Underwater unmanned ship, path planning method thereof, electronic equipment, medium and program

Abstract

The embodiment of the invention discloses an underwater unmanned ship and a path planning method thereof, electronic equipment, a medium and a program, and a path planning module of the underwater unmanned ship comprises a target maintenance object three-dimensional point cloud map construction unit, an initial inspection path generation unit and a global smooth inspection path generation unit; the target maintenance object three-dimensional point cloud map construction unit is used for generating a laying path three-dimensional model of a target maintenance object according to the multi-dimensional positioning associated position information of the unmanned underwater vehicle; the initial inspection path generation unit is used for generating an initial inspection path of the target maintenance object according to the laying path three-dimensional model; and the global smooth inspection path generation unit is used for generating a global smooth inspection path of the target maintenance object according to the discrete path points included in the initial inspection path. According to the technical scheme of the embodiment of the invention, the accuracy, real-time performance and reliability of the path planning of the underwater unmanned ship can be improved, and the adaptability to the complex environment in the path planning process is improved.

Description

Technical Field

[0001] The embodiments of the present invention relate to the technical fields of marine engineering intelligent equipment and underwater robots, and in particular to an underwater unmanned boat and its path planning method, electronic equipment, storage medium and program. Background Art

[0002] With the rapid development of the global offshore wind power industry, the variety of underwater facilities is increasing, and the number of years these facilities operate underwater is also increasing. Underwater facilities are exposed to complex river or ocean environments for extended periods of time. These facilities are susceptible to damage, fractures, and surface corrosion, resulting in serious threats to the operational safety and reliability of these facilities and associated surface equipment.

[0003] At present, traditional underwater facility operation and maintenance mainly relies on manual diving inspection or towed remotely operated underwater robots (ROV).

[0004] In the process of realizing the present invention, the inventors found that the operation and maintenance of existing underwater facilities have the following defects: manual diving operations are limited by the physical fitness of divers and underwater operation time, the scope of a single inspection is limited, and it is difficult to implement in high-risk environments (such as strong currents or low visibility, etc.), and the operation and maintenance efficiency is extremely low. Towed remote-controlled underwater robots need to rely on the traction of the mother ship, have poor maneuverability, and have extremely high deployment and maintenance costs. At the same time, the traditional global path planning algorithms used by existing underwater robots, such as Dijkstra's algorithm, do not fully consider the curved laying paths of underwater facilities, dynamic ocean current interference and the complexity of the seabed terrain, resulting in unmanned boats easily deviating from the inspection object and heading towards obstacles, or colliding with obstacles (such as reefs or shipwrecks, etc.). According to statistics, the obstacle avoidance failure rate of existing underwater robot systems in complex sea conditions is as high as 25%, and their path planning and obstacle avoidance capabilities are relatively weak. Summary of the Invention

[0005] The embodiments of the present invention provide an underwater unmanned vehicle and its path planning method, electronic equipment, storage medium and program, which can improve the accuracy, real-timeness and reliability of the underwater unmanned vehicle's path planning capabilities, and improve its adaptability to complex environments during the path planning process.

[0006] According to one aspect of the present invention, an underwater unmanned vehicle is provided, comprising a path planning module, wherein the path planning module comprises a target maintenance object three-dimensional point cloud map construction unit, an initial inspection path generation unit, and a global smooth inspection path generation unit; wherein:

[0007] The target maintenance object three-dimensional point cloud map construction unit is used to generate a three-dimensional model of the laying path of the target maintenance object according to the multi-dimensional positioning correlation position information of the underwater unmanned boat;

[0008] The initial inspection path generating unit is used to generate an initial inspection path for the target inspection object according to the three-dimensional model of the laying path of the target inspection object;

[0009] The global smooth inspection path generation unit is used to generate a global smooth inspection path of the target maintenance object according to the discrete path points included in the initial inspection path;

[0010] The underwater unmanned boat performs maintenance tasks on the target maintenance object according to the global smooth inspection path.

[0011] According to another aspect of the present invention, a path planning method for an underwater unmanned vehicle is provided, comprising:

[0012] Generate a three-dimensional model of the laying path of the target maintenance object based on the multi-dimensional positioning and correlation position information of the underwater unmanned vehicle;

[0013] Generate an initial inspection path for the target maintenance object based on the three-dimensional model of the laying path of the target maintenance object;

[0014] Generating a global smooth inspection path for the target maintenance object according to the discrete path points included in the initial inspection path;

[0015] The underwater unmanned boat performs maintenance tasks on the target maintenance object according to the global smooth inspection path.

[0016] According to another aspect of the present invention, an electronic device is provided, comprising:

[0017] at least one processor; and

[0018] a memory communicatively connected to the at least one processor; wherein,

[0019] The memory stores a computer program that can be executed by the at least one processor, and the computer program is executed by the at least one processor so that the at least one processor can execute the path planning method for the underwater unmanned vehicle described in any embodiment of the present invention.

[0020] According to another aspect of the present invention, a computer-readable storage medium is provided, wherein the computer-readable storage medium stores computer instructions, and the computer instructions are used to enable a processor to implement the path planning method for an underwater unmanned vehicle according to any embodiment of the present invention when executed.

[0021] According to another aspect of the present invention, a computer program product is provided, comprising a computer program, which, when executed by a processor, implements the path planning method for an underwater unmanned vehicle according to any embodiment of the present invention.

[0022] The embodiment of the present invention forms a path planning module for an underwater unmanned vehicle by constructing a three-dimensional point cloud map of a target maintenance object, generating an initial inspection path, and generating a global smooth inspection path. The three-dimensional point cloud map construction unit of the target maintenance object generates a three-dimensional model of the laying path of the target maintenance object according to the multi-dimensional positioning and correlation position information of the underwater unmanned vehicle, and the initial inspection path generation unit generates an initial inspection path of the target maintenance object according to the three-dimensional model of the laying path of the target maintenance object. Then, the global smooth inspection path generation unit generates a global smooth inspection path of the target maintenance object according to the discrete path points included in the initial inspection path, thereby realizing the path planning function for the underwater unmanned vehicle. Accordingly, the underwater unmanned vehicle can perform the maintenance task on the target maintenance object according to the global smooth inspection path finally output by the path planning module. The above technical solution can solve the problems of low accuracy and reliability of the existing underwater unmanned vehicle path planning, and can realize dynamic path planning, thereby improving the accuracy, real-timeness and reliability of the underwater unmanned vehicle path planning, and improving the adaptability to complex environments during the path planning process.

[0023] It should be understood that the content described in this section is not intended to identify the key or important features of the embodiments of the present invention, nor is it intended to limit the scope of the present invention. Other features of the present invention will become readily understood through the following description. BRIEF DESCRIPTION OF THE DRAWINGS

[0024] In order to more clearly illustrate the technical solutions in the embodiments of the present invention, the following briefly introduces the drawings required for use in the description of the embodiments. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without creative work.

[0025] Figure 1 This is a structural diagram of an underwater unmanned boat provided by an embodiment of the present invention;

[0026] Figure 2 1 is a schematic structural diagram of another underwater unmanned vehicle provided by an embodiment of the present invention;

[0027] Figure 3 This is a schematic structural diagram of an unmanned boat system for intelligent inspection and autonomous maintenance of underwater cables in offshore wind farms, provided by an embodiment of the present invention;

[0028] Figure 4 This is a flow chart of a path planning method for an underwater unmanned vehicle provided by an embodiment of the present invention;

[0029] Figure 5 A schematic structural diagram of an electronic device provided by an embodiment of the present invention. DETAILED DESCRIPTION

[0030] In order to enable those skilled in the art to better understand the solutions of the present invention, the technical solutions in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the embodiments described are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative efforts should fall within the scope of protection of the present invention.

[0031] It should be noted that the terms "include" and "have" and any variations thereof are intended to cover non-exclusive inclusions. For example, a process, method, system, product or apparatus that includes a series of steps or units is not necessarily limited to those steps or units explicitly listed, but may include other steps or units that are not explicitly listed or are inherent to these processes, methods, products or apparatus.

[0032] Figure 1 FIG. 1 is a structural diagram of an underwater unmanned boat provided by an embodiment of the present invention. Figure 1 As shown, the underwater unmanned vehicle 10 includes a path planning module 110, which includes a target maintenance object three-dimensional point cloud map construction unit 111, an initial inspection path generation unit 112 and a global smooth inspection path generation unit 113; wherein:

[0033] The target maintenance object three-dimensional point cloud map construction unit 111 is used to generate a three-dimensional model of the laying path of the target maintenance object according to the multi-dimensional positioning correlation position information of the underwater unmanned vehicle;

[0034] The initial inspection path generating unit 112 is used to generate an initial inspection path for the target maintenance object according to the three-dimensional model of the laying path of the target maintenance object;

[0035] The global smooth inspection path generating unit 113 is used to generate a global smooth inspection path of the target maintenance object according to the discrete path points included in the initial inspection path;

[0036] Among them, the underwater unmanned boat 10 performs maintenance tasks on the target maintenance object according to the global smooth inspection path.

[0037] Among them, the path planning module 110 is mainly used to perform real-time dynamic planning of the cruising path of the underwater unmanned boat 10 on the bottom of the water. The target maintenance object three-dimensional point cloud map construction unit 111 is used to generate three-dimensional model information of the laying path of the target maintenance object. The initial inspection path generation unit 112 is used to generate the initial inspection path for the underwater unmanned boat 10 cruising on the bottom of the water, and the initial inspection path is also the initialized path. The global smooth inspection path generation unit 113 is used to generate a global smooth inspection path for the underwater unmanned boat 10 cruising on the bottom of the water, and the global smooth inspection path is also the inspection path finally adopted by the underwater unmanned boat 10. The target maintenance object can be an underwater facility laid on the bottom of the water that needs to be inspected. The embodiment of the present invention does not limit the facility type of the target maintenance object. For example, the target maintenance object can include but is not limited to underwater power transmission cables or oil and gas production system equipment.

[0038] Specifically, the target maintenance object three-dimensional point cloud map construction unit 111 can generate a three-dimensional model of the laying path of the target maintenance object based on the multi-dimensional positioning associated position information of the underwater unmanned boat 10. Among them, the multi-dimensional positioning associated position information can be the position information obtained by the underwater unmanned boat 10 using a variety of different positioning methods. The laying path three-dimensional model is also a three-dimensional model corresponding to the laying path of the target maintenance object. That is, the target maintenance object three-dimensional point cloud map construction unit 111 can obtain in real time the multi-dimensional positioning associated position information obtained by the underwater unmanned boat 10 positioning the target maintenance object through a variety of different positioning methods, and fuse the multi-dimensional positioning associated position information, thereby effectively using the multi-dimensional positioning position information to generate the laying path three-dimensional model of the target maintenance object, which can improve the accuracy of the laying path three-dimensional model of the target maintenance object.

[0039] Accordingly, after the target maintenance object three-dimensional point cloud map construction unit 111 generates a three-dimensional model of the laying path of the target maintenance object, the generated three-dimensional model of the laying path of the target maintenance object can be sent to the initial inspection path generation unit 112. The initial inspection path generation unit 112 can then parse the three-dimensional model of the laying path of the target maintenance object, and can combine various related interference factors on the bottom of the water, such as the gravitational field of the target maintenance object, the ocean current interference field, and the obstacle repulsion field, etc., to add corresponding constraints to the parsing results of the three-dimensional model of the laying path of the target maintenance object, and then generate an initial inspection path for inspecting the target maintenance object based on the three-dimensional model of the laying path of the target maintenance object. Optionally, the initial inspection path usually includes discrete path points.

