A robot path planning method based on multi-strategy improved walrus optimization algorithm

Abstract

The invention belongs to the field of robot technology, and discloses a robot path planning method based on a multi-strategy improved walrus optimization algorithm, comprising: constructing a plurality of initial paths, taking the shortest path as a goal, constructing a fitness function, and generating a walrus population based on the fitness function, calculating the fitness value of each walrus, iteratively updating the walrus position through the search phase of the walrus algorithm, adopting the best neighbor dynamic adjustment method and the directed evolution mutation strategy to update the walrus population, and performing population iteration in combination with the search behavior of the red-tailed hawk algorithm; performing random walk and optimal value enhancement, and finally updating the global optimal walrus position in combination with the attack behavior of the red-tailed hawk algorithm. The technical scheme of the invention improves the convergence speed and solution quality of the path planning algorithm, and maintains the robustness in high-dimensional and nonlinear problem spaces.

Description

Technical Field

[0001] The invention belongs to the technical field of robots, and in particular relates to a robot path planning method based on a multi-strategy improved walrus optimization algorithm. Background Art

[0002] In recent years, in the wave of rapid development of automation and intelligent technology, the path planning problem of mobile robots has attracted widespread attention from academia and industry, especially in the fields of service robots, driverless cars and industrial automation, the importance of path planning has become increasingly prominent. The path planning problem refers to the mobile robot finding an optimal path from the starting point to the end point in the workspace after receiving the task assignment information, taking into account the environmental information and according to one or some optimization criteria, such as the shortest time, the shortest distance, and the least energy consumption. Effective path planning can not only ensure that the robot completes the task safely and efficiently, but also significantly enhance its adaptability and safety in complex environments. At present, the algorithms for dealing with the path planning problem of mobile robots can be divided into two categories: one is the classic algorithms such as A*, fast search random tree, artificial potential field method, etc., but these traditional algorithms often have the disadvantages of large computational complexity, low search efficiency, and slow convergence speed. The other is the swarm intelligence algorithm designed recently inspired by the biological behavior of nature, such as the gray wolf optimization algorithm (GWO), the dung beetle optimization algorithm (DBO), the snake optimization algorithm (SO), etc. This type of algorithm draws inspiration from group behavior in nature and solves complex optimization problems by simulating the social interactions of these organisms. It has the characteristics of simple structure, fast speed, and strong optimization ability. However, it also has shortcomings such as reduced population diversity in the later stages of algorithm iteration, easy to fall into local optimal solutions, and weak convergence ability. When performing mobile robot path planning, it is difficult to find the optimal path in the current environment. Summary of the invention

[0003] The purpose of the present invention is to provide a robot path planning method based on a multi-strategy improved walrus optimization algorithm to solve the problems existing in the above-mentioned prior art.

[0004] To achieve the above object, the present invention provides a robot path planning method based on a multi-strategy improved walrus optimization algorithm, comprising:

[0005] Step 1: Determine a path starting point and a path end point, and construct a number of initial paths based on the path starting point and the path end point;

[0006] Step 2: constructing a fitness function with the shortest path as the goal, and generating a walrus population with each of the initial paths as the walrus;

[0007] Step 3: Calculate the fitness value corresponding to each of the walruses based on the fitness function, and update and replace the position of each of the walruses in the walrus population based on the fitness value of each of the walruses combined with the exploration phase of the walrus optimization algorithm;

[0008] Step 4: performing position updating and mutation operations on the walrus population based on the optimal neighbor dynamic adjustment method and directed evolution mutation strategy;

[0009] Step 5: Combine the search behavior of the red-tailed hawk algorithm to iterate the position update of the walrus population obtained in step 4, and record the first optimal fitness value and the corresponding first optimal walrus position;

[0010] Step 6: updating the first optimal walrus position based on the optimal value enhanced random walk strategy to obtain the second optimal walrus position;

[0011] Step 7: Update the second optimal walrus position in combination with the attack behavior of the red-tailed hawk algorithm to obtain the third optimal walrus position, and output the global optimal walrus position.

[0012] Repeat steps 4 to 7 for iterative update until the iterative update reaches the preset number of iterations, and output the global optimal walrus position, that is, the optimal path value.

[0013] Optionally, the step 1 specifically includes:

[0014] A plurality of non-obstruction area navigation points are generated between the path starting point and the path end point, and the path starting point, each non-obstruction area navigation point and the path end point are connected to obtain a plurality of initial paths.

