Underwater robotic fish full-coverage path planning method and system based on multi-strategy collaborative optimization

Abstract

The invention discloses an underwater robotic fish full-coverage path planning method and system based on multi-strategy collaborative optimization, and belongs to the technical field of underwater robot path planning. Aiming at the problems of low path planning efficiency, poor dynamic environment adaptability and difficulty in algorithm parameter adjustment in the prior art, a dynamic grid coding and multi-resolution collaborative modeling method is provided, an obstacle probability region is divided through sonar point cloud data, the grid refining depth is dynamically adjusted in combination with a greedy optimization algorithm, and a multi-layer environment representation map is generated; an improved genetic algorithm framework is adopted, path fragments are recombined through self-adaptive variation rate adjustment and a three-point crossover operator, a conflict resolution mechanism is introduced to correct a countercurrent path direction, and a motion trail is optimized in combination with path smoothness constraint. According to the invention, high-efficiency and low-energy-consumption full-coverage path planning can be realized in a complex underwater environment, and the method is suitable for scenes of ocean exploration, pipeline inspection, ecological monitoring and the like.

Description

Technical Field

[0001] The present invention belongs to the technical field, and in particular relates to a method and system for full-coverage path planning of an underwater robotic fish based on multi-strategy collaborative optimization. Background Art

[0002] In existing technologies, full-coverage path planning for underwater robotic fish primarily relies on global static planning and dynamic environment switching. For example, existing technology 1 uses a modified Bayazit algorithm to segment regions of interest and combines it with multi-threaded optimization to generate coverage paths. However, this relies on high-performance hardware and suffers from low computational efficiency, making it difficult to handle the real-time processing of large-scale obstacle data in complex underwater environments. Existing technology 2 proposes a solution that switches between global and local planning based on real-time data. However, this approach lacks intelligent decision-making capabilities in dynamic environments and cannot adjust the path in real time based on changes in water flow or obstacle movement, resulting in poor adaptability of the planning results.

[0003] After analysis, the existing traditional methods generally have the following problems:

[0004] First, existing path planning methods often face the problem of huge computational complexity when dealing with complex or large-scale underwater environments, especially when a large amount of environmental data and obstacles need to be considered during the path planning process. These methods often require high computing configurations or a large amount of computing resources to solve the problem, resulting in a slow and inefficient path planning process. In practical applications, especially in complex underwater environments, the extension of task completion time not only wastes a lot of energy, but may also affect the effective execution of the task. Existing algorithms may not fully optimize these problems and still rely on high-performance hardware, unable to complete tasks with limited resources;

[0005] Secondly, most existing path planning algorithms are based on fixed rules and preset strategies, lacking sufficient flexibility and intelligent decision-making capabilities. These methods usually do not make intelligent adjustments based on the characteristics of the target area, changes in the environment, or the dynamic requirements of the task, which limits the adaptability and efficiency of path planning. For example, if certain areas in the underwater environment change, the existing algorithms may not be able to respond quickly or adjust the planned path to adapt to the new situation, resulting in low time efficiency and even possible failure to complete the task. Therefore, the path planning algorithms of the existing technology lack sufficient intelligent decision-making capabilities when facing a dynamically changing environment, and it is difficult to make real-time optimization according to different task requirements;

[0006] Third, some existing path planning methods (such as genetic algorithms) have significant limitations in parameter settings. Traditional genetic algorithms require manual adjustment of multiple parameters, but these parameters are often closely related to the specific task environment. Existing methods often lack flexibility when dealing with different environments and task requirements, making it difficult to dynamically adjust to actual conditions. This results in suboptimal path planning results. As task requirements change, if algorithm parameters cannot be optimized in a timely manner, the adaptability and effectiveness of path planning will be limited.

[0007] The root cause of these problems is that existing technologies fail to effectively balance environmental modeling accuracy and computing resource consumption, and lack adaptive parameter optimization mechanisms, resulting in limited path planning efficiency and reliability in dynamic environments. Summary of the Invention

[0008] In order to solve the above technical problems, the present invention proposes a full-coverage path planning method and system for underwater robotic fish based on multi-strategy collaborative optimization to solve the problems existing in the above-mentioned prior art.

[0009] In a first aspect, to achieve the above-mentioned objectives, the present invention provides a method for full-coverage path planning of an underwater robotic fish based on multi-strategy collaborative optimization, comprising the following steps:

[0010] Construct an initial two-dimensional grid matrix based on the minimum detection distance of the robotic fish sensor and establish a global coordinate system;

[0011] The obstacle coverage probability of each grid cell is calculated using sonar point cloud data, and the grid cell state is divided into free area and complete obstacle area;

[0012] With available memory as the constraint, a greedy optimization algorithm is used to dynamically adjust the grid refinement depth of each region to generate a multi-layer grid map including a basic grid matrix and a quadtree index.