[0040] After the initial inspection path generation unit 112 generates the initial inspection path for inspecting the target inspection object, the generated initial inspection path can be further sent to the global smooth inspection path generation unit 113. The global smooth inspection path generation unit 113 can further analyze the discrete path points in the initial inspection path to fit the discrete path points in the initial inspection path into a continuously differentiable "snake-shaped" inspection path as a global smooth inspection path. It can be seen that the global smooth inspection path can be a continuously differentiable curved path that can eliminate sharp turns and jitter in the initial inspection path, providing the best inspection path solution for the underwater unmanned vehicle 10.

[0041] Accordingly, after the path planning module 110 of the underwater unmanned vehicle 10 generates a global smooth inspection path through the global smooth inspection path generation unit 113, the underwater unmanned vehicle 10 can perform corresponding maintenance tasks on the target maintenance object according to the global smooth inspection path.

[0042] For example, underwater transmission cables, as core infrastructure connecting wind turbines and power grids, undertake the key task of power transmission. Therefore, underwater transmission cables can be used as target maintenance objects. Accordingly, the target maintenance object three-dimensional point cloud map construction unit in the path planning module of the underwater unmanned vehicle obtains in real time the multi-dimensional positioning associated position information of the underwater unmanned vehicle for positioning the underwater transmission cable, and generates a three-dimensional model of the laying path of the underwater transmission cable based on the multi-dimensional positioning associated position information. Then, the initial inspection path generation unit generates the initial inspection path of the underwater transmission cable based on the three-dimensional model of the laying path of the underwater transmission cable. Since the initial inspection path usually includes sharp turns and jitters, which are not conducive to the maintenance operation of the underwater unmanned vehicle, the discrete path points included in the initial inspection path can be smoothed by the global smooth inspection path generation unit, thereby generating a global smooth inspection path of the target maintenance object, so as to drive the underwater unmanned vehicle to autonomously perform the cable maintenance operation process on the underwater transmission cable according to the global smooth inspection path of the target maintenance object.

[0043] It can be seen that the global smooth inspection path of the target maintenance object generated based on the multi-dimensional positioning and correlation position information of the underwater unmanned vehicle integrates the real-time positioning position information of different dimensions. Its positioning result is more accurate, which can effectively improve the accuracy and real-time performance of path planning. At the same time, the global smooth inspection path eliminates the adverse conditions such as jitter and sharp turns in the initial inspection path, making the final generated path smoother and more reliable, which can ensure the stability of the underwater unmanned vehicle's movement process. In addition, the global smooth inspection path generated based on the real-time positioning position information of different dimensions also optimizes the adaptability of the path planning process to the complex underwater environment, allowing the underwater unmanned vehicle to operate efficiently and stably in any type of underwater environment.

[0044] The embodiment of the present invention forms a path planning module for an underwater unmanned vehicle by constructing a three-dimensional point cloud map of a target maintenance object, generating an initial inspection path, and generating a global smooth inspection path. The three-dimensional point cloud map construction unit of the target maintenance object generates a three-dimensional model of the laying path of the target maintenance object according to the multi-dimensional positioning and correlation position information of the underwater unmanned vehicle, and the initial inspection path generation unit generates an initial inspection path of the target maintenance object according to the three-dimensional model of the laying path of the target maintenance object. Then, the global smooth inspection path generation unit generates a global smooth inspection path of the target maintenance object according to the discrete path points included in the initial inspection path, thereby realizing the path planning function for the underwater unmanned vehicle. Accordingly, the underwater unmanned vehicle can perform the maintenance task on the target maintenance object according to the global smooth inspection path finally output by the path planning module. The above technical solution can solve the problems of low accuracy and reliability of the existing underwater unmanned vehicle path planning, and can realize dynamic path planning, thereby improving the accuracy, real-timeness and reliability of the underwater unmanned vehicle path planning, and improving the adaptability to complex environments during the path planning process.

[0045] Figure 2 This is a structural diagram of another underwater unmanned vehicle provided by an embodiment of the present invention. This embodiment is concretized based on the above embodiment. In this embodiment, the specific implementation method of each unit in the path planning module and various specific optional implementation methods of other functional modules of the underwater unmanned vehicle are given.

[0046] Optionally, the multi-dimensional positioning associated position information includes millimeter-level three-dimensional positioning classification information, three-dimensional point cloud scanning information, and sonar echo information.

[0047] The millimeter-level 3D positioning classification information can be millimeter-level 3D positioning information obtained by locating different types of defects on the target maintenance object. The 3D point cloud scanning information can be positioning information obtained by performing a 3D point cloud scan of the target maintenance object. The sonar echo information can be positioning information obtained by performing a sonar echo detection on the target maintenance object.

[0048] In the embodiment of the present invention, a variety of positioning function modules can be used to perform multimodal positioning detection on various defects of the target maintenance object to obtain millimeter-level three-dimensional positioning classification information. At the same time, the LiDAR (Light Laser Detection and Ranging) equipment ( Figure 2 (not shown) performs a three-dimensional point cloud scan on the underwater environment where the underwater unmanned boat 10 is currently located to obtain three-dimensional point cloud scanning information. Optionally, the sonar detection equipment ( Figure 2(not shown) performs sonar detection on the underwater environment where the underwater unmanned boat 10 is currently located, thereby receiving the echo signal reflected by the underwater object and obtaining the sonar echo information. The path planning module 110 can obtain millimeter-level three-dimensional positioning classification information, three-dimensional point cloud scanning information and sonar echo information, and adopts SLAM (Simultaneous Localization and Mapping, simultaneous positioning and mapping) technology to fuse the obtained millimeter-level three-dimensional positioning classification information, three-dimensional point cloud scanning information and sonar echo information, thereby generating a centimeter-level precision three-dimensional model of the laying path of the target maintenance object. Among them, the core step of SLAM technology is point cloud registration. The underwater unmanned boat 10 will continuously collect multiple frames of LiDAR point clouds during the movement, and the point clouds of different frames have posture deviations due to the movement of the boat body, and the offset needs to be eliminated through registration. Therefore, the multiple frames of point clouds can be aligned first, and then the error function can be defined. Optionally, the point cloud registration error function can be:

[0049]

[0050] Among them, E(R,t) represents the point cloud registration error, R is the rotation matrix, t is the translation vector; x i ,y i are the coordinates of the matching point pairs in the point clouds of adjacent frames; ω i is the weight coefficient calculated based on the point cloud density and curvature, and Nd is the number of point clouds.

[0051] Correspondingly, such as Figure 2 As shown, the underwater unmanned boat 10 may also include a multimodal state detection module 120, which may include an underwater electromagnetic sensor 121, a flexible fiber optic sonar array 122, a multispectral imager 123 and a data fusion unit 124; wherein: the underwater electromagnetic sensor 121 is used to measure the surrounding magnetic field gradient data of the target maintenance object; the flexible fiber optic sonar array 122 is used to measure the damage position positioning data of the target maintenance object; the multispectral imager 123 is used to measure the surface corrosion biological attachment data of the target maintenance object; the data fusion unit 124 is used to perform weighted fusion on the surrounding magnetic field gradient data, damage position positioning data and surface corrosion biological attachment data, and generate millimeter-level three-dimensional positioning classification information according to the weighted fusion result.

[0052] The surrounding magnetic field gradient data may be the gradient data of the magnetic field surrounding the target detection object. The damage location positioning data may be the positioning data of the damage location of the target detection object. The surface corrosion organism attachment data may be the data of the organism attachment situation in the water on the target detection object.

[0053] Existing technologies typically use a single sensor (such as sonar or camera) to detect defects in a target object. This makes it difficult to simultaneously detect multiple defects in the target object, such as current leakage, insulation damage, and micro-corrosion in underwater cables. Single-sensor defect detection also suffers from insufficient accuracy. For example, sonar misses over 30% of minor surface damage (<5mm), while the risk of optical sensors failing increases significantly in turbid waters.

[0054] This embodiment of the present invention employs a multimodal state detection module 120 to perform multi-dimensional defect detection on target inspection objects. Specifically, the multimodal state detection module 120 can employ multiple detection modules to detect various defects in the target inspection object in real time. For example, when an underwater cable is used as the inspection target, the multimodal state detection module 120 can detect various defect types in the underwater cable in real time, including current leakage, insulation damage, mechanical damage, and surface corrosion defects.

[0055] Specifically, the multimodal state detection module 120 may include an underwater electromagnetic sensor 121 to measure the surrounding magnetic field gradient data of the target inspection object through the underwater electromagnetic sensor 121, perform electromagnetic gradient field analysis on the target inspection object, and thus detect current leakage and insulation damage. The multimodal state detection module 120 may also include a flexible fiber optic sonar array 122 to measure damage location positioning data on the target inspection object through the flexible fiber optic sonar array 122, perform sonar beam analysis, and locate mechanical damage to the target inspection object. Optionally, the flexible fiber optic sonar array 122 may be composed of M×N basic sensing unit array elements, each of which may be composed of a fiber optic sonar sensor capable of independently receiving acoustic wave reflection signals. The multimodal state detection module 120 may also include a multispectral imager 123 to identify surface corrosion and biological adhesion of the target inspection object through multispectral imaging, thereby measuring surface corrosion and biological adhesion data in the form of spectral characteristics. Accordingly, the multimodal state detection module 120 can perform a weighted fusion of the surrounding magnetic field gradient data, damage location data, and surface corrosion biofilm attachment data through its internal data fusion unit 124, and generate millimeter-level three-dimensional positioning and classification information based on the weighted fusion results. Thus, the multimodal state detection module 120 combines electromagnetic gradient field analysis, sonar beamforming, and multispectral imaging to achieve millimeter-level positioning and classification of defects in target inspection objects, with a positioning error typically less than 5 cm and a defect false alarm rate of less than 1%.

[0056] In an optional embodiment of the present invention, the underwater electromagnetic sensor 121 may also be used to measure the surrounding magnetic field gradient data of the target maintenance object based on the following formula:

[0057]

[0058] in, represents the surrounding magnetic field gradient, B x represents the component of the magnetic field in the x direction of the Cartesian coordinate system, B y represents the component of the magnetic field in the y direction of the Cartesian coordinate system, B z represents the component of the magnetic field in the z direction of the Cartesian coordinate system. (B th When the magnetic field gradient threshold is set, it can be determined that the target maintenance object has insulation defects, and the coordinates of the abnormal area of ​​the target maintenance object (x d ,y d ,z d The coordinates of the abnormal area (x d ,y d ,z d ) can be used as the input of the dynamic potential field model and input into the initial inspection path generation unit 112, so as to guide the underwater unmanned vehicle 10 to avoid the defective area when cruising along the laying path of the target inspection object during the dynamic path planning process. At the same time, the coordinates of the abnormal area of ​​the target inspection object located by the underwater electromagnetic sensor 121 (x d ,y d ,z d ) can also be used as the operation basis of the bionic manipulator of the underwater unmanned boat 10 to drive the bionic manipulator of the underwater unmanned boat 10 to move according to the coordinates of the abnormal area (x d ,y d ,z d ) locates defective points, such as the location of insulation damage, so as to perform laser cleaning or conductive tape application operations to repair abnormal locations.

