Unmanned aerial vehicle intelligent path planning method and system

Abstract

The invention discloses an intelligent path planning method and system for an unmanned aerial vehicle. The method comprises the following steps: initializing related parameters and optimization algorithm parameters of the unmanned aerial vehicle; coding the path points of the unmanned aerial vehicle; calculating fitness values of all individuals in the population; generating a new population through selection, crossover and mutation operations; carrying out local search on individuals with better fitness in the population; dynamically adjusting algorithm parameters according to population fitness changes; and outputting the path with the highest fitness value as the optimal flight path of the unmanned aerial vehicle. According to the method, a local search mechanism and a dynamic adjustment strategy are introduced on the basis of a traditional genetic algorithm, the method is suitable for a combinatorial optimization problem, excellent performance is shown in unmanned aerial vehicle path planning, and a path with the shortest time can be efficiently found.

Description

Technical Field

[0001] The present invention relates to a method for unmanned aerial vehicle (UAV) path planning, and in particular to a method and system for intelligent UAV path planning, and belongs to the field of UAV path planning. Background Art

[0002] With the rapid development of drone technology, its application in logistics, agriculture, environmental monitoring and disaster relief is becoming more and more extensive. In these application scenarios, planning the flight path of drones, especially finding the optimal path in complex environments, has become a key technical issue. The core goal of path planning is to determine the best path for drones from the starting point to the target point, and the shortest path problem is one of the most common requirements. How to efficiently solve the shortest path problem has become the core content of the research on drone autonomous navigation systems.

[0003] Traditional path planning algorithms are often limited in computational efficiency and result quality when faced with complex environments or large-scale path planning problems, especially when the solution space is large or there are many path points. In order to improve the efficiency of path planning and optimize the path quality, many studies have begun to adopt intelligent optimization algorithms. Genetic algorithms, as a classic heuristic optimization algorithm, have been widely used in path planning problems. The advantage of genetic algorithms is that they can perform global searches in complex solution spaces and continuously optimize the quality of solutions through fitness functions. They are suitable for various types of optimization problems and have strong adaptability and flexibility, especially in multi-objective optimization and large-scale problems. However, traditional genetic algorithms have problems such as slow convergence, poor local search capabilities, and sensitive parameter settings.

[0004] Based on genetic algorithms, the present invention proposes an optimization method combining local search mechanism and dynamic adjustment strategy for the shortest path problem of unmanned aerial vehicles. Different from the traditional genetic algorithm that mainly relies on global search (such as selection, crossover and mutation operations), the improved genetic algorithm organically combines local search with global search, which not only improves the quality of the solution, but also significantly accelerates the convergence speed, making it perform better in complex optimization problems. At the same time, the meta-learning module dynamically adjusts the algorithm parameters according to the performance of the population, so that the algorithm can be optimized adaptively and improve its adaptability at different stages, so as to find the optimal path more effectively. Summary of the invention

[0005] Purpose of the invention: To address the actual needs of the shortest path problem in drone path planning, a drone intelligent path planning method and system is proposed. This method can overcome the problems of slow convergence speed, insufficient local search capability, and sensitive parameter settings of traditional genetic algorithms, efficiently search for the optimal solution, and achieve a dual improvement in path planning efficiency and optimization quality, thereby meeting the needs of drones for efficient path planning in logistics, agriculture, environmental monitoring, and other fields.

[0006] Technical solution: To achieve the above-mentioned invention object, the present invention adopts the following technical solution:

[0007] The method and system for intelligent path planning of unmanned aerial vehicles include the following steps:

[0008] S1: Initialize the parameter set, including drone related parameters and optimization algorithm parameters;

[0009] S2: Encode the path points and use a coding method suitable for genetic algorithms to represent the planned path of the UAV;

[0010] S3: Taking the inverse of the drone flight time as the objective function, calculate the fitness values ​​of all individuals in the population;

[0011] S4: Generate new populations through selection, crossover and mutation operations;

[0012] S5: Recalculate the fitness value of the population, perform local search on individuals with better fitness in the population, repeat S4 on those with poorer fitness, and continue global search;

[0013] S6: When iterating to the fifth generation, the algorithm parameters are dynamically adjusted according to the changes in population fitness;

[0014] S7: Execute S5-S6 in a loop until the termination condition is met, and output the path with the highest fitness value as the optimal flight path of the UAV, that is, the path with the shortest flight time.

