A topology-aware spectral clustering path planning method and system for multi-unmanned ship cooperative coverage detection and a storage medium
By employing the topology-sensing spectral clustering path planning method, the problem of incomplete sonar imaging in multi-unmanned surface vessel systems was solved, achieving efficient data fusion and path optimization, and generating highly adaptable coverage detection paths.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- HARBIN ENG UNIV
- Filing Date
- 2026-02-12
- Publication Date
- 2026-06-09
Smart Images
Figure CN122170909A_ABST
Abstract
Description
Technical Field
[0001] This invention belongs to the field of marine surveying and multi-unmanned system coverage path planning technology, specifically involving a topological sensing spectrum clustering path planning method, system and storage medium for multi-unmanned surface vessel collaborative coverage detection. Background Technology
[0002] With the development of the marine economy, unmanned surface vessels (USVs) equipped with side-scan sonar for seabed topography exploration, search and rescue, and underwater target identification have become the mainstream operational mode. Multi-USV Coverage Path Planning (MCPP) is a key technology for these mission areas. This invention studies an MCPP method for underwater target search missions, which have high requirements for coverage path quality.
[0003] However, due to the inherent physical limitations of underwater acoustic propagation, sonar imaging often suffers from incomplete imaging, severe pixel intensity attenuation, and reduced resolution at the edges of scanning strips. To compensate for this deficiency and improve the confidence level of target detection, in practical operations, the overlapping area between adjacent survey lines can be utilized to observe the same target multiple times from different azimuth angles, thereby improving detection accuracy through multi-view data fusion technology. This data fusion process places extremely high demands on the "spatial compactness" of the coverage path. Only when high-density adjacent coverage trajectories are assigned to the same unmanned surface vessel and operate continuously in time and space can the overlapping data for fusion be acquired efficiently, avoiding data acquisition discontinuities caused by cross-regional navigation.
[0004] Existing MCPP methods have significant shortcomings:
[0005] 1. Grid-based allocation methods (such as DARP and STC): These methods discretize the region into small grid cells for allocation. In pursuit of optimal load balancing, the algorithms often forcibly fragment physically continuous regions, generating elongated sub-regions with jagged boundaries. This not only increases the number of turns required by the unmanned surface vessel (USV) but also disrupts the continuity of the survey lines, causing adjacent survey lines to be assigned to different USVs. Therefore, data fusion using overlapping regions becomes extremely difficult with this type of method.
[0006] 2. Heuristic neural network-based methods (such as GBNN and BINN): These methods typically introduce predefined motion templates or excitation mechanisms to attempt to reduce turning and obstacle avoidance. Therefore, they rely on local decisions and lack optimization of the global path structure. When dealing with large-scale complex environments, these methods are easily limited by the rigidity of the predefined templates, resulting in irregular shapes and loose boundaries in the generated task sub-regions. It is difficult to maintain the spatial compactness and trajectory continuity of the sub-regions while ensuring load balancing.
[0007] In summary, there is an urgent need for a new method for area allocation and path planning that can take "maintaining the continuity of the survey line" as the core principle, while strictly ensuring the load balance of each unmanned surface vessel, maximize the spatial compactness of the sub-region, and ensure that adjacent survey lines can be used for each other, thereby providing a high-quality trajectory foundation for sonar data fusion. Summary of the Invention
[0008] The purpose of this invention is to solve the problems of difficulty in sonar edge data fusion and low detection accuracy caused by neglecting spatial compactness in existing technologies. It proposes a topological sensing spectrum clustering path planning method, system and storage medium for multi-unmanned surface vessel collaborative coverage detection.
[0009] The objective of this invention is achieved through the following technical solution:
[0010] A topology-sensing spectral clustering path planning method for collaborative coverage detection by multiple unmanned surface vessels includes the following steps:
[0011] Step 1: Decompose the acquired task area into a set of rectangles as the basic task unit;
[0012] Step 2: Using the area of the rectangle as the node weight, calculate the topological adjacency strength as the edge weight based on the consistency of the effective common boundary length and extension direction between adjacent rectangles, and construct a weighted undirected connected graph that represents the topological adjacency strength graph.
[0013] Step 3: Calculate the Laplacian matrix and its eigenvectors of the weighted undirected connected graph, and map all rectangular nodes to the low-dimensional spectral space;
[0014] Step 4: Based on the weight of each rectangular node, discretize each rectangular node into one or more virtual nodes with equal weight, and determine the load capacity constraint of each unmanned surface vessel based on the total number of virtual nodes and the number of unmanned surface vessels.
