Seabed topography and landform mapping method and system based on underwater robot

Abstract

The present invention discloses a method and system for surveying and mapping seabed topography based on an underwater robot, and relates to the technical field of underwater robots. Based on an original measuring point acoustic wave data set S, reliable basic data is provided for measurement, a multipath disturbance feature set F of each sonar scanning point is extracted, multipath interference in an underwater environment is effectively identified and characterized, data is clustered through density clustering analysis, dense clusters and discrete points are successfully identified, and disturbance residuals Cc are obtained and compared with a preset interference judgment threshold Omap, a multipath label is assigned to each sonar scanning point, and a time series and amplitude series of the sonar scanning point are combined to accurately obtain the main path echo time of each measuring point, effectively judge the continuity of the topography, and ensure the high precision and stability of underwater topography survey, especially in the elimination of multipath interference, reconstruction and judgment of the main path, and the final accurate reflection of the topography.

Description

Technical Field

[0001] The present invention relates to the technical field of underwater robots, and in particular to a method and system for surveying and mapping seabed topography based on underwater robots. Background Art

[0002] Underwater mapping, as a core technology within the broader field of ocean information perception and intelligent detection, has long been responsible for understanding the spatial structure and topography of the seafloor. With the increasing demand for deep-sea resource development, submarine communications deployment, and marine ecosystem monitoring, the demand for automation and data accuracy in underwater spatial mapping is rapidly increasing. At the same time, manual operation or towed sonar systems are increasingly unable to meet the high-efficiency and high-precision mission objectives.

[0003] To adapt to the challenges of three-dimensional mapping in complex seabed environments, underwater robots have gradually become the main operating platforms, especially in geomorphic areas such as submarine canyons, fault zones, coral reefs and shipwrecks. These systems, which have autonomous motion capabilities and high-precision sensor integration capabilities, can achieve large-scale and fine-grained geomorphic data acquisition.

[0004] Although current seafloor mapping systems, using equipment such as high-resolution multibeam sonar, can construct point cloud images and complete 3D modeling in underwater environments, significant technical bottlenecks remain in achieving the goal of "true terrain reconstruction." For example, multipath interference, caused by the diversity of sound wave propagation paths in complex terrain, is particularly pronounced in areas with drastic geometric variations, such as steep cliffs, overhanging structures, and cracks and crevices. In these scenarios, in addition to the "primary path" that travels directly from the robot to the target seafloor, sound waves also reflect off various underwater interfaces or structures, forming a series of delayed "secondary paths" of echoes. These multipath echoes can be misinterpreted as actual target structures, resulting in the appearance of artificial highlands, faults, or even continuous ridges or troughs in the mapping data. This, in turn, leads to structural artifacts in the final 3D terrain map. Current methods struggle to effectively remove the spurious terrain information introduced by multipath noise, especially during automated modeling, as data filtering strategies rely on time windows and signal amplitude discrimination. Summary of the Invention

[0005] In response to the deficiencies of the prior art, the present invention provides a method and system for surveying and mapping seabed topography based on an underwater robot, which solves the problems mentioned in the background art.

[0006] To achieve the above objectives, the present invention is implemented through the following technical solutions: A method for surveying and mapping seabed topography based on an underwater robot comprises the following steps:

[0007] S1, based on the underwater robot performing sonar scanning at fixed intervals on the preset mapping path, obtaining time series reflection data at each sonar scanning point to form the original measurement point acoustic wave dataset S;

[0008] S2. Extract features from the original measurement point acoustic wave dataset S to obtain highly correlated feature parameters of multipath disturbance at each sonar scanning point to form a multipath disturbance feature set F.

[0009] S3. Map the multipath disturbance feature set F into a three-dimensional feature space and perform cluster analysis using a density clustering algorithm to identify dense clusters and discrete points. Obtain the disturbance residual Cc and compare it with the preset interference judgment threshold Omap. Assign a multipath label to each sonar scanning point to obtain the residual set A and the multipath label set Lmp.

[0010] S4, extract the time series and amplitude series from the original measurement point acoustic wave data set S of each sonar scanning point in the multipath label set Lmp, recalculate the main path echo time Tmain, and form the main path time set Hmain;

[0011] S5. Based on the sound wave propagation speed v, the main path time set Hmain is converted into the depth coordinate Z, and the height difference sequence △Z and the horizontal distance difference △D are obtained to evaluate the consistency index G of the landform. Based on the consistency index G, the reconstruction of the main path is triggered.

[0012] Preferably, said S1 includes S11 and S12;

[0013] S11, based on the underwater robot performing sonar scanning at fixed intervals on a preset mapping path, sequentially determining sonar scanning points i, and obtaining a sonar scanning point set P = {1, 2, ..., i|i∈N}, where N represents the total number of sonar scanning points;

[0014] S12. Perform high-resolution directional scanning on each sonar scanning point i based on the sonar scanning point set P to obtain complete time series echo data, which constitute the original acoustic wave data S(i) of each sonar scanning point i. Integrate the original acoustic wave data S of all sonar scanning points in the sonar scanning point set P to obtain the original measurement point acoustic wave data set S={S(1), S(2), …, S(i)|i∈N}.

[0015] Preferably, the time series echo data includes the echo time series T(i) of the sonar scanning point i, the amplitude series A(i) of the sonar scanning point i, the effective signal duration D(i) of the sonar scanning point i, the main echo peak time Tpeak and the waveform oscillation coefficient γ(i) of the sonar scanning point i;

[0016] The original sound wave data of sonar scanning point i S(i) = {T(i), A(i), D(i), Tpeak(i), γ(i)};

[0017] The echo time series T(i) of sonar scanning point i specifically represents the time axis of the entire echo receiving window, constituting the horizontal coordinate of the time domain. The echo time series T(i) of sonar scanning point i = {t1, t2, ..., tj|j∈k}, where tj represents the echo time series of the jth sampling time point, and k represents the total number of time points;

[0018] The amplitude sequence A(i) of sonar scanning point i specifically represents the amplitude of the acoustic echo that matches the echo time sequence T(i) of sonar scanning point i. The amplitude sequence A(i) of sonar scanning point i = {a1, a2, …, aj|j∈k}, where aj represents the amplitude sequence at the jth sampling time point.

