Unmanned vehicle path planning method and system based on air-ground collaborative awareness, and medium

Abstract

The invention provides an unmanned vehicle path planning method and system based on air-ground collaborative awareness, and a medium, and the method comprises the following steps: building a grid map of a ground environment based on a sensing system and a positioning navigation system carried by an unmanned aerial vehicle, enabling the unmanned vehicle to obtain the grid map built by the unmanned aerial vehicle, and correcting and supplementing the grid map in combination with environment information sensed by the unmanned vehicle; according to the corrected and supplemented grid map, establishing an unmanned vehicle path planning target function based on air-ground collaborative awareness with the shortest moving path, or the shortest consumed time, or the least consumed energy as a constraint; according to the objective function, carrying out optimal position updating through an improved blackbone gull optimization algorithm, and determining an optimal blackbone gull position; and determining an optimal path planning result according to the optimal blackbone gull position which is sequentially updated according to a preset maximum number of iterations. According to the method, the problems of local optimum and convergence precision of the Warner gull optimization algorithm are solved, and the unmanned vehicle path planning effect can be remarkably improved.

Description

Technical Field

[0001] The present invention relates to the technical field of unmanned vehicles, and in particular to an unmanned vehicle path planning method based on air-ground collaborative perception. Background Art

[0002] The air-ground heterogeneous robot system composed of aerial drones and ground unmanned vehicles is a hot topic in the research of distributed artificial intelligence technology. The organic coordination and cross-domain collaboration of drones and unmanned vehicles will lead the new model of future robot technology and application. Unmanned vehicles can accurately locate ground targets at close range, but when environmental information is unknown or partially known, the on-board sensors have great limitations in their ability to perceive the environment and can only achieve local path planning. Drones have a wider field of view and can obtain global information about the surrounding environment at a specific height, but they also lose a lot of local information due to the height. Through the collaboration of the two and the complementary advantages, global path planning of unmanned vehicles can be achieved.

[0003] The path planning problem of unmanned vehicles based on air-ground collaborative perception can actually be regarded as a complex optimization problem with constraints. Therefore, some intelligent optimization algorithms have played a positive role in improving the effect of unmanned vehicle path planning, and many scholars have carried out a lot of research work. For example, Chen Xuejun et al. improved the intelligent water drop algorithm and proposed an unmanned vehicle obstacle avoidance path planning method based on the improved water drop algorithm (Chen Xuejun, Bei Shaoyi. An unmanned vehicle obstacle avoidance path planning method based on an improved intelligent water drop algorithm [P]. Jiangsu Province: CN110703767A, 2020-01-17.); Liu Hongdan et al. proposed a quantum wolf pack algorithm and used it for automatic obstacle avoidance research of unmanned intelligent vehicles (Liu Sheng, Zhang Lanyong, Ding Yixuan, Li Bing, Li Peng, Sun Yue. Automatic collision avoidance method for unmanned intelligent vehicles based on quantum wolf pack algorithm [P]. Heilongjiang Province: CN110471426A, 2019-11-19.); Jiang Pengcheng et al. proposed a path planning method for unmanned vehicles based on ant colony algorithm (Jiang Pengcheng, Cong Hua, Qiu Mianhao, Zhang Nan, Feng Fuzhou, Zhang Chuanqing, Liu Xixia, Zhang Lixia, He Jiawu, Zhang Xiaoming, Wang Zhirong, Yang Changwei. Hybrid path planning algorithm for unmanned vehicles [P]. Beijing: CN110609557A, 2019-12-24.); Ge Hongwei et al. proposed an improved multi-objective particle swarm algorithm, and used it as an optimization method to study the path planning of unmanned vehicles (Ge Hongwei, Qian Xiaoyu, Ge Yang. Unmanned vehicle path planning method based on improved multi-objective particle swarm algorithm [P]. Jiangsu: CN107992051A, 2018-05-04.).

[0004] According to current research results, intelligent optimization algorithm is an effective path planning method. The Sooty tern optimization algorithm (STOA) is a new intelligent optimization algorithm that simulates the foraging behavior of sooty terns and can also be applied to path planning problems. However, the Sooty tern optimization algorithm still has some defects, which makes the algorithm easy to fall into local optimality and low convergence accuracy. When performing path planning, it often fails to achieve the ideal path planning effect. Summary of the invention

[0005] To solve the above problems, the present invention provides an unmanned vehicle path planning method based on air-ground collaborative perception. The method introduces the position update mechanism of the reptile search algorithm and the satin bowerbird optimization algorithm, which effectively solves the problem of mediocre planning effect caused by easy falling into local optimality during path planning.