[0015] Optionally, the fitness function is specifically:

[0016]

[0017] f X ′=E1.*E2.*n B .*1+α·Density(G)+ρ·Curbature(P)

[0018]

[0019] In the formula, f x is the path optimization function, f X ′ is the path penalty function, which is used to optimize the path, N is the number of path points, (x i ,y i ) is the i-th point on the path, E1 and E2 represent the map boundary values, n Bis the number of obstacles, α and ρ are weight coefficients between (0, 1), P is the original path, G is the obstacle matrix, Density(G) is the obstacle density function, which is the ratio of the number of obstacles to the total area of ​​the map, Curvature(P) is the path curvature function, which is used to measure the curvature of path P on the entire path, p i+1 -p i With p i+2 p i+1 is the vector between adjacent points.

[0020] Optionally, the first walrus population is generated using each of the initial paths as a walrus, specifically comprising: using each of the initial paths as a walrus individual, initializing the walrus population position through Sine-Tent-Cosine chaotic mapping based on the walrus individuals in combination with a walrus algorithm, and generating the walrus population.

[0021] Optionally, the step 4 specifically includes:

[0022] The position of each walrus in the walrus population is updated based on the optimal neighbor dynamic adjustment method, and the population is iterated based on the fitness value corresponding to each walrus position after the position update to obtain the walrus population after the position update; and the position of each walrus in the walrus population after the position update is mutated based on the directed evolution mutation strategy to obtain the position of each walrus after the mutation operation.

[0023] Optionally, the position of each walrus in the walrus population is updated based on the optimal neighbor dynamic adjustment method, and the specific calculation formula is:

[0024]

[0025]

[0026] In the formula, h is a dynamic parameter, t is the current number of iterations, and X i,j is the position of each walrus, X new is the newly generated position of each walrus, β is the dynamic adjustment factor, X best is the current global optimal walrus position, Used to calculate variation strength.

[0027] Optionally, the positions of each walrus in the walrus population after position update are mutated based on the directed evolution mutation strategy. The specific calculation formula is:

[0028]

[0029] Where, X DEMS(i,:) is the position of each walrus after the mutation disturbance, X i,j is the position of each walrus, t is the current iteration number, It is used to calculate the mutation intensity, 2rand()-1 is used to generate a random number in the interval [-1,1], dim is the problem dimension, and randn(1, dim) is used to generate a random vector that conforms to the standard normal distribution, multiplied by Used to balance the intensity of variation in different dimensions.

[0030] Optionally, the position of the walrus population obtained in step 4 is updated iteratively in combination with the search behavior of the red-tailed hawk algorithm. The specific calculation formula is:

[0031]

[0032] In the formula, is the newly generated position of the i-th walrus in the j-th dimension in the fourth stage, X i,j For each walrus position, history i，j is the historical empirical position of the walrus, X best is the current global optimal solution, b is the step size control parameter in the search behavior, 2·rand()-1 is used to generate a random number in the interval [-1,1], ub and lb represent the upper and lower bounds of the solution space respectively.

[0033] Optionally, the first optimal walrus position is updated based on the optimal value enhanced random walk strategy, and the specific calculation formula is:

[0034] The random step length of Levy flight is:

[0035] Optimal value enhanced random walk strategy:

[0036]

[0037] Where, X i,j is the position of each walrus, μ and v are random numbers that follow a normal distribution, is a constant with a value range of Box-muller standard normal distribution of random numbers, rand() is used to generate a random number between (0,1), randn() is used to generate a random number that conforms to the standard normal distribution, X best is the current global optimal walrus position, and step_size is the dynamic factor of Levy flight.

[0038] Optionally, the second optimal walrus position is updated in combination with the attack behavior of the red-tailed hawk algorithm, and the specific calculation formula is:

[0039]

[0040] In the formula, is the global optimal solution updated in the fourth stage, a is the step length control parameter in the attack behavior, X i,j is the position of the ith walrus in the jth dimension, rand() is a random number uniformly distributed from 0 to 1, t is the current iteration number, X best is the current global optimal solution, 2·rand()-1 is used to generate a random number in the interval [-1,1], ub and lb represent the upper and lower bounds of the solution space respectively, history i，j For the historical experience position of walrus.