[0019] The effective signal duration D(i) of the sonar scanning point i specifically represents the time duration of the echo signal. The effective signal duration D(i) of the sonar scanning point i is obtained by the tk-t1 calculation formula;

[0020] The main echo peak time Tpeak specifically represents the time point of the maximum amplitude of the main path reflection signal, which is used for multipath reflection comparison and is obtained by extracting the maximum value in the amplitude sequence A(i) of the sonar scanning point i;

[0021] The waveform oscillation coefficient γ(i) of the sonar scanning point i specifically represents the degree of waveform mutation. The waveform oscillation coefficient γ(i) of the sonar scanning point i is obtained by the following calculation formula:

[0022] ;

[0023] Where A(i, j+1) and A(i, j) represent the amplitude sequence of the sonar scanning point i at the j+1th sampling point and the amplitude sequence of the jth sampling point, respectively. δ represents the preset amplitude mutation judgment threshold. When A(i, j+1)-A(i, j)>amplitude mutation judgment threshold δ, the value returns 1; otherwise, it returns 0.

[0024] Preferably, said S2 includes S21;

[0025] S21. Extract features from the original measurement point acoustic wave data set S to obtain the multipath disturbance height-correlated feature parameters of each sonar scanning point i in the original measurement point acoustic wave data set S = {S (1), S (2), ..., S (i) | i∈N}, and form a multipath disturbance feature set F = {F (1), F (2), ..., F (i) | i∈N};

[0026] The multipath disturbance highly correlated characteristic parameters include the maximum echo delay time offset δT(i) of the sonar scanning point i, the amplitude standard deviation σA(i) of the sonar scanning point i, and the waveform jump density η(i) per unit time of the sonar scanning point i;

[0027] The maximum echo delay time offset δT(i) of sonar scanning point i is obtained by the following calculation formula:

[0028] ;

[0029] Where max(T(i)) represents the maximum value in the echo time series T of the sonar scanning point i, specifically the last bit of the sequence, reflecting the arrival time of the last residual reflection;

[0030] The standard deviation of the amplitude of sonar scanning point i, σA(i), is obtained by the following calculation formula:

[0031] ;

[0032] Where, std represents the standard deviation function;

[0033] The waveform jump density η(i) per unit time at the sonar scanning point i is calculated by Obtain the calculation formula.

[0034] Preferably, said S3 includes S31 and S32;

[0035] S31, mapping the multipath disturbance feature set F into a three-dimensional feature space, performing cluster analysis using a density clustering algorithm, and identifying dense clusters and discrete points;

[0036] Among them, dense clusters and discrete points are divided into clusters by using the DBSCAN density clustering algorithm to divide the multipath disturbance highly correlated characteristic parameter density of each sonar scanning point in the multipath disturbance feature set F. Sonar scanning points that are close to each other are clustered into a cluster, while discrete sonar scanning points are marked as noise. The sonar scanning points marked as noise are simultaneously removed from the three-dimensional feature space and do not participate in the calculation of the disturbance residual.

[0037] S32, calculating the perturbation residual of the sonar scanning point clusters in the three-dimensional feature space after the elimination process, obtaining the perturbation residual Cc(i) of each sonar scanning point i, and comparing it with the preset interference judgment threshold Omap, assigning a multipath label to each sonar scanning point, and obtaining a residual set A and a multipath label set Lmp;

[0038] Among them, the residual set A = {Cc(1), Cc(2), ..., Cc(i) | i∈N};

[0039] The multipath label set Lmp={L(1), L(2), …, L(i)|i∈N}.

[0040] Preferably, the calculation of the disturbance residual specifically calculates the Euclidean distance from each sonar scanning point to the center of the sonar scanning point cluster, and the Euclidean distance is marked as the disturbance residual Cc, which reflects the degree of deviation of each sonar scanning point i from the normal sonar scanning point cluster center;

[0041] Multipath labels are assigned using the following comparison method:

[0042] When the disturbance residual Cc(i) of sonar scanning point i is greater than the interference judgment threshold Omap, the multipath interference point label of sonar scanning point i is assigned, sonar scanning point i is marked as multipath interference, and the multipath label L(i)=1 of sonar scanning point i is generated;

[0043] When the disturbance residual Cc(i) of sonar scanning point i is less than the interference judgment threshold Omap, the multipath interference point label assignment of sonar scanning point i is not performed, sonar scanning point i is marked as a non-interference point, and the multipath label L(i) = 0 of sonar scanning point i is generated.

[0044] Preferably, the S4 includes S41;

[0045] S41, extract the time series T and amplitude series A of each sonar scanning point in the original measurement point acoustic wave data set S of each sonar scanning point in the multipath label set Lmp, recalculate the main path echo time Tmain(i) of the sonar scanning point i, and form the main path time set Hmain={Tmain(1), Tmain(2), ..., Tmain(i)|i∈N};

[0046] The main path echo time Tmain(i) of sonar scanning point i is obtained by the following calculation formula:

[0047] ;

[0048] Where argmax(A(i, j)) represents the maximum amplitude sequence of sonar scanning point i at sampling time point j, and the sampling time point j of the acquired amplitude sequence is used as the sampling time point of the echo time series of sonar scanning point i.

[0049] The sampling time point j of the echo time series of the sonar scanning point i is determined to be the sampling time point corresponding to the maximum amplitude, and is marked as the main path echo time Tmain(i) of the sonar scanning point i.

[0050] Preferably, said S5 includes S51 and S52;

[0051] S51, transforming each sonar scanning point in the main path time set Hmain based on the sound wave propagation velocity v, obtaining the depth coordinate Z(i) of the sonar scanning point i, and calculating the height difference sequence △Z and the horizontal distance difference △D between each pair of adjacent sonar scanning points based on all the transformed depth coordinates Z;

[0052] The depth coordinate Z(i) of the sonar scanning point i is obtained by the following calculation formula:

[0053] ;

[0054] Where Tmain(i) represents the main path echo time of the i-th sonar scanning point;

[0055] Among them, the height difference sequence △Z is obtained by the calculation formula △Z=Z(i+1)-Z(i);

[0056] The horizontal distance difference △D is obtained by extracting the interval between the fixed interval (i+1)th sonar scanning point and the i-th sonar scanning point on the preset mapping path.

