Path planning method based on hybrid strategy sparrow simulated annealing

Abstract

The invention discloses a path planning method based on hybrid strategy sparrow simulated annealing, and belongs to the field of unmanned aerial vehicle navigation of automatic driving and low-altitude economy. The method comprises the following steps: firstly, performing population initialization processing on a sparrow algorithm, and initializing the position of a sparrow population by adopting a cross recombination mechanism and an elite reverse learning strategy of a class genetic algorithm, so that the sparrow population is more abundant in distribution in a matrix space; secondly, the diversity of position solutions, the search probability of optimal solutions and the overall fitness are improved through a'wisdom 'sparrow trust mechanism of a nonlinear convergence inertia factor; and finally, selecting a'wisdom 'sparrow group with higher fitness according to a proportion, and jumping out of local optimum by using a sequence disturbance simulated annealing algorithm to achieve the shortest path of path planning. According to the method, search precision and search efficiency are considered, and it is guaranteed that the simulation mobile robot or the unmanned aerial vehicle has better stability and convergence capacity in the path planning optimization process; and when the map environment is complex or multiple machines work cooperatively, a shorter optimization path is provided.

Description

Technical Field

[0001] The present invention provides a path planning method, specifically relates to a path planning method based on a hybrid strategy sparrow simulated annealing algorithm, which takes into account global and local search performances while satisfying solution set stability, optimal matrix distribution, etc., and belongs to the field of automatic driving and low-altitude economical unmanned aerial vehicle navigation. Background Art

[0002] With the popularization of intelligent algorithm technology in various fields, traditional manufacturing is developing towards smart factories. Autonomous driving technology has attracted extensive research in the industry due to its advantages of flexibility, intelligence and efficiency. The development of autonomous driving technology is inseparable from the breakthroughs of multiple key technologies, including perception technology, positioning and navigation, path planning, control execution, etc. Among them, perception technology based on deep learning promotes the rapid perception and recognition of road environment by autonomous driving systems; high-precision positioning and navigation technology provides a basis for accurate vehicle positioning and path planning; continuous optimization of path planning algorithms enables autonomous driving systems to have more complex response capabilities; and the improvement of control execution technology ensures the safe and stable operation of vehicles. The mutual coordination and progress of these key technologies provide solid technical support for the evolution of autonomous driving technology.

[0004] Path planning is a constrained optimization problem that uses intelligent search algorithms to enable drones or mobile robots to plan a smooth path from the starting point to the target point without collision in a map environment full of obstacles. This problem has important applications in the fields of autonomous driving and drone navigation. It can help select the best path and control costs to improve efficiency and maximize benefits. In path planning problems, the following aspects usually need to be considered: the size and dimension of the map; the number of simulated mobile robots or drones; the constraints between the starting point and the end point, such as the path trajectory selection method and whether the path smoothness needs to be considered. The goal of solving the path planning problem is usually the shortest path length and the smoothness of the path. This involves challenges such as how to determine whether the path is the shortest and whether the path trajectory conforms to the actual situation. Existing technologies cannot solve these problems. Summary of the invention

[0005] The present invention provides a path planning method based on hybrid strategy sparrow simulated annealing, which is used to solve the situation that the existing technology cannot solve the problem that the path planning cannot find the shortest path, and the method takes into account both global search and local search capabilities.

[0006] The technical solution adopted by the present invention to solve the above technical problems is: a path planning method based on mixed strategy sparrow simulated annealing, the specific steps of the method are as follows:

[0007] Step 1: Read the basic data of the map and establish a two-dimensional or three-dimensional coordinate system according to the map dimensions;

[0008] Step 2: Read the starting position and arrival position of the mobile robot or drone;

[0009] Step 3: According to the information of population classification, the initial sparrow population is randomly generated, and the matrix distribution encoding method is adopted, and each sparrow individual is stipulated to include the matrix distribution and encoding order;

[0010] Step 4: Initialize the parameters of the sparrow algorithm, including the population size SearchAgents_no, the maximum number of iterations Max_iteration, the warning value ST, the current iteration number t, the discoverer group value PDNumber and the vigilant group value SDNumber;