[0015] Step 5: In the spectral space, using virtual nodes as units, solve the linear sum allocation problem that satisfies the load capacity constraint iteratively, assign the virtual nodes to the sub-regions corresponding to each unmanned surface vessel, and determine the sub-region to which each original rectangular node belongs based on the allocation results, thus obtaining the initial region allocation scheme;
[0016] Step 6: Perform connectivity checks and repairs on the initial region allocation scheme, and fine-tune the rectangles at the boundaries of sub-regions based on a gradient-based boundary refinement strategy with load balancing as the goal, to obtain the final region allocation scheme and output the coverage path.
[0017] Further, in step 2, the topological adjacency strength is calculated as the edge weight based on the consistency between the effective common boundary length and the extension direction between adjacent rectangles. This specifically includes:
[0018] Calculate any two adjacent rectangles and Edge weights between ,
[0019]
[0020] Among them, node weight Rectangle Geometric area; node weight Rectangle The geometric area;
[0021] for and The effective shared boundary length, if and If the extension directions are consistent, then equal to rectangle and Geometric shared boundary length ;like and If the directions of extension are orthogonal, then The total length of the shared boundary between the connected components in the same direction to which one rectangle belongs and the other rectangle. ,
[0022]
[0023] in, Represents a rectangle The direction of extension is The connected components, These represent the horizontal and vertical directions, respectively. Represents a rectangle The direction of extension is The connected components; For connected components All rectangles in the middle The sum of the directly shared boundary lengths; For connected components All rectangles in the middle The sum of the directly shared boundary lengths.
[0024] Furthermore, step 3 specifically includes,
[0025] Calculate the normalized Laplacian matrix ,in, It is the identity matrix; It is a degree matrix; Let the edge weight matrix be the edge weight matrix.
[0026] For the normalized Laplace matrix Perform feature decomposition. Let the total number of unmanned surface vessels be... Then it is necessary Regions are divided based on cluster centers. Based on the fundamental principles of spectral embedding, regions are selected... The former The eigenvectors corresponding to the smallest eigenvalues constitute the spectral embedding matrix. ,in, The total number of rectangles in the rectangle set input in step 1;
[0027] Each rectangle Mapped to A point in 3D space In this space, the closer the points are, the more tightly they are physically topologically combined, and the more suitable they are to be assigned to the same unmanned surface vessel.
[0028] Further, step 4 discretizes each rectangular node into one or more virtual nodes with equal weights, specifically including:
[0029] Based on the total number of target virtual nodes Sum of the weights of all rectangular nodes Calculate scaling factor ,in The weights of the rectangle nodes;
[0030] Each rectangle node Split into virtual nodes ,in ,symbol Indicates rounding up; each virtual node Inherit from its parent node coordinate vector in spectral space .
[0031] Furthermore, the determination of the load capacity constraints for each unmanned surface vessel specifically involves:
[0032] No. Payload capacity constraints of unmanned surface vessels ,in, This represents the total number of virtual nodes. Number of unmanned surface vessels; symbol Indicates rounding down; Assign an indicator variable to the remainder to ensure that the sum of the capacities of all unmanned surface vessels equals the total number of nodes. .
[0033] Furthermore, step 5 includes the following steps:
[0034] (1) The cluster centers are initialized using the maximum distance dispersion strategy, where the first cluster center is the first cluster center. The cluster center corresponds to the first... Both subregions belong to the first... Unmanned surface vessel;
[0035] (2) in the In the next iteration, the cost matrix is constructed. , among which, element Indicates the first virtual nodes Assigned to the The cost of an unmanned surface vessel:
[0036]
[0037] in, For virtual nodes spectral space coordinates; For the first Each cluster of centers; As a connectivity penalty, if a node Belonging rectangle and the first If a subregion is topologically disconnected, then it is infinitely large;
[0038] (3) Solve the linear sum distribution problem and find the current optimal distribution matrix. ,in :
[0039]
[0040] (4) Update the spectral embedding coordinates of each cluster center according to the current allocation matrix, and repeat (2)-(3) until the cluster center positions converge or the maximum number of iterations is reached.
[0041] (5) According to the majority voting principle, count the original rectangles. Determine the ownership status of all included virtual nodes. The final sub-region :
[0042]
[0043] in, It is a rectangle The corresponding set of virtual nodes.