[0057] Preferably, S52, based on analyzing the linear relationship between the height difference sequence △Z(i) and the horizontal distance difference △D(i) obtained from adjacent sonar scanning points i, a consistency index G for evaluating the landform is constructed to measure the continuity and stability of the landform, and the consistency index G is compared with a preset reconstruction threshold Gthe, and then the reconstruction of the main path is triggered based on the comparison result;

[0058] Consistency index G passed Obtain calculation formula;

[0059] The reconstruction of the main path is judged by the following comparison method:

[0060] When the consistency index G> the reconstruction threshold Gthe, the terrain is determined to be discontinuous, the path is re-evaluated, and the main path reconstruction of the main path time set Hmain is triggered;

[0061] When the consistency index G ≤ reconstruction threshold Gthe, the landform is judged to be continuous and the main path time set Hmain is applied.

[0062] The seabed topography mapping system based on underwater robots includes a scanning data acquisition module, a scanning feature extraction module, a feature mapping marking module, a main path acquisition module, and a decision optimization module;

[0063] The scanning data acquisition module is based on the underwater robot performing sonar scanning at fixed intervals on the preset mapping path, acquiring time series reflection data at each sonar scanning point to form the original measurement point acoustic wave dataset S;

[0064] The scanning feature extraction module extracts features from the original measurement point acoustic wave data set S, obtains the multipath disturbance height-related feature parameters of each sonar scanning point, and forms a multipath disturbance feature set F;

[0065] The feature mapping and labeling module maps the multipath disturbance feature set F into a three-dimensional feature space, performs cluster analysis using a density clustering algorithm, identifies dense clusters and discrete points, obtains the disturbance residual Cc, and compares it with the preset interference judgment threshold Omap. It then assigns a multipath label to each sonar scanning point, obtaining the residual set A and the multipath label set Lmp.

[0066] The main path acquisition module extracts the time series and amplitude series from the original measurement point acoustic wave dataset S of each sonar scanning point in the multi-path label set Lmp, recalculates the main path echo time Tmain, and forms the main path time set Hmain. The decision optimization module converts the main path time set Hmain into depth coordinates Z based on the sound wave propagation velocity v, obtains the height difference sequence △Z and the horizontal distance difference △D, evaluates the consistency index G of the terrain, and judges the triggering of main path reconstruction based on the consistency index G.

[0067] The present invention provides a method and system for mapping seabed topography based on an underwater robot, which has the following beneficial effects:

[0068] (1) Based on the original measurement point acoustic wave data set S, a reliable basic data is provided for the measurement. The multipath disturbance feature set F of each sonar scanning point is extracted, and the multipath interference in the underwater environment is effectively identified and characterized. The data is clustered by density cluster analysis, and dense clusters and discrete points are successfully identified. Then, the disturbance residual Cc is obtained and compared with the preset interference judgment threshold Omap. A multipath label is assigned to each sonar scanning point, and finally a residual set A and a multipath label set Lmp are generated. In the calculation of the main path echo time Tmain, the time series and amplitude series of the sonar scanning point are combined to accurately obtain the main path echo time of each measuring point, effectively judge the continuity of the terrain, and then judge whether to trigger the reconstruction of the main path based on the consistency index G, ensuring the high precision and stability of underwater terrain survey. In particular, it provides effective technical support in the elimination of multipath interference, the reconstruction judgment of the main path, and the final accurate reflection of the terrain.

[0069] (2) Complete echo data is acquired through high-resolution directional scanning and integrated to form the original measurement point acoustic wave data set S, which lays a solid foundation for accurate data processing. In particular, the calculation of the waveform oscillation coefficient γ(i) can capture the waveform mutation caused by multipath interference in the acoustic signal, thereby effectively identifying potential multipath interference areas. Through this series of precise data acquisition and processing, a comprehensive and accurate description of the seabed topography is provided. At the same time, the errors that may occur in traditional methods are effectively avoided, the accuracy of the measurement results is improved, and the efficiency and high quality of underwater mapping are ensured. This method not only improves the integrity of the data and the ability to capture details, but also optimizes the detection and processing of multipath interference, reduces the errors caused by signal noise, and thus ensures the reliability and stability of the final results.

[0070] (3) The density clustering algorithm DBSCAN effectively distinguishes core points, boundary points, and noise points. Leveraging its efficient clustering capabilities, it can identify potential multipath interference in complex underwater environments and eliminate noise points. This measure improves the accuracy of subsequent perturbation residual calculations. By calculating perturbation residuals for the eliminated sonar scan point clusters and further combining them with the preset interference determination threshold Omap, each sonar scan point is assigned an accurate multipath label, ensuring the precise distinction between multipath interference points and non-interference points. BRIEF DESCRIPTION OF THE DRAWINGS

[0071] Figure 1 This is a schematic diagram of the steps of the seabed topography and landform mapping method based on an underwater robot according to the present invention;

[0072] Figure 2 This is a schematic diagram of a block diagram of a seabed topography and landform mapping system based on an underwater robot according to the present invention;

[0073] Figure 3 Schematic diagram of the relationship between depth coordinate (Z(i)) and sonar scanning point. DETAILED DESCRIPTION

[0074] The following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, rather than all the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative work are within the scope of protection of the present invention.

[0075] Example 1

[0076] The present invention provides a method for mapping seabed topography based on underwater robots. Figure 1 , including the following steps:

[0077] S1, based on the underwater robot performing sonar scanning at fixed intervals on the preset mapping path, obtaining time series reflection data at each sonar scanning point to form the original measurement point acoustic wave dataset S;

[0078] S2. Extract features from the original measurement point acoustic wave dataset S to obtain highly correlated feature parameters of multipath disturbance at each sonar scanning point to form a multipath disturbance feature set F.

[0079] S3. Map the multipath disturbance feature set F into a three-dimensional feature space and perform cluster analysis using a density clustering algorithm to identify dense clusters and discrete points. Obtain the disturbance residual Cc and compare it with the preset interference judgment threshold Omap. Assign a multipath label to each sonar scanning point to obtain the residual set A and the multipath label set Lmp.