[0011] Step 5: Use the crossover recombination mechanism and elite reverse learning strategy of genetic algorithms to further expand the matrix space distribution range of the current sparrow population;

[0012] Step 6: Improve the diversity of matrix space position solutions, the search probability of the optimal solution and the overall fitness through the "wise man" sparrow trust mechanism of nonlinear convergence inertia factor, update each sparrow individual, and arrange the individual fitness values ​​in descending order;

[0013] Step 7: Initialize the simulated annealing algorithm parameters, initial temperature T0, termination temperature Tend, cooling coefficient α, select sparrow individuals according to a certain ratio for simulated annealing operation to adjust the state in the search process and avoid the algorithm from falling into the local optimum. Output the sparrow individuals after the simulated annealing operation and judge whether the fitness after the simulated annealing search mechanism is better than the original solution set, and select the optimal solution set as the new sparrow population;

[0014] Step 8: Determine whether the termination condition is met. If so, terminate and output the optimal individual and its fitness; otherwise, output the current number of iterations and the current optimal fitness value, and return to step 5; the termination condition is that the maximum number of iterations is reached or the fitness value converges, and terminate if either is met.

[0015] Specifically, step 1 needs to read the basic data of the map including the dimensions and its length, width, and height, and establish a two-dimensional coordinate system with the length as the x-axis, the width as the y-axis, and the starting position as the origin, or establish a three-dimensional coordinate system with the length as the x-axis, the width as the y-axis, and the height as the z-axis, and the lower left corner of the starting position as the origin.

[0016] Specifically, in step 2, it is necessary to read the distribution information of the mobile robots or drones, the number of mobile robots or drones in the map, and the starting position and arrival position of each mobile robot or drone on the map.

[0017] Specifically, the genetic algorithm-like crossover recombination mechanism and elite reverse learning strategy in step 5 include the following specific steps:

[0018] Step 5.1: Input the parameters of each sparrow individual, i.e., simulate the matrix distribution and coding order of the individual, and initialize the matrix space;

[0019] Step 5.2: Read the distribution information of the current group and select the discoverer group and vigilant group for this iteration;

[0020] Step 5.3: Using a crossover recombination mechanism similar to a genetic algorithm, perform matrix distribution crossover recombination on the discoverers and vigilant agents selected in step 5.2, and record the new matrix distribution into the matrix list;

[0021] Step 5.4: Use the elite reverse learning strategy for each updated matrix distribution to determine whether the strategy optimization has achieved expectations, and record the matrix distribution after strategy optimization into the matrix list.

[0022] Specifically, the crossover recombination mechanism in step 5.3 has the following steps:

[0023] Step 5.3.1: Assume that the sparrow population is X, where the discoverer population is X 1 For the father, vigilant group X 2 is the mother parent, and the position information P is the chromosome set;

[0024] Step 5.3.2: Since the number of discoverers and vigilant people in the sparrow algorithm is different, direct crossover recombination will cause the number of the two groups to change. Assume that the proportion of discoverers and vigilant people in the population is a i , (i=1,2), then the number of groups to be crossed is X c for:

[0025] Step 5.3.3: Record the matrix distribution of the discoverer group and the vigilant group after crossover recombination into the matrix list.

[0026] Specifically, the specific steps of step 5.4 using the elite reverse learning strategy for each updated matrix distribution are as follows:

[0027] Step 5.4.1: According to the current matrix distribution, introduce the reverse solution, expand the search area of ​​the algorithm, and search the reverse area for the individuals of the original solution;

[0028] Step 5.4.2: Determine the superiority of the original solution and the reverse solution. If the reverse solution is better than the original solution, the reverse solution is the elite solution. Otherwise, the original solution is the elite solution. Record the matrix distribution after the elite reverse strategy.

[0029] Step 5.4.3: Use this matrix distribution as the final initialization of the sparrow population.