[0041] The technical effects of the present invention are:

[0042] (1) The present invention integrates the attack and search phases of the red-tailed hawk algorithm into the walrus optimization algorithm, which can effectively improve the global search capability and local development accuracy of the walrus algorithm. The attack behavior increases the ability of the algorithm to break through the local optimal solution by imitating the swiftness and directness of the red-tailed hawk's hunting, while the search behavior ensures the extensive exploration of the algorithm in the solution space, thereby avoiding premature convergence and significantly improving the algorithm's ability to handle path planning problems.

[0043] (2) Based on the dynamic Levy variable step size strategy, the present invention proposes an optimal value enhanced random walk strategy; this strategy abandons the traditional local search at each individual walrus position and replaces it with a more centralized and efficient search method, that is, only a detailed search is performed at the current global optimal position (i.e., the individual position with the best performance among all walruses). The innovative improvement of local search focuses on the detailed search of the global optimal position, which increases the probability of finding a better solution. Overall, the application of this strategy greatly improves the convergence speed of the algorithm and the quality of the solution, while maintaining robustness in high-dimensional and nonlinear problem spaces.

[0044] (3) The present invention combines the best neighbor dynamic adjustment method with the walrus optimization algorithm to achieve a secondary update of the walrus position. This strategy adds additional exploration power to the algorithm and effectively improves the diversity of the population. In this way, the algorithm can maintain its focus on the high-quality solution area while maintaining a broad exploration of the global solution space, thereby more efficiently discovering the global optimal solution or a solution close to the global optimal solution in complex and changeable optimization problems.

[0045] (4) The present invention uses a directed evolution mutation strategy to update the position of each walrus, which can provide a more refined search space and effectively improve the local search capability and search accuracy of the walrus algorithm. BRIEF DESCRIPTION OF THE DRAWINGS

[0046] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the following briefly introduces the drawings required to be used in the embodiments. In the drawings:

[0047] Figure 1 is a flow chart of a method in an embodiment of the present invention;

[0048] Figure 2 : is a comparison chart of optimization iteration curves of the improved walrus optimization algorithm (IWaOA) in the embodiment of the present invention and other algorithms in 12 benchmark functions; wherein, Figure 2 (a) is a comparison diagram of optimization iteration curves of the improved walrus optimization algorithm (IWaOA) in an embodiment of the present invention and other algorithms in the F1 benchmark function; Figure 2 (b) is a comparison diagram of optimization iteration curves of the improved walrus optimization algorithm (IWaOA) in an embodiment of the present invention and other algorithms in the F2 benchmark function; Figure 2 (c) is a comparison diagram of optimization iteration curves of the improved walrus optimization algorithm (IWaOA) in an embodiment of the present invention and other algorithms in the F3 benchmark function; Figure 2 (d) is a comparison diagram of optimization iteration curves of the improved walrus optimization algorithm (IWaOA) in an embodiment of the present invention and other algorithms in the F4 benchmark function; Figure 2 (e) is a comparison diagram of optimization iteration curves of the improved walrus optimization algorithm (IWaOA) in an embodiment of the present invention and other algorithms in the F5 benchmark function; Figure 2 (f) is a comparison diagram of optimization iteration curves of the improved walrus optimization algorithm (IWaOA) in an embodiment of the present invention and other algorithms in the F6 benchmark function; Figure 2 (g) is a comparison diagram of optimization iteration curves of the improved walrus optimization algorithm (IWaOA) in an embodiment of the present invention and other algorithms in the F7 benchmark function; Figure 2 (h) is a comparison diagram of optimization iteration curves of the improved walrus optimization algorithm (IWaOA) in an embodiment of the present invention and other algorithms in the F8 benchmark function; Figure 2 (i) is a comparison diagram of optimization iteration curves of the improved walrus optimization algorithm (IWaOA) in an embodiment of the present invention and other algorithms in the F9 benchmark function; Figure 2 (j) is a comparison diagram of optimization iteration curves of the improved walrus optimization algorithm (IWaOA) in an embodiment of the present invention and other algorithms in the F10 benchmark function; Figure 2 (k) is a comparison diagram of optimization iteration curves of the improved walrus optimization algorithm (IWaOA) in an embodiment of the present invention and other algorithms in the F11 benchmark function; Figure 2 (1) is a comparison diagram of optimization iteration curves of the improved walrus optimization algorithm (IWaOA) in an embodiment of the present invention and other algorithms in the F12 benchmark function;

[0049] Figure 3 In the embodiment of the present invention, Figure 1 Schematic diagram of the path planning effect in;

[0050] Figure 4 In the embodiment of the present invention, Figure 1 Schematic diagram of the fitness value of the path planning function in;