[0015] Furthermore, the encoding method described in step S2 is as follows: Let Pa = {pa1, pa2, ..., pa m} represents the target points of m drones, where pa k =(x ak ,y ak , z ak ), represents the coordinates of the kth target point. Let the positive integer set PA = {PA1, PA2, PA3, ..., PA m} represents the path of the drone, where the gene value PA i represents the label of the target point, and m represents the sequence length.

[0016] Furthermore, the flight time T of the drone described in step S3 A Specifically expressed as:

[0017]

[0018] Where pa0 represents the starting point of the drone, S a，b represents the distance between points a and b, v arepresents the flight speed of the drone. Then the fitness value of the population can be expressed as:

[0019]

[0020] Furthermore, the operation described in step S4 includes:

[0021] S41: Selection, first evaluate the fitness of each individual. The higher the fitness, the greater the proportion of the individual on the roulette wheel. Then add up the fitness of all individuals to get the total fitness. Then calculate the selection probability. The selection probability of each individual is equal to the ratio of its fitness to the total fitness. Then build a roulette wheel, map the fitness value to the roulette wheel according to the selection probability of each individual, and form a continuous probability distribution. Finally, randomly select, generate a random number between 0 and 1, and select individuals according to the interval on the roulette wheel. If the random number falls within the interval corresponding to an individual, the individual is selected.

[0022] S42: Crossover, for two given parent solutions, first generate two random numbers r1 and r2, and then exchange some genes in the two parent solutions between r1 and r2. The crossover operation may result in duplicate genes in the solution. In this case, the exchanged genes in the offspring solution are retained, and the duplicate genes in other fragments are mapped to the corresponding genes of the conflicting genes in the exchanged fragments.

[0023] S43: Mutation. For each solution in the population, first generate two random numbers r1 and r2, and then perform a mutation operation to exchange the first gene and the second gene in the solution.

[0024] Furthermore, the local search in step S5 includes:

[0025] S51: Select several individuals with higher fitness from the current population to form a sub-population to be optimized;

[0026] S52: randomly generate three different integers r1, r2 and r3, representing three position indexes in the current individual, and remove the three edges at the corresponding positions in the current individual;

[0027] S53: After removing the above three edges, reverse the intermediate segments between the index positions and reconnect the disconnected path segments to form a new individual;

[0028] Furthermore, the dynamic adjustment parameters in step S6 include:

[0029] S61: Record the performance of each generation of the population, including the fitness value of the population and the crossover rate p c , mutation rate p m , and the proportion of individuals participating in local search in the population pl ;

[0030] S62: When iterating to the fifth generation and beyond, calculate the average fitness value of the last five generations

[0031]

[0032] Where ftness(t) represents the fitness value of the tth generation;

[0033] S63: Compare the current fitness value with the average fitness value. If the fitness of the current generation deteriorates (i.e. Lower p c With p l , improve p m , specifically expressed as:

[0034]

[0035] If the fitness improves (i.e. Lower p m , increase p c With p l , specifically expressed as:

[0036]

[0037] In order to balance the exploration and stability of the algorithm, the three parameter settings are constrained as follows:

[0038]

[0039] Based on the same inventive concept, the present invention also provides a method and system for intelligent path planning of a UAV, including:

[0040] Initialization module, used to initialize the parameter set, including drone-related parameters and optimization algorithm parameters;

[0041] The global search module, implemented through a genetic algorithm, uses selection, crossover, and mutation operations to explore the global solution space and preliminarily generate feasible paths;

[0042] The local search module optimizes some paths with higher fitness, improves the quality of the solution by locally adjusting the path structure, and avoids falling into the local optimum, so that the individual gradually approaches the global optimal solution;

[0043] As well, the meta-learning module dynamically adjusts the genetic algorithm parameters and optimizes key parameters such as the crossover rate and mutation rate according to the population performance, ensuring the adaptability and stability of the algorithm at different stages.

[0044] Based on the same inventive concept, the present invention provides a computer system, including a memory, a processor, and a computer program stored in the memory and executable on the processor. When the computer program is loaded into the processor, the method and system for intelligent path planning of a UAV is implemented.