[0030] The method of updating each individual sparrow in step 6 includes the following specific steps:

[0031] Step 6.1: Select the Levy flight strategy to simulate the process of many animals in nature trying to find food in an uncertain environment, ensuring that the individual sparrow can carefully search a small area around itself in a short distance, and occasionally walk a longer distance to ensure that it can enter another area and search in a wider range;

[0032] Step 6.2: The inertia factor p of the roulette probability is composed based on the sigmoid function, where the unknown number distribution of the sigmoid function is the sum of the softplus function and the iteration number as its own unknown number distribution;

[0033] Step 6.2: Determine whether to use the Levy flight strategy. When p is less than the value of the rand function, use the Levy flight search method to update. When p is greater than the value of the rand function, directly proceed to the next step of the "wise man" sparrow trust mechanism search method;

[0034] Step 6.3: When p is greater than the value of the rand function, assume that the most intelligent sparrow individual is the "wise" sparrow individual, Sη is the degree of trust of the "wise" sparrow individual in the sparrow population, sη 0 is the maximum trust coefficient of the “wise” sparrow individual, α is the influence coefficient, Sr i,j is the relative distance position of different sparrows, then the formula for the influence of this individual on the population is as follows:

[0035] Step 6.4: Taking Sη as the benchmark, in the tth iteration, the position of the "wise" sparrow individual in the sparrow population is and sparrow population location If the sparrow's position S is updated i,j The formula is as follows:

[0036] Step 6.5: Repeat steps 6.2 to 6.4 until each individual is updated.

[0037] Specifically, the specific steps of the simulated annealing operation performed on the sparrow individual in step 7 are as follows:

[0038] Step 7.1: After completing step 6 and updating each individual sparrow, calculate the fitness value of each individual sparrow, sort them in descending order, and initialize the simulated annealing algorithm parameters;

[0039] Step 7.2: Take out the top 25% sparrow individuals with better fitness as the "wise" sparrow set, and the remaining 75% as the strategic sparrow set;

[0040] Step 7.2: Perform simulated annealing operation on the top 25% of sparrow individuals, conduct neighborhood transformation search, and add 1-2ps, 2-2ps, and 1-2opt perturbation mechanisms;

[0041] Step 7.3: Make the following judgment on the new sparrow individuals after disturbance: if the fitness value of the new solution is better than the old solution, then accept the new solution; if the fitness value of the new solution is lower than the old solution, then accept the new solution according to the Metropolis criterion, perturb each sparrow individual L times, and perform T = T0*α cooling operation after the disturbance is completed until the temperature drops to Tend, and output the final sparrow individual;

[0042] Step 7.3: Determine whether the fitness of the sparrow solution set is better than the original solution set after the simulated annealing search mechanism. If it is better, accept the new solution; otherwise, maintain the original solution set, that is, the solution set optimized by the "wise" sparrow trust mechanism of the nonlinear convergence inertia factor.

[0043] Compared with the prior art, the beneficial effects of the present invention are as follows: firstly, the population of the sparrow algorithm is initialized, and the crossover recombination mechanism and elite reverse learning strategy of the genetic algorithm are adopted to initialize the position of the sparrow population, so that its distribution in the matrix space is richer; secondly, the diversity of the position solution, the search probability of the optimal solution and the overall fitness are improved through the "wise" sparrow trust mechanism of the nonlinear convergence inertia factor; finally, the "wise" sparrow population with higher fitness is selected according to a certain proportion, and the simulated annealing algorithm of sequence perturbation is used to jump out of the local optimum, so as to avoid the algorithm from falling into the local optimum. It can be seen that the path planning method based on the mixed strategy sparrow simulated annealing designed by the present invention can take into account both local and global search capabilities, and has better search accuracy and shorter path length in the face of complex map environments. BRIEF DESCRIPTION OF THE DRAWINGS

[0044] Figure 1 The flowchart of the algorithm of the present invention as a whole;

[0045] Figure 2 It is a schematic map of a two-dimensional or three-dimensional coordinate system, wherein Figure (a) is a two-dimensional simulation map, and Figure (b) is a three-dimensional simulation map;