[0013] Initialize the population of the genetic algorithm, dynamically adjust the mutation probability according to the average fitness of the population, and retain the individuals with the highest fitness to directly enter the next generation population;

[0014] Randomly select three crossover points on the parent chromosome to divide the fragments and exchange them to generate daughter chromosomes;

[0015] Reversely adjust or regenerate the path segments in the daughter chromosomes that conflict with the water flow direction, and reduce the number of turns through path smoothing;

[0016] The mutation rate is dynamically adjusted according to the mean and standard deviation of the population fitness, and the mutation operation is performed in combination with the path smoothness constraint and the water flow compliance constraint.

[0017] Optionally, the process of calculating the obstacle coverage probability of each grid cell using sonar point cloud data includes:

[0018] According to the distribution of sonar point clouds collected by sensors, the obstacle reflection signal strength in each grid unit is counted;

[0019] Map the signal strength to a probability value and set a probability threshold to distinguish between free areas and completely obstructed areas;

[0020] After updating the grid state, probabilistically dynamically encoded environment representation data is generated.

[0021] Optionally, the process of dynamically adjusting the grid refinement depth of each region using the greedy optimization algorithm includes:

[0022] Set the maximum number of refinement levels based on available memory capacity;

[0023] Prioritize the grid areas with obstacle probability in the middle range for refinement;

[0024] The refinement depth is iteratively adjusted until the memory limit is reached or the coverage reaches a preset threshold.

[0025] Optionally, the process of dynamically adjusting the mutation probability includes:

[0026] Calculate the mean and standard deviation of the current population fitness;

[0027] Dynamically adjust the scaling factor of the initial mutation rate based on the ratio of the mean to the standard deviation;

[0028] When the mean population fitness decreases or the standard deviation increases, increase the mutation rate to enhance global search.

[0029] Optionally, the process of dividing the parent chromosome into segments and exchanging them includes:

[0030] Three crossover points are randomly selected to split the parent chromosome into four segments;

[0031] Swap the middle two segments of two parent chromosomes to generate daughter chromosomes;

[0032] Check the continuity of path points in the offspring chromosomes and delete duplicate nodes.

[0033] Optionally, the process of reversely adjusting the conflicting path segments includes:

[0034] Detect the movement direction in the child path segment whose angle with the water flow direction exceeds the threshold;

[0035] Adjust the direction of the upstream path segment to the downstream direction;

[0036] The adjusted path is smoothed by spline interpolation to eliminate sharp turning points.

[0037] In a second aspect, the present invention further provides a full-coverage path planning system for underwater robotic fish based on multi-strategy collaborative optimization, which is used to implement a full-coverage path planning method for underwater robotic fish based on multi-strategy collaborative optimization. The system includes:

[0038] The environmental modeling module is used to construct an initial two-dimensional grid matrix and establish a global coordinate system based on the minimum detection distance of the robotic fish sensor. It calculates the obstacle coverage probability of each grid cell using sonar point cloud data, divides the free area into a completely obstructed area, and dynamically adjusts the grid refinement depth based on available memory constraints to generate a multi-layer grid map.

[0039] The path planning module is used to initialize the population of the genetic algorithm, dynamically adjust the mutation probability according to the average fitness of the population, retain the individuals with the highest fitness to enter the next generation population, perform a three-point crossover operation on the parent chromosome to generate the child chromosome, and adjust the direction and smooth the conflicting path segments;

[0040] The dynamic optimization module is used to dynamically adjust the mutation rate according to the mean and standard deviation of the population fitness, perform mutation operations in combination with the path smoothness constraint and the water flow compliance constraint, and output the final coverage path.

[0041] Optionally, the environment modeling module includes:

[0042] A grid division unit is used to automatically calculate the basic grid size and construct an initial two-dimensional grid matrix according to the minimum detection distance of the sensor;

[0043] The probability coding unit is used to count the obstacle reflection signal strength of each grid cell in the sonar point cloud data, map it into a probability value and set a threshold to distinguish the area type;

[0044] A dynamic refinement unit is used to iteratively adjust the grid refinement depth based on a greedy optimization algorithm, prioritizing areas in the middle range of obstacle probability until the memory limit or coverage threshold is reached.

[0045] In a third aspect, the present invention further provides a computer terminal device, comprising:

[0046] one or more processors;

[0047] a memory, coupled to the processor, for storing one or more programs;

[0048] When the one or more programs are executed by the one or more processors, the one or more processors implement a full-coverage path planning method for underwater robotic fish based on multi-strategy collaborative optimization.