[0080] S4, extract the time series and amplitude series from the original measurement point acoustic wave data set S of each sonar scanning point in the multipath label set Lmp, recalculate the main path echo time Tmain, and form the main path time set Hmain;

[0081] S5. Based on the sound wave propagation speed v, the main path time set Hmain is converted into the depth coordinate Z, and the height difference sequence △Z and the horizontal distance difference △D are obtained to evaluate the consistency index G of the landform. Based on the consistency index G, the reconstruction of the main path is triggered.

[0082] In this embodiment, based on the underwater robot performing sonar scanning at fixed intervals along a preset mapping path, the system can obtain time-series reflection data at each sonar scanning point, forming an original measurement point acoustic wave data set S, providing reliable basic data for measurement. The multipath disturbance feature set F of each sonar scanning point is extracted, effectively identifying and characterizing the multipath interference in the underwater environment. The data is clustered through density clustering analysis, successfully identifying dense clusters and discrete points, and then obtaining the disturbance residual Cc. This is compared with the preset interference judgment threshold Omap, and a multipath label is assigned to each sonar scanning point, ultimately generating a residual set A and a multipath label set Lmp. In the calculation of the main path echo time Tmain, the time series and amplitude series of the sonar scanning points are combined to accurately obtain the main path echo time of each measurement point. The main path time set Hmain is converted into a depth coordinate Z using the sound wave propagation velocity v, and the height difference sequence ΔZ and the horizontal distance difference ΔD are calculated. By evaluating the landform consistency index G, effectively determining landform continuity and then, based on this consistency index, determining whether to trigger the reconstruction of the main path, this ensures high precision and stability in underwater topographic surveys. This provides effective technical support for eliminating multipath interference, reconstructing the main path, and ultimately accurately reflecting the terrain. This approach not only improves the accuracy and efficiency of underwater measurements, but also resolves measurement errors caused by multipath interference, ensuring the reliability and integrity of seabed topographic data.

[0083] Example 2

[0084] This embodiment is explained in Example 1, please refer to Figure 1 , specifically: the S1 includes S11 and S12;

[0085] S11, based on the underwater robot performing sonar scanning at fixed intervals on a preset mapping path, sequentially determining sonar scanning points i, and obtaining a sonar scanning point set P = {1, 2, ..., i|i∈N}, where N represents the total number of sonar scanning points;

[0086] S12. Perform high-resolution directional scanning on each sonar scanning point i based on the sonar scanning point set P to obtain complete time series echo data, which constitute the original acoustic wave data S(i) of each sonar scanning point i. Integrate the original acoustic wave data S of all sonar scanning points in the sonar scanning point set P to obtain the original measurement point acoustic wave data set S={S(1), S(2), …, S(i)|i∈N}.

[0087] Among them, the time series echo data includes the echo time series T(i) of sonar scanning point i, the amplitude series A(i) of sonar scanning point i, the effective signal duration D(i) of sonar scanning point i, the main echo peak time Tpeak and the waveform oscillation coefficient γ(i) of sonar scanning point i;

[0088] The original sound wave data of sonar scanning point i S(i) = {T(i), A(i), D(i), Tpeak(i), γ(i)};

[0089] The echo time series T(i) of sonar scanning point i specifically represents the time axis of the entire echo receiving window, constituting the horizontal coordinate of the time domain. The echo time series T(i) of sonar scanning point i = {t1, t2, ..., tj|j∈k}, where tj represents the echo time series of the jth sampling time point, and k represents the total number of time points;

[0090] The amplitude sequence A(i) of sonar scanning point i specifically represents the amplitude of the acoustic echo that matches the echo time sequence T(i) of sonar scanning point i. The amplitude sequence A(i) of sonar scanning point i = {a1, a2, …, aj|j∈k}, where aj represents the amplitude sequence at the jth sampling time point.

[0091] The effective signal duration D(i) of the sonar scanning point i specifically represents the time duration of the echo signal. The effective signal duration D(i) of the sonar scanning point i is obtained by the tk-t1 calculation formula;

[0092] The main echo peak time Tpeak specifically represents the time point of the maximum amplitude of the main path reflection signal, which is used for multipath reflection comparison and is obtained by extracting the maximum value in the amplitude sequence A(i) of the sonar scanning point i;

[0093] The waveform oscillation coefficient γ(i) at sonar scanning point i specifically indicates the degree of waveform mutation. The more unstable the waveform and the more jumps there are, the more likely multipath interference is. The waveform oscillation coefficient γ(i) at sonar scanning point i is obtained by the following calculation formula:

[0094] ;

[0095] Where A(i, j+1) and A(i, j) represent the amplitude sequence of the j+1th sampling point and the amplitude sequence of the jth sampling point of sonar scanning point i, respectively. δ represents the preset amplitude mutation judgment threshold, which is usually 0.1*max(A(i)). When A(i, j+1)-A(i, j)>amplitude mutation judgment threshold δ, the value is returned to 1, otherwise it returns 0.

[0096] In this embodiment, by setting a predetermined mapping path and performing sonar scanning at fixed intervals, a set of sonar scanning points P is sequentially acquired, ensuring a uniform distribution of measurement points and providing a basis for subsequent detailed data analysis. In step S12, the echo time series T(i), amplitude series A(i), effective signal duration D(i), main echo peak time Tpeak, and waveform oscillation coefficient γ(i) of each sonar scanning point are combined to acquire complete echo data through high-resolution directional scanning and integrate them to form the original measurement point acoustic wave data set S, laying a solid foundation for accurate data processing. In particular, the calculation of the waveform oscillation coefficient γ(i) can capture waveform mutations in the acoustic wave signal caused by multipath interference, thereby effectively identifying potential multipath interference areas. Through this series of precise data acquisition and processing, a comprehensive and accurate description of the seabed topography is provided, while effectively avoiding errors that may occur in traditional methods, improving the accuracy of measurement results, and ensuring the efficiency and high quality of underwater mapping. This approach not only improves data integrity and detail capture capabilities, but also optimizes the detection and processing of multipath interference, reduces errors caused by signal noise, and thus ensures the reliability and stability of the final results.