[0045] Beneficial effects: Based on genetic algorithms, the present invention proposes an optimization method that combines local search mechanisms and dynamic adjustment strategies for the shortest path problem of unmanned aerial vehicles. Unlike traditional genetic algorithms that mainly rely on global search (such as selection, crossover, and mutation operations), the improved genetic algorithm organically combines local search with global search, which not only improves the quality of the solution, but also significantly accelerates the convergence speed, making it perform better in complex optimization problems. At the same time, the meta-learning module dynamically adjusts the algorithm parameters according to the performance of the population, so that the algorithm can be optimized adaptively and improve its adaptability at different stages, so as to more effectively find the optimal path. BRIEF DESCRIPTION OF THE DRAWINGS

[0046] Figure 1 is a flow chart of an embodiment of the present invention.

[0047] Figure 2 Schematic diagram of the crossover process.

[0048] Figure 3 Schematic diagram of the mutation process.

[0049] Figure 4 Schematic diagram of the local search process. DETAILED DESCRIPTION

[0050] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0051] like Figure 1 As shown, an intelligent path planning method and system for a drone disclosed in an embodiment of the present invention specifically comprises the following steps:

[0052] S1: Initialization parameter set. Mainly includes UAV mission scenarios, as well as optimization algorithm related parameters such as improved genetic algorithm population and maximum number of iterations, including:

[0053] The present invention sets the task scene size to L X ×L Y ×L Z , the coordinates of any point in space are (x, y, z), where x∈{1, 2, …, L X}, y∈{1, 2, …, L Y}, z∈{1, 2, …, L Z}. Use V = {v1, v2, ..., v n1} represents all buildings in the mission scene, v i ∈V, i=1,2,…,n represents the i-th building, n represents the number of buildings, Pa={pa1,pa2,…,pa m} represents the target point of m drones, pa k =(x ak ,y ak , z ak ), k = 1, 2, ..., m, represents the kth target point of the drone. During the mission, the drone can fly freely in all directions with a flight speed of v a The population size of the improved genetic algorithm is S pop ; The maximum number of iterations is G max ; The crossover probability is p c ; The mutation probability is p m ; The proportion of individuals performing local search is p l ;

[0054] S2: Encode the path points and use the following encoding method to represent the planned path of the drone:

[0055] Let Pa = {pa1, pa2, ..., pa m} represents the target points of m drones, where pa k =(x ak ,y ak , z ak ), represents the coordinates of the kth target point. Let the positive integer set PA = {PA1, PA2, PA3, ..., PA m} represents the path of the drone, where the gene value PA i represents the label of the target point, and m represents the sequence length. For example, the randomly generated sequence is (3, 9, 6, 2, 1, 5, 4, 10, 7, 8). This means that there are 10 targets that need to be visited, and the order of visit is 3, 9, 6, 2, 1, 5, 4, 10, 7, 8.

[0056] S3: First, combine the UAV’s path PA = {PA1, PA2, PA3, …, PA m}, calculate the flight time T of the drone A :

[0057]

[0058] Where pa0 represents the starting point of the drone, S a，b represents the distance between points a and b, va Represents the flight speed of the drone. The basic principle of the genetic algorithm is to promote the evolution of the entire population towards a high-quality solution by selecting individuals with higher fitness for reproduction. If the flight time is directly used as the fitness value, the algorithm may tend to select individuals with longer flight time, which is completely opposite to the goal of the shortest path. Therefore, the reciprocal of the flight time is used as the objective function to calculate the fitness value of all individuals in the population, which can be specifically expressed as:

[0059]

[0060] This design can clearly align the algorithm's search direction with the optimization goal, thereby improving the efficiency and accuracy of the algorithm. By adopting the inverse, shorter flight times will produce higher fitness values, thereby enhancing the fitness difference between high-quality solutions and low-quality solutions. This difference plays an important role in the selection operation, making it easier for the algorithm to identify and retain high-quality individuals and accelerate the convergence process.

[0061] S4: In genetic algorithms, generating new populations through selection, crossover, and mutation operations is one of its core mechanisms. These operations explore the solution space by simulating the natural evolution process, thereby driving the algorithm to evolve toward the global optimal solution or approximate optimal solution. Specifically, they include:

[0062] S41: Selection. The purpose of the selection operation is to select individuals with higher fitness from the current population according to their fitness values ​​as parents for the next generation. A common selection method is roulette selection. In the UAV path planning problem, the fitness value is usually proportional to the inverse of the flight time, that is, the shorter the flight time, the higher the fitness of the individual (path). In the selection process, individuals with higher fitness have a greater probability of being selected, thereby increasing their chances of being inherited by the next generation.