[0044] Furthermore, step (1) initializes the cluster centers using a maximum distance dispersion strategy, specifically as follows:
[0045] From the set of spectral space coordinates of all virtual nodes In the process, one is randomly selected as the first cluster center. ;
[0046] According to the Each cluster of hearts Calculate all unselected nodes Shortest distance to the currently selected center set :
[0047]
[0048] Selecting makes The largest node is the first Cluster centers are selected until one is chosen. Each cluster of hearts.
[0049] Furthermore, step 6, based on a gradient-based boundary refinement strategy, fine-tunes the rectangles at the boundaries of sub-regions with load balancing as the objective, specifically including:
[0050] (1) Detect whether there are isolated island rectangles in each task sub-region. If so, reassign them to the adjacent sub-region with the longest shared boundary.
[0051] (2) For leaf rectangles located at the boundary of a subregion Calculate and remove it from the current sub-region Move to adjacent sub-region Load variance gain after ,like Furthermore, the movement operation does not damage the original area. If the connectivity is such that the move operation is executed;
[0052]
[0053] in, and Sub-regions and The sum of the areas of all rectangles within the current load is the current load. The target average load for all sub-regions.
[0054] A topology-sensing spectral clustering path planning system for collaborative coverage detection by multiple unmanned surface vessels includes modules for executing processing instructions for each step of the method.
[0055] Data input module: used to receive rectangular decomposition data and unmanned surface vessel quantity parameters;
[0056] Weighted graph construction module: used to calculate the effective adjacency strength between rectangles and construct a weighted undirected connected graph that represents the topological adjacency strength graph;
[0057] Spectral embedding module: used to perform Laplacian feature mapping, transforming topological relationships into spectral space coordinates;
[0058] Weight discretization and allocation module: used to perform virtual node splitting and linear allocation solution, and output the initial task partitioning scheme;
[0059] The optimization and adjustment module is used to perform connectivity repair and gradient-based boundary load fine-tuning, and outputs the final region allocation scheme and coverage path.
[0060] A computer-readable storage medium having a computer program stored thereon, wherein when the computer program is executed by a processor, the steps of a topology-aware spectral clustering path planning method for multi-unmanned surface vessel cooperative coverage detection are disclosed.
[0061] A computer program product includes a computer program that, when executed by a processor, implements steps of a topology-aware spectral clustering path planning method for collaborative coverage detection of multiple unmanned surface vessels.
[0062] The beneficial effects of this invention are as follows:
[0063] This invention overcomes the limitations of traditional methods that only focus on indicators such as path length and coverage. By using topological clustering based on rectangular units, it maximizes the path compactness within sub-regions, ensuring that adjacent survey lines belong to the same operational unit. This provides a high-quality trajectory foundation for utilizing overlapping areas for sonar data fusion and compensating for edge imaging defects, thus helping to improve the coverage detection path quality of multi-UAV systems in underwater target search missions. Simultaneously, this invention achieves relatively accurate regional balanced allocation, ensuring that the mission area size and coverage path length of the UAVs tend to be consistent. Furthermore, this invention can adapt to different numbers of UAVs and complex environments, generating well-connected and compact operational zones in both open water and complex island / reef environments. Attached Figure Description
[0064] Figure 1 This is a flowchart illustrating the principle of the present invention;
[0065] Figure 2 This is a schematic diagram of the rectangular allocation unit in this invention;
[0066] Figure 3 This is the rectangular division result used in the embodiments of the present invention;
[0067] Figure 4 This is a simulation result diagram of graph partitioning and region allocation in an embodiment of the present invention;
[0068] Figure 5 This is a simulation result diagram of the coverage trajectory of multiple unmanned surface vessels in an embodiment of the present invention. Detailed Implementation
[0069] The present invention will now be further described with reference to the accompanying drawings.
[0070] A topology-aware spectral clustering path planning method, system, and storage medium for collaborative coverage detection using multiple unmanned surface vessels (USVs) abandons the traditional grid cell allocation mode and instead uses "rectangle decomposition results" as the basic operational unit. Simultaneously, combining the ideas of the STC method, connecting the internal grids of a rectangle forms a complete edge of the spanning tree, ensuring that a long, straight rectangle naturally contains a set of continuous reciprocating scan lines, thus maintaining the integrity of the rectangle and preserving the integrity of the local detection data. Based on this, the invention transforms the "maximizing spatial compactness" problem into a weighted graph partitioning problem in graph theory by quantifying the topological adjacency strength between rectangles, and utilizes a weight discretization strategy to solve the load balancing problem.
[0071] like Figure 1 As shown, the system involved in this invention includes a data input module, a weighted graph construction module, a spectral embedding module, a weight discretization and allocation module, and an optimization and adjustment module. Specifically, it includes the following steps:
[0072] Step 1: Cell construction based on rectangular decomposition.