[0046] Figure 3 The coordinate system shows the starting and ending points of the map, where Figure (a) shows the starting and ending points of the two-dimensional simulation map, and Figure (b) shows the starting and ending points of the three-dimensional simulation map;

[0047] Figure 4 Schematic diagram of the coding sequence for the discoverer group and the vigilant group;

[0048] Figure 5 Schematic diagram of crossover recombination of coding sequences for discoverers and vigilant;

[0049] Figure 6 Schematic diagram of the coordinate system of the sigmoid function and the softplus function, where (a) is the image of the sigmoid function and (b) is the image of the softplus function;

[0050] Figure 7 Schematic diagrams of the perturbation mechanisms of 1-2ps, 2-2ps, and 1-2opt, wherein (a) is a schematic diagram of the perturbation mechanism of 1-2ps, (b) is a schematic diagram of the perturbation mechanism of 2-2ps, and (c) is a schematic diagram of the perturbation mechanism of 1-2opt;

[0051] Figure 8 This is a grid simulation diagram of a warehouse, where (a) is a two-dimensional comparison diagram of six algorithms, and (b) is a two-dimensional map multi-machine planning comparison diagram;

[0052] Fig. 9 This is a spatial simulation diagram of a mountain range, where Figure (a) is a comparison of six three-dimensional algorithms, and Figure (b) is a comparison of multi-machine planning of three-dimensional maps. DETAILED DESCRIPTION

[0053] In order to make the purpose, technical solution and advantages of the present invention more clearly understood, the present invention is further described in detail below in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only for explaining the present invention and are not intended to limit the present invention.

[0054] Example 1: Figures 1 to 9 As shown, a path planning method based on hybrid strategy sparrow simulated annealing includes the following steps (the overall process is as follows Figure 1 shown):

[0055] Step 1: Read the basic data of the map and establish a two-dimensional or three-dimensional coordinate system according to the map dimensions;

[0056] Step 2: Read the starting position and arrival position of the mobile robot or drone;

[0057] Step 3: According to the information of population classification, the initial sparrow population is randomly generated, and the matrix distribution encoding method is adopted, and each sparrow individual is stipulated to include the matrix distribution and encoding order;

[0058] Step 4: Initialize the parameters of the sparrow algorithm, including the population size SearchAgents_no, the maximum number of iterations Max_iteration, the warning value ST, the current iteration number t, the discoverer group value PDNumber and the vigilant group value SDNumber;

[0059] Step 5: Use the crossover recombination mechanism and elite reverse learning strategy of genetic algorithms to further expand the matrix space distribution range of the current sparrow population;

[0060] Step 6: Improve the diversity of matrix space position solutions, the search probability of the optimal solution and the overall fitness through the "wise man" sparrow trust mechanism of nonlinear convergence inertia factor, update each sparrow individual, and arrange the individual fitness values ​​in descending order;

[0061] Step 7: Initialize the simulated annealing algorithm parameters, initial temperature T0, termination temperature Tend, cooling coefficient α, select sparrow individuals according to a certain ratio for simulated annealing operation to adjust the state in the search process and avoid the algorithm from falling into the local optimum. Output the sparrow individuals after the simulated annealing operation and judge whether the fitness after the simulated annealing search mechanism is better than the original solution set, and select the optimal solution set as the new sparrow population;

[0062] Step 8: Determine whether the termination condition is met. If so, terminate and output the optimal individual and its fitness; otherwise, output the current number of iterations and the current optimal fitness value, and return to step 5; the termination condition is that the maximum number of iterations is reached or the fitness value converges, and terminate if either is met.

[0063] The specific implementation process is as follows: In order to solve the problem that the current path planning algorithm cannot guarantee the search accuracy and path length of the simulated mobile robot or drone in the decision-making and planning process, and cannot take into account the local and global search capabilities at the same time. First, the crossover recombination mechanism and elite reverse learning strategy of the genetic algorithm are adopted to initialize the position of the sparrow population, so that its distribution in the matrix space is richer; secondly, the "wise" sparrow trust mechanism of the nonlinear convergence inertia factor is used to improve the diversity of position solutions, the search probability of the optimal solution and the overall fitness, so as to simulate the optimization process of the mobile robot or drone; finally, the "wise" sparrow population with higher fitness is selected according to a certain proportion, and the simulated annealing algorithm of sequence perturbation is used to jump out of the local optimum to achieve the shortest path of the current iteration.