[0051] Figure 5 In the embodiment of the present invention, Figure 2 Schematic diagram of the path planning effect in;

[0052] Figure 6 In the embodiment of the present invention, Figure 2 Schematic diagram of the fitness value of the path planning function in . DETAILED DESCRIPTION

[0053] Now, various exemplary embodiments of the present invention are described in detail, and this detailed description should not be considered as a limitation of the present invention, but should be understood as a more detailed description of certain aspects, characteristics and embodiments of the present invention. It should be noted that, in the absence of conflict, the embodiments in this application and the features in the embodiments can be combined with each other. The present application will be described in detail below with reference to the accompanying drawings and in conjunction with the embodiments.

[0054] Embodiment 1

[0055] like Figure 1 - Figure 6 As shown, this embodiment provides a robot path planning method based on a multi-strategy improved walrus optimization algorithm, including:

[0056] Step 1: Determine a path starting point and a path end point, and construct a number of initial paths based on the path starting point and the path end point;

[0057] Step 2: constructing a fitness function with the shortest path as the goal, and generating a walrus population with each of the initial paths as the walrus;

[0058] Step 3: Calculate the fitness value corresponding to each of the walruses based on the fitness function, and update and replace the position of each of the walruses in the walrus population based on the fitness value of each of the walruses combined with the exploration phase of the walrus optimization algorithm;

[0059] Step 4: performing position updating and mutation operations on the walrus population based on the optimal neighbor dynamic adjustment method and directed evolution mutation strategy;

[0060] Step 5: Combine the search behavior of the red-tailed hawk algorithm to iterate the position update of the walrus population obtained in step 4, and record the first optimal fitness value and the corresponding first optimal walrus position;

[0061] Step 6: updating the first optimal walrus position based on the optimal value enhanced random walk strategy to obtain the second optimal walrus position;

[0062] Step 7: Update the second optimal walrus position in combination with the attack behavior of the red-tailed hawk algorithm to obtain the third optimal walrus position, and output the global optimal walrus position.

[0063] Repeat steps 4 to 7 for iterative update until the iterative update reaches the preset number of iterations, and output the global optimal walrus position, that is, the optimal path value.

[0064] This embodiment optimizes the path search process by integrating multiple improvement strategies, improves the path planning effect, and realizes accurate planning of the robot path. A robot path planning method based on a multi-strategy improved walrus optimization algorithm in this embodiment specifically includes the following steps:

[0065] S1: Set the map boundary, initialization parameters, population size pop, maximum number of iterations Max_iter, and number of navigation points d on the path.

[0066] S2: Determine the starting point and end point of the path: Generate d-2 non-obstruction navigation points between the determined starting point and end point, connect the starting point, the generated d-2 non-obstruction navigation points and the end point to generate the final complete path. Design the objective function in combination with constraints such as path continuity, path smoothness, shortest path, environmental boundaries and obstacles. Path continuity: The generated path cannot have circuitous parts. For example, when the current node is (x1, y1) and moves to the next node (x2, y2), x2>x1 or y2>y1 should be satisfied. Path smoothness: The generated planned path should be as smooth as possible, and the tortuous parts should be minimized to reduce the redundant parts of the path. Constraints such as environmental boundaries and obstacles: The planned moving path must be limited to the grid area of ​​the given map boundary, and it is forbidden to cross the obstacle grid between each node of the moving path. Shortest path condition: While satisfying the above constraints, when selecting a path, the function with the shortest path should be selected, that is, the minimum fitness function value obtained by the improved walrus algorithm after path planning. The fitness function (path length) can be expressed as:

[0067]

[0068] f X ′=E1.*E2.*n B .*1+α·Density(G)+ρ·Curbature(P)

[0069]

[0070] In the formula, f x is the path optimization function, used for basic path planning, f X ′ is the path penalty function, which is used to optimize the path. N is the number of path points, (x i ,yi ) is the i-th point on the path, E1 and E2 represent the map boundary values, n B is the number of obstacles. α and ρ are weight coefficients between (0, 1). In this embodiment, α=0.3 and ρ=0.3 are taken. P is the original path, and G is the obstacle matrix. Density(G) is the obstacle density function, which is the ratio of the number of obstacles to the total area of ​​the map. Curvature(P) is the path curvature function, which is used to measure the curvature of path P on the entire path. i+1 p i With p i+2 p i+1 Represents the vectors between adjacent points.