[0059] The flexible fiber optic sonar array 122 can also be used to measure the damage location data of the target maintenance object based on the following formula:

[0060]

[0061] Wherein, Δt represents the time difference of sound wave reflection, d represents the damage depth of the target maintenance object, c represents the speed of sound, (x m ,y m ,z m ) represents the coordinate position of the damage point, (x ij ,y ij ,0) represents the coordinates of the (i,j)th element in the flexible fiber optic sonar array, Δt ij Indicates the time difference of array element reception.

[0062] Specifically, the flexible fiber optic sonar array 122 can be based on the formula: The damage depth d of the mechanical damage of the target repair object is located, and the damage position (x m ,y m ,z m ) can be solved by the path planning module 110 through the beam forming algorithm. The specific solution process of the beam forming algorithm of the path planning module 110 is as follows: First, the position of the array element is calibrated, that is, assuming that the coordinates of the (i, j)th array element in the flexible fiber optic sonar array 122 are (x ij ,y ij ,0), the array element coordinate system can be based on the array plane, and then the damage point (x m ,y m ,z m ) to each array element produces a time delay Δt ij , establish the equation system:

[0063]

[0064] in, ω ij is the element weight coefficient. Finally, the least square method can be used to solve the overdetermined equation using multiple array data to obtain (x m ,y m ). It can be seen that each element in the flexible fiber optic sonar array 122 achieves three-dimensional positioning of the damaged position of the target inspection object by measuring the time difference Δt of the arrival of the sound waves and combining it with the beamforming algorithm.

[0065] The multispectral imager 123 may also be used to measure the surface corrosion organism attachment data of the target maintenance object based on the following formula:

[0066]

[0067] Among them, D λ represents the spectral distortion index, S ref (λ) represents the standard spectral reflectance of the uncorroded surface of the target inspection object, S obs (λ) represents the spectral reflectance detected in real time, and λ is the wavelength of the light wave.

[0068] The multispectral imager 123 covers the visible to near infrared band (λ∈[400,1000]nm) and uses the spectral distortion index D λ It can identify the surface corrosion and biological attachment of the target maintenance object. λ Greater than D th (set spectral distortion index threshold), it is determined that the target maintenance object indicates the presence of corrosion or biological attachment.

[0069] The data fusion unit 124 may also be configured to perform weighted fusion of the surrounding magnetic field gradient data, the damage location data, and the surface corrosion organism attachment data based on the following formula:

[0070]

[0071] Among them, C fused Denotes the comprehensive confidence of defects, W k is the dynamic weight, C k represents the independent confidence of the underwater electromagnetic sensor, the flexible fiber optic sonar array, and the multispectral imager, B th represents the magnetic field gradient threshold, z max Indicates the maximum value of the coordinate position of the damage point in the z-axis direction, Δt th Indicates the sound wave reflection time difference threshold, D th Indicates the spectral distortion index threshold. When k = 1, represents the independent confidence of the underwater electromagnetic sensor 121; when k=2, represents the independent confidence of the flexible fiber optic sonar array 122; when k=3, represents the independent confidence of the multispectral imager 123.

[0072] Optionally, the data fusion unit 124 may use a covariance matrix weighting algorithm to fuse the output data of the electromagnetic, acoustic, and optical submodules to generate a comprehensive defect confidence C fused . Accordingly, when C fused ≥C th (preset comprehensive confidence threshold), it is determined that the target maintenance object has defects that need maintenance, and the defect type classification is determined based on the confidence of the dominant positioning module. For example, when C1 is the highest among C1, C2 and C3, the underwater electromagnetic sensor 121 is determined to be the dominant positioning module, and the currently detected defect type is insulation damage. Three-dimensional positioning can be combined with the coordinate output of each positioning module (x d ,y d ,z d ) and (x m ,y m ,z m ) coordinate values ​​are weighted averaged to obtain the final defect position coordinates.

[0073] The multimodal state detection module 120 includes an underwater electromagnetic sensor 121, a flexible fiber optic sonar array 122, and a multispectral imager 123 as parallel sensing units, and a data fusion unit 124 as the collaborative control core. Data fusion unit 124 weightedly fuses multi-source data to output millimeter-level, three-dimensional positioning and classification information for defects in the target repair object. This provides input for the subsequent path planning process and maintenance module of the path planning module 110, achieving millimeter-level, three-dimensional positioning and classification of defects in the target repair object.

[0074] In an optional embodiment of the present invention, the initial inspection path generation unit 112 is specifically used to: generate a comprehensive potential field function based on the three-dimensional model of the laying path of the target inspection object, the gravitational field of the target inspection object, the ocean current interference field and the obstacle repulsion field; drive the underwater unmanned boat to cruise along the laying direction of the target inspection object according to the comprehensive potential field function to obtain the initial inspection path of the target inspection object.

[0075] Among them, the comprehensive potential field function can be a function formed by superimposing multiple potential functions of different forms, which is used to describe the potential energy distribution and interaction in complex systems.

[0076] The initial inspection path generation unit 112 can embed a comprehensive potential field model of the gravitational field, ocean current interference field and obstacle repulsion field of the target inspection object to generate a comprehensive potential field function, and drive the underwater unmanned boat 10 to cruise along the laying direction of the target inspection object according to the comprehensive potential field function, thereby obtaining the initial inspection path of the target inspection object.

[0077] In an optional embodiment of the present invention, the initial inspection path generating unit 112 is specifically configured to generate the comprehensive potential field function based on the following formula:

[0078] U total (P)=U target (P)+U current (P)+U obstacle (P)

[0079] U target (P) = k1·ln(1+||PP target ||)

[0080] U current (P)=k2·||V current ||(1-cosθ)

[0081]

[0082] Among them, U total (P) represents the comprehensive potential field function, U target (P) represents the gravitational field of the target maintenance object, Ucurrent (P) represents the ocean current interference field, which is used to compensate for the influence of the flow velocity on the heading of the underwater unmanned vehicle 10; U obstacle (P) represents the obstacle repulsion field, which is used to prevent the underwater unmanned vehicle 10 from contacting underwater (or seabed) obstacles; P represents the current three-dimensional position coordinates of the underwater unmanned vehicle, which can be determined based on the millimeter-level three-dimensional positioning classification information output by the multimodal state detection module 120; k1 is the gravity gain coefficient, P target represents the coordinates of the nearest point on the center line of the target maintenance object, k2 is the current compensation coefficient, V current is the real-time ocean current vector, θ is the angle between the heading of the underwater unmanned vehicle and the ocean current, k3 is the repulsive gain coefficient, ε is the smoothing factor, P obs,j is the jth obstacle position, and M represents the number of obstacles.

[0083] In an optional embodiment of the present invention, the global smooth inspection path generation unit 113 is specifically configured to generate a global smooth inspection path of the target maintenance object according to the discrete path points included in the initial inspection path based on the following formula:

[0084]

[0085] Wherein, P(s) represents the global smooth inspection path, N i,k (s) represents the B-spline basis function, s is the path parameter, Q i represents the discrete path point, k represents the curve order, and n represents the number of the discrete path points.

[0086] In the embodiment of the present invention, the path planning module 110 generates a preliminary inspection path according to the initial inspection path generation unit 112, and then generates a global smooth inspection path according to the B-spline basis function N i,k (s) The discrete path points Q in the preliminary inspection path i The path planning module 110 can then fit a continuous and differentiable "snake" inspection path P(s) to eliminate sharp turns and jitter, ultimately obtaining a globally smooth inspection path. In this way, the path planning module 110 can ensure that the underwater unmanned vehicle 10 autonomously cruises along the "snake" path of the target inspection object based on the globally smooth inspection path.

[0087] In an optional embodiment of the present invention, Figure 2As shown, the path planning module 110 may also include a speed and attitude adaptive adjustment unit 114, which is used to adjust the speed of the underwater unmanned vehicle based on the global smooth inspection path of the target maintenance object; wherein the underwater unmanned vehicle performs the maintenance task on the target maintenance object based on the adjusted speed and the global smooth inspection path. By adjusting the speed of the underwater unmanned vehicle based on the global smooth inspection path of the target maintenance object, the rationality and accuracy of the underwater unmanned vehicle's speed can be improved, avoiding the omission of maintenance tasks due to excessive speed or the excessive delay of maintenance operations due to excessive speed.

[0088] In an optional embodiment of the present invention, the speed and attitude adaptive adjustment unit 114 is specifically configured to adjust the speed of the underwater unmanned vehicle according to the global smooth inspection path of the target maintenance object based on the following formula:

[0089] ν=ν max ·exp(-β·κ-γ·||V current ||)

[0090] Wherein, ν represents the adjusted speed of the underwater unmanned vehicle, ν max is the maximum speed, β and γ are adjustment coefficients, k represents the curve order, V current is the real-time ocean current vector. Optionally, the parameter k can be determined based on the B-spline basis function used in the global smooth inspection path P(s).

[0091] Therefore, the speed and attitude adaptive adjustment unit 114 can adjust the speed and attitude of the ship according to the curvature κ of the B-spline path and the current intensity || V current Dynamically adjust the speed v and attitude, with the speed adaptive adjustment range reaching 0.5-3m / s, thereby avoiding entanglement in the curved section of the target inspection object and ensuring local motion stability.

[0092] In the above technical solution, the path planning module 110 generates a three-dimensional model of the laying path through the target maintenance object three-dimensional point cloud map construction unit 111. The laying path three-dimensional model can be a three-dimensional model with centimeter-level accuracy, which can help the underwater unmanned boat 10 accurately identify and plan the path in a complex water (or sea) environment, and can provide a spatial reference for path planning. The initial path planning is realized by the initial inspection path generation unit 112, and the initial planned path is optimized according to the position and curvature of the target maintenance object through the global smoothing inspection path generation unit 113 to realize global path generation and smoothing, which can improve the accuracy and stability of the target maintenance object detection. Finally, the speed and attitude are adaptively adjusted by the speed and attitude adaptive adjustment unit 114 to ensure local motion stability.

[0093] In an optional embodiment of the present invention, the initial inspection path generation unit 112 can also be used to: update the comprehensive potential field function according to the curvature penalty factor; the global smooth inspection path generation unit is also used to: optimize the global smooth inspection path of the target maintenance object according to the curvature of the global smooth inspection path of the target maintenance object and the average position of the center line of the target maintenance object in the three-dimensional model of the laying path.

[0094] The curvature penalty factor may be a factor used to refer to the corrected curvature.

[0095] In an embodiment of the present invention, in order to further improve the accuracy of path planning, a curvature penalty factor can be introduced to monitor and optimize the global smooth inspection path in real time to ensure the efficient completion of the maintenance task of the target maintenance object. Specifically, the curvature penalty factor can be introduced according to the initial inspection path generation unit 112 to update the comprehensive potential field function. When the comprehensive potential field function is updated, the initial inspection path and the global smooth inspection path of the target maintenance object are also updated synchronously. At this time, the global smooth inspection path generation unit 113 can be used to optimize the global smooth inspection path of the target maintenance object based on the curvature of the global smooth inspection path of the target maintenance object and the average position of the center line of the target maintenance object in the three-dimensional model of the laying path.