[0063] Specifically, first evaluate the fitness of each individual. The higher the fitness, the greater the proportion of the individual on the roulette wheel. Then add up the fitness of all individuals to get the total fitness. Then calculate the selection probability. The selection probability of each individual is equal to the ratio of its fitness to the total fitness. Then build a roulette wheel, and map the fitness value to the roulette wheel according to the selection probability of each individual to form a continuous probability distribution. Finally, randomly select and generate a random number between 0 and 1, and select individuals according to the interval on the roulette wheel. If the random number falls within the interval corresponding to a certain individual, the individual is selected. The role of the selection operation is to retain the high-quality solutions in the solution space and provide reproduction opportunities for these solutions so that excellent genes can be passed on to the next generation. By continuously screening and retaining the optimal solution, the genetic algorithm can gradually narrow the solution space and converge to the optimal solution.

[0064] S42: Crossover, crossover operation is a key genetic operation in genetic algorithms, which is used to generate new offspring by inheriting and recombining the genes of the parents, thereby creating a new solution. This process imitates the mechanism of biological inheritance in nature, and produces a potentially better offspring solution by recombining the characteristics of the two parents, so that the algorithm retains certain excellent characteristics when exploring the solution space, and can fully develop new potential solutions.

[0065] The partial mapping crossover operation is a commonly used crossover method. It is widely used to solve combinatorial optimization problems such as the traveling salesman problem because of its simple implementation and good inheritance of the genetic characteristics of the parental solutions. Specifically, for two given parental solutions, two crossover points r1 and r2 are first randomly generated. These points divide the solution path into three parts. Then, some of the genes between the crossover points are exchanged between the two parental solutions. However, the crossover operation may cause duplicate genes or loss of some necessary genes in the offspring solution. To solve this problem, a mapping-based conflict resolution mechanism is adopted: the genes of the exchanged fragments are retained, and the duplicate genes in other fragments are mapped, and the duplicate genes are replaced with their corresponding genes in the exchanged fragments. The crossover process is as follows: Figure 2 As shown in the figure. The offspring after crossover can not only retain the excellent characteristics of the parental solution, but also produce a better solution through gene recombination. This operation plays an important role in the process of exploring the solution space, providing the genetic algorithm with a variety of solution candidates, thereby improving the algorithm's global search ability and solution efficiency.

[0066] S43: Mutation, mutation operation is another important genetic operation in genetic algorithms. Its main function is to introduce new characteristics by randomly adjusting the genes of the solution, thereby increasing genetic diversity. Unlike the crossover operation, the mutation operation does not generate offspring through the genetic recombination of the parents, but directly modifies the genes of the individual solution. This process not only simulates the gene mutation phenomenon in biological evolution, but also provides the algorithm with the ability to explore new solution space, which helps to avoid the population from falling into the local optimal solution.

[0067] Exchange mutation is a common and efficient mutation method. It is widely used because of its simple implementation, flexible operation and ability to effectively increase the genetic diversity of the population. Specifically, for each individual solution in the population, the execution steps of exchange mutation are as follows: First, two gene positions r1 and r2 are randomly generated. These positions represent two randomly selected genes in the current individual solution. Then, in the mutation operation, the gene values ​​at these two positions are exchanged. The mutation process is as follows: Figure 3 This simple exchange of genes changes the arrangement structure of the current solution, thus introducing new potential solutions to the population.

[0068] S5: The improved genetic algorithm significantly enhances the performance of the algorithm by introducing a local search mechanism based on the traditional genetic algorithm. Traditional genetic algorithms mainly rely on global search operations, such as selection, crossover, and mutation, to find better solutions in the solution space. However, this method may be limited in search efficiency and solution quality, and it is easy to fall into local optimality. To overcome this problem, the improved genetic algorithm combines a local search strategy, which not only effectively improves the quality of the solution, but also significantly accelerates the convergence speed of the algorithm. Specifically, the fitness value of the population is calculated, and local searches are performed on individuals with better fitness in the population. S4 is repeated for those with poorer fitness, and global search continues. The local search includes:

[0069] S51: Select several individuals with higher fitness from the current population to form a sub-population to be optimized;

[0070] S52: randomly generate three different integers r1, r2 and r3, representing three position indexes in the current individual, and remove the three edges at the corresponding positions in the current individual;

[0071] S53: After removing the above three edges, the intermediate segments between the index positions are reversed and the disconnected path segments are reconnected to form a new individual.