[0049] In a fourth aspect, the present invention also provides a computer-readable storage medium having a computer program stored thereon. When the computer program is executed by a processor, it implements a method for full-coverage path planning of an underwater robot fish based on multi-strategy collaborative optimization.

[0050] Compared with the prior art, the present invention has the following advantages and technical effects:

[0051] This invention provides a full-coverage path planning method and system for underwater robotic fish based on multi-strategy collaborative optimization. Through dynamic grid encoding and a multi-resolution collaborative mechanism, this invention significantly improves the efficiency and accuracy of complex underwater environment modeling, achieving a balance between environmental characterization and computational load under limited hardware resources. The improved multi-strategy genetic algorithm framework integrates an adaptive mutation rate, a three-point crossover operator, and a dynamic constraint mechanism to enhance the global search capability and local optimization effect of path planning, effectively reducing redundant path turns and countercurrent energy consumption. Path smoothing based on conflict resolution and water flow compliance constraints further ensures path feasibility and motion efficiency. Compared to traditional methods, this solution offers significant advantages in dynamic environmental adaptability, algorithm parameter self-optimization capabilities, and resource consumption control. BRIEF DESCRIPTION OF THE DRAWINGS

[0052] The accompanying drawings, which constitute part of the present invention, are provided to provide a further understanding of the present invention. The exemplary embodiments of the present invention and their descriptions are provided to explain the present invention and do not constitute an undue limitation of the present invention. In the accompanying drawings:

[0053] Figure 1 A schematic diagram of task area modeling according to an embodiment of the present invention;

[0054] Figure 2 A schematic diagram of the optimization genetic algorithm flow in an embodiment of the present invention;

[0055] Figure 3 A schematic diagram of an optimized genetic algorithm according to an embodiment of the present invention;

[0056] Figure 4 A schematic diagram of randomly selecting three intersections according to an embodiment of the present invention;

[0057] Figure 5 A schematic diagram of segment division according to an embodiment of the present invention;

[0058] Figure 6 A schematic diagram of an exchange segment according to an embodiment of the present invention;

[0059] Figure 7 This is a schematic diagram of conflict resolution according to an embodiment of the present invention;

[0060] Figure 8Schematic diagram of the dynamic interval adaptive mutation strategy of an embodiment of the present invention, wherein (a) is a schematic diagram of the randomly selected mutation interval, (b) is a schematic diagram of the mutation operator in the early stage of iteration, and (c) is a schematic diagram of the mutation operator in the late stage of iteration;

[0061] Figure 9 DEGA coverage diagram of an embodiment of the present invention;

[0062] Figure 10 This is a GA coverage map of an embodiment of the present invention. DETAILED DESCRIPTION

[0063] It should be noted that, in the absence of conflict, the embodiments and features of the embodiments of the present invention can be combined with each other. The present invention will be described in detail below with reference to the accompanying drawings and in combination with the embodiments.

[0064] It should be noted that the steps shown in the flowcharts of the accompanying drawings can be executed in a computer system such as a set of computer-executable instructions, and that, although a logical order is shown in the flowcharts, in some cases, the steps shown or described can be executed in an order different from that shown here.

[0065] Example 1

[0066] This embodiment provides a full-coverage path planning method for underwater robotic fish based on multi-strategy collaborative optimization, including:

[0067] Construct an initial two-dimensional grid matrix based on the minimum detection distance of the robotic fish sensor and establish a global coordinate system;

[0068] The obstacle coverage probability of each grid cell is calculated using sonar point cloud data, and the grid cell state is divided into free area and complete obstacle area;

[0069] With available memory as the constraint, a greedy optimization algorithm is used to dynamically adjust the grid refinement depth of each region to generate a multi-layer grid map including a basic grid matrix and a quadtree index.

[0070] Initialize the population of the genetic algorithm, dynamically adjust the mutation probability according to the average fitness of the population, and retain the individuals with the highest fitness to directly enter the next generation population;

[0071] Randomly select three crossover points on the parent chromosome to divide the fragments and exchange them to generate daughter chromosomes;

[0072] Reversely adjust or regenerate the path segments in the daughter chromosomes that conflict with the water flow direction, and reduce the number of turns through path smoothing;

[0073] The mutation rate is dynamically adjusted according to the mean and standard deviation of the population fitness, and the mutation operation is performed in combination with the path smoothness constraint and the water flow compliance constraint.

[0074] As an implementation method of this embodiment, the process of calculating the obstacle coverage probability of each grid cell using sonar point cloud data includes:

[0075] According to the distribution of sonar point clouds collected by sensors, the obstacle reflection signal strength in each grid unit is counted;

[0076] Map the signal strength to a probability value and set a probability threshold to distinguish between free areas and completely obstructed areas;

[0077] After updating the grid state, probabilistically dynamically encoded environment representation data is generated.