[0071] For three consecutive path points P in the path i , P i+1 , P i+2 , if the following conditions exist:

[0072]

[0073] In the formula, k represents the node of the path curve, x1 and x3 represent the x coordinates of the path, y1 and y3 represent their y coordinates, and G = 0 means that there is no obstacle. If the condition is met, then point P is considered i+1 Can be removed as it does not add to the smoothness or feasibility of the path.

[0074] S3: Determine new path location information by improving the walrus optimization algorithm through multiple strategies;

[0075] Determine whether the new path exceeds the boundary of the map. If it is found that the path exceeds the boundary, the algorithm will adjust the path position to the boundary and recalculate its fitness value. Compare with the current group optimal or individual historical optimal path to evaluate the fitness value of the new path. If the fitness value of the new path is lower, which means that the effect is better, then the corresponding optimal path will be replaced by the new path, and this better location information will be saved. If the fitness value of the new path does not exceed the existing optimal value, then the status quo will be maintained without replacement. The entire search process will continue as above until the preset maximum number of iterations is reached. Once the number of iterations meets the conditions, the algorithm will stop running and the final stored group optimal path location information will be identified as the optimal solution.

[0076] S4: Based on the walrus algorithm, the walrus population position is initialized through the Sine-Tent-Cosine chaotic mapping; and according to the objective function, the initial optimal fitness value and the optimal walrus position are calculated.

[0077] The random number is generated by Sine-Tent-Cosine chaotic mapping, where r is a parameter controlling the chaotic mapping, r∈[0,1], and the value of r in this embodiment is 0.5. t is the initial chaos seed used to generate random numbers between (0, 1):

[0078]

[0079] The random numbers generated by the Sine-Tent-Cosine chaotic map are used to initialize the positions of the walruses:

[0080] X i =lb+(ub-lb)*x t

[0081] In the formula, X i represents the i-th individual, lb and ub represent the lower bound and upper bound of the solution space respectively.

[0082] S5: Update the population by imitating the exploration phase of walrus foraging. The update formula for phase 1 is:

[0083]

[0084] weight = tanh(rand())

[0085]

[0086] In the formula, represents the new position of the i-th walrus generated in the first stage. weight is the hyperbolic tangent random weight, and tanh is the hyperbolic tangent function. SW represents the best candidate solution for the current optimal walrus position, and I i,j represents a randomly selected integer between 1 and 2. i,j = 1, the positions of the walruses are updated normally. i,j = 2, the algorithm can search the solution space more extensively and produce more significant changes in the walrus position update. After this stage is completed, the newly generated walrus positions are substituted into the objective function, and the walrus positions are maintained or updated according to the results.

[0087] S6: Update the population by imitating the walrus migration behavior. The update formula for stage 2 is as follows:

[0088]

[0089] In the formula, F represents the new position of the i-th walrus in the second stage, k is the position of the target walrus selected for the i-th walrus migration, k∈1,2,…,N, and k≠i, N is the number of walrus populations. kRepresents the fitness value after migration. After this stage is completed, the newly generated positions of each walrus are substituted into the objective function, and the positions of each walrus are maintained or updated according to the results.

[0090] This formula simulates the migration behavior of walruses in a vast search space. It describes each walrus randomly migrating to the location of another walrus in another area. In this process, the location of the walrus is randomly selected, simulating the irregular migration in nature. The location of the walrus will only be updated if the new location can bring a significant increase in the value of the objective function, thus moving towards a better solution.

[0091] S7: Update the walrus position by imitating the walrus's behavior of escaping and fighting predators. The specific formula is:

[0092]

[0093] Local bounds:

[0094]

[0095] In the formula, is the newly generated position of the i-th walrus in the j-th dimension in the third stage, is the fitness function value of the position, and t is the current iteration number. and Represent the local upper bound and local lower bound allowed for the j-th dimension respectively.

[0096] Local upper and lower bounds are used to create a neighborhood radius that changes with algorithm iterations. This allows the algorithm to have a larger exploration range at the beginning of the search, and gradually narrows the search range as the iterations progress, focusing on areas that are more likely to contain the optimal solution. Within the central neighborhood of each walrus, new positions are randomly generated using the above formula to simulate the behavior of walruses escaping predators. After this stage is completed, the newly generated positions of each walrus are substituted into the objective function, and the positions of each walrus are maintained or updated according to the results.