[0072] The local search process is as follows Figure 4 As shown in the figure. In the improved genetic algorithm, some individuals with better fitness perform local search because they are already in the better area of ​​the solution space. Further optimization of individuals close to the global optimal solution is conducive to finding a more accurate local optimal solution. The remaining individuals repeat the global search and continue to explore more extensively in the solution space to maintain the diversity of the population.

[0073] S6: After the introduction of local search, the performance of the algorithm depends on several key parameters, including the crossover rate p c , mutation rate p m , and the proportion of individuals in the population participating in local search p l Traditional GA usually uses fixed parameter values. However, the optimal parameter settings may change over time. Fixed parameter settings may cause the algorithm to fall into a local optimal solution or converge slowly. Therefore, the algorithm parameters are dynamically adjusted according to the changes in population fitness, including:

[0074] S61: Record the performance of each generation of the population, including the fitness value of the population and the crossover rate p c , mutation rate p m , and the proportion of individuals participating in local search in the population p l ;

[0075] S62: When iterating to the fifth generation and beyond, calculate the average fitness value of the last five generations Specifically, it can be expressed as:

[0076]

[0077] Where fitness(t) represents the fitness value of the tth generation;

[0078] S63: Compare the current fitness value with the average fitness value. If the fitness of the current generation deteriorates (i.e. Lower p c With p l , improve p m , specifically expressed as:

[0079]

[0080] If the fitness improves (i.e. Lower p m , increase p c With p l , specifically expressed as:

[0081]

[0082] In order to balance the exploration and stability of the algorithm, the three parameter settings are constrained as follows:

[0083]

[0084] This dynamic parameter adjustment method based on changes in population fitness enables the algorithm to self-optimize according to performance at different stages, thereby improving solution efficiency, avoiding premature convergence, and exploring better solutions in a larger range.

[0085] S7: Loop end judgment: If the termination condition (maximum number of iterations) is reached, the complete path of the UAV is planned; otherwise, return to step 5 and continue to iterate; finally, the UAV's shortest flight path is obtained.

[0086] Based on the same inventive concept, an intelligent path planning method and system for unmanned aerial vehicles disclosed in an embodiment of the present invention includes: an initialization module for initializing a parameter set, including unmanned aerial vehicle related parameters and optimization algorithm parameters; a global search module, implemented by a genetic algorithm, using selection, crossover and mutation operations to explore the global solution space and preliminarily generate feasible paths; a local search module, optimizing some paths with higher fitness, improving the quality of the solution by locally adjusting the path structure, while avoiding falling into the local optimum, so that the individual gradually approaches the global optimal solution; and a meta-learning module, dynamically adjusting the genetic algorithm parameters, optimizing key parameters such as the crossover rate and mutation rate according to the population performance, and ensuring the adaptability and stability of the algorithm at different stages. The specific implementation details of each module refer to the above method embodiment, which will not be repeated here.

[0087] Based on the same inventive concept, an embodiment of the present invention discloses a computer system, including a memory, a processor, and a computer program stored in the memory and executable on the processor. When the computer program is loaded into the processor, the method and system for intelligent path planning of a UAV is implemented.

Claims

1. A method and system for intelligent path planning of unmanned aerial vehicles, characterized in that: The steps include: S1: Initialize the parameter set, including drone related parameters and optimization algorithm parameters; S2: Encode the path points and use a coding method suitable for genetic algorithms to represent the planned path of the UAV; S3: Taking the inverse of the drone flight time as the objective function, calculate the fitness values ​​of all individuals in the population; S4: Generate new populations through selection, crossover and mutation operations; S5: Recalculate the fitness value of the population, perform local search on individuals with better fitness in the population, repeat S4 on those with poorer fitness, and continue global search; S6: When iterating to the fifth generation, the algorithm parameters are dynamically adjusted according to the changes in population fitness; S7: Execute S5-S6 in a loop until the termination condition is met, and output the path with the highest fitness value as the optimal flight path of the UAV, that is, the path with the shortest flight time.