[0078] Specifically, the specific process of calculating the obstacle coverage probability includes:

[0079] This paper proposes an underwater environment modeling method based on multi-resolution collaborative rasterization, which achieves the coordinated optimization of environmental representation accuracy and computational efficiency through a dynamic layering mechanism. Figure 1 As shown in the figure, the basic grid size is automatically calculated based on the characteristics of the robotic fish sensor (minimum detection distance d_min), an initial two-dimensional grid matrix is ​​constructed, and a global coordinate system is established. A probabilistic dynamic encoding mechanism is then used to calculate the obstacle coverage probability of each grid cell using sonar point cloud data, subdividing the cell status into two categories: free area (Type = 0) and complete obstacle (Type = 1). To ensure real-time performance, a greedy optimization algorithm is established with available memory as a constraint, dynamically adjusting the refinement depth of each region to maximize environmental coverage within hardware resource constraints. The final output is a multi-layer grid map consisting of a basic grid matrix and a detailed quadtree index. This supports rapid topological queries and risk quantification assessment in the subsequent path planning module, providing an accurate representation of the environmental state for multi-objective optimization.

[0080] As an implementation method of this embodiment, the process of dynamically adjusting the grid refinement depth of each region using the greedy optimization algorithm includes:

[0081] Set the maximum number of refinement levels based on available memory capacity;

[0082] Prioritize the grid areas with obstacle probability in the middle range for refinement;

[0083] The refinement depth is iteratively adjusted until the memory limit is reached or the coverage reaches a preset threshold.

[0084] As an implementation method of this embodiment, the process of dynamically adjusting the mutation probability includes:

[0085] Calculate the mean and standard deviation of the current population fitness;

[0086] Dynamically adjust the scaling factor of the initial mutation rate based on the ratio of the mean to the standard deviation;

[0087] When the mean population fitness decreases or the standard deviation increases, increase the mutation rate to enhance global search.

[0088] Specifically, the specific process includes: Figure 2 、 Figure 3 As shown in the figure, traditional genetic algorithms can suffer from slow convergence and easy trapping in local optima when planning full-coverage paths for underwater robotic fish. To address this, a collaborative optimization framework integrating multiple strategies is proposed. By introducing an adaptive mutation rate, a three-point crossover operator, and a dynamic weight adjustment mechanism, it aims to improve the efficiency of path planning and the quality of the solution.

[0089] 1) Adaptive mutation rate: Dynamically adjust the mutation probability according to the diversity and fitness of the population. Set the mutation probability p m Average fitness value The specific relationship is:

[0090]

[0091] Among them, p m0 is the initial mutation probability, f max This strategy increases the mutation rate when the population fitness is low, promoting global search; it reduces the mutation rate when the fitness increases, strengthening the use of excellent solutions.

[0092] 2) Elite retention strategy: In each generation of inheritance, the individuals with the highest fitness are retained to ensure that excellent genes are passed on to the next generation. The specific operations are:

[0093] Fitness calculation: Calculate the fitness value f of each individual in the population i .

[0094] Elite selection: select the top N with the highest fitness values e Individuals directly enter the next generation population.

[0095] Genetic operations: The remaining individuals are generated through selection, crossover, and mutation and enter the next generation population.

[0096] As an implementation method in this embodiment, the process of dividing the parent chromosome into segments and performing the exchange includes:

[0097] Three crossover points are randomly selected to split the parent chromosome into four segments;

[0098] Swap the middle two segments of two parent chromosomes to generate daughter chromosomes;

[0099] Check the continuity of path points in the offspring chromosomes and delete duplicate nodes.

[0100] Specifically, the specific process includes:

[0101] 3) Improved crossover operator (three-point crossover): Three-point crossover is a multi-point crossover method that can more fully explore the gene space compared to traditional single-point or two-point crossover. Specifically, three-point crossover first randomly selects three crossover points, then divides the parent chromosome into multiple segments based on these three points, and then exchanges these segments to generate offspring individuals. Therefore, in order to improve the effect and quality of path planning, the present invention proposes a three-point crossover operator that enhances the exchange and recombination capabilities of genes, improves the exploratory nature of the solution space, and avoids invalid turns and over-optimization problems in path planning. The specific operations are:

[0102] Select the intersection point: Figure 4 As shown in Figure 1, in each generation, three crossover points c1, c2, and c3 are randomly selected from the parent chromosome. These crossover points divide the chromosome into four segments. The choice of crossover points affects the effectiveness of the crossover operation, so the crossover points should be distributed as evenly as possible to generate diverse paths across the entire path space.