[0097] S8: This embodiment proposes an optimal neighboring dynamic adjustment method and combines it with the walrus optimization algorithm to update the walrus position. The formula is as follows:

[0098]

[0099] h is a dynamic parameter, and its value in this embodiment is 0.5. t represents the current iteration number, X i,j Indicates the position of each walrus, X newRepresents the newly generated position of each walrus. After this stage is completed, the newly generated positions of each walrus are substituted into the objective function, and the positions of each walrus are maintained or updated according to the results. β is the dynamic adjustment factor.

[0100] S9: Directed evolution mutation strategy combined with walrus optimization algorithm, the formula is as follows:

[0101]

[0102] Where, X DEMS(i,:) Represents the position of each walrus after directed evolution. t represents the current iteration number.

[0103] Used to calculate the mutation intensity. 2. rand()-1 is used to generate a random number in the interval [-1,1] to achieve directed evolution mutation. randn(1, dim) is used to generate a random vector that conforms to the standard normal distribution. Multiply by It is used to balance the intensity of variation in different dimensions. After this stage is completed, the newly generated positions of each walrus are substituted into the objective function, and the positions of each walrus are maintained or updated according to the results.

[0104] S10: This embodiment introduces a dynamic Levy flight strategy and proposes an optimal value enhanced random walk strategy to update the position of each walrus. The specific formula is as follows:

[0105] The random step length of Levy flight is:

[0106] Among them, μ~N(0,σ 2 ) and v~N(0,1) are random numbers that follow a normal distribution. The value range is usually This embodiment takes The value of σ is:

[0107]

[0108] Nonlinear decreasing dynamic adaptive step size factor:

[0109] The optimal value enhanced random walk strategy proposed in this embodiment is:

[0110]

[0111] In the formula, Γ() is the standard gamma function, Box-Muller is a method for generating normal random numbers, which can generate standard normal distribution random number pairs from uniformly distributed random numbers. rand() is used to generate random numbers between (0,1). randn() is used to generate a random number that conforms to the standard normal distribution, Xbest is the current global optimal walrus position, and step_size is the dynamic factor of Levy flight.

[0112] S11: This embodiment incorporates the attack behavior and search behavior stages of the red-tailed hawk algorithm on the basis of the original walrus algorithm as the fourth stage to enhance the global search capability of the original algorithm and accelerate the convergence speed.

[0113] The specific formula for the fourth stage is as follows:

[0114] Aggressive behavior:

[0115]

[0116] in, is the global optimal solution updated in the fourth stage, a is the step length control parameter in the attack behavior, and its value in this embodiment is 1. i,j is the position of the ith walrus in the jth dimension, rand() is a random number uniformly distributed from 0 to 1. t is the current iteration number, X best is the current global optimal solution. ub and lb represent the upper and lower bounds of the solution space respectively. i，j For the historical experience position of walrus.

[0117] Search Behavior:

[0118] b is a step length control parameter in the search behavior, and its value in this embodiment is 2.

[0119]

[0120] After this stage is completed, the newly generated positions of each walrus are substituted into the objective function, and the positions of each walrus are maintained or updated according to the results obtained.

[0121] S12: Use the above multi-strategy improved walrus optimization algorithm to iterate, and when the preset maximum number of iterations is reached, stop the iteration and output the optimal path planning result. The test function table is shown in Table 1:

[0122] Table 1 Test function table

[0123]

[0124]

[0125] In the test, the population planning of each algorithm is 30, the maximum number of iterations is 1000, and each test function is tested 30 times. The statistical results are shown in Table 2, and the algorithm results are ranked in Table 3. Table 4 shows the shortest paths of each algorithm under different environments.

[0126] Table 2 Comparison of test results

[0127]

[0128]

[0129]

[0130]

[0131]

[0132] Table 3. Ranking of algorithm results

[0133] Test Function IWaOA WOA PSO NGO GWO NOA GA <![CDATA[F1]]> 1 2 5 3 4 7 6 <![CDATA[F2]]> 1 2 5 3 4 7 6 <![CDATA[F3]]> 1 2 5 3 4 7 6 <![CDATA[F4]]> 1 2 5 3 4 7 6 <![CDATA[F5]]> 1 2 5 3 4 7 6 <![CDATA[F6]]> 1 2 5 3 4 7 6 <![CDATA[F7]]> 1 2 5 3 4 7 6 <![CDATA[F8]]> 1 3 4 2 5 6 7 <![CDATA[F9]]> 1 2 5 3 4 7 6 <![CDATA[F 10 ]]> 1 2 5 3 4 7 6 <![CDATA[F 11 ]]> 1 2 5 3 4 7 6 <![CDATA[F 12 ]]> 1 2 5 3 4 7 6 Average ranking 1 2.08 4.92 2.92 4.08 6.92 6.08