[0097] Example 3

[0098] This embodiment is explained in Example 2, please refer to Figure 1 , specifically: said S2 includes S21;

[0099] S21. Extract features from the original measurement point acoustic wave data set S to obtain the multipath disturbance height-correlated feature parameters of each sonar scanning point i in the original measurement point acoustic wave data set S = {S (1), S (2), ..., S (i) | i∈N}, and form a multipath disturbance feature set F = {F (1), F (2), ..., F (i) | i∈N};

[0100] The multipath disturbance highly correlated characteristic parameters include the maximum echo delay time offset δT(i) of the sonar scanning point i, the amplitude standard deviation σA(i) of the sonar scanning point i, and the waveform jump density η(i) per unit time of the sonar scanning point i;

[0101] The maximum echo delay time offset δT(i) of sonar scanning point i is obtained by the following calculation formula:

[0102] ;

[0103] Where max(T(i)) represents the maximum value in the echo time series T of the sonar scanning point i, specifically the last bit of the sequence, reflecting the arrival time of the last residual reflection;

[0104] The standard deviation of the amplitude of sonar scanning point i, σA(i), is obtained by the following calculation formula:

[0105] ;

[0106] Where, std represents the standard deviation function;

[0107] The waveform jump density η(i) per unit time at the sonar scanning point i is calculated by Obtain the calculation formula.

[0108] Said S3 includes S31 and S32;

[0109] S31, mapping the multipath disturbance feature set F into a three-dimensional feature space, performing cluster analysis using a density clustering algorithm, and identifying dense clusters and discrete points;

[0110] Among them, dense clusters and discrete points are divided into clusters by using the DBSCAN density clustering algorithm to divide the multipath disturbance highly correlated characteristic parameter density of each sonar scanning point in the multipath disturbance feature set F. Sonar scanning points that are close to each other are clustered into a cluster, while discrete sonar scanning points are marked as noise. The sonar scanning points marked as noise are simultaneously removed from the three-dimensional feature space and do not participate in the calculation of the disturbance residual.

[0111] The DBSCAN density clustering algorithm distinguishes core points, boundary points, and noise points based on the neighborhood radius bj and the minimum number of points minPts;

[0112] Core point If a sonar scan point has at least the minimum number of points minPtsd within the radius bj, the sonar scan point is marked as a core point;

[0113] The number of sonar scan points in the neighborhood of the boundary point is less than the minimum number of points minPtsd, but is located in the neighborhood of the core point;

[0114] Noise points are sonar scan points that do not belong to the neighborhood of any core point, and the DBSCAN density clustering algorithm will mark them as noise;

[0115] S32, calculating the perturbation residual of the sonar scanning point clusters in the three-dimensional feature space after the elimination process, obtaining the perturbation residual Cc(i) of each sonar scanning point i, and comparing it with the preset interference judgment threshold Omap, assigning a multipath label to each sonar scanning point, and obtaining a residual set A and a multipath label set Lmp;

[0116] Among them, the residual set A = {Cc(1), Cc(2), ..., Cc(i) | i∈N};

[0117] The multipath label set Lmp={L(1), L(2), …, L(i)|i∈N}.

[0118] The calculation of the perturbation residual specifically calculates the Euclidean distance from each sonar scanning point to the center of the sonar scanning point cluster. The Euclidean distance is marked as the perturbation residual Cc, which reflects the degree of deviation of each sonar scanning point i from the normal sonar scanning point cluster center.

[0119] Multipath labels are assigned using the following comparison method:

[0120] When the disturbance residual Cc(i) of sonar scanning point i is greater than the interference judgment threshold Omap, the multipath interference point label of sonar scanning point i is assigned, sonar scanning point i is marked as multipath interference, and the multipath label L(i)=1 of sonar scanning point i is generated;

[0121] When the disturbance residual Cc(i) of sonar scanning point i is less than the interference judgment threshold Omap, the multipath interference point label assignment of sonar scanning point i is not performed, sonar scanning point i is marked as a non-interference point, and the multipath label L(i) = 0 of sonar scanning point i is generated.

[0122] In this embodiment, by extracting features from the original acoustic wave dataset S, highly correlated characteristic parameters of multipath interference are extracted for each sonar scan point i, including the maximum echo delay time offset δT(i), the amplitude standard deviation σA(i), and the waveform jump density η(i), forming a multipath interference feature set F. This process accurately captures the waveform characteristics of each measuring point, effectively identifying and characterizing multipath interference in underwater environments. In particular, the waveform jump density η(i) is used to measure signal variations, thereby more accurately detecting interference sources. The density clustering algorithm DBSCAN effectively distinguishes core points, boundary points, and noise points. Leveraging its efficient clustering capabilities, it can identify potential multipath interference in complex underwater environments and eliminate noise points, improving the accuracy of subsequent disturbance residual calculations. By performing disturbance residual calculations on the eliminated sonar scan point clusters and combining them with a preset interference determination threshold Omap, each sonar scan point is assigned an accurate multipath label, ensuring precise distinction between multipath interference points and non-interference points. Ultimately, the entire process ensures the accuracy of the measurement results and maintains high stability and credibility under the influence of multipath interference, providing effective path reconstruction and interference elimination methods, thereby improving the overall quality and efficiency of seabed topography mapping.

[0123] Example 4

[0124] This embodiment is explained in Example 3, please refer to Figure 1 Specifically: the S4 includes S41;

[0125] S41, extract the time series T and amplitude series A of each sonar scanning point in the original measurement point acoustic wave data set S of each sonar scanning point in the multipath label set Lmp, recalculate the main path echo time Tmain(i) of the sonar scanning point i, and form the main path time set Hmain={Tmain(1), Tmain(2), ..., Tmain(i)|i∈N};

[0126] The main path echo time Tmain(i) of sonar scanning point i is obtained by the following calculation formula:

[0127] ;

[0128] Where argmax(A(i, j)) represents the maximum amplitude sequence of sonar scanning point i at sampling time point j, and the sampling time point j of the acquired amplitude sequence is used as the sampling time point of the echo time series of sonar scanning point i.