[0103] Divide into segments: Divide the parent chromosome into four segments using the selected crossover points, e.g. Figure 5 shown.

[0104] Exchange fragments: such as Figure 6 As shown, the segments of the two parent chromosomes are crossed and exchanged to generate two daughter chromosomes. Assuming that the parent chromosome 1 and the parent chromosome 2 are P1 and P2 respectively, the crossover operation generates two daughter chromosomes S1 and S2:

[0105] This step ensures that after crossover, each offspring retains multiple pieces of information from its parent, thereby enhancing the diversity of knowledge and the exploration of the optimization space.

[0106] As an implementation manner in this embodiment, the process of reversely adjusting the conflicting path segments includes:

[0107] Detect the movement direction in the child path segment whose angle with the water flow direction exceeds the threshold;

[0108] Adjust the direction of the upstream path segment to the downstream direction;

[0109] The adjusted path is smoothed by spline interpolation to eliminate sharp turning points.

[0110] Specifically, the specific process of conflict resolution includes: Figure 7As shown in the figure, for the path segments in the daughter chromosomes, there may be path duplication or inappropriateness. In particular, in underwater path planning, the kinematic constraints of the underwater robotic fish need to be considered. In order to avoid generating invalid paths (such as countercurrent paths or cross paths), the daughter chromosomes need to undergo conflict resolution. For each path segment, if the path direction is opposite to the water flow direction, it is necessary to make a reverse adjustment or regenerate the path segment to ensure that the path follows the water flow direction and reduce energy consumption. By smoothing the path, the sharp turns in the path are reduced, thereby improving the movement efficiency of the robotic fish.

[0111] 4) Dynamic interval adaptive mutation strategy: Mutation operation is a key step to promote population diversity and avoid falling into local optimality. Traditional mutation strategies usually use a fixed mutation rate, which may lead to two problems: first, the mutation rate is too low, which cannot effectively explore the solution space and causes the algorithm to fall into local optimality; second, the mutation rate is too high, which may destroy the excellent individuals in the current population and cause the algorithm to converge slowly. Therefore, if Figure 8 As shown in the figure, a dynamic interval adaptive mutation strategy is proposed, which aims to automatically adjust the mutation rate and the weight of target optimization according to the fitness change of the population, thereby enhancing the global search ability of the algorithm and improving the convergence efficiency. The adjustment principle is:

[0112] Average value of population fitness: When the fitness value of a population is low, it means that the quality of the solution of the current population is poor, and the mutation rate should be increased to explore more of the solution space.

[0113] Standard deviation of population fitness: The standard deviation reflects the diversity of the population. A large standard deviation indicates that the solutions in the population are still diverse, and the mutation rate can be moderately reduced. A small standard deviation indicates that the population is converging, and the mutation rate should be increased to avoid falling into a local optimum.

[0114] Based on these factors, the adjustment of the mutation rate can be expressed by the following formula:

[0115]

[0116] Among them, p m (t) is the mutation rate of the current generation t, p m0 is the initial mutation rate, σ f (t) is the standard deviation of the current population fitness, is the average value of the current population fitness, and σ is the adjustment factor that controls the impact of the standard deviation on the mutation rate.

[0117] The steps are:

[0118] Fitness calculation: Calculate the fitness value f of all individuals in the current population i , and calculate the average fitness and standard deviation of the population according to the following formula:

[0119]

[0120] Where N is the population size, f i is the fitness of individual i.

[0121] Adjust the mutation rate: according to the average fitness of the current population and standard deviation σ f (t), use the above formula to adjust the mutation rate p m (t) When the population fitness is low and the standard deviation is large, the mutation rate increases to promote more exploration of the solution space; when the population fitness is high and the standard deviation is small, the mutation rate decreases to maintain excellent solutions in the population.

[0122] Perform mutation operation: according to the dynamically adjusted mutation rate p m (t) Perform mutation operations on the selected individuals. Mutation can include genetic changes (such as exchanging path points, changing path directions, etc.) to explore new solution spaces.

[0123] In traditional genetic algorithms, mutation operations are usually performed by randomly selecting individuals and performing random transformations on them. However, simple random mutations may lead to fluctuations in path quality, and specific constraints (such as obstacles in the underwater environment, water flow direction, etc.) need to be considered. Therefore, the present invention incorporates a constraint processing mechanism for path planning problems into the mutation operation, making the mutation operation more reasonable:

[0124] Path smoothness constraint: When performing mutation, if the mutated path has too many sharp turns, the path is corrected using the smoothing algorithm Spline interpolation to ensure that the mutated path is kinematically feasible.

[0125] Water flow compliance constraints: If the mutated path goes against the current or becomes unable to follow the current, the system will automatically adjust the path to align with the water flow as much as possible. By adjusting the path direction, the impact of the water flow on the robot fish is reduced, improving the feasibility and efficiency of the path.