[0134] Table 4 Comparison of path planning test results

[0135]

[0136]

[0137] This embodiment proposes a mobile robot path planning method based on a multi-strategy improved walrus optimization algorithm. The beneficial effects of each improved strategy are as follows:

[0138] (1) The attack and search phases of the red-tailed hawk algorithm are integrated into the walrus optimization algorithm as the algorithm update phase 4, which can effectively improve the global search capability and local development accuracy of the walrus algorithm. The attack behavior increases the algorithm's ability to break through the local optimal solution by imitating the swiftness and directness of the red-tailed hawk's predation, while the search behavior ensures that the algorithm explores extensively in the solution space, thereby avoiding premature convergence. The algorithm's ability to handle path planning problems is significantly improved.

[0139] (2) By combining the dynamic Levy variable step size strategy, an optimal value enhanced random walk strategy is proposed; this strategy abandons the traditional local search at each individual walrus position and replaces it with a more centralized and efficient search method, that is, only a detailed search is performed at the current global optimal position (i.e., the individual position with the best performance among all walruses), and the Box-Muller transformation is introduced at the global optimal position, and random numbers are incorporated into the dynamic step size. The innovative improvement of local search focuses on the detailed search of the global optimal position, which increases the probability of finding a better solution. In addition, the introduction of the Box-Muller transformation adds randomness to the algorithm, which helps the algorithm escape the local optimal trap and achieve a more comprehensive global exploration. The addition of combined random numbers further enriches the diversity of the search, allowing the algorithm to flexibly deal with various complex problems. Overall, the application of this strategy greatly improves the convergence speed of the algorithm and the quality of the solution, while maintaining robustness in high-dimensional and nonlinear problem spaces.

[0140] (3) The optimal neighbor dynamic adjustment method is combined with the walrus optimization algorithm to achieve a secondary update of the walrus position. This strategy adds additional exploration power to the algorithm and effectively improves the diversity of the population. In this way, the algorithm can maintain its focus on the high-quality solution area while not losing the extensive exploration of the global solution space, so that it can more efficiently find the global optimal solution or a solution close to the global optimal solution in complex and changing optimization problems.

[0141] (4) Using the directed evolution mutation strategy to update the position of each walrus can provide a more refined search space and effectively improve the local search capability and search accuracy of the walrus algorithm.

[0142] The above is only a preferred specific implementation of the present application, but the protection scope of the present application is not limited thereto. Any changes or substitutions that can be easily thought of by a person skilled in the art within the technical scope disclosed in the present application should be included in the protection scope of the present application. Therefore, the protection scope of the present application should be based on the protection scope of the claims.

Claims

1. A robot path planning method based on a multi-strategy improved walrus optimization algorithm, characterized in that: include: Step 1: Determine a path starting point and a path end point, and construct a number of initial paths based on the path starting point and the path end point; Step 2: constructing a fitness function with the shortest path as the goal, and generating a walrus population with each of the initial paths as the walrus; The first walrus population is generated by taking each of the initial paths as a walrus, specifically comprising: taking each of the initial paths as a walrus individual, initializing the walrus population position by Sine-Tent-Cosine chaotic mapping based on the walrus individual and the walrus algorithm, and generating the walrus population; Step 3: Calculate the fitness value corresponding to each of the walruses based on the fitness function, and update and replace the position of each of the walruses in the walrus population based on the fitness value of each of the walruses combined with the exploration phase of the walrus optimization algorithm; Step 4: performing position updating and mutation operations on the walrus population based on the optimal neighbor dynamic adjustment method and directed evolution mutation strategy; The step 4 specifically includes: Based on the optimal neighbor dynamic adjustment method, the position of each walrus in the walrus population is updated, and the population is iterated based on the fitness value corresponding to each walrus position after the position update to obtain the walrus population after the position update; based on the directed evolution mutation strategy, the position of each walrus in the walrus population after the position update is mutated to obtain the position of each walrus after the mutation operation; The position of each walrus in the walrus population is updated based on the optimal neighboring dynamic adjustment method, and the specific calculation formula is: In the formula, h is a dynamic parameter, t is the current number of iterations, and X i,j is the position of each walrus, X new is the newly generated position of each walrus, β is the dynamic adjustment factor, X best is the current global optimal walrus position, Used to calculate variation strength; Step 5: Combine the search behavior of the red-tailed hawk algorithm to iterate the position update of the walrus population obtained in step 4, and record the first optimal fitness value and the corresponding first optimal walrus position; Step 6: updating the first optimal walrus position based on the optimal value enhanced random walk strategy to obtain the second optimal walrus position; Step 7: Update the second optimal walrus position in combination with the attack behavior of the red-tailed hawk algorithm to obtain the third optimal walrus position, and output the global optimal walrus position; Repeat steps 4 to 7 for iterative update until the iterative update reaches the preset number of iterations, and output the global optimal walrus position, that is, the optimal path value.