[0096] In an optional embodiment of the present invention, the initial inspection path generating unit 112 is specifically configured to update the comprehensive potential field function according to the curvature penalty factor based on the following formula:

[0097] U total (P)=U target (P)+U current (P)+U obstacle (P)+U curvature (P)

[0098] U curvature (P) = k4·κ(P) 2

[0099] Among them, U curvature (P) represents the curvature penalty term, k4 is the curvature penalty coefficient, and κ(P) is the path curvature of the current three-dimensional position coordinate P of the underwater unmanned vehicle.

[0100] The global smooth inspection path generation unit 113 is specifically configured to optimize the global smooth inspection path of the target maintenance object based on the following formula:

[0101]

[0102] Among them, E path represents the path optimization objective function value, Pi represents the three-dimensional position coordinates of the i-th path point; C represents the average position coordinates of the center line of the target maintenance object in the three-dimensional model of the laying path, which can be updated in real time through SLAM technology. curvature (P i ) represents the global smooth inspection path at point P i The curvature κ(s) at point P can be obtained by using the curvature formula of the B-spline curve path P(s). i The corresponding curvature value is obtained. A represents the weight coefficient for adjusting the path smoothness, N represents the number of path points included in the global smooth inspection path, κ(s) represents the curvature of the global smooth inspection path, P(s) represents the global smooth inspection path, P′(s) represents the first-order derivative of P(S), p”(s) represents the second-order derivative of P(S), E curvature represents the total curvature integral of the path.

[0103] It can be seen that the above-mentioned global smooth inspection path optimization method can automatically generate the optimal inspection path for the target maintenance object based on the real-time input data. The real-time input data can include the obstacle position coordinates P updated by LiDAR and sonar. obs,j , Real-time ocean current vector V current , the current three-dimensional position coordinates P of the underwater unmanned vehicle and the path curvature parameter, i.e., the curve order κ. The above-mentioned global smooth inspection path optimization method can avoid excessive bending by optimizing the curvature of the path, thereby ensuring the stability and reliability of the target inspection object. The above-mentioned global smooth inspection path optimization method is divided into two levels. In the first level, the curvature of the B-spline curve path P(s) is first calculated. Secondly, the optimization goal is to minimize the total curvature integral of the path In the second level, in the integrated potential field U total Add curvature penalty term U in (P) curvature (P) = k4·κ(P) 2 , and then the comprehensive potential field formula is updated to U total (P)=U target (P)+U current (P)+U obstacle (P)+U curvature (P). Guided by the integrated potential field gradient, the unmanned vehicle will gravitate toward a low-curvature path, thus avoiding motion instability or entanglement risks caused by excessive curvature. This global smooth inspection path optimization method not only improves the accuracy of path planning for the target inspection object, but also effectively avoids redundant paths, reducing the risk of damage to the target inspection object, thereby improving the efficiency of maintenance operations for the target inspection object.

[0104] In an optional embodiment of the present invention, the underwater unmanned boat 10 may further include a real-time underwater terrain obstacle avoidance module 130, which may include a multimodal obstacle perception unit 131 and a reinforcement learning obstacle avoidance decision unit 132; wherein: the multimodal obstacle perception unit 131 is used to: obtain the three-dimensional coordinate point cloud data of the obstacle in the underwater area where the target maintenance object is located and the magnetic field anomaly detection data; generate an obstacle probability map based on the three-dimensional coordinate point cloud data of the obstacle and the magnetic field anomaly detection data; the reinforcement learning obstacle avoidance decision unit 132 is used to: generate obstacle avoidance instructions based on the obstacle probability map and the current state data of the underwater unmanned boat in combination with reinforcement learning; and perform real-time dynamic adjustment of the global smooth inspection path of the target maintenance object through the obstacle avoidance instructions.

[0105] The multimodal obstacle sensing unit 131 can combine multimodal data to achieve real-time perception of underwater obstacles. The reinforcement learning obstacle avoidance decision unit 132 can dynamically adjust the global smooth inspection path of the target maintenance object in real time based on the obstacle perception results of the multimodal obstacle sensing unit 131. The obstacle three-dimensional coordinate point cloud data can be the coordinate data obtained by the underwater unmanned vehicle 10 performing three-dimensional point cloud detection of underwater obstacles. The magnetic field anomaly detection data can be the anomaly data detected by the underwater electromagnetic sensor 121 when measuring the surrounding magnetic field gradient data of the target maintenance object. The obstacle probability map can be a map representing the corresponding probability of the existence of an obstacle. The obstacle avoidance instruction can instruct the underwater unmanned vehicle 10 to avoid the obstacle.

[0106] Specifically, the multimodal obstacle perception unit 131 in the underwater terrain real-time obstacle avoidance module 130 can obtain the three-dimensional coordinate point cloud data of the obstacles in the underwater area where the target maintenance object is located and the magnetic field anomaly detection data from the path planning module 110 or the multimodal state detection module 120. Furthermore, the multimodal obstacle perception unit 131 in the underwater terrain real-time obstacle avoidance module 130 can generate an obstacle probability map based on the obtained three-dimensional coordinate point cloud data of the obstacles and the magnetic field anomaly detection data, and send the obstacle probability map to the reinforcement learning obstacle avoidance decision unit 132. The reinforcement learning obstacle avoidance decision unit 132 can generate obstacle avoidance instructions based on the obstacle probability map and the current state data of the underwater unmanned vehicle 10, such as the current position, current speed, and current heading angle, combined with the reinforcement learning method. After the obstacle avoidance instructions are generated, the global smooth inspection path of the target maintenance object can be dynamically adjusted in real time through the obstacle avoidance instructions.

[0107] In an optional embodiment of the present invention, the multimodal obstacle sensing unit 131 is specifically configured to generate an obstacle probability map based on the obstacle three-dimensional coordinate point cloud data and the magnetic field anomaly detection data based on the following formula:

[0108]

[0109] Among them, P obs (x,y) represents the probability that there is an obstacle at the position (x,y), B is the sensitivity coefficient, represents the magnetic field gradient at position (x,y), B th Represents the magnetic field gradient threshold. Greater than B th , it is determined that there is a metal obstacle.

[0110] It can be seen that the multimodal obstacle perception unit 131 can obtain the three-dimensional coordinate point cloud data of underwater obstacles (such as reefs or shipwrecks, etc.) obtained by real-time scanning of the underwater terrain by LiDAR, and can detect magnetic field anomalies such as metal obstacles through the magnetometer positioning module in the multimodal state detection module 120, and then fuse the LiDAR and magnetometer data, that is, construct an obstacle probability map based on the obstacle three-dimensional coordinate point cloud data and the magnetic field anomaly detection data, providing environmental perception input for obstacle avoidance decisions.

[0111] In an optional embodiment of the present invention, the reinforcement learning obstacle avoidance decision unit 132 is specifically configured to generate an obstacle avoidance instruction based on the following formula:

[0112] Q(s,a)←Q(s,a)+α[r+γmax a′ Q(s',a')-Q(s,a)]

[0113]

[0114] Reward(s,a)=w1·Distance obstacle +w2 Distance targete +w3·cos(δ)

[0115] Among them, Q(s,a) represents the expected cumulative reward value of taking action a in state s, s represents the state space, which can be defined as the distance between the LiDAR point cloud feature and the target inspection object. a represents the action space, which can be used to adjust the heading angle. Optionally, the action a∈{turn left 10°, turn right 10°, accelerate, decelerate}. α is the learning rate, r is the immediate reward, calculated by the reward function Reward(s,a), and γ is the discount factor; a' represents the action in the next state, max a′ Q(s',a') represents the maximum expected reward for all actions in the next state s', Distance obstacle Indicates the distance between the current obstacle and the underwater unmanned vehicle, Distance targeterepresents the distance between the target maintenance object and the underwater unmanned vehicle, δ represents the approach angle between the underwater unmanned vehicle and the obstacle, and w1, w2 and w3 are weight coefficients.

[0116] Optionally, the reinforcement learning obstacle avoidance decision unit 132 can update the action value function through the Q-learning algorithm based on the obstacle probability map and the current state data of the underwater unmanned vehicle to generate optimal obstacle avoidance instructions, such as turning left 10°, turning right 10°, accelerating and decelerating, etc.

[0117] It can be seen that the reward function Reward(s,a) mainly consists of three main components: obstacle distance Distance obstacle , which measures the distance between the current obstacle and the underwater unmanned vehicle. The longer the distance, the higher the reward given to avoid collision; the distance between the target inspection object and the underwater unmanned vehicle is Distance targete , which represents the distance between the target inspection object and the unmanned boat. The closer the distance is to the target inspection object, the higher the reward is, indicating that the target inspection object is in the correct inspection area. The angle cos(δ) is used to measure the proximity between the underwater unmanned boat and the obstacle. The smaller the angle, the higher the risk of approaching the obstacle and the lower the reward. The weight coefficients w1, w2, and w3 can control the influence of the above three factors on the final reward. By optimizing the reward function, the reinforcement learning algorithm enables the underwater unmanned boat to adjust the heading angle in real time in the underwater environment, avoid obstacles and optimize path selection. As reinforcement learning continues, the algorithm can continuously adjust its strategy, improve obstacle avoidance capabilities, and ensure that the underwater unmanned boat can safely complete inspection tasks in a dynamic environment.

[0118] The underwater terrain real-time obstacle avoidance module 130 achieves synergy between multimodal perception and reinforcement learning decision-making through a multimodal obstacle perception unit 131 and a reinforcement learning obstacle avoidance decision-making unit 132. Using a reinforcement learning strategy, it adjusts the heading angle to avoid obstacles and ensures the underwater unmanned vehicle can complete its mission safely and efficiently, thus achieving efficient obstacle avoidance in complex underwater environments. An obstacle probability map provides environmental perception input, and a Q-learning algorithm optimizes heading and speed through a dynamic reward mechanism, ultimately ensuring that the unmanned vehicle autonomously cruises along the target inspection object within a safe distance. The dynamically optimized obstacle avoidance strategy, generated based on the Q-learning algorithm and LiDAR point cloud data, has been verified to have an obstacle avoidance success rate exceeding 95% in complex terrain.

[0119] It is understandable that the robotic arm system of existing underwater robots is prone to causing secondary damage to the target maintenance object during the maintenance task due to problems such as rigid structure and low control accuracy (±10mm).

[0120] In order to solve the above problems, in an optional embodiment of the present invention, the underwater unmanned boat 10 can also be configured with a flexible bionic robotic arm collaboration module 140. The flexible bionic robotic arm collaboration module 140 may include a flexible bionic robotic arm 141 and a robotic arm maintenance module 142 at the end of the robotic arm; the robotic arm maintenance module 142 may include a laser cleaning head and a conductive tape attaching device; wherein, the flexible bionic robotic arm collaboration module 140 is used for: when the underwater unmanned boat performs a maintenance task on the target maintenance object, the flexible bionic robotic arm 141 drives the end of the robotic arm to reach the target maintenance position of the global smooth inspection path; according to the defect type of the target maintenance position, the laser cleaning head or the conductive tape attaching device is started to perform maintenance on the target maintenance position.