[0126] The mutated path cost function can be evaluated by the following formula:

[0127] Cost(S)=w1×Length(S)+w2×Smoothness(S)+w3×Energy(S)

[0128] Among them, S is the mutated path, w1, w2, and w3 are weight coefficients, and Length(S), Smoothness(S), and Energy(S) represent the length, smoothness, and energy consumption of the path, respectively.

[0129] To verify the effectiveness and superiority of the proposed multi-strategy collaborative optimization-based full-coverage path planning method for underwater robotic fish, a series of simulation experiments were conducted. The core goal of the experiments was to ensure that the path planning effectively covered the entire mission area while optimizing the path, reducing energy consumption, and avoiding ineffective turns. The algorithm's optimization level was demonstrated by comparing experimental data.

[0130] In the simulation experiment, it is necessary to focus on comparing the following data indicators:

[0131] Path coverage: Calculates the percentage of the area covered by the planned path, ideally 100%.

[0132]

[0133] A total Represents the total number of grid cells in the task area, A cov Indicates the number of grid cells covered during path planning.

[0134] Path Length: The total length of the path from the start point to the end point. A shorter path means more efficient coverage.

[0135]

[0136] Assume that the path consists of N consecutive path points P1, P2, ..., P N Composition, where each point P i =(x i ,y i ), the path length can be defined as the sum of the distances between adjacent points.

[0137] Energy Consumption: Calculates the total energy consumption of a path. The energy consumption of each unit of the path is related to the path length, the direction of the water flow, and the curvature of the path. Considering the energy consumption of an underwater robotic fish during movement, energy consumption is generally related to the path length, the degree of turning, and whether it complies with the water flow.

[0138]

[0139] L i represents the length of path segment i, θ i is the motion direction of the path segment, θ flow For the water flow direction of this section, the adjustment coefficient γ is introduced to characterize the impact of water flow on energy consumption. It is assumed that the angle consumption is proportional to the size of the angle difference, and its proportional coefficient is recorded as β.

[0140] Smoothness: This measures the smoothness of a path. The fewer turns a path has, the higher its smoothness. This can be assessed by calculating the total curvature of the path.

[0141]

[0142] Among them, K i A larger value indicates that the path changes drastically, thus reducing the path smoothness. A smaller Smoothness value indicates a smoother path.

[0143] As shown in Table 1, Figure 9 、 Figure 10 The following are the results and schematic diagram of the simulation experiment.

[0144] Table 1

[0145]

[0146] Based on this, an embodiment of the present invention provides a full-coverage path planning method for underwater robotic fish based on multi-strategy collaborative optimization. The present invention significantly improves the efficiency and accuracy of complex underwater environment modeling through dynamic grid coding and multi-resolution collaborative mechanisms, achieving a balance between environmental characterization and computational load under limited hardware resources. The improved multi-strategy genetic algorithm framework integrates adaptive mutation rate, three-point crossover operator and dynamic constraint mechanism to enhance the global search capability and local optimization effect of path planning, effectively reducing redundant path steering and countercurrent energy consumption. Path smoothing processing based on conflict resolution and water flow compliance constraints further ensures the feasibility and motion efficiency of the path. Compared with traditional methods, this solution has significant advantages in dynamic environment adaptability, algorithm parameter self-optimization capability and resource consumption control.

[0147] Example 2

[0148] In this embodiment, a computer terminal device is provided, including:

[0149] one or more processors;

[0150] a memory, coupled to the processor, for storing one or more programs;

[0151] When the one or more programs are executed by the one or more processors, the one or more processors implement the methods in the above embodiments.

[0152] In this embodiment, a computer-readable storage medium is further provided, on which a computer program is stored. When the computer program is executed by a processor, the method in the above embodiment is implemented.

[0153] In this embodiment, an electronic device is further provided, including a memory and a processor. The memory stores a computer program, and the processor is configured to run the computer program to execute the method in the above embodiment.

[0154] The above program can be run in the processor, or it can be stored in the memory (or computer-readable medium), which includes permanent and non-permanent, removable and non-removable media and can be implemented by any method or technology to store information. The information can be computer-readable instructions, data structures, program modules or other data. Examples of computer storage media include, but are not limited to, phase change memory (PRAM), static random access memory (SRAM), dynamic random access memory (DRAM), other types of random access memory (RAM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), flash memory or other memory technology, read-only compact disc read-only memory (CD-ROM), digital versatile disc (DVD) or other optical storage, magnetic cassettes, tape disk storage or other magnetic storage devices or any other non-transmission media that can be used to store information that can be accessed by a computing device.