2. A robot path planning method based on a multi-strategy improved walrus optimization algorithm according to claim 1, characterized in that: The step 1 specifically includes: A plurality of non-obstruction area navigation points are generated between the path starting point and the path end point, and the path starting point, each non-obstruction area navigation point and the path end point are connected to obtain a plurality of initial paths.

3. The robot path planning method based on the multi-strategy improved walrus optimization algorithm according to claim 1 is characterized in that: The fitness function is specifically: f x ′=E1·E2·n B ·(1+α·Density(G)+ρ·Curbature(P)) In the formula, fx is the path optimization function, f X ￠ is the path penalty function used to optimize the path, N is the number of path points, (xi, yi) is the i-th point on the path, E1 and E2 represent the map boundary values, n B is the number of obstacles, α and ρ are weight coefficients between (0, 1), P is the original path, G is the obstacle matrix, Density(G) is the obstacle density function, which is the ratio of the number of obstacles to the total area of ​​the map, Curvature(P) is the path curvature function, which is used to measure the curvature of path P on the entire path, p i+1 -p i With p i+2 -p i+1 is the vector between adjacent points.

4. The robot path planning method based on the multi-strategy improved walrus optimization algorithm according to claim 1 is characterized in that: Based on the directed evolution mutation strategy, the positions of each walrus in the walrus population after position update are mutated. The specific calculation formula is: Where, X DEMS(i,:) is the position of each walrus after the mutation disturbance, X i,j is the position of each walrus, t is the current iteration number, Used to calculate the mutation strength, 2×rand()-1 is used to generate a random number in the interval [-1,1], dim is the problem dimension, randn(1, dim) is used to generate a random vector that conforms to the standard normal distribution, multiplied by Used to balance the intensity of variation in different dimensions.

5. The robot path planning method based on the multi-strategy improved walrus optimization algorithm according to claim 1 is characterized in that: Combined with the search behavior of the red-tailed hawk algorithm, the position of the walrus population obtained in step 4 is updated iteratively. The specific calculation formula is: In the formula, is the newly generated position of the i-th walrus in the j-th dimension in the fourth stage, X i,j For each walrus position, history i，j is the historical empirical position of the walrus, X best is the current global optimal solution, b is the step size control parameter in the search behavior, 2×rand()-1 is used to generate a random number in the interval [-1,1], ub and lb represent the upper and lower bounds of the solution space respectively.

6. The robot path planning method based on the multi-strategy improved walrus optimization algorithm according to claim 1 is characterized in that: The first optimal walrus position is updated based on the optimal value enhanced random walk strategy. The specific calculation formula is: The random step length of Levy flight is: Optimal value enhanced random walk strategy: Where, X i,j is the position of each walrus, μ and v are random numbers that follow a normal distribution, is a constant with a value range of Box-Muller standard normal distribution of random numbers, rand() is used to generate a random number between (0,1), randn() is used to generate a random number that conforms to the standard normal distribution, X best is the current global optimal walrus position, and step_size is the dynamic factor of Levy flight.

7. The robot path planning method based on the multi-strategy improved walrus optimization algorithm according to claim 1 is characterized in that: The second optimal walrus position is updated in combination with the attack behavior of the red-tailed hawk algorithm. The specific calculation formula is: In the formula, is the global optimal solution updated in the fourth stage, a is the step length control parameter in the attack behavior, X i,j is the position of the ith walrus in the jth dimension, rand() is a random number uniformly distributed from 0 to 1, t is the current iteration number, X best is the current global optimal solution, 2×rand()-1 is used to generate a random number in the interval [-1,1], ub and lb represent the upper and lower bounds of the solution space respectively, history i，j For the historical experience position of walrus.

Images