[0064] Furthermore, for step 1, the basic data of the map including the dimensions and its length, width and height need to be read, and a two-dimensional coordinate system is established with the length as the x-axis, the width as the y-axis, and the starting position as the origin, or a three-dimensional coordinate system is established with the length as the x-axis, the width as the y-axis, and the height as the z-axis, and the lower left corner of the starting position as the origin (such as Figure 2 shown).

[0065] Furthermore, for step 2, it is necessary to read the simulated mobile robot or drone distribution information, the number of mobile robots or drones in the map, the starting position and arrival position of each mobile robot or drone on the map (such as Figure 3 shown).

[0066] Furthermore, the crossover recombination mechanism and elite reverse learning strategy of the genetic algorithm in step 5 include the following specific steps:

[0067] Step 5.1: Input the parameters of each sparrow individual, i.e., simulate the matrix distribution and coding order of the individual, and initialize the matrix space;

[0068] Step 5.2: Read the distribution information of the current group and select the discoverer group and vigilant group for this iteration (such as Figure 4 shown);

[0069] Step 5.3: Using a crossover recombination mechanism similar to a genetic algorithm, the discoverers and vigilants selected in step 5.2 are cross-recombined in coding order, and the new matrix distribution is recorded in the matrix list (such as Figure 5 shown);

[0070] Step 5.4: Use the elite reverse learning strategy for each updated matrix distribution to determine whether the strategy optimization has achieved expectations, and record the matrix distribution after strategy optimization into the matrix list.

[0071] Furthermore, the specific steps of the crossover recombination mechanism in step 5.3 are as follows:

[0072] Step 5.3.1: Assume that the sparrow population is X, where the discoverer population is X 1 For the father, vigilant group X 2 is the mother parent, and the position information P is the chromosome set;

[0073] Step 5.3.2: Since the number of discoverers and vigilant people in the sparrow algorithm is different, direct crossover recombination will cause the number of the two groups to change. Assume that the proportion of discoverers and vigilant people in the population is a i , (i=1,2), then the number of groups to be crossed is X c for:

[0074] Step 5.3.3: Record the matrix distribution of the discoverer group and the vigilant group after crossover recombination into the matrix list.

[0075] Furthermore, the specific steps of step 5.4 for applying the elite reverse learning strategy to each updated matrix distribution are as follows:

[0076] Step 5.4.1: According to the current matrix distribution, introduce the reverse solution, expand the search area of ​​the algorithm, and search the reverse area for the individuals of the original solution;

[0077] Step 5.4.2: Determine the superiority of the original solution and the reverse solution. If the reverse solution is better than the original solution, the reverse solution is the elite solution. Otherwise, the original solution is the elite solution. Record the matrix distribution after the elite reverse strategy.

[0078] Step 5.4.3: Use this matrix distribution as the final initialization of the sparrow population;

[0079] Furthermore, the method of updating each individual sparrow in step 6 mainly includes the following specific steps:

[0080] Step 6.1: Select the Levy flight strategy to simulate the process of many animals in nature trying to find food in an uncertain environment, ensuring that the individual sparrow can carefully search a small area around itself in a short distance, and occasionally walk a longer distance to ensure that it can enter another area and search in a wider range;

[0081] Step 6.2: The inertia factor p of the roulette probability is composed of the sigmoid function as the main body, where the unknown number distribution of the sigmoid function is the sum of the softplus function and the iteration number as its own unknown number distribution; (such as Figure 6 (shown)

[0082] Step 6.2: Determine whether to use the Levy flight strategy. When p is less than the value of the rand function, use the Levy flight search method to update. When p is greater than the value of the rand function, directly proceed to the next step of the "wise man" sparrow trust mechanism search method;

[0083] Step 6.3: When p is greater than the value of the rand function, assume that the most intelligent sparrow individual is the "wise" sparrow individual, Sη is the degree of trust of the "wise" sparrow individual in the sparrow population, sη 0 is the maximum trust coefficient of the “wise” sparrow individual, α is the influence coefficient, Sr i,j is the relative distance position of different sparrows, then the formula for the influence of this individual on the population is as follows:

[0084] Step 6.4: Taking Sη as the benchmark, in the tth iteration, the position of the "wise" sparrow individual in the sparrow population is and sparrow population location If the sparrow's position S is updated i,j The formula is as follows:

[0085] Step 6.5: Repeat steps 6.2 to 6.4 until each individual is updated.