[0155] These computer programs can also be loaded onto a computer or other programmable data processing device so that a series of operational steps are executed on the computer or other programmable device to produce a computer-implemented process, thereby providing instructions for executing on the computer or other programmable device to implement the process. Figure 1 a process or multiple processes and / or boxes Figure 1 The steps of the functions specified in one or more blocks can be implemented by different modules corresponding to different steps.

[0156] This embodiment provides such a device or system. The system is called an underwater robotic fish full coverage path planning system based on multi-strategy collaborative optimization, and includes:

[0157] The environmental modeling module is used to construct an initial two-dimensional grid matrix and establish a global coordinate system based on the minimum detection distance of the robotic fish sensor. It calculates the obstacle coverage probability of each grid cell using sonar point cloud data, divides the free area into a completely obstructed area, and dynamically adjusts the grid refinement depth based on available memory constraints to generate a multi-layer grid map.

[0158] The path planning module is used to initialize the population of the genetic algorithm, dynamically adjust the mutation probability according to the average fitness of the population, retain the individuals with the highest fitness to enter the next generation population, perform a three-point crossover operation on the parent chromosome to generate the child chromosome, and adjust the direction and smooth the conflicting path segments;

[0159] The dynamic optimization module is used to dynamically adjust the mutation rate according to the mean and standard deviation of the population fitness, perform mutation operations in combination with the path smoothness constraint and the water flow compliance constraint, and output the final coverage path.

[0160] As an implementation method in this embodiment, the environment modeling module includes:

[0161] A grid division unit is used to automatically calculate the basic grid size and construct an initial two-dimensional grid matrix according to the minimum detection distance of the sensor;

[0162] The probability coding unit is used to count the obstacle reflection signal strength of each grid cell in the sonar point cloud data, map it into a probability value and set a threshold to distinguish the area type;

[0163] A dynamic refinement unit is used to iteratively adjust the grid refinement depth based on a greedy optimization algorithm, prioritizing areas in the middle range of obstacle probability until the memory limit or coverage threshold is reached.

[0164] As an implementation method of this embodiment, the path planning module includes:

[0165] Population initialization unit, used to randomly generate the initial population and calculate individual fitness;

[0166] The three-point crossover unit is used to randomly select three crossover points to split the parent chromosome, exchange the two middle segments to generate the daughter chromosome, and check the continuity of the path points;

[0167] The conflict resolution unit is used to detect the countercurrent path in the child path segment that conflicts with the water flow direction, adjust the direction and perform spline interpolation smoothing.

[0168] As an implementation method of this embodiment, the dynamic optimization module includes:

[0169] Fitness analysis unit, used to calculate the average and standard deviation of the fitness of the current population;

[0170] A mutation rate adjustment unit is used to dynamically scale the initial mutation rate based on the ratio of the mean to the standard deviation, increasing the mutation rate when the fitness decreases or the standard deviation increases;

[0171] The constrained mutation unit is used to introduce path smoothness constraints and water flow compliance constraints in the mutation operation to eliminate invalid turns and countercurrent paths.

[0172] As an implementation manner in this embodiment, the three-point intersection unit further includes:

[0173] The crossover point generation subunit is used to randomly and evenly distribute three crossover points in the parent chromosome;

[0174] The segmental recombination subunit is used to exchange the middle two segments of the parent chromosome and delete the duplicate nodes;

[0175] The path verification subunit is used to verify the continuity and kinematic feasibility of the offspring path.

[0176] As an implementation manner in this embodiment, the conflict resolution unit further includes:

[0177] a countercurrent detection subunit, for identifying the direction of motion in a path segment whose angle with the water flow direction exceeds a preset threshold;

[0178] A direction adjustment subunit, used to correct the direction of the upstream path segment to the downstream direction;

[0179] The interpolation smoothing subunit is used to eliminate sharp turning points in the path through spline interpolation.

[0180] The system or device is used to implement the functions of the method in the above-mentioned embodiment. Each module in the system or device corresponds to each step in the method, which has been explained in the method and will not be repeated here.

[0181] Through the above implementation, the problem of full coverage path planning of underwater robot fish based on multi-strategy collaborative optimization in the related technology is solved, thereby ensuring that the problems existing in the existing technology are solved.

[0182] The above are merely preferred embodiments of the present invention, but the scope of protection of the present invention is not limited thereto. Any changes or substitutions that can be easily conceived by a person skilled in the art within the technical scope disclosed in the present invention should be included in the scope of protection of the present invention. Therefore, the scope of protection of the present invention should be based on the scope of protection of the claims.