[0073] This invention uses rectangular decomposition results As the basic unit for area allocation, rather than the traditional grid unit. Because side-scan sonar operations rely on reciprocating scans along long straight lines, a complete rectangular grid, when its centers are connected, forms an independent long straight side, which can be considered as a set of closely spaced survey lines with consistent orientation. For example... Figure 2 As shown, after the region allocation is completed, each sub-region constructs a minimum spanning tree based on the rectangle allocation results. At this point, the coverage path around the spanning tree will inevitably form a complete round-trip traversal within each rectangle. This complete round-trip path forms a long and continuous overlapping band within the rectangle and at the boundaries of adjacent rectangles, laying the foundation for multi-view sonar data fusion. The blue filled area represents the rectangle decomposition result; the dark blue box is used to distinguish different rectangles; the red solid circle represents the grid center; the red dashed lines represent the edges formed by connecting the grid centers within the rectangles; the red dashed lines are the connecting lines between different rectangles, thus forming the minimum spanning tree; the shaded area is the region where data fusion can be formed.
[0074] Step 2: Construct a topology-aware weighted adjacency graph.
[0075] (1) Let the weighted undirected graph be constructed as follows: Define node weights Rectangle The geometric area (i.e., the number of grid cells contained). Define the edge weights. Representation of rectangle and The topological adjacency strength between them.
[0076] (2) Subsequently, based on the consistency of the extension directions of adjacent rectangles, the topological adjacency strength is calculated by category. For any two adjacent rectangles and If rectangle and If the extension directions are consistent, the length of the shared boundary between the rectangles can be calculated directly. and make If the two rectangles are not geometrically adjacent, then .
[0077] If rectangle and If the extension directions are orthogonal, then the effective shared boundary length is used. :
[0078]
[0079] in, Represents a rectangle The corresponding connected component in the same direction, index depending on The direction of extension; Similarly.
[0080] Defined as a connected component All rectangles in the middle The sum of the lengths of the directly shared boundaries:
[0081]
[0082] in, Rectangle and belonging to connected components rectangle in The shared boundary length between them. The solution is similar.
[0083] (3) Final definition of edge weights Representation of rectangle and The compactness of the space after aggregation:
[0084]
[0085] This edge weight definition introduces a connectivity enhancement mechanism when orthogonally adjoining regions, ensuring that regions sharing long boundaries and extending in the same direction are aggregated first, thus preserving overlapping regions for data fusion to the greatest extent possible.
[0086] Step 3: Spectral space mapping.
[0087] The spectral clustering algorithm is used to capture the global topological structure and transform discrete topological relationships into continuous geometric coordinates.
[0088] (1) Calculate the normalized Laplace matrix:
[0089]
[0090] in, It is the identity matrix; It is a degree matrix; Let be the edge weight matrix.
[0091] (2) For the normalized Laplace matrix Perform feature decomposition and select the top features. The eigenvectors corresponding to the smallest eigenvalues constitute the spectral embedding matrix. .
[0092] (3) Each rectangle Mapped to A point in 3D space In this space, the closer the points are, the more tightly they are physically topologically connected, and the more suitable they are to be assigned to the same unmanned surface vessel.
[0093] Step 4: Weight discretization allocation strategy.
[0094] (1) Set the total number of virtual nodes Calculate the scaling factor:
[0095]
[0096] Among them, the summation term This is the area of all rectangles.
[0097] Each original rectangle node Split into virtual nodes :
[0098]
[0099] Among them, symbols This indicates rounding up to the nearest integer.
[0100] (2) All by Split virtual nodes They all inherit the coordinates of their parent node in spectral space. Then, the calculations for each unmanned surface vessel were performed. Load capacity constraints :
[0101]
[0102] in, This represents the total number of virtual nodes. Number of unmanned surface vessels; symbol Indicates rounding down; Assign an indicator variable to the remainder to ensure that the sum of the capacities of all unmanned surface vessels is exactly equal to the total number of nodes. .
[0103] At this point, the problem becomes... Virtual particles of a unit mass are uniformly distributed into Given a subregion, we need to minimize the spectral spatial distance between particles within that subregion. This problem is a linear sum distribution problem and can be solved directly using the Hungarian algorithm.
[0104] Step 5: Solve the linear and distributive problems.