[0121] The flexible bionic robotic arm 141 may be a multi-degree-of-freedom flexible bionic robotic arm. Optionally, the flexible bionic robotic arm 141 may be composed of seven articulated joints, with the rotation angle θ of each joint being i The inverse kinematics model can be used to solve the problem, and then the end of the manipulator is driven to reach the target position based on the target point coordinates, where the target point coordinates can be the defect position coordinates located by the multimodal state detection module 120 using the multimodal positioning information.

[0122] When the underwater unmanned vehicle 10 performs an inspection task on a target inspection object, the flexible bionic manipulator 141 can be used to drive the end of the manipulator to reach the target inspection position of the global smooth inspection path. Optionally, the inverse kinematics model formula of the flexible bionic manipulator 141 is as follows:

[0123]

[0124] Among them, T target represents the target pose matrix of the end effector of the flexible bionic manipulator 141, T base is the coordinate system of the robot base; θ i is the rotation angle of the i-th joint of the flexible bionic robotic arm 141; S i is the spiral motion generator corresponding to the joint axis. Optionally, the joint angle θ can be solved by gradient descent method. i , so that the end effector of the flexible bionic robotic arm 141 can accurately reach the target maintenance position, and n1 represents the number of joints of the flexible bionic robotic arm 141.

[0125] The flexible bionic robotic arm collaborative module 140 determines a specific repair method based on the defect type at the target repair location. If the defect type at the target repair location is biological adhesion, the laser cleaning head can be activated to repair the target repair location. If the defect type at the target repair location is insulation damage, the conductive tape application device can be activated to repair the target repair location.

[0126] Optionally, the laser cleaning head can be based on the laser energy density E laser The target repair object surface is cleaned of biological attachments. For example, the laser cleaning head can repair the target repair position based on the following formula:

[0127]

[0128] Among them, E laser Represents the laser energy density, P laser is the laser power, r is the spot radius, t exp is the exposure time. laser ≥5J / cm 2 , the cleaning action is triggered.

[0129] Optionally, the conductive tape applying device can adjust the applying pressure based on the insulation damage coordinates and size corresponding to the target repair location through a PID (Proportional Integral Derivative) force control model to ensure that the tape is tightly attached. For example, the conductive tape applying device can inspect the target repair location based on the following formula:

[0130]

[0131] Among them, F adhesive represents the tape application pressure of the conductive tape application device, e(t) = F desired -F actuale is the pressure error, F desired Expressed as preset ideal pressure, F actuale is the tactile sensor feedback force, K p , K i and K d is the control parameter.

[0132] In an optional embodiment of the present invention, the flexible bionic manipulator collaborative module 140 may further include a tactile feedback-visual servo collaborative control unit ( Figure 2 (not shown), the tactile feedback-visual servoing collaborative control unit can be integrated into the flexible bionic robotic arm 141 to drive the end of the robotic arm to reach the target maintenance position based on the following formula:

[0133] ΔP=K v (P target0 -P visual )+K f (F desired -F actuale )

[0134] Among them, ΔP represents the corrected displacement of the end effector of the robot arm, Ptarget0 represents the target point coordinates, P visual represents the visual positioning coordinates, F desired Indicates the preset ideal pressure, F actuale is the tactile sensor feedback force, K v is the visual weight coefficient, K f is the tactile weight coefficient.

[0135] Among them, the tactile feedback-visual servo collaborative control unit can integrate the tactile sensor feedback force F actuale and visual positioning coordinates P visual , generating correction displacement of the end of the flexible bionic robotic arm to prevent the robotic arm from applying excessive force or deviating from the target.

[0136] In summary, the flexible bionic robotic arm collaborative module drives the end of the robotic arm to the target maintenance position through the multi-degree-of-freedom flexible bionic robotic arm 141, and then uses a laser cleaning head to remove biological attachments on the cable surface according to the defect type and location, or uses a conductive tape attaching device to attach tape to the damaged insulation location. At the same time, a tactile feedback-visual servo collaborative control unit is used to prevent the robotic arm from applying excessive force or deviating from the target.

[0137] The underwater unmanned vehicle (UUV) consists of a multimodal state detection module, a path planning module, a real-time underwater terrain obstacle avoidance module, and a flexible bionic robotic arm collaboration module. The multimodal state detection module, serving as the system's "perception core," is used to detect current leakage, insulation damage, mechanical damage, and surface corrosion defects of underwater inspection targets in real time. It incorporates high-precision underwater electromagnetic sensors, a flexible fiber-optic sonar array, and a multispectral imager. By integrating electromagnetic gradient fields, acoustic wave reflection time differences, and multispectral feature data, it achieves millimeter-level positioning and classification of defects on the target inspection object. The path planning module, based on SLAM technology and the millimeter-level positioning and classification information output by the multimodal state detection module, constructs a three-dimensional point cloud map of the target inspection object's path. Combining dynamic potential field functions with B-spline curves, it generates a "snake-like" inspection path, driving the UUV to autonomously cruise along the target inspection object, adaptively adjusting its speed and attitude to avoid entanglement in curved sections of the target inspection object. The flexible bionic robotic arm collaboration module utilizes a multi-degree-of-freedom flexible bionic robotic arm with an integrated laser cleaning head and conductive tape application device at its end. As the underwater unmanned vehicle autonomously navigates the target object, it uses a tactile feedback force control model and a visual servoing algorithm to remove biological deposits from the target object's surface and temporarily repair damaged insulation. The underwater terrain real-time obstacle avoidance module integrates underwater LiDAR and a reinforcement learning algorithm. Using a dynamic Q-value update strategy, it optimizes the local path of the autonomous navigation path, avoiding obstacles such as reefs and shipwrecks while maintaining a safe distance of 1.5 meters or greater from the target object. This multi-module collaborative control system enables autonomous operation of the entire "inspection-diagnosis-maintenance" process for underwater target objects. It can perform defect detection, autonomous maintenance, and terrain obstacle avoidance on target objects in strong currents and turbid waters, significantly improving the operational efficiency and safety of underwater maintenance operations.

[0138] Specific application scenarios

[0139] Figure 3 This is a schematic diagram of the structure of an unmanned boat system for intelligent inspection and autonomous maintenance of underwater cables in offshore wind farms provided by an embodiment of the present invention. In a specific example, Figure 3 As shown in the figure, taking the underwater cable as the target maintenance object as an example, the autonomous maintenance process of the underwater cable by the underwater unmanned vehicle is specifically explained.

[0140] like Figure 3As shown in the figure, the multimodal cable status detection module is the multimodal status detection module of the underwater unmanned boat. As the "perception core" of the system, this module is used to perceive the current leakage, insulation damage, mechanical damage and surface corrosion defects of underwater cables in real time. It includes underwater high-precision electromagnetic sensors, flexible fiber optic sonar arrays and multispectral imagers. By fusing electromagnetic gradient fields, sound wave reflection time differences and multispectral feature data, it can achieve millimeter-level positioning and classification of cable defects. Specifically, the multimodal cable status detection module uses the magnetic field gradient formula to Locate current leakage and insulation damage when The system triggers an alarm when a defect occurs. It then calculates the damage depth d based on the acoustic wave reflection time difference, and uses a beamforming algorithm to determine the damage location. Spectral distortion index is then used to detect surface corrosion and biological attachment. A covariance matrix weighted algorithm is used to output a comprehensive defect confidence score, achieving a false alarm rate of less than 1%.

[0141] like Figure 3 As shown in the figure, the cable topology adaptive path planning module is the path planning module of the underwater unmanned vehicle. This module constructs a three-dimensional point cloud map of the cable laying path based on the millimeter-level positioning classification information output by the SLAM technology and the multimodal cable status detection module. It combines the dynamic potential field function and the B-spline curve to generate a "snake-shaped" inspection path, drives the unmanned vehicle to cruise autonomously along the cable, and adaptively adjusts the speed and posture to avoid the risk of entanglement in the curved section of the cable.

[0142] like Figure 3 As shown, the bionic robotic arm collaborative maintenance module is the flexible bionic robotic arm collaborative module of the underwater unmanned vehicle. This module uses a multi-degree-of-freedom flexible bionic robotic arm to enable the end effector to accurately reach the detection target point. The end is integrated with a laser cleaning head and a conductive tape application device. Laser cleaning and tape application are used to remove biological adhesion and adjust the adhesion pressure, respectively, with an adhesion accuracy of ±2mm. During the underwater unmanned vehicle's autonomous navigation along the cable, a tactile feedback force control model and a visual servo algorithm are used to correct displacement to ensure non-destructive operation, thereby removing biological adhesion from the cable surface and temporarily repairing insulation damage.

[0143] like Figure 3 As shown in the figure, the real-time seabed terrain obstacle avoidance module is the underwater unmanned vehicle's real-time underwater terrain obstacle avoidance module. This module obtains an obstacle probability map constructed by fusing LiDAR and magnetometers, and the obstacle probability map has a detection resolution of 5cm. At the same time, the real-time seabed terrain obstacle avoidance module integrates seabed LiDAR and reinforcement learning algorithms. Through a dynamic Q-value update strategy, it optimizes the local path of the autonomous cruising path, avoiding obstacles such as reefs and shipwrecks, while maintaining a safe distance from the cable (e.g., ≥1.5m). The obstacle avoidance success rate exceeds 95%.

[0144] The underwater unmanned boat solves the problems of low efficiency and poor real-time performance of underwater cable inspection by integrating multimodal perception, adaptive path planning and bionic maintenance technology, thereby improving the efficiency and real-time performance of underwater cable inspection; the deep integration of electromagnetic sensors, flexible fiber optic sonar arrays and multispectral imagers provides millimeter-level positioning and classification, which can solve the problems of insufficient accuracy and high false alarm rate of cable defect detection, ensure high precision and high efficiency of cable maintenance, and thus improve the accuracy of underwater cable defect detection; path planning based on SLAM technology and dynamic potential field guidance, combined with reinforcement learning algorithm for real-time obstacle avoidance, ensures a safe distance from the cable, ensures safe cruising of the unmanned boat, solves the problem of weak traditional path planning and obstacle avoidance capabilities and inability to adapt to dynamic environments, and can optimize dynamic environment adaptability; based on the bionic robotic arm collaborative maintenance module, it adds automated maintenance equipment, uses laser cleaning and conductive tape attaching devices to achieve efficient and non-destructive cable maintenance and repair operations, and solves the existing problem of lack of non-destructive cable maintenance equipment and reliance on manual intervention.

[0145] It can be seen that the above-mentioned unmanned boat system for intelligent inspection and autonomous maintenance of underwater cables in offshore wind farms can complete cable defect detection, autonomous maintenance and terrain obstacle avoidance in strong currents and turbid waters through multi-module collaborative control, significantly improving the operation and maintenance efficiency and safety of offshore wind farm transmission cables, and realizing the full process of "detection-diagnosis-maintenance" autonomous operation of underwater cables, providing a fully autonomous solution for the efficient operation and maintenance of underwater cables in offshore wind farms, significantly reducing the need for manual intervention and operation and maintenance costs, while improving detection accuracy and environmental adaptability, and has important engineering application value.