[0006] To achieve the above object, the present invention provides the following technical solutions.

[0007] A method for unmanned vehicle path planning based on air-ground collaborative perception includes the following steps:

[0008] A grid map of the ground environment is established based on the perception system and positioning navigation system carried by the drone. The unmanned vehicle obtains the grid map established by the drone and corrects and supplements the grid map based on the environmental information perceived by itself.

[0009] According to the modified and supplemented grid map, the starting point is determined, and an objective function with the shortest moving path, the shortest time consumption or the least energy consumption as constraints is established;

[0010] According to the objective function, the optimal position is updated in the grid map by using the improved sooty tern optimization algorithm to determine the optimal sooty tern position; wherein, the improved sooty tern optimization algorithm introduces the position update mechanism of the reptile search algorithm and the satin bowerbird optimization algorithm into the position update method of the original sooty tern optimization algorithm, and comprehensively considers the factors of switching different position update modes with different iteration times, the optimal position of the sooty tern in this iteration, the average value of each dimension of the current solution, the difference percentage between each dimension of the current solution and the optimal solution, the positions of other sooty terns within the population, and the fitness value of the courtship nest to update the sooty tern position;

[0011] The optimal path planning result is determined by sequentially updating the optimal sooty tern position according to the preset maximum number of iterations.

[0012] Preferably, the updating of the optimal position by using the improved Sooty Tern optimization algorithm comprises the following steps:

[0013] Set the size of the sooty tern population, the maximum number of iterations, and the upper and lower bounds of the sooty tern optimization;

[0014] Introduce Gaussian mapping to initialize the population location of sooty terns;

[0015] The migration and attack behaviors of sooty terns were simulated, and the position update mechanism of reptile search algorithm and satin bowerbird optimization algorithm was introduced to improve the position update method of sooty terns, perform position update, and record the optimal position of sooty terns in the iteration;

[0016] The optimal position of the sooty tern is further updated through bidirectional sine mutation in each dimension, and the optimal position in the iteration is recorded;

[0017] The optimal position is updated in sequence according to the preset number of iterations, and the optimal path result is output.

[0018] Preferably, the introducing of Gaussian mapping to initialize the position of the sooty tern population comprises the following steps:

[0019] Determine the population size N, the lower boundary LB and the upper boundary UB of the sooty tern;

[0020] Generate random number x through Gaussian mapping t :

[0021]

[0022] In the formula, mod(·) is the remainder function, x t+1 is the next random number;

[0023] Use the generated Gaussian random numbers to initialize the positions of the sooty terns:

[0024] P s (t) = (UB - LB) × x t +LB.

[0025] Preferably, the position update mechanism of introducing the reptile search algorithm and the satin bowerbird optimization algorithm to improve the position update mode of the sooty tern, and performing position update, comprises the following steps:

[0026] Simulating the migration behavior of sooty terns, including conflict avoidance, aggregation, and regeneration;

[0027] Conflict avoidance: The conflict avoidance behavior process of the sooty tern is simulated and expressed as follows:

[0028] C s (t) = S A ×P s (t);

[0029] Where: P s (t) represents the position of the sooty tern at the current iteration t; C s(t) represents the new position of the sooty tern without colliding with other sooty terns; S A Represents a variable factor to avoid collision, which is used to calculate the position after avoiding collision. Its constraint formula is as follows:

[0030] S A =C f -(t×C f / Miter);

[0031] Where: C f To adjust S A The control variable; t represents the current iteration number; S A As the number of iterations increases, from C f Gradually decreases to 0; if we assume that C f =2, S A will gradually decrease from 2 to 0; Miter is the number of iterations;

[0032] Aggregation: Aggregation means that the current sooty tern moves closer to the best position among the adjacent sooty terns under the premise of avoiding conflict, that is, moves closer to the optimal position. Its mathematical expression is as follows:

[0033] M s (t) = C B ×(P bs (t)-P s (t));

[0034] Where: P bs (t) is the optimal position of the sooty tern in the tth iteration; M s (t) represents the position P at different locations s (t) Towards the optimal position P bs (t) The process of moving; C B is a random variable that makes exploration more comprehensive, and varies according to the following formula:

[0035] C B =0.5×rand;

[0036] Where: rand is a random number in the range of [0,1];