[0086] Further, step 7.1: after completing step 6 and updating each individual sparrow, the fitness value of each individual sparrow is calculated and sorted in descending order, and the simulated annealing algorithm parameters are initialized;

[0087] Step 7.2: Take out the top 25% sparrow individuals with better fitness as the "wise" sparrow set, and the remaining 75% as the strategic sparrow set;

[0088] Step 7.2: Perform simulated annealing operation on the top 25% of sparrow individuals, perform neighborhood transformation search, and add 1-2ps, 2-2ps, and 1-2opt perturbation mechanisms; (e.g. Figure 7 (shown)

[0089] Step 7.3: Make the following judgment on the new sparrow individuals after disturbance: if the fitness value of the new solution is better than the old solution, then accept the new solution; if the fitness value of the new solution is lower than the old solution, then accept the new solution according to the Metropolis criterion, perturb each sparrow individual L times, and perform T=T0*α cooling operation after the disturbance until the temperature drops to Tend, and output the final sparrow individual.

[0090] Step 7.3: Determine whether the fitness of the sparrow solution set is better than the original solution set after the simulated annealing search mechanism. If it is better, accept the new solution; otherwise, maintain the original solution set, that is, the solution set optimized by the "wise" sparrow trust mechanism of the nonlinear convergence inertia factor.

[0091] To further illustrate the accuracy and reliability of the present invention, two groups of experiments are conducted to demonstrate its superiority in algorithm performance and path planning.

[0092] Experiment 1: 12 algorithm performance test functions widely used internationally are used as the test set of the present invention to show the comparative results of the algorithm in this paper (LFSSA), sparrow algorithm (SSA), improved mayfly algorithm (LMA), random walk sparrow algorithm (RSSA) and two intelligent search optimization algorithms (PSO, GWO). The 12 test functions are composed of a mixture of unimodal functions or multimodal functions, which can well reflect the actual performance of the algorithm. The test function summary is shown in Table 1, and the test results are compared in Table 2:

[0093] Table 1 Summary of test functions

[0094]

[0095] The optimal values ​​of the algorithm comparison in Table 2 are bolded.

[0096] Table 2 Comparison of test results

[0097]

[0098] As can be seen from Table 2, the hybrid strategy of the sparrow simulated annealing algorithm (LFSSA) of the present invention shows more accurate optimal value search, better solution accuracy, and very stable optimization ability in the search performance experiment of the test function compared with the other five algorithms. It has better search ability and search accuracy in both single-peak test functions and multi-peak test functions.

[0099] Experiment 2: Using a grid simulation map of a warehouse (such as Figure 8 ) and a spatial simulation map of a mountain range (as shown in Fig. 9 ) for example verification.

[0100] 1. A grid simulation diagram of a warehouse, such as Figure 8 As shown, Figure 8 (a) is a two-dimensional comparison chart of six algorithms. Figure 8 (a) is a comparison chart of multi-machine planning on a two-dimensional map.

[0101] All algorithms were simulated 15 times, and the results are shown in Table 3:

[0102] Table 3

[0103]

[0104] 2. A spatial simulation map of a mountain range, such as Fig. 9 As shown, Fig. 9 (a) is a comparison chart of six three-dimensional algorithms. Fig. 9 (b) is a comparison chart of multi-machine planning on a three-dimensional map.

[0105] All algorithms were simulated 15 times, and the results are shown in Table 4:

[0106] Table 4

[0107]

[0108] The above experiments prove that the method of the present invention can find shorter paths in different maps and multi-machine simulations, which proves that the improvement strategy of the algorithm is reasonable and stable.