[0105] (1) From the set of spectral space coordinates of all virtual nodes In the process, one is randomly selected as the first cluster center. For the subsequent... Each cluster of hearts Calculate all unselected nodes Shortest distance to the currently selected center set :
[0106]
[0107] Selecting makes The largest node is the first Cluster centers are selected until one is chosen. A cluster center is used to ensure the dispersion of the initial center in the topological space.
[0108] (2) Construct the cost matrix And iteratively solve, its elements Indicates the first In the nth iteration virtual nodes Assigned to the The cost of an unmanned surface vessel:
[0109]
[0110] in, For virtual nodes spectral space coordinates; For the first Each cluster of centers; As a connectivity penalty, if a node Belonging rectangle and the first If a subregion is topologically disconnected, then it is infinitely large.
[0111] (3) Construct and solve the linear sum allocation problem to find the optimal allocation matrix. ,in :
[0112]
[0113] (5) This problem can be solved directly using the Hungarian algorithm to obtain the optimal allocation matrix for the current iteration. Subsequently, according to Update the spectral embedding coordinates of each cluster center Then, the next round of iterations begins, continuing until the cluster center position converges or the maximum number of iterations is reached.
[0114] (5) Since the virtual nodes corresponding to the same rectangle are not necessarily completely assigned to the same sub-region, the original rectangles are finally counted according to the majority voting principle. Determine the ownership status of all included virtual nodes. The final sub-region :
[0115]
[0116] in, It is a rectangle The corresponding set of virtual nodes.
[0117] Step 6: Gradient-based boundary refinement.
[0118] After completing the above steps, a preliminary sub-region for multi-unmanned surface vessel (USV) missions is obtained, satisfying load balancing and possessing high spatial compactness. However, spectral partitioning inevitably leads to isolated components separated from the main body within some sub-regions. Furthermore, this invention proposes a gradient-based boundary refinement strategy to address this problem:
[0119] (1) First, check whether there are isolated island rectangles in each task sub-region. If so, reassign them to the adjacent sub-region with the longest shared boundary.
[0120] (2) Secondly, the load balancing after the rectangle movement in connectivity repair needs to be adjusted. This applies to leaf rectangles located at the boundaries of sub-regions. The calculation removes it from the current sub-region. Move to adjacent sub-region Load variance gain after :
[0121]
[0122] in, and Sub-regions and The sum of the areas of all rectangles within the current load; The target average load for all sub-regions.
[0123] like Furthermore, the movement operation does not damage the original area. If the connectivity is confirmed, the move operation is performed. This process simulates discrete gradient descent, which further converges the system load variance to a minimum, achieving a globally optimal balanced distribution.
[0124] (3) Finally, based on the rectangle allocation results, a minimum spanning tree is constructed between the rectangles in each sub-region. Then, a coverage path is generated around the spanning tree, and the Dubins algorithm is used to process the path to generate a smooth coverage trajectory.
[0125] This invention discloses a topology-sensing spectral clustering path planning system for collaborative coverage detection using multiple unmanned surface vessels, comprising:
[0126] Data input module: used to receive rectangular decomposition data and unmanned surface vessel quantity parameters;
[0127] Weighted graph construction module: used to calculate the effective adjacency strength between rectangles and construct a topology-aware weighted graph;
[0128] Spectral embedding module: used to perform Laplacian feature mapping, transforming topological relationships into spectral space coordinates;
[0129] Weight discretization and allocation module: used to perform virtual node splitting and linear allocation solution, and output the initial task partitioning scheme;
[0130] The optimization and adjustment module is used to perform connectivity repair and gradient-based boundary load fine-tuning, and outputs the final region allocation scheme and coverage path.
[0131] Implementation Case:
[0132] This case study uses a real map of the Changshan Islands in China, measuring 14.4 km × 27.2 km, with a resolution of 90 × 170. The minimum spanning tree level (rectangular partitioning level) has a grid size of 160m × 160m, and the coverage path level has a grid size of 80m. Simulated turning experiments determined the minimum turning radius of the unmanned surface vessels in this case study to be 20m, which was used to generate the Dubins trajectory.
[0133] First, we input the rectangular division result of the map, such as... Figure 3 As shown in the diagram. Red lines represent horizontal rectangles, and blue lines represent vertical rectangles. Each rectangle has a solid gray dot at its center. The adjacency between rectangles is represented by a connected graph composed of these points, with gray lines connecting the points indicating an adjacency relationship between two rectangles.