[0037] Update: Update means that the current sooty tern moves towards the direction of the optimal position and updates the position. Its mathematical expression is:

[0038] D s (t) = C s (t)+M s (t);

[0039] Where: D s (t) is the distance the sooty tern moves from the current position to the optimal position;

[0040] Simulating the attack behavior of sooty terns: During migration, sooty terns can increase their flight altitude by using their wings, and can also adjust their speed and attack angle. When attacking prey, their hovering behavior in the air can be defined as the following mathematical model:

[0041]

[0042] Where: r is the radius of each spiral; θ is a random angle value in the range [0, 2π]; u and v are related constants that define the shape of the spiral, both of which can be set to 1; e is the base of the natural logarithm;

[0043] By introducing the reptile search algorithm and the satin bowerbird optimization algorithm, the improved sooty tern position update formula is as follows:

[0044]

[0045] Where:

[0046] η (j) (t) = P bs(j) (t)×P (j) (t);

[0047]

[0048] Among them, P s(j) (t+1) is the sooty tern P after the t+1th iteration s The position of the jth dimension of (t+1); P s(j) (t) is the number of sooty terns after the tth iteration s The position of the jth dimension of (t); P bs(j) (t) represents the optimal position P after the current tth iteration bs The jth dimension of (t); D s(j) (t) is the distance D that the sooty tern moves to its new location s (t) is the jth dimension; γ is a random number between [0,1]; rand is a random number between [0,1]; η (j) (t) represents the sooty tern P at the tth iteration s (t) is the j-th dimension hunting operator; ε is a small positive number; R (j) (t) is a reduction function used to reduce the search area; M(P s (t)) indicates Sooty Tern P s (t) is the average position of all j-th dimensions; n represents the dimension of the planning problem to be solved; P (j) (t) represents the best solution P of the sooty tern after the tth iteration bs (t) and the current solution P sThe percentage difference of the j-th dimension position of (t); P s (t) is the position of the sooty tern after the tth iteration; r1 and r2 are random integers between [1,N], represents the r1th sooty tern after the tth iteration The j-th dimension position of represents the r2th sooty tern after the tth iteration The j-th dimension position of (j) and LB (j) denote the upper and lower bounds of the j-th dimension position of the sooty tern respectively; ES(t) is the evolution factor, which randomly decreases between 2 and -2 during the entire iteration process; r3 denotes a random integer between [-1,1]; α is a sensitive parameter used to control the search accuracy of hunting cooperation during the iteration process; β is a sensitive parameter that controls the exploration accuracy of the encirclement phase during the iteration process; F f (·) is the fitness function when calculating the fitness value; F t is the fitness value of the courtship nest;

[0049] Calculate the fitness value:

[0050] fitness(t)=F f (P s (t+1));

[0051] In the formula, F f (·) is the fitness function when calculating the fitness value;

[0052] Record the best sooty tern in the iteration.

[0053] Preferably, further updating the optimal position by bidirectional sine mutation dimension by dimension comprises the following steps:

[0054] For dimension j, the sine chaos value is calculated according to the current number of iterations, and the positive and negative directions are switched with equal probability:

[0055] sinValue = sin(πx0);

[0056]

[0057] In the formula, rand is a random number between 0 and 1; x0 is the iteration sequence value;

[0058] Perform mutation perturbation on the optimal position:

[0059] P bs(j) (t+1)′=P bs(j) (t+1)+SinValue×P bs(j) (t+1)

[0060] Where: P bs(j) (t+1) represents the optimal position P of the t+1th iteration bs The jth dimension of (t+1);

[0061] Greedy Update:

[0062]

[0063] After each dimension has been mutated, stop mutation.

[0064] A path planning system for an unmanned vehicle based on air-ground collaborative perception, the system comprising:

[0065] processor;

[0066] a memory having stored thereon a computer program executable on the processor;

[0067] Among them, when the computer program is executed by the processor, the steps of the unmanned vehicle path planning method based on air-ground collaborative perception are implemented.

[0068] A computer-readable storage medium stores a data processing program, which, when executed by a processor, implements the steps of the unmanned vehicle path planning method based on air-ground collaborative perception.