[0146] The embodiments of the present invention aim to solve the problems of low efficiency, poor real-time performance and insufficient environmental adaptability of existing underwater cable inspection technologies. It provides a fully autonomous unmanned boat system to achieve high-precision detection, dynamic path planning, autonomous maintenance and complex terrain obstacle avoidance of underwater cables in offshore wind farms. It can be expanded to other marine engineering scenarios such as cross-sea cables and underwater facilities of offshore oil platforms, providing core technical support for the research and development of high-precision, fully autonomous and anti-interference underwater intelligent equipment, and significantly improving the safety and economy of operation and maintenance.

[0147] Figure 4 This is a flow chart of a path planning method for an underwater unmanned boat provided by an embodiment of the present invention. This embodiment is applicable to the case of dynamically planning a path for an underwater unmanned boat. The method can be executed by a path planning module of the underwater unmanned boat, and the path planning module is integrated into the underwater unmanned boat, for example, it can be specifically integrated into the control device of the underwater unmanned boat. The control device of the underwater unmanned boat can be a terminal control module, which serves as the control center of the underwater unmanned boat and controls the underwater unmanned boat to perform various operations, including but not limited to data collection, path planning, real-time obstacle avoidance, and execution of maintenance tasks. Accordingly, if Figure 4 As shown, the method includes the following operations:

[0148] S110: Generate a three-dimensional model of the laying path of the target maintenance object based on the multi-dimensional positioning correlation position information of the underwater unmanned vehicle.

[0149] S120: Generate an initial inspection path for the target maintenance object based on the three-dimensional model of the installation path for the target maintenance object.

[0150] S130. Generate a global smooth inspection path for the target maintenance object based on the discrete path points included in the initial inspection path; wherein the underwater unmanned vehicle performs the maintenance task on the target maintenance object according to the global smooth inspection path.

[0151] The embodiment of the present invention realizes the path planning function for the underwater unmanned boat by generating a three-dimensional model of the laying path of the target maintenance object based on the multi-dimensional positioning correlation position information of the underwater unmanned boat, and generating an initial inspection path of the target maintenance object based on the three-dimensional model of the laying path of the target maintenance object, and then generating a global smooth inspection path of the target maintenance object based on the discrete path points included in the initial inspection path. Accordingly, the underwater unmanned boat can perform maintenance tasks on the target maintenance object based on the final global smooth inspection path. The above technical solution can solve the problems of low accuracy and reliability in the existing underwater unmanned boat path planning, and can realize dynamic path planning, thereby improving the accuracy, real-timeness and reliability of the underwater unmanned boat path planning, and improving the adaptability to complex environments during the path planning process.

[0152] Optionally, after generating the global smooth inspection path of the target maintenance object based on the discrete path points included in the initial inspection path, it may also include: adjusting the speed of the underwater unmanned vehicle according to the global smooth inspection path of the target maintenance object; wherein the underwater unmanned vehicle performs the maintenance task on the target maintenance object according to the adjusted speed and the global smooth inspection path.

[0153] Optionally, the multi-dimensional positioning associated position information includes millimeter-level three-dimensional positioning classification information, three-dimensional point cloud scanning information, and sonar echo information; before generating a three-dimensional model of the laying path of the target maintenance object based on the multi-dimensional positioning associated position information of the underwater unmanned boat, it may also include: obtaining the surrounding magnetic field gradient data measured by the underwater electromagnetic sensor for the target maintenance object; obtaining the damage position positioning data measured by the flexible fiber optic sonar array for the target maintenance object; obtaining the surface corrosion biological attachment data measured by the multi-spectral imager for the target maintenance object; performing weighted fusion on the surrounding magnetic field gradient data, the damage position positioning data, and the surface corrosion biological attachment data, and generating the millimeter-level three-dimensional positioning classification information based on the weighted fusion result.

[0154] Optionally, obtaining the surrounding magnetic field gradient data of the target maintenance object measured by the underwater electromagnetic sensor may include: obtaining the surrounding magnetic field gradient data measured by the target maintenance object based on the following formula:

[0155]

[0156] in, represents the surrounding magnetic field gradient, B x represents the component of the magnetic field in the x direction of the Cartesian coordinate system, B y represents the component of the magnetic field in the y direction of the Cartesian coordinate system, B z represents the component of the magnetic field in the z direction of the Cartesian coordinate system;

[0157] The obtaining of the damage position location data measured by the flexible fiber optic sonar array on the target maintenance object may include: obtaining the damage position location data measured on the target maintenance object based on the following formula:

[0158]

[0159] Wherein, Δt represents the time difference of sound wave reflection, d represents the damage depth of the target maintenance object, c represents the speed of sound, (x m ,y m ,z m ) represents the coordinate position of the damage point, (x ij ,y ij ,0) represents the coordinates of the (i,j)th element in the flexible fiber optic sonar array, Δt ij Indicates the time difference of array element reception;

[0160] The acquiring of the surface corrosion organism attachment data measured by the multispectral imager on the target maintenance object may include: acquiring the surface corrosion organism attachment data measured on the target maintenance object based on the following formula:

[0161]

[0162] Among them, D λ represents the spectral distortion index, S ref (λ) represents the standard spectral reflectance of the surface of the target object under uncorroded condition, S obs (λ) represents the spectral reflectance detected in real time, and λ is the wavelength of light;

[0163] The weighted fusion of the surrounding magnetic field gradient data, the damage position positioning data, and the surface corrosion organism attachment data, and generating the millimeter-level three-dimensional positioning classification information according to the weighted fusion result may include: weighted fusion of the surrounding magnetic field gradient data, the damage position positioning data, and the surface corrosion organism attachment data based on the following formula:

[0164]

[0165] Among them, C fused Denotes the comprehensive confidence of defects, W k is the dynamic weight, C k represents the independent confidence of the underwater electromagnetic sensor, the flexible fiber optic sonar array, and the multispectral imager, B th represents the magnetic field gradient threshold, z max Indicates the maximum value of the coordinate position of the damage point in the z-axis direction, Δt th Indicates the sound wave reflection time difference threshold, D th Indicates the spectral distortion index threshold.

[0166] Optionally, generating the initial inspection path of the target maintenance object based on the three-dimensional model of the laying path of the target maintenance object may include: generating a comprehensive potential field function based on the three-dimensional model of the laying path of the target maintenance object, the gravitational field of the target maintenance object, the ocean current interference field and the obstacle repulsion field; and driving the underwater unmanned boat to cruise along the laying direction of the target maintenance object according to the comprehensive potential field function to obtain the initial inspection path of the target maintenance object.

[0167] Optionally, generating a comprehensive potential field function based on the three-dimensional model of the laying path of the target maintenance object, the gravitational field of the target maintenance object, the ocean current interference field, and the obstacle repulsion field includes:

[0168] The comprehensive potential field function is generated based on the following formula:

[0169] U total (P)=U target (P)+U current (P)+U obstacle (P)

[0170] U target (P) = k1·ln(1+||PP target ||)

[0171] U current (P)=k2·||V current ||(1-cosθ)

[0172]

[0173] Among them, U total (P) represents the comprehensive potential field function, U target (P) represents the gravitational field of the target maintenance object, U current (P) represents the ocean current disturbance field, U obstacle (P) represents the obstacle repulsion field, P represents the current three-dimensional position coordinates of the underwater unmanned vehicle, k1 is the gravity gain coefficient, P target represents the coordinates of the nearest point on the center line of the target maintenance object, k2 is the current compensation coefficient, V current is the real-time ocean current vector, θ is the angle between the heading of the underwater unmanned vehicle and the ocean current, k3 is the repulsive gain coefficient, ε is the smoothing factor, P obs,j is the jth obstacle position, and M represents the number of obstacles.

[0174] Optionally, generating the global smooth inspection path of the target maintenance object according to the discrete path points included in the initial inspection path may include: generating the global smooth inspection path of the target maintenance object according to the discrete path points included in the initial inspection path based on the following formula:

[0175]

[0176] Wherein, P(s) represents the global smooth inspection path, N i,k (s) represents the B-spline basis function, s is the path parameter, Q i represents the discrete path point, k represents the curve order, and n represents the number of the discrete path points.

[0177] Optionally, adjusting the speed of the underwater unmanned vehicle according to the global smooth inspection path of the target maintenance object includes:

[0178] The speed of the underwater unmanned vehicle is adjusted according to the global smooth inspection path of the target maintenance object based on the following formula:

[0179] υ=υ max ·exp(-β·κ-γ·||V current ||)

[0180] Wherein, υ represents the adjusted speed of the underwater unmanned vehicle, υ max is the maximum speed, β and γ are adjustment coefficients, k represents the curve order, V current is the real-time ocean current vector.

[0181] Optionally, after generating the global smooth inspection path of the target maintenance object based on the discrete path points included in the initial inspection path, it may also include: updating the comprehensive potential field function based on the curvature penalty factor; optimizing the global smooth inspection path of the target maintenance object based on the curvature of the global smooth inspection path of the target maintenance object and the average position of the center line of the target maintenance object in the three-dimensional model of the laying path.

[0182] Optionally, updating the comprehensive potential field function according to the curvature penalty factor may include: updating the comprehensive potential field function according to the curvature penalty factor based on the following formula:

[0183] U total (P)=U target (P)+U current (P)+U obstacle (P)+U curvature (P)

[0184] U curvature (P) = k4·κ(P) 2

[0185] Among them, U curvature (P) represents the curvature penalty term, k4 is the curvature penalty coefficient, and κ(P) is the path curvature of the current three-dimensional position coordinate P of the underwater unmanned vehicle;

[0186] Optimizing the global smooth inspection path of the target maintenance object according to the curvature of the global smooth inspection path of the target maintenance object and the average position of the center line of the target maintenance object in the three-dimensional model of the laying path may include: optimizing the global smooth inspection path of the target maintenance object based on the following formula:

[0187]

[0188]

[0189] Among them, E path represents the path optimization objective function value, P i represents the three-dimensional position coordinates of the i-th path point; C represents the average position coordinates of the center line of the target maintenance object in the three-dimensional model of the laying path, C curvature (P i ) represents the global smooth inspection path at point P iThe curvature at , A represents the weight coefficient for adjusting the path smoothness, N represents the number of path points included in the global smooth inspection path, κ(s) represents the curvature of the global smooth inspection path, P(s) represents the global smooth inspection path, P′(s) represents the first-order derivative of P(S), P”(s) represents the second-order derivative of P(S), E curvature represents the total curvature integral of the path.

[0190] Optionally, the above method may also include: obtaining three-dimensional coordinate point cloud data and magnetic field anomaly detection data of obstacles in the underwater area where the target maintenance object is located; generating an obstacle probability map based on the three-dimensional coordinate point cloud data of the obstacle and the magnetic field anomaly detection data; generating obstacle avoidance instructions based on the obstacle probability map and the current state data of the underwater unmanned boat in combination with reinforcement learning; and performing real-time dynamic adjustment on the global smooth inspection path of the target maintenance object through the obstacle avoidance instructions.

[0191] Optionally, generating an obstacle probability map based on the obstacle three-dimensional coordinate point cloud data and the magnetic field anomaly detection data may include: generating an obstacle probability map based on the obstacle three-dimensional coordinate point cloud data and the magnetic field anomaly detection data based on the following formula:

[0192]

[0193] Among them, P obs (x,y) represents the probability that there is an obstacle at the position (x,y), B is the sensitivity coefficient, represents the magnetic field gradient at position (x, y), B th represents the magnetic field gradient threshold.