[0134] Based on the above solution steps, we obtain the following: Figure 4 The region allocation results are shown. (3D) Figure 4 (a) showed Figure 3 The image shows the rectangle partitioning results after inputting the rectangle partitioning results, along with the corresponding rectangle allocation results. Different colors represent different sub-regions. Figure 4 (b) shows the linear sum distribution of virtual nodes after weight discretization of the weighted connected graph. Each gray dashed envelope contains all virtual nodes corresponding to the current rectangle after weight discretization; the gray connections between different envelopes inherit the connectivity of the weighted connected graph.
[0135] Finally, based on the rectangle allocation results, a minimum spanning tree is constructed between the rectangles in each sub-region, and a Dubins coverage path (total coverage trajectory length is 3461.22 km) is generated around the spanning tree, as shown below. Figure 5 As shown in the figure. The filled area between the trajectories represents a long and continuous sensory fusion region formed between adjacent round trip traversal paths, with a total length of 3091.04 km.
[0136] In particular, in some preferred embodiments of the present invention, a computer device is also provided, including a memory and a processor and a computer program stored in the memory, wherein the processor executes the computer program to implement the steps of the topology-aware spectrum clustering path planning method for multi-unmanned surface vessel cooperative coverage detection described in any of the above embodiments.
[0137] In some other preferred embodiments of the present invention, a computer-readable storage medium is also provided, on which a computer program / instruction is stored, wherein when the computer program is executed by a processor, the steps of the topology-sensing spectrum clustering path planning method for multi-unmanned surface vessel cooperative coverage detection described in any of the above embodiments are implemented.
[0138] Those skilled in the art will understand that all or part of the processes in the above embodiments can be implemented by a computer program instructing related hardware. The computer program can be stored in a non-volatile computer-readable storage medium. When the computer program is executed, it can include the processes of the above embodiments of the topology sensing spectrum clustering path planning method for multi-unmanned surface vessel cooperative coverage detection, which will not be repeated here.
[0139] In the description of this specification, the references to terms such as "one embodiment," "some embodiments," "example," "specific example," or "some examples," etc., refer to specific features, structures, materials, or characteristics described in connection with that embodiment or example, which are included in at least one embodiment or example of the present invention. In this specification, the illustrative expressions of the above terms do not necessarily refer to the same embodiment or example. Furthermore, the specific features, structures, materials, or characteristics described may be combined in any suitable manner in one or more embodiments or examples. Moreover, without contradiction, those skilled in the art can combine and integrate the different embodiments or examples described in this specification, as well as the features of different embodiments or examples.
[0140] Furthermore, the terms "first" and "second" are used for descriptive purposes only and should not be construed as indicating or implying relative importance or implicitly specifying the number of indicated technical features. Thus, a feature defined as "first" or "second" may explicitly or implicitly include at least one of that feature. In the description of this invention, "N" means at least two, such as two, three, etc., unless otherwise explicitly specified.
[0141] Any process or method description in the flowchart or otherwise herein can be understood as representing a module, segment, or portion of code comprising one or more N executable instructions for implementing custom logic functions or processes, and the scope of preferred embodiments of the invention includes additional implementations in which functions may be performed not in the order shown or discussed, including substantially simultaneously or in reverse order depending on the functions involved, as should be understood by those skilled in the art to which embodiments of the invention pertain.
[0142] The logic and / or steps represented in the flowchart or otherwise described herein, for example, can be considered as a sequenced list of executable instructions for implementing logical functions, and can be embodied in any computer-readable medium for use by, or in conjunction with, an instruction execution system, apparatus, or device (such as a computer-based system, a processor-included system, or other system that can fetch and execute instructions from, an instruction execution system, apparatus, or device). For the purposes of this specification, "computer-readable medium" can be any means that can contain, store, communicate, propagate, or transmit programs for use by, or in conjunction with, an instruction execution system, apparatus, or device. More specific examples (a non-exhaustive list) of computer-readable media include: an electrical connection having one or more wires (electronic device), a portable computer disk drive (magnetic device), random access memory (RAM), read-only memory (ROM), erasable and editable read-only memory (EPROM or flash memory), fiber optic devices, and portable optical disc read-only memory (CDROM). Alternatively, the computer-readable medium may be paper or other suitable media on which the program can be printed, since the program can be obtained electronically, for example, by optically scanning the paper or other medium, followed by editing, interpreting, or otherwise processing as necessary, and then stored in a computer memory.
[0143] It should be understood that various parts of the present invention can be implemented in hardware, software, firmware, or a combination thereof. In the above embodiments, the N steps or methods can be implemented in software or firmware stored in memory and executed by a suitable instruction execution system. For example, if implemented in hardware as in another embodiment, it can be implemented using any one or a combination of the following techniques known in the art: discrete logic circuits having logic gates for implementing logical functions on data signals, application-specific integrated circuits (ASICs) having suitable combinational logic gates, programmable gate arrays (PGAs), field-programmable gate arrays (FPGAs), etc.