[0109] The specific implementation modes of the present invention are described in detail above in conjunction with the accompanying drawings, but the present invention is not limited to the above implementation modes, and various changes can be made within the knowledge scope of ordinary technicians in this field without departing from the purpose of the present invention.

Claims

1. A path planning method based on hybrid strategy sparrow simulated annealing, characterized in that: The specific steps are as follows: Step 1: Read the basic data of the map and establish a two-dimensional or three-dimensional coordinate system according to the map dimensions; Step 2: Read the starting position and arrival position of the mobile robot or drone; Step 3: According to the information of population classification, the initial sparrow population is randomly generated, and the matrix distribution encoding method is adopted, and each sparrow individual is stipulated to include the matrix distribution and encoding order; Step 4: Initialize the parameters of the sparrow algorithm, including the population size SearchAgents_no, the maximum number of iterations Max_iteration, the warning value ST, the current iteration number t, the discoverer group value PDNumber and the vigilant group value SDNumber; Step 5: Use the crossover recombination mechanism and elite reverse learning strategy of genetic algorithms to further expand the matrix space distribution range of the current sparrow population; Step 6: Improve the diversity of matrix space position solutions, the search probability of the optimal solution and the overall fitness through the "wise man" sparrow trust mechanism of nonlinear convergence inertia factor, update each sparrow individual, and arrange the individual fitness values ​​in descending order; Step 7: Initialize the simulated annealing algorithm parameters, initial temperature T0, termination temperature Tend, cooling coefficient α, select sparrow individuals according to a certain ratio for simulated annealing operation to adjust the state in the search process and avoid the algorithm from falling into the local optimum. Output the sparrow individuals after the simulated annealing operation and judge whether the fitness after the simulated annealing search mechanism is better than the original solution set, and select the optimal solution set as the new sparrow population; Step 8: Determine whether the termination condition is met, if it is met, it ends and outputs the optimal individual and its fitness; Otherwise, the current number of iterations and the current optimal fitness value are output, and the process returns to step 5; the termination condition is that the maximum number of iterations is reached or the fitness value converges, and the process terminates if either of the conditions is met.

2. A path planning method based on hybrid strategy sparrow simulated annealing according to claim 1, characterized in that: Step 1 requires reading the basic data of the map including the dimensions and its length, width, and height, and establishing a two-dimensional coordinate system with the length as the x-axis, the width as the y-axis, and the starting position as the origin, or establishing a three-dimensional coordinate system with the length as the x-axis, the width as the y-axis, and the height as the z-axis, and the lower left corner of the starting position as the origin.

3. A path planning method based on hybrid strategy sparrow simulated annealing according to claim 1, characterized in that: In step 2, it is necessary to read the simulated mobile robot or drone distribution information, the number of mobile robots or drones in the map, and the starting position and arrival position of each mobile robot or drone on the map.

4. A path planning method based on hybrid strategy sparrow simulated annealing according to claim 1, characterized in that: The crossover recombination mechanism and elite reverse learning strategy of the genetic algorithm in step 5 include the following specific steps: Step 5.1: Input the parameters of each sparrow individual, i.e., simulate the matrix distribution and coding order of the individual, and initialize the matrix space; Step 5.2: Read the distribution information of the current group and select the discoverer group and vigilant group for this iteration; Step 5.3: Using a crossover recombination mechanism similar to a genetic algorithm, perform matrix distribution crossover recombination on the discoverers and vigilant agents selected in step 5.2, and record the new matrix distribution into the matrix list; Step 5.4: Use the elite reverse learning strategy for each updated matrix distribution to determine whether the strategy optimization has achieved expectations, and record the matrix distribution after strategy optimization into the matrix list.