[0129] The sampling time point j of the echo time series of the sonar scanning point i is determined to be the sampling time point corresponding to the maximum amplitude, and is marked as the main path echo time Tmain(i) of the sonar scanning point i.

[0130] Said S5 includes S51 and S52;

[0131] S51, transforming each sonar scanning point in the main path time set Hmain based on the sound wave propagation velocity v, obtaining the depth coordinate Z(i) of the sonar scanning point i, and calculating the height difference sequence △Z and the horizontal distance difference △D between each pair of adjacent sonar scanning points based on all the transformed depth coordinates Z;

[0132] The depth coordinate Z(i) of the sonar scanning point i is obtained by the following calculation formula:

[0133] ;

[0134] Where Tmain(i) represents the main path echo time of the i-th sonar scanning point;

[0135] Among them, the height difference sequence △Z is obtained by the calculation formula △Z=Z(i+1)-Z(i);

[0136] The horizontal distance difference △D is obtained by extracting the interval between the fixed interval (i+1)th sonar scanning point and the i-th sonar scanning point on the preset mapping path.

[0137] S52. Based on the linear relationship between the height difference sequence △Z(i) and the horizontal distance difference △D(i) obtained from adjacent sonar scanning points i, a consistency index G for evaluating the landform is constructed to measure the continuity and stability of the landform. The consistency index G is compared with a preset reconstruction threshold Gthe, and the reconstruction of the main path is triggered based on the comparison result.

[0138] Consistency index G passed Obtain calculation formula;

[0139] The reconstruction of the main path is judged by the following comparison method:

[0140] When the consistency index G> the reconstruction threshold Gthe, the terrain is determined to be discontinuous, the path is re-evaluated, and the main path reconstruction of the main path time set Hmain is triggered;

[0141] When the consistency index G ≤ reconstruction threshold Gthe, the landform is judged to be continuous and the main path time set Hmain is applied.

[0142] In this embodiment, the main path echo time Tmain(i) is recalculated by extracting the time series T and amplitude series A from the original measurement point dataset S for each sonar scan point in the multipath tag set Lmp, and a main path time set Hmain is generated. This calculation process ensures the accuracy of the main path echo time for each sonar scan point by extracting the maximum value of the amplitude series, providing precise time data for subsequent depth coordinate conversion and path analysis. Combined with the acoustic wave propagation velocity v, the depth coordinate Z(i) of each sonar scan point is calculated based on the main path time set Hmain. Furthermore, the height difference series ΔZ and horizontal distance difference ΔD between adjacent sonar scan points are calculated, providing important data for terrain consistency assessment. By analyzing the linear relationship between the height difference and horizontal distance difference between adjacent sonar scan points, a terrain consistency index G is constructed, which effectively measures the continuity and stability of the terrain. Comparison with the preset reconstruction threshold Gthresh determines whether to trigger main path reconstruction. The greatest advantage of this method is that it can dynamically judge changes in underwater terrain, ensure the timing and accuracy of path reconstruction, and ensure that the final generated main path time set Hmain has extremely high terrain consistency and reliability, effectively avoiding path errors caused by irregular terrain and significantly improving measurement accuracy and stability.

[0143] Example 5

[0144] For the seabed topography and landform mapping system based on underwater robots, please refer to Figure 2 ,Specifically: including scanning data acquisition module, scanning feature extraction module, feature mapping marking module, main path acquisition module and decision optimization module;

[0145] The scanning data acquisition module is based on the underwater robot performing sonar scanning at fixed intervals on the preset mapping path, acquiring time series reflection data at each sonar scanning point to form the original measurement point acoustic wave dataset S;

[0146] The scanning feature extraction module extracts features from the original measurement point acoustic wave data set S, obtains the multipath disturbance height-related feature parameters of each sonar scanning point, and forms a multipath disturbance feature set F;

[0147] The feature mapping and labeling module maps the multipath disturbance feature set F into a three-dimensional feature space, performs cluster analysis using a density clustering algorithm, identifies dense clusters and discrete points, obtains the disturbance residual Cc, and compares it with the preset interference judgment threshold Omap. It then assigns a multipath label to each sonar scanning point, obtaining the residual set A and the multipath label set Lmp.

[0148] The main path acquisition module extracts the time series and amplitude series from the original measurement point acoustic wave dataset S of each sonar scanning point in the multi-path label set Lmp, recalculates the main path echo time Tmain, and forms the main path time set Hmain. The decision optimization module converts the main path time set Hmain into depth coordinates Z based on the sound wave propagation velocity v, obtains the height difference sequence △Z and the horizontal distance difference △D, evaluates the consistency index G of the terrain, and judges the triggering of main path reconstruction based on the consistency index G.

[0149] A specific example of calculating the depth coordinate Z(i) from the main path echo time Tmain(i):

[0150] Assume that the underwater robot performs a sonar scan on a preset path, and the data is as follows:

[0151] The speed of sound wave propagation v=1500;

[0152] The echo time data of the echo time series T(i) for the i-th sonar scanning point are as follows:

[0153] Sonar scan point i:

[0154] i=1:1; echo time sequence T(i): 0.3; amplitude sequence A(i): 0.8;

[0155] i=2:2; echo time sequence T(i):0.32; amplitude sequence A(i):0.75;

[0156] i=3:3; echo time series T(i): 0.31; amplitude series A(i): 0.85;

[0157] i=4:4; echo time series T(i): 0.29; amplitude series A(i): 0.82;

[0158] Calculate the main path echo time Tmain(i):

[0159] Sonar scan point i:

[0160] i=1: 1; main path echo time Tmain(i): 0.31;

[0161] i=2: 2; main path echo time Tmain(i): 0.32;

[0162] i=3:3; main path echo time Tmain(i): 0.31;

[0163] i=4:4; main path echo time Tmain(i): 0.29;

[0164] Calculate the depth coordinate Z(i):

[0165] Sonar scan point i:

[0166] i=1:1; main path echo time Tmain(i): 0.31; depth coordinate Z(i): 232.5;

[0167] i=2:2; main path echo time Tmain(i): 0.32; depth coordinate Z(i): 240;

[0168] i=3:3; main path echo time Tmain(i): 0.31; depth coordinate Z(i): 232.5;

[0169] i=4:4; main path echo time Tmain(i): 0.29; depth coordinate Z(i): 217.5;

[0170] Calculate the height difference sequence ΔZ, horizontal distance difference ΔD=5:

[0171] Sonar scan point i:

[0172] i=1-2: 1→2; height difference ΔZi: 7.5;

[0173] i=2-3: 2→3; height difference ΔZi: -7.5;

[0174] i=3-4: 3→4; height difference ΔZi: 15.0;

[0175] The calculated geomorphic consistency index G=[7.5+(-7.5)+15.0] / [5+5+5]=1.0.