[0144] Those skilled in the art will understand that all or part of the steps of the methods in the above embodiments can be implemented by a program instructing related hardware. The program can be stored in a computer-readable storage medium, and when executed, the program includes one or a combination of the steps of the method embodiments.
[0145] Furthermore, the functional units in the various embodiments of the present invention can be integrated into a processing module, or each unit can exist physically separately, or two or more units can be integrated into a module. The integrated module can be implemented in hardware or as a software functional module. If the integrated module is implemented as a software functional module and sold or used as an independent product, it can also be stored in a computer-readable storage medium.
[0146] The storage medium mentioned above can be a read-only memory, a disk, or an optical disk, etc. Although embodiments of the present invention have been shown and described above, it is to be understood that the above embodiments are exemplary and should not be construed as limiting the present invention. Those skilled in the art can make changes, modifications, substitutions, and variations to the above embodiments within the scope of the present invention.
[0147] The above description is merely a preferred embodiment of the present invention and is not intended to limit the invention. Various modifications and variations can be made to the present invention by those skilled in the art. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the scope of protection of the present invention.
Claims
1. A topology-sensing spectrum clustering path planning method for collaborative coverage detection by multiple unmanned surface vessels, characterized in that, Includes the following steps: Step 1: Decompose the acquired task area into a set of rectangles as the basic task unit; Step 2: Using the area of the rectangle as the node weight, calculate the topological adjacency strength as the edge weight based on the consistency of the effective common boundary length and extension direction between adjacent rectangles, and construct a weighted undirected connected graph that represents the topological adjacency strength graph. Step 3: Calculate the Laplacian matrix and its eigenvectors of the weighted undirected connected graph, and map all rectangular nodes to the low-dimensional spectral space; Step 4: Based on the weight of each rectangular node, discretize each rectangular node into one or more virtual nodes with equal weight, and determine the load capacity constraint of each unmanned surface vessel based on the total number of virtual nodes and the number of unmanned surface vessels. Step 5: In the spectral space, using virtual nodes as units, solve the linear sum allocation problem that satisfies the load capacity constraint iteratively, assign the virtual nodes to the sub-regions corresponding to each unmanned surface vessel, and determine the sub-region to which each original rectangular node belongs based on the allocation results, thus obtaining the initial region allocation scheme; Step 6: Perform connectivity checks and repairs on the initial region allocation scheme, and fine-tune the rectangles at the boundaries of sub-regions based on a gradient-based boundary refinement strategy with load balancing as the goal, to obtain the final region allocation scheme and output the coverage path.
2. The topology-sensing spectrum clustering path planning method for multi-unmanned surface vessel collaborative coverage detection according to claim 1, characterized in that, Step 2, which calculates the topological adjacency strength as the edge weight based on the consistency between the effective common boundary length and the extension direction of adjacent rectangles, specifically includes: Calculate any two adjacent rectangles and Edge weights between , Among them, node weight Rectangle Geometric area; node weight Rectangle The geometric area; for and The effective shared boundary length, if and If the extension directions are consistent, then equal to rectangle and Geometric shared boundary length ;like and If the directions of extension are orthogonal, then The total length of the shared boundary between the connected components in the same direction to which one rectangle belongs and the other rectangle. , in, Represents a rectangle The direction of extension is The connected components, These represent the horizontal and vertical directions, respectively. Represents a rectangle The direction of extension is The connected components; For connected components All rectangles in the middle The sum of the directly shared boundary lengths; For connected components All rectangles in the middle The sum of the directly shared boundary lengths.
3. The topology-sensing spectrum clustering path planning method for multi-unmanned surface vessel cooperative coverage detection according to claim 2, characterized in that, Step 3 specifically includes, Calculate the normalized Laplacian matrix ,in, It is the identity matrix; It is a degree matrix; Let the edge weight matrix be the edge weight matrix. For the normalized Laplace matrix Perform feature decomposition. Let the total number of unmanned surface vessels be... Then it is necessary Regions are divided based on cluster centers. Based on the fundamental principles of spectral embedding, regions are selected... The former The eigenvectors corresponding to the smallest eigenvalues constitute the spectral embedding matrix. ,in, The total number of rectangles in the rectangle set input in step 1; Each rectangle Mapped to A point in 3D space In this space, the closer the points are, the more tightly they are physically topologically combined, and the more suitable they are to be assigned to the same unmanned surface vessel.