[0194] Optionally, generating obstacle avoidance instructions based on the obstacle probability map and the current state data of the underwater unmanned vehicle in combination with reinforcement learning may include: generating obstacle avoidance instructions based on the following formula:

[0195] Q(s,a)←Q(s,a)+α[r+γmax a′ Q(s',a')-Q(s,a)]

[0196]

[0197] Reward(s,a)=w1·Distance obstacle +w2 Distance targete +w3·cos(δ)

[0198] Among them, Q(s,a) represents the expected cumulative reward value of taking action a in state s, s represents the state space, a represents the action space, α is the learning rate, r is the immediate reward, which is calculated by the reward function Reward(s,a), and γ is the discount factor; a' represents the action in the next state, max a′ Q(s',a') represents the maximum expected reward for all actions in the next state s', Distance obstacle Indicates the distance between the current obstacle and the underwater unmanned vehicle, Distance targete represents the distance between the target maintenance object and the underwater unmanned vehicle, δ represents the approach angle between the underwater unmanned vehicle and the obstacle, and w1, w2 and w3 are weight coefficients.

[0199] Optionally, the above method may also include: when the underwater unmanned boat performs the maintenance task on the target maintenance object, the flexible bionic robotic arm drives the end of the robotic arm to reach the target maintenance position of the global smooth inspection path; according to the defect type of the target maintenance position, the laser cleaning head or the conductive tape attaching device is started to inspect the target maintenance position.

[0200] Optionally, the step of driving the end of the flexible bionic robotic arm to reach the target inspection position of the global smooth inspection path may include:

[0201] The end of the robotic arm is driven to reach the target maintenance position based on the following formula:

[0202] ΔP=K v (P target0 -P visual )+K f (F desired -F actuale )

[0203] Among them, ΔP represents the corrected displacement of the end effector of the robot arm, P target0 represents the target point coordinates, P visual represents the visual positioning coordinates, F desired Indicates the preset ideal pressure, F actuale is the tactile sensor feedback force, K v is the visual weight coefficient, K f is the tactile weight coefficient.

[0204] In the technical solutions disclosed herein, the collection, storage, use, processing, transmission, provision and disclosure of user personal information are in compliance with the relevant laws and regulations and do not violate public order and good morals.

[0205] It should be noted that the relevant information (including but not limited to user device information, user personal information, etc.) and data (including but not limited to data for display, data for analysis, etc.) involved in this disclosure are all information and data authorized by the user or fully authorized by all parties, and the collection, use and processing of relevant data comply with relevant laws, regulations and standards in the relevant regions.

[0206] It should be noted that any arrangement and combination of the technical features in the above embodiments also falls within the protection scope of the present invention.

[0207] Figure 5 A schematic structural diagram of an electronic device 100 that can be used to implement an embodiment of the present invention is shown. The electronic device is intended to represent various forms of digital computers, such as laptop computers, desktop computers, workstations, personal digital assistants, servers, blade servers, mainframe computers, and other suitable computers. The electronic device can also represent various forms of mobile devices and other similar computing devices. The components shown herein, their connections and relationships, and their functions are merely examples and are not intended to limit the implementation of the present invention described and / or claimed herein. Optionally, the electronic device can be integrated into an underwater unmanned vehicle as a terminal control device to perform the path planning of the underwater unmanned vehicle provided in the above embodiments.

[0208] like Figure 5 As shown, the electronic device 100 includes at least one processor 11, and a memory, such as a read-only memory (ROM) 12, a random access memory (RAM) 13, etc., which is communicatively connected to the at least one processor 11. The memory stores a computer program that can be executed by the at least one processor, and the processor 11 can perform various appropriate actions and processes according to the computer program stored in the read-only memory (ROM) 12 or the computer program loaded from the storage unit 18 into the random access memory (RAM) 13. Various programs and data required for the operation of the electronic device 100 can also be stored in the RAM 13. The processor 11, ROM 12, and RAM 13 are connected to each other via a bus 14. An input / output (I / O) interface 15 is also connected to the bus 14.

[0209] Multiple components in the electronic device 100 are connected to the I / O interface 15, including an input unit 16, such as a keyboard, a mouse, etc.; an output unit 17, such as various types of displays, speakers, etc.; a storage unit 18, such as a magnetic disk, an optical disk, etc.; and a communication unit 19, such as a network card, a modem, a wireless communication transceiver, etc. The communication unit 19 allows the electronic device 100 to exchange information / data with other devices via a computer network such as the Internet and / or various telecommunication networks.

[0210] The processor 11 can be any general-purpose and / or specialized processing component with processing and computing capabilities. Some examples of the processor 11 include, but are not limited to, a central processing unit (CPU), a graphics processing unit (GPU), various specialized artificial intelligence (AI) computing chips, various processors running machine learning model algorithms, a digital signal processor (DSP), and any other suitable processor, controller, microcontroller, etc. The processor 11 executes the various methods and processes described above, such as the path planning method for an underwater unmanned vehicle.

[0211] Optionally, the path planning method of the underwater unmanned vehicle may include: generating a three-dimensional model of the laying path of the target maintenance object based on the multi-dimensional positioning associated position information of the underwater unmanned vehicle; generating an initial inspection path of the target maintenance object based on the three-dimensional model of the laying path of the target maintenance object; generating a global smooth inspection path of the target maintenance object based on the discrete path points included in the initial inspection path; wherein, the underwater unmanned vehicle performs the maintenance task on the target maintenance object according to the global smooth inspection path.

[0212] In some embodiments, the path planning method for the underwater unmanned vehicle can be implemented as a computer program, which is tangibly contained in a computer-readable storage medium, such as the storage unit 18. In some embodiments, part or all of the computer program can be loaded and / or installed on the electronic device 100 via the ROM 12 and / or the communication unit 19. When the computer program is loaded into the RAM 13 and executed by the processor 11, one or more steps of the path planning method for the underwater unmanned vehicle described above can be performed. Alternatively, in other embodiments, the processor 11 can be configured to execute the path planning method for the underwater unmanned vehicle by any other appropriate means (for example, by means of firmware).

[0213] Various embodiments of the systems and techniques described herein can be implemented in digital electronic circuit systems, integrated circuit systems, field programmable gate arrays (FPGAs), application specific integrated circuits (ASICs), application specific standard products (ASSPs), system-on-chip systems (SOCs), programmable logic devices (CPLDs), computer hardware, firmware, software, and / or combinations thereof. These various embodiments can include being implemented in one or more computer programs that are executable and / or interpreted on a programmable system that includes at least one programmable processor, which can be a special purpose or general purpose programmable processor that can receive data and instructions from a storage system, at least one input device, and at least one output device, and transmit data and instructions to the storage system, the at least one input device, and the at least one output device.

[0214] Computer programs for implementing the methods of the present invention may be written in any combination of one or more programming languages. These computer programs may be provided to a processor of a general-purpose computer, a special-purpose computer, or other programmable data processing device, such that when the computer program is executed by the processor, the functions / operations specified in the flowcharts and / or block diagrams are implemented. The computer program may be executed entirely on the machine, partially on the machine, as a stand-alone software package, partially on the machine and partially on a remote machine, or entirely on a remote machine or server.

[0215] In the context of the present invention, computer-readable storage media can be tangible media that can contain or store a computer program for use with an instruction execution system, device or equipment or used in combination with an instruction execution system, device or equipment. Computer-readable storage media can include but are not limited to electronic, magnetic, optical, electromagnetic, infrared or semiconductor systems, devices or equipment, or any suitable combination of the foregoing. Alternatively, computer-readable storage media can be machine-readable signal media. More specific examples of machine-readable storage media can include electrical connections based on one or more lines, portable computer disks, hard disks, random access memories (RAM), read-only memories (ROM), erasable programmable read-only memories (EPROM or flash memory), optical fibers, portable compact disk read-only memories (CD-ROM), optical storage devices, magnetic storage devices, or any suitable combination of the foregoing.

[0216] To provide interaction with a user, the systems and techniques described herein can be implemented on an electronic device having: a display device (e.g., a CRT (cathode ray tube) or LCD (liquid crystal display) monitor) for displaying information to the user; and a keyboard and pointing device (e.g., a mouse or trackball) through which the user can provide input to the electronic device. Other types of devices can also be used to provide interaction with the user; for example, the feedback provided to the user can be any form of sensory feedback (e.g., visual feedback, auditory feedback, or tactile feedback); and the input from the user can be received in any form (including acoustic input, voice input, or tactile input).

[0217] The systems and techniques described herein can be implemented in a computing system that includes back-end components (e.g., as a data server), or a computing system that includes middleware components (e.g., an application server), or a computing system that includes front-end components (e.g., a user computer with a graphical user interface or web browser through which a user can interact with implementations of the systems and techniques described herein), or a computing system that includes any combination of such back-end components, middleware components, or front-end components. The components of the system can be interconnected by any form or medium of digital data communication (e.g., a communication network). Examples of communication networks include: a local area network (LAN), a wide area network (WAN), a blockchain network, and the Internet.

[0218] A computing system may include clients and servers. The clients and servers are typically remote from each other and typically interact via a communication network. This client-server relationship arises through computer programs running on the respective computers, creating a client-server relationship. The server may be a cloud server, also known as a cloud computing server or cloud host. This server is a hosting product within the cloud computing service ecosystem that addresses the management difficulties and limited scalability of traditional physical hosting and VPS services.

[0219] It should be understood that the various forms of the processes shown above can be used to reorder, add, or delete steps. For example, the steps described in this disclosure can be performed in parallel, sequentially, or in a different order, as long as the desired results of the technical solutions disclosed in this disclosure can be achieved. This is not a limitation herein.

[0220] The above specific embodiments do not constitute a limitation on the scope of protection of this disclosure. Those skilled in the art will appreciate that various modifications, combinations, sub-combinations, and substitutions may be made based on design requirements and other factors. Any modifications, equivalent substitutions, and improvements made within the spirit and principles of this disclosure shall be included within the scope of protection of this disclosure.

Claims

1. An underwater unmanned boat, characterized in that: It includes a path planning module, which includes a target maintenance object three-dimensional point cloud map construction unit, an initial inspection path generation unit and a global smooth inspection path generation unit; wherein: The target maintenance object three-dimensional point cloud map construction unit is used to generate a three-dimensional model of the laying path of the target maintenance object according to the multi-dimensional positioning correlation position information of the underwater unmanned boat; The initial inspection path generating unit is used to generate an initial inspection path for the target inspection object according to the three-dimensional model of the laying path of the target inspection object; The global smooth inspection path generation unit is used to generate a global smooth inspection path of the target maintenance object according to the discrete path points included in the initial inspection path; The underwater unmanned boat performs maintenance tasks on the target maintenance object according to the global smooth inspection path.

2. The underwater unmanned vehicle according to claim 1, characterized in that: The path planning module also includes a speed and attitude adaptive adjustment unit for: Adjusting the speed of the underwater unmanned vehicle according to the global smooth inspection path of the target maintenance object; The underwater unmanned boat performs maintenance tasks on the target maintenance object according to the adjusted speed and the global smooth inspection path.