Claims

1. A full-coverage path planning method for underwater robotic fish based on multi-strategy collaborative optimization, characterized in that: The following steps are involved: Construct an initial two-dimensional grid matrix based on the minimum detection distance of the robotic fish sensor and establish a global coordinate system; The obstacle coverage probability of each grid cell is calculated using sonar point cloud data, and the grid cell state is divided into free area and complete obstacle area; With available memory as the constraint, a greedy optimization algorithm is used to dynamically adjust the grid refinement depth of each region to generate a multi-layer grid map including a basic grid matrix and a quadtree index. Initialize the population of the genetic algorithm, dynamically adjust the mutation probability according to the average fitness of the population, and retain the individuals with the highest fitness to directly enter the next generation population; Randomly select three crossover points on the parent chromosome to divide the fragments and exchange them to generate daughter chromosomes; Reversely adjust or regenerate the path segments in the daughter chromosomes that conflict with the water flow direction, and reduce the number of turns through path smoothing; The mutation rate is dynamically adjusted according to the mean and standard deviation of the population fitness, and the mutation operation is performed in combination with the path smoothness constraint and the water flow compliance constraint.

2. The method according to claim 1, characterized in that The process of calculating the obstacle coverage probability of each grid cell using sonar point cloud data includes: According to the distribution of sonar point clouds collected by sensors, the obstacle reflection signal strength in each grid unit is counted; Map the signal strength to a probability value and set a probability threshold to distinguish between free areas and completely obstructed areas; After updating the grid state, probabilistically dynamically encoded environment representation data is generated.

3. The method according to claim 1, characterized in that The process of dynamically adjusting the grid refinement depth of each region using the greedy optimization algorithm includes: Set the maximum number of refinement levels based on available memory capacity; Prioritize the grid areas with obstacle probability in the middle range for refinement; The refinement depth is iteratively adjusted until the memory limit is reached or the coverage reaches a preset threshold.

4. The method according to claim 1, wherein The process of dynamically adjusting the mutation probability includes: Calculate the mean and standard deviation of the current population fitness; Dynamically adjust the scaling factor of the initial mutation rate based on the ratio of the mean to the standard deviation; When the mean population fitness decreases or the standard deviation increases, increase the mutation rate to enhance global search.

5. The method according to claim 1, wherein The process of dividing the parent chromosome into segments and exchanging them includes: Three crossover points are randomly selected to split the parent chromosome into four segments; Swap the middle two segments of two parent chromosomes to generate daughter chromosomes; Check the continuity of path points in the offspring chromosomes and delete duplicate nodes.

6. The method according to claim 1, characterized in that The process of reversely adjusting the conflicting path segments includes: Detect the movement direction in the child path segment whose angle with the water flow direction exceeds the threshold; Adjust the direction of the upstream path segment to the downstream direction; The adjusted path is smoothed by spline interpolation to eliminate sharp turning points.

7. A full-coverage path planning system for underwater robotic fish based on multi-strategy collaborative optimization, characterized by: The system comprises: The environmental modeling module is used to construct an initial two-dimensional grid matrix and establish a global coordinate system based on the minimum detection distance of the robotic fish sensor. It calculates the obstacle coverage probability of each grid cell using sonar point cloud data, divides the free area into a completely obstructed area, and dynamically adjusts the grid refinement depth based on available memory constraints to generate a multi-layer grid map. The path planning module is used to initialize the population of the genetic algorithm, dynamically adjust the mutation probability according to the average fitness of the population, retain the individuals with the highest fitness to enter the next generation population, perform a three-point crossover operation on the parent chromosome to generate the child chromosome, and adjust the direction and smooth the conflicting path segments; The dynamic optimization module is used to dynamically adjust the mutation rate according to the mean and standard deviation of the population fitness, perform mutation operations in combination with the path smoothness constraint and the water flow compliance constraint, and output the final coverage path.

8. The system according to claim 7, characterized in that The environment modeling module includes: A grid division unit is used to automatically calculate the basic grid size and construct an initial two-dimensional grid matrix according to the minimum detection distance of the sensor; The probability coding unit is used to count the obstacle reflection signal strength of each grid cell in the sonar point cloud data, map it into a probability value and set a threshold to distinguish the area type; A dynamic refinement unit is used to iteratively adjust the grid refinement depth based on a greedy optimization algorithm, prioritizing areas in the middle range of obstacle probability until the memory limit or coverage threshold is reached.

9. A computer terminal device, characterized in that: include: one or more processors; a memory, coupled to the processor, for storing one or more programs; When the one or more programs are executed by the one or more processors, the one or more processors implement the underwater robotic fish full coverage path planning method based on multi-strategy collaborative optimization as described in any one of claims 1 to 6.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the method for full-coverage path planning of an underwater robotic fish based on multi-strategy collaborative optimization as described in any one of claims 1 to 6 is implemented.

Images