4. The topology-sensing spectrum clustering path planning method for multi-unmanned surface vessel cooperative coverage detection according to claim 3, characterized in that, Step 4 discretizes each rectangular node into one or more virtual nodes with equal weights, specifically including: Based on the total number of target virtual nodes Sum of the weights of all rectangular nodes Calculate scaling factor ,in The weights of the rectangle nodes; Each rectangle node Split into virtual nodes ,in ,symbol Indicates rounding up; each virtual node Inherit from its parent node coordinate vector in spectral space .
5. The topology-sensing spectrum clustering path planning method for multi-unmanned surface vessel cooperative coverage detection according to claim 4, characterized in that, The determination of the load capacity constraints for each unmanned surface vessel is specifically as follows: No. Payload capacity constraints of unmanned surface vessels ,in, This represents the total number of virtual nodes. Number of unmanned surface vessels; symbol Indicates rounding down; Assign an indicator variable to the remainder to ensure that the sum of the capacities of all unmanned surface vessels equals the total number of nodes. .
6. A topology-sensing spectrum clustering path planning method for collaborative coverage detection by multiple unmanned surface vessels as described in claim 4, characterized in that, Step 5 includes the following steps: (1) The cluster centers are initialized using the maximum distance dispersion strategy, where the first cluster center is the first cluster center. The cluster center corresponds to the first... Both subregions belong to the first... Unmanned surface vessel; (2) in the In the next iteration, the cost matrix is constructed. , among which, element Indicates the first virtual nodes Assigned to the The cost of an unmanned surface vessel: in, For virtual nodes spectral space coordinates; For the first Each cluster of centers; As a connectivity penalty, if a node Belonging rectangle and the first If a subregion is topologically disconnected, then it is infinitely large; (3) Solve the linear sum distribution problem and find the current optimal distribution matrix. ,in : (4) Update the spectral embedding coordinates of each cluster center according to the current allocation matrix, and repeat (2)-(3) until the cluster center positions converge or the maximum number of iterations is reached. (5) According to the majority voting principle, count the original rectangles. Determine the ownership status of all included virtual nodes. The final sub-region : in, It is a rectangle The corresponding set of virtual nodes.
7. The topology-sensing spectrum clustering path planning method for multi-unmanned surface vessel cooperative coverage detection according to claim 6, characterized in that, Step (1) uses a maximum distance dispersion strategy to initialize the cluster centers. The specific steps are as follows: From the set of spectral space coordinates of all virtual nodes In the process, one is randomly selected as the first cluster center. ; According to the Each cluster of hearts Calculate all unselected nodes Shortest distance to the currently selected center set : Selecting makes The largest node is the first Cluster centers are selected until one is chosen. Each cluster of hearts.
8. The topology-sensing spectrum clustering path planning method for multi-unmanned surface vessel cooperative coverage detection according to claim 1, characterized in that, Step 6, based on a gradient-based boundary refinement strategy, fine-tunes the rectangles at the boundaries of sub-regions with load balancing as the objective. Specifically, this includes: (1) Detect whether there are isolated island rectangles in each task sub-region. If so, reassign them to the adjacent sub-region with the longest shared boundary. (2) For leaf rectangles located at the boundary of a subregion Calculate and remove it from the current sub-region Move to adjacent sub-region Load variance gain after ,like Furthermore, the movement operation does not damage the original area. If the connectivity is such that the move operation is executed; in, and Sub-regions and The sum of the areas of all rectangles within the current load is the current load. The target average load for all sub-regions.
9. A topology-sensing spectrum clustering path planning system for collaborative coverage detection by multiple unmanned surface vessels, characterized in that, Includes a module for executing the processing instructions for each step of the topology-sensing spectral clustering path planning method for multi-unmanned surface vessel cooperative coverage detection as described in any one of claims 1-8. Data input module: used to receive rectangular decomposition data and unmanned surface vessel quantity parameters; Weighted graph construction module: used to calculate the effective adjacency strength between rectangles and construct a weighted undirected connected graph that represents the topological adjacency strength graph; Spectral embedding module: used to perform Laplacian feature mapping, transforming topological relationships into spectral space coordinates; Weight discretization and allocation module: used to perform virtual node splitting and linear allocation solution, and output the initial task partitioning scheme; The optimization and adjustment module is used to perform connectivity repair and gradient-based boundary load fine-tuning, and outputs the final region allocation scheme and coverage path.
10. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by a processor, it implements the steps of the method according to any one of claims 1 to 8.