ﬂow gradually. The surface deformation rate of the dam body is ﬁtted by linear regression analysis, and the interpolated rate results are compared and veriﬁed with the plumb line measurements. Finally, the consistency of the dam deformation average rate based on the PS time series analysis technology by GB-SAR and plumb lines is veriﬁed in this article, demonstrated the excellent performance of the proposed method for remote multipoint displacement measurements of the dam.
Analyzing the Characteristics of Soil Moisture Using GLDAS Data: A Case Study in Eastern China
ABSTRACT: In this paper, we use GLDAS (Global Land Data Assimilation System) to analyze the effects of air temperature and precipitation on the characteristics of soil moisture in the eastern region of
China from 1961 to 2011. We ﬁnd that the temperature and precipitation in different seasons have different degrees of inﬂuence on the characteristics of soil moisture in each layer. The results show that over the last 50 years, the soil moisture in eastern China has a tendency to dry out, especially between the late 1970s to the early 1980s. The change of soil moisture with the depth of soil layer has similar inter-annual and seasonal patterns. Soil moisture in different depths (0–200 cm) positively correlates with the air temperatures of spring, summer and autumn, but negatively correlates with the air temperature in winter at 0–10 cm, 40–100 cm and 100–200 cm. Similarly, soil moisture is
positively related to the precipitation in spring, summer and autumn, and the opposite is true in winter. The results also show that precipitation has a signiﬁcant effect on the shallow soil moisture (0–10 cm), while air temperature most affects the deep soil moisture (100–200 cm).
Force and torque exerted by internal solitary waves in background
parabolic current on cylindrical tendon leg by numerical simulation
ABSTRACT: An internal gravity wave (IGW) model is employed to simulate the generation of internal solitary waves (ISWs) over a sill by tidal ﬂows, and it is shown that the simulated ISW-induced current ﬁeld agrees basically with that observed. Then we use this model to study the force and torque exerted by ISWs in background parabolic current on small-diameter cylindrical tendon leg of the oil platform. Eight numerical experiments are designed and the results are compared. It is found that, no matter whether a background parabolic current is considered or not, the maximum force lies at the depth of turning point (where the horizontal current in upper layer begins to turn to zero and ﬂow against that in lower layer), a negative extremum torque appears at the depth of turning point, and the maximum torque appears at the bottom of tendon leg. With background parabolic currents, the depth of turning point becomes shallower, and the magnitude of force decreases with depth from the depth of turning point either upward or downward. In case the affecting depths of background parabolic currents are the same, both the maximum force and its appearing depth decrease with increasing current curvature. If the maximum current velocities are the same, the maximum force decreases whilst the depths of the maximum force and turning point increase with increasing current curvature. If the background current curvatures are the same, both the maximum force and its appearing depth decrease with increasing affecting depth of parabolic current.