[0176] While embodiments of the present invention have been shown and described, it will be appreciated by those skilled in the art that various changes, modifications, substitutions, and variations may be made to these embodiments without departing from the principles and spirit of the invention, and that the scope of the invention is defined by the appended claims and their equivalents.

Claims

1. A method for mapping seabed topography based on an underwater robot, characterized by: The following steps are involved: S1, based on the underwater robot performing sonar scanning at fixed intervals on the preset mapping path, obtaining time series reflection data at each sonar scanning point to form the original measurement point acoustic wave dataset S; S2. Extract features from the original measurement point acoustic wave dataset S to obtain highly correlated feature parameters of multipath disturbance at each sonar scanning point to form a multipath disturbance feature set F. S3. Map the multipath disturbance feature set F into a three-dimensional feature space and perform cluster analysis using a density clustering algorithm to identify dense clusters and discrete points. Obtain the disturbance residual Cc and compare it with the preset interference judgment threshold Omap. Assign a multipath label to each sonar scanning point to obtain the residual set A and the multipath label set Lmp. S4, extract the time series and amplitude series from the original measurement point acoustic wave data set S of each sonar scanning point in the multipath label set Lmp, recalculate the main path echo time Tmain, and form the main path time set Hmain; S5. Convert the main path time set Hmain into a depth coordinate Z based on the acoustic wave propagation velocity v, obtain the height difference sequence △Z and the horizontal distance difference △D, evaluate the consistency index G of the landform, and determine the triggering of the main path reconstruction based on the consistency index G; Said S5 includes S51 and S52; S51, transforming each sonar scanning point in the main path time set Hmain based on the sound wave propagation velocity v, obtaining the depth coordinate Z(i) of the sonar scanning point i, and calculating the height difference sequence △Z and the horizontal distance difference △D between each pair of adjacent sonar scanning points based on all the transformed depth coordinates Z; The depth coordinate Z(i) of the sonar scanning point i is obtained by the following calculation formula: ； Where Tmain(i) represents the main path echo time of the i-th sonar scanning point; Among them, the height difference sequence △Z is obtained by the calculation formula △Z=Z(i+1)-Z(i); The horizontal distance difference △D is obtained by extracting the interval between the fixed interval i+1th sonar scanning point and the ith sonar scanning point on the preset mapping path; S52. Based on the linear relationship between the height difference sequence △Z(i) and the horizontal distance difference △D(i) obtained from adjacent sonar scanning points i, a consistency index G for evaluating the landform is constructed to measure the continuity and stability of the landform. The consistency index G is compared with a preset reconstruction threshold Gthe, and the reconstruction of the main path is triggered based on the comparison result. Consistency index G passed Obtain calculation formula; The reconstruction of the main path is judged by the following comparison method: When the consistency index G> the reconstruction threshold Gthe, the terrain is determined to be discontinuous, the path is re-evaluated, and the main path reconstruction of the main path time set Hmain is triggered; When the consistency index G ≤ reconstruction threshold Gthe, the landform is judged to be continuous and the main path time set Hmain is applied.

2. The method for surveying and mapping seabed topography based on an underwater robot according to claim 1, characterized in that: Said S1 includes S11 and S12; S11, based on the underwater robot performing sonar scanning at fixed intervals on a preset mapping path, sequentially determining sonar scanning points i, and obtaining a sonar scanning point set P = {1, 2, ..., i|i∈N}, where N represents the total number of sonar scanning points; S12. Perform high-resolution directional scanning on each sonar scanning point i based on the sonar scanning point set P to obtain complete time series echo data, which constitute the original acoustic wave data S(i) of each sonar scanning point i. Integrate the original acoustic wave data S of all sonar scanning points in the sonar scanning point set P to obtain the original measurement point acoustic wave data set S={S(1), S(2), …, S(i)|i∈N}.

3. The method for surveying and mapping seabed topography based on an underwater robot according to claim 2, characterized in that: in, The time series echo data includes the echo time series T(i) of sonar scanning point i, the amplitude series A(i) of sonar scanning point i, the effective signal duration D(i) of sonar scanning point i, the main echo peak time Tpeak and the waveform oscillation coefficient γ(i) of sonar scanning point i; The original sound wave data of sonar scanning point i S(i) = {T(i), A(i), D(i), Tpeak(i), γ(i)}; The echo time series T(i) of sonar scanning point i specifically represents the time axis of the entire echo receiving window, constituting the horizontal coordinate of the time domain. The echo time series T(i) of sonar scanning point i = {t1, t2, ..., tj|j∈k}, where tj represents the echo time series of the jth sampling time point, and k represents the total number of time points; The amplitude sequence A(i) of sonar scanning point i specifically represents the amplitude of the acoustic echo that matches the echo time sequence T(i) of sonar scanning point i. The amplitude sequence A(i) of sonar scanning point i = {a1, a2, …, aj|j∈k}, where aj represents the amplitude sequence at the jth sampling time point. The effective signal duration D(i) of the sonar scanning point i specifically represents the time duration of the echo signal. The effective signal duration D(i) of the sonar scanning point i is obtained by the tk-t1 calculation formula; The main echo peak time Tpeak specifically represents the time point of the maximum amplitude of the main path reflection signal, which is used for multipath reflection comparison and is obtained by extracting the maximum value in the amplitude sequence A(i) of the sonar scanning point i; The waveform oscillation coefficient γ(i) of the sonar scanning point i specifically represents the degree of waveform mutation. The waveform oscillation coefficient γ(i) of the sonar scanning point i is obtained by the following calculation formula: ； Where A(i, j+1) and A(i, j) represent the amplitude sequence of the sonar scanning point i at the j+1th sampling point and the amplitude sequence of the jth sampling point, respectively. δ represents the preset amplitude mutation judgment threshold. When A(i, j+1)-A(i, j)>amplitude mutation judgment threshold δ, the value returns 1; otherwise, it returns 0.