3. The underwater unmanned vehicle according to claim 1, characterized in that: The multi-dimensional positioning associated position information includes millimeter-level three-dimensional positioning classification information, three-dimensional point cloud scanning information, and sonar echo information; the underwater unmanned vehicle also includes a multi-modal state detection module, which includes an underwater electromagnetic sensor, a flexible fiber optic sonar array, a multi-spectral imager, and a data fusion unit; wherein: The underwater electromagnetic sensor is used to measure the surrounding magnetic field gradient data of the target maintenance object; The flexible fiber optic sonar array is used to measure damage location data of the target maintenance object; The multispectral imager is used to measure surface corrosion organism attachment data of the target maintenance object; The data fusion unit is used to perform weighted fusion on the surrounding magnetic field gradient data, the damage position positioning data and the surface corrosion biological attachment data, and generate the millimeter-level three-dimensional positioning classification information according to the weighted fusion result.

4. The underwater unmanned vehicle according to claim 3, characterized in that: The underwater electromagnetic sensor is further used to measure the surrounding magnetic field gradient data of the target maintenance object based on the following formula: in, represents the surrounding magnetic field gradient, B x represents the component of the magnetic field in the x direction of the Cartesian coordinate system, B y represents the component of the magnetic field in the y direction of the Cartesian coordinate system, B z represents the component of the magnetic field in the z direction of the Cartesian coordinate system; The flexible fiber optic sonar array is further used to measure the damage location data of the target maintenance object based on the following formula: Wherein, Δt represents the time difference of sound wave reflection, d represents the damage depth of the target maintenance object, c represents the speed of sound, (x m ,y m ,z m ) represents the coordinate position of the damage point, (x ij ,y ij ,0) represents the coordinates of the (i,j)th element in the flexible fiber optic sonar array, Δt ij Indicates the time difference of array element reception; The multispectral imager is further used to measure the surface corrosion organism attachment data of the target maintenance object based on the following formula: Among them, D λ represents the spectral distortion index, S ref (λ) represents the standard spectral reflectance of the surface of the target object under uncorroded condition, S obs (λ) represents the spectral reflectance detected in real time, and λ is the wavelength of light; The data fusion unit is further configured to perform weighted fusion of the surrounding magnetic field gradient data, the damage position location data, and the surface corrosion organism attachment data based on the following formula: Among them, C fused Denotes the comprehensive confidence of defects, W k is the dynamic weight, C k represents the independent confidence of the underwater electromagnetic sensor, the flexible fiber optic sonar array, and the multispectral imager, B th represents the magnetic field gradient threshold, z max Indicates the maximum value of the coordinate position of the damage point in the z-axis direction, Δt th Indicates the sound wave reflection time difference threshold, D th Indicates the spectral distortion index threshold.

5. The underwater unmanned vehicle according to claim 1, characterized in that: The initial inspection path generating unit is further configured to: Generate a comprehensive potential field function based on the three-dimensional model of the laying path of the target maintenance object, the gravitational field of the target maintenance object, the ocean current interference field and the obstacle repulsion field; The underwater unmanned vehicle is driven to cruise along the laying direction of the target maintenance object according to the comprehensive potential field function to obtain an initial inspection path of the target maintenance object.

6. The underwater unmanned vehicle according to claim 5, characterized in that: The initial inspection path generating unit is further configured to: The comprehensive potential field function is generated based on the following formula: IN total (P)=U target (P)+U current (P)+U obstacle (P) U target (P)=k1·ln(1+||PP target ||) U current (P)=k2·||V current ||(1-cosθ) Among them, U total (P) represents the comprehensive potential field function, U target (P) represents the gravitational field of the target maintenance object, U current (P) represents the ocean current disturbance field, U obstacle (p) represents the obstacle repulsion field, P represents the current three-dimensional position coordinates of the underwater unmanned vehicle, k1 is the gravity gain coefficient, P target represents the coordinates of the nearest point on the center line of the target maintenance object, k2 is the current compensation coefficient, V current is the real-time ocean current vector, θ is the angle between the heading of the underwater unmanned vehicle and the ocean current, k3 is the repulsive gain coefficient, ε is the smoothing factor, P obs,j is the jth obstacle position, and M represents the number of obstacles.

7. The underwater unmanned vehicle according to claim 1, characterized in that: The global smooth inspection path generation unit is further used for: A global smooth inspection path for the target maintenance object is generated according to the discrete path points included in the initial inspection path based on the following formula: Wherein, P(s) represents the global smooth inspection path, N i,k (s) represents the B-spline basis function, s is the path parameter, Q i represents the discrete path point, k represents the curve order, and n represents the number of the discrete path points.

8. The underwater unmanned vehicle according to claim 2, characterized in that: The speed and attitude adaptive adjustment unit is also used for: The speed of the underwater unmanned vehicle is adjusted according to the global smooth inspection path of the target maintenance object based on the following formula: υ=υ max ·exp(-β·κ-γ·||V current ||) Wherein, υ represents the adjusted speed of the underwater unmanned vehicle, υ max is the maximum speed, β and γ are adjustment coefficients, k represents the curve order, V current is the real-time ocean current vector.

9. The underwater unmanned vehicle according to claim 6, characterized in that: The initial inspection path generating unit is further configured to: Updating the comprehensive potential field function according to the curvature penalty factor; The global smooth inspection path generation unit is further used for: The global smooth inspection path of the target maintenance object is optimized according to the curvature of the global smooth inspection path of the target maintenance object and the average position of the center line of the target maintenance object in the three-dimensional model of the laying path.

10. The underwater unmanned vehicle according to claim 9, characterized in that: The initial inspection path generating unit is further configured to: The comprehensive potential field function is updated according to the curvature penalty factor based on the following formula: IN total (P)=U target (P)+U current (P)+U obstacle (P)+U curvature (P) U curvature (P)=k4·κ(P) 2 Among them, U curvature (P) represents the curvature penalty term, k4 is the curvature penalty coefficient, and κ(P) is the path curvature of the current three-dimensional position coordinate P of the underwater unmanned vehicle; The global smooth inspection path generation unit is further used for: The global smooth inspection path of the target maintenance object is optimized based on the following formula: Among them, E path represents the path optimization objective function value, P i represents the three-dimensional position coordinates of the i-th path point; C represents the average position coordinates of the center line of the target maintenance object in the three-dimensional model of the laying path, C curvature (P i ) represents the global smooth inspection path at point P i The curvature at , A represents the weight coefficient for adjusting the path smoothness, N represents the number of path points included in the global smooth inspection path, κ(s) represents the curvature of the global smooth inspection path, P(s) represents the global smooth inspection path, P′(s) represents the first-order derivative of P(S), P”(s) represents the second-order derivative of P(S), E curvature represents the total curvature integral of the path.

11. The underwater unmanned vehicle according to claim 1, characterized in that: It also includes an underwater terrain real-time obstacle avoidance module, which includes a multimodal obstacle perception unit and a reinforcement learning obstacle avoidance decision unit; wherein: The multimodal obstacle sensing unit is configured to: obtain three-dimensional coordinate point cloud data of obstacles in the underwater area where the target maintenance object is located and magnetic field anomaly detection data; and generate an obstacle probability map based on the three-dimensional coordinate point cloud data of obstacles and the magnetic field anomaly detection data; The reinforcement learning obstacle avoidance decision unit is used to: generate obstacle avoidance instructions based on the obstacle probability map and the current state data of the underwater unmanned vehicle in combination with reinforcement learning; and dynamically adjust the global smooth inspection path of the target maintenance object in real time through the obstacle avoidance instructions.

12. The underwater unmanned vehicle according to claim 11, characterized in that: The multimodal obstacle sensing unit is further configured to: An obstacle probability map is generated based on the obstacle three-dimensional coordinate point cloud data and the magnetic field anomaly detection data based on the following formula: Among them, P obs (x,y) represents the probability that there is an obstacle at the position (x,y), B is the sensitivity coefficient, represents the magnetic field gradient at position (x, y), B th represents the magnetic field gradient threshold.

13. The underwater unmanned vehicle according to claim 11, characterized in that: The reinforcement learning obstacle avoidance decision unit is further used to: Generate obstacle avoidance instructions based on the following formula: Q(s,a)←Q(s,a)+α[r+γmax a′ Q(s',a')-Q(s,a)] Reward(s,a)=w1·Distance obstacle +w2·Distance targete +w3·cos(δ) Among them, Q(s,a) represents the expected cumulative reward value of taking action a in state s, s represents the state space, a represents the action space, α is the learning rate, r is the immediate reward, which is calculated by the reward function Reward(s,a), and γ is the discount factor; a' represents the action in the next state, max a′ Q(s',a') represents the maximum expected reward for all actions in the next state s', Distance obstacle Indicates the distance between the current obstacle and the underwater unmanned vehicle, Distance targete represents the distance between the target maintenance object and the underwater unmanned vehicle, δ represents the approach angle between the underwater unmanned vehicle and the obstacle, and w1, w2 and w3 are weight coefficients.

14. The underwater unmanned vehicle according to claim 1, characterized in that: The system also includes a flexible bionic robotic arm collaborative module, which includes a flexible bionic robotic arm and a robotic arm maintenance module at the end of the robotic arm; the robotic arm maintenance module includes a laser cleaning head and a conductive tape attaching device; wherein the flexible bionic robotic arm collaborative module is used to: When the underwater unmanned boat performs the maintenance task on the target maintenance object, the flexible bionic manipulator drives the end of the manipulator arm to reach the target maintenance position of the global smooth inspection path; The laser cleaning head or the conductive tape attaching device is activated according to the defect type of the target repair position to repair the target repair position.

15. The underwater unmanned vehicle according to claim 14, characterized in that: The flexible bionic robotic arm collaborative module is also used for: The end of the robotic arm is driven to reach the target maintenance position based on the following formula: ΔP=K v (P target0 -P visual )+K f (F desired -F actuale ) Among them, ΔP represents the corrected displacement of the end effector of the robot arm, P target0 represents the target point coordinates, P visual represents the visual positioning coordinates, F desired Indicates the preset ideal pressure, F actuale is the tactile sensor feedback force, K v is the visual weight coefficient, K f is the tactile weight coefficient.

16. A path planning method for an underwater unmanned vehicle, characterized in that: include: Generate a three-dimensional model of the laying path of the target maintenance object based on the multi-dimensional positioning and correlation position information of the underwater unmanned vehicle; Generate an initial inspection path for the target maintenance object based on the three-dimensional model of the laying path of the target maintenance object; Generating a global smooth inspection path for the target maintenance object according to the discrete path points included in the initial inspection path; The underwater unmanned boat performs maintenance tasks on the target maintenance object according to the global smooth inspection path.

17. An electronic device, characterized in that: The electronic device comprises: at least one processor; and a memory communicatively connected to the at least one processor; wherein, The memory stores a computer program executed by the at least one processor, and the computer program is executed by the at least one processor so that the at least one processor can execute the path planning method for the underwater unmanned vehicle according to claim 16.

18. A computer-readable storage medium, characterized in that The computer-readable storage medium stores computer instructions, and the computer instructions are used to enable the processor to implement the path planning method for the underwater unmanned vehicle according to claim 16 when executed.

19. A computer program product comprising a computer program / instructions, wherein: When the computer program / instructions are executed by the processor, the path planning method for the underwater unmanned vehicle according to claim 16 is implemented.

Images