2. The intelligent path planning method and system for unmanned aerial vehicles according to claim 1, characterized in that: The encoding method described in step S2 is specifically as follows: Let Pa = {pa1, pa2, ..., pa m } represents the target points of m drones, where pa k =(x ak ,y ak , z ak ), represents the coordinates of the kth target point. Let the positive integer set PA = {PA1, PA2, PA3, ..., PA m } represents the path of the drone, where the gene value PA i represents the label of the target point, and m represents the sequence length.

3. The intelligent path planning method and system for unmanned aerial vehicles according to claim 1, characterized in that: The flight time T of the drone described in step S3 A Specifically expressed as: Where pa0 represents the starting point of the drone, S a，b represents the distance between points a and b, v a represents the flight speed of the drone. Then the fitness value of the population can be expressed as:

4. The UAV intelligent path planning method and system according to claim 1, characterized in that: The operation described in step S4 includes: S41: Selection, first evaluate the fitness of each individual. The higher the fitness, the greater the proportion of the individual on the roulette wheel. Then add up the fitness of all individuals to get the total fitness. Then calculate the selection probability. The selection probability of each individual is equal to the ratio of its fitness to the total fitness. Then build a roulette wheel, and map the fitness value to the roulette wheel according to the selection probability of each individual to form a continuous probability distribution. Finally, randomly select, generate a random number between 0 and 1, and select individuals according to the interval on the roulette wheel. If the random number falls within the interval corresponding to an individual, the individual is selected; S42: Crossover, for two given parent solutions, first generate two random numbers r1 and r2, and then exchange some genes in the two parent solutions between r1 and r2. The crossover operation may result in duplicate genes in the solution. In this case, the exchanged genes in the offspring solution are retained, and the duplicate genes in other fragments are mapped to the corresponding genes of the conflicting genes in the exchanged fragment; S43: Mutation. For each solution in the population, first generate two random numbers r1 and r2, and then perform a mutation operation to exchange the first gene and the second gene in the solution.

5. The intelligent path planning method and system for unmanned aerial vehicles according to claim 1, characterized in that: The local search described in step S5 includes: S51: Select several individuals with higher fitness from the current population to form a sub-population to be optimized; S52: randomly generate three different integers r1, r2 and r3, representing three position indexes in the current individual, and remove the three edges at the corresponding positions in the current individual; S53: After removing the above three edges, the intermediate segments between the index positions are reversed and the disconnected path segments are reconnected to form a new individual.

6. The UAV intelligent path planning method and system according to claim 1, characterized in that: The dynamically adjusted parameters in step S6 include: S61: Record the performance of each generation of the population, including the fitness value of the population and the crossover rate p c , mutation rate p m , and the proportion of individuals participating in local search in the population p l ; S62: When iterating to the fifth generation and beyond, calculate the average fitness value of the last five generations Specifically, it can be expressed as: Where fitness(t) represents the fitness value of the tth generation; S63: Compare the current fitness value with the average fitness value. If the fitness of the current generation deteriorates (i.e. Lower p c With p l , improve p m , specifically expressed as: If the fitness improves (i.e. ), reduce p m , increase p c With p l , specifically expressed as: In order to balance the exploration and stability of the algorithm, the three parameter settings are constrained as follows:

7. A method and system for intelligent path planning of unmanned aerial vehicles, characterized in that: include: Initialization module, used to initialize the parameter set, including drone-related parameters and optimization algorithm parameters; The global search module, implemented through a genetic algorithm, uses selection, crossover, and mutation operations to explore the global solution space and preliminarily generate feasible paths; The local search module optimizes some paths with higher fitness, improves the quality of the solution by locally adjusting the path structure, and avoids falling into the local optimum, so that the individual gradually approaches the global optimal solution; As well, the meta-learning module dynamically adjusts the genetic algorithm parameters and optimizes key parameters such as the crossover rate and mutation rate according to the population performance, ensuring the adaptability and stability of the algorithm at different stages.

8. A computer system, characterized in that: It includes a memory, a processor, and a computer program stored in the memory and executable on the processor. When the computer program is loaded into the processor, the method and system for intelligent path planning of a drone according to any one of claims 1 to 6 are implemented.

Images