4. The method for surveying and mapping seabed topography based on an underwater robot according to claim 3, characterized in that: Said S2 includes S21; S21. Extract features from the original measurement point acoustic wave data set S to obtain the multipath disturbance height-correlated feature parameters of each sonar scanning point i in the original measurement point acoustic wave data set S = {S (1), S (2), ..., S (i) | i∈N}, and form a multipath disturbance feature set F = {F (1), F (2), ..., F (i) | i∈N}; The multipath disturbance highly correlated characteristic parameters include the maximum echo delay time offset δT(i) of the sonar scanning point i, the amplitude standard deviation σA(i) of the sonar scanning point i, and the waveform jump density η(i) per unit time of the sonar scanning point i; The maximum echo delay time offset δT(i) of sonar scanning point i is obtained by the following calculation formula: ； Where max(T(i)) represents the maximum value in the echo time series T of the sonar scanning point i, specifically the last bit of the sequence, reflecting the arrival time of the last residual reflection; The standard deviation of the amplitude of sonar scanning point i, σA(i), is obtained by the following calculation formula: ； Where, std represents the standard deviation function; The waveform jump density η(i) per unit time at the sonar scanning point i is calculated by Obtain the calculation formula.

5. The method for surveying and mapping seabed topography based on an underwater robot according to claim 4, characterized in that: Said S3 includes S31 and S32; S31. Map the multipath disturbance feature set F into the three-dimensional feature space, use the density clustering algorithm to perform cluster analysis, and identify dense clusters and discrete points. Among them, dense clusters and discrete points are divided into clusters by using the DBSCAN density clustering algorithm to divide the multipath disturbance highly correlated characteristic parameter density of each sonar scanning point in the multipath disturbance feature set F. Sonar scanning points that are close to each other are clustered into a cluster, while discrete sonar scanning points are marked as noise. The sonar scanning points marked as noise are simultaneously removed from the three-dimensional feature space and do not participate in the calculation of the disturbance residual. S32, calculating the perturbation residual of the sonar scanning point clusters in the three-dimensional feature space after the elimination process, obtaining the perturbation residual Cc(i) of each sonar scanning point i, and comparing it with the preset interference judgment threshold Omap, assigning a multipath label to each sonar scanning point, and obtaining a residual set A and a multipath label set Lmp; Among them, the residual set A = {Cc(1), Cc(2), ..., Cc(i) | i∈N}; The multipath label set Lmp={L(1), L(2), …, L(i)|i∈N}.

6. The method for surveying and mapping seabed topography based on an underwater robot according to claim 5, characterized in that: in, The calculation of the perturbation residual specifically calculates the Euclidean distance from each sonar scanning point to the center of the sonar scanning point cluster. The Euclidean distance is marked as the perturbation residual Cc, which reflects the degree of deviation of each sonar scanning point i from the normal sonar scanning point cluster center. Multipath labels are assigned using the following comparison method: When the disturbance residual Cc(i) of sonar scanning point i is greater than the interference judgment threshold Omap, the multipath interference point label of sonar scanning point i is assigned, sonar scanning point i is marked as multipath interference, and the multipath label L(i)=1 of sonar scanning point i is generated; When the disturbance residual Cc(i) of sonar scanning point i is less than the interference judgment threshold Omap, the multipath interference point label assignment of sonar scanning point i is not performed, sonar scanning point i is marked as a non-interference point, and the multipath label L(i) = 0 of sonar scanning point i is generated.

7. The method for surveying and mapping seabed topography based on an underwater robot according to claim 6, characterized in that: Said S4 includes S41; S41, extract the time series T and amplitude series A of each sonar scanning point in the original measurement point acoustic wave data set S of each sonar scanning point in the multipath label set Lmp, recalculate the main path echo time Tmain(i) of the sonar scanning point i, and form the main path time set Hmain={Tmain(1), Tmain(2), ..., Tmain(i)|i∈N}; The main path echo time Tmain(i) of sonar scanning point i is obtained by the following calculation formula: ； Where argmax(A(i, j)) represents the maximum amplitude sequence of sonar scanning point i at sampling time point j, and the sampling time point j of the acquired amplitude sequence is used as the sampling time point of the echo time series of sonar scanning point i. The sampling time point j of the echo time series of the sonar scanning point i is determined to be the sampling time point corresponding to the maximum amplitude, and is marked as the main path echo time Tmain(i) of the sonar scanning point i.

8. A seabed topography and landform mapping system based on an underwater robot, applied to the seabed topography and landform mapping method based on an underwater robot according to any one of claims 1 to 7, characterized in that: It includes scanning data acquisition module, scanning feature extraction module, feature mapping marking module, main path acquisition module and decision optimization module; The scanning data acquisition module is based on the underwater robot performing sonar scanning at fixed intervals on the preset mapping path, acquiring time series reflection data at each sonar scanning point to form the original measurement point acoustic wave dataset S; The scanning feature extraction module extracts features from the original measurement point acoustic wave data set S, obtains the multipath disturbance height-related feature parameters of each sonar scanning point, and forms a multipath disturbance feature set F; The feature mapping and labeling module maps the multipath disturbance feature set F into a three-dimensional feature space, performs cluster analysis using a density clustering algorithm, identifies dense clusters and discrete points, obtains the disturbance residual Cc, and compares it with the preset interference judgment threshold Omap. It then assigns a multipath label to each sonar scanning point, obtaining the residual set A and the multipath label set Lmp. The main path acquisition module extracts the time series and amplitude series from the original measurement point acoustic wave dataset S of each sonar scanning point in the multi-path label set Lmp, recalculates the main path echo time Tmain, and forms the main path time set Hmain. The decision optimization module converts the main path time set Hmain into the depth coordinate Z based on the sound wave propagation velocity v, obtains the height difference sequence △Z and the horizontal distance difference △D, evaluates the consistency index G of the terrain, and judges the triggering of the reconstruction of the main path based on the consistency index G.

Images