5. A path planning method based on mixed strategy sparrow simulated annealing according to claim 4, characterized in that: The specific steps of the crossover recombination mechanism in step 5.3 are as follows: Step 5.3.1: Assume that the sparrow population is X, where the discoverer population is X 1 For the father, vigilant group X 2 is the mother parent, and the position information P is the chromosome set; Step 5.3.2: Since the number of discoverers and vigilant people in the sparrow algorithm is different, direct crossover recombination will cause the number of the two groups to change. Assume that the proportion of discoverers and vigilant people in the population is a i , (i=1,2), then the number of groups to be crossed is X c for: Step 5.3.3: Record the matrix distribution of the discoverer group and the vigilant group after crossover recombination into the matrix list.

6. A path planning method based on mixed strategy sparrow simulated annealing according to claim 4, characterized in that: The specific steps of step 5.4 using the elite reverse learning strategy for each updated matrix distribution are as follows: Step 5.4.1: According to the current matrix distribution, introduce the reverse solution, expand the search area of ​​the algorithm, and search the reverse area for the individuals of the original solution; Step 5.4.2: Determine the superiority of the original solution and the reverse solution. If the reverse solution is better than the original solution, the reverse solution is the elite solution. Otherwise, the original solution is the elite solution. Record the matrix distribution after the elite reverse strategy. Step 5.4.3: Use this matrix distribution as the final initialization of the sparrow population.

7. A path planning method based on mixed strategy sparrow simulated annealing according to claim 4, characterized in that: Updating each sparrow individual in step 6 includes the following specific steps: Step 6.1: Select the Levy flight strategy to simulate the process of many animals in nature trying to find food in an uncertain environment, ensuring that the individual sparrow can carefully search a small area around itself in a short distance, and occasionally walk a longer distance to ensure that it can enter another area and search in a wider range; Step 6.2: The inertia factor p of the roulette probability is composed based on the sigmoid function, where the unknown number distribution of the sigmoid function is the sum of the softplus function and the iteration number as its own unknown number distribution; Step 6.2: Determine whether to use the Levy flight strategy. When p is less than the value of the rand function, use the Levy flight search method to update. When p is greater than the value of the rand function, directly proceed to the next step of the "wise man" sparrow trust mechanism search method; Step 6.3: When p is greater than the value of the rand function, assume that the most intelligent sparrow individual is the "wise" sparrow individual, Sη is the degree of trust of the "wise" sparrow individual in the sparrow population, sη0 is the maximum trust coefficient of the "wise" sparrow individual, α is the influence coefficient, Sr i,j is the relative distance position of different sparrows, then the formula for the influence of this individual on the population is as follows: Step 6.4: Taking Sη as the benchmark, in the tth iteration, the position of the "wise" sparrow individual in the sparrow population is and sparrow population location If the sparrow's position S is updated i,j The formula is as follows: Step 6.5: Repeat steps 6.2 to 6.4 until each individual is updated.

8. The path planning method based on hybrid strategy sparrow simulated annealing according to claim 1 is characterized by: The specific steps of the simulated annealing operation performed on the sparrow individuals in step 7 are as follows: Step 7.1: After completing step 6 and updating each individual sparrow, calculate the fitness value of each individual sparrow, sort them in descending order, and initialize the simulated annealing algorithm parameters; Step 7.2: Take out the top 25% of sparrow individuals with better fitness as the "wise" sparrow set, and the remaining 75% as the strategic sparrow set; Step 7.2: Perform simulated annealing operation on the top 25% of sparrow individuals, conduct neighborhood transformation search, and add 1-2ps, 2-2ps, and 1-2opt perturbation mechanisms; Step 7.3: Make the following judgment on the new sparrow individuals after disturbance: if the fitness value of the new solution is better than the old solution, then accept the new solution; if the fitness value of the new solution is lower than the old solution, then accept the new solution according to the Metropolis criterion, perturb each sparrow individual L times, and perform T = T0*α cooling operation after the disturbance is completed until the temperature drops to Tend, and output the final sparrow individual; Step 7.3: Determine whether the fitness of the sparrow solution set is better than the original solution set after the simulated annealing search mechanism. If it is better, accept the new solution; otherwise, maintain the original solution set, that is, the solution set optimized by the "wise" sparrow trust mechanism of the nonlinear convergence inertia factor.

Images