[0069] Beneficial effects of the present invention:

[0070] The present invention proposes an unmanned vehicle path planning method based on air-ground collaborative perception. The method improves the position update method of the sooty tern, introduces the position update mechanism of the reptile search algorithm and the satin bowerbird optimization algorithm to improve the position update method of the sooty tern, and comprehensively considers factors such as switching different position update modes with different iteration times, the optimal position of the sooty tern in this iteration, the average value of each dimension of the current solution, the difference percentage between each dimension of the current solution and the optimal solution, the position of other sooty terns within the population, and the fitness value of the courtship bird nest to update the position of the sooty tern, thereby increasing the search range and avoiding falling into the local optimum. Combined with the objective function of path planning, better path planning points can be obtained, and further better path planning results are obtained. BRIEF DESCRIPTION OF THE DRAWINGS

[0071] Figure 1 is a flow chart of a method for unmanned vehicle path planning based on air-ground collaborative perception according to an embodiment of the present invention;

[0072] Figure 2 is a path planning result of an unmanned vehicle path planning method based on air-ground collaborative perception in an embodiment of the present invention;

[0073] Figure 3It is an iterative process curve of an unmanned vehicle path planning method based on air-ground collaborative perception in an embodiment of the present invention. DETAILED DESCRIPTION

[0074] 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 used to explain the present invention and are not intended to limit the present invention.

[0075] Example 1

[0076] The present invention proposes an unmanned vehicle path planning method based on air-ground collaborative perception. Aiming at several problems existing in STOA, an improved sooty tern optimization algorithm (ISTOA) is proposed and used for unmanned vehicle path planning based on air-ground collaborative perception. The specific process is as follows Figure 1 As shown, the specific steps are as follows:

[0077] S1: A grid map of the ground environment is established based on the perception system and positioning navigation system carried by the drone. The unmanned vehicle obtains the grid map established by the drone and corrects and supplements the grid map based on the environmental information it perceives.

[0078] S2: Establish the objective function of the unmanned vehicle path planning based on air-ground information coordination (which can be set to the shortest moving path, shortest time consumption, least energy consumption, etc. according to actual needs), and set the corresponding constraints and the number of key nodes D of the path.

[0079] S3: Parameter setting, mainly including: the size of the sooty tern population (i.e. the number of sooty tern individuals) N; the maximum number of iterations (i.e. the condition for stopping iteration) Miter; the lower boundary LB of the sooty tern optimization; the upper boundary UB of the sooty tern optimization.

[0080] S4: Determine the size of the population N, the optimization lower boundary LB of the black tern and the optimization upper boundary UB of the black tern; introduce Gaussian mapping to initialize the position of the black tern population, including the following steps:

[0081] Determine the population size N, the lower boundary LB and the upper boundary UB of the sooty tern;

[0082] Generate random number x through Gaussian mapping t :

[0083]

[0084] In the formula, mod(·) is the remainder function, x t+1 The next random number.

[0085] Use the generated Gaussian random numbers to initialize the positions of the sooty terns:

[0086] P s (t) = (UB - LB) × x t +LB.

[0087] By introducing Gaussian mapping to initialize the position of the sooty tern population, the uniformity and diversity of the population position distribution can be improved, thereby enhancing the stability of the algorithm.

[0088] S5: Conflict avoidance: Simulate the conflict avoidance behavior of the sooty tern, expressed as follows:

[0089] C s (t) = S A ×P s (t);

[0090] Where: P s (t) represents the position of the sooty tern at the current iteration t; C s (t) represents the new position of the sooty tern without colliding with other sooty terns; S A Represents a variable factor to avoid collision, which is used to calculate the position after avoiding collision. Its constraint formula is as follows:

[0091] S A =C f -(t×C f / Miter);

[0092] Where: C f To adjust S A The control variable; t represents the current iteration number; S A As the number of iterations increases, from C f Gradually decreases to 0; if we assume that C f =2, S A Will gradually decrease from 2 to 0; Miter is the number of iterations.

[0093] Aggregation: Aggregation means that the current sooty tern moves closer to the best position among the adjacent sooty terns under the premise of avoiding conflict, that is, moves closer to the optimal position. Its mathematical expression is as follows:

[0094] M s (t) = C B ×(P bs (t)-P s (t));

[0095] Where: P bs (t) is the optimal position of the sooty tern in the tth iteration; M s (t) represents the position P at different locations s (t) Towards the optimal position P bs (t) The process of moving; CB is a random variable that makes exploration more comprehensive, and varies according to the following formula:

[0096] C B =0.5×rand;

[0097] Where: rand is a random number in the range [0,1].

[0098] Update: Update means that the current sooty tern moves towards the direction of the optimal position and updates the position. Its mathematical expression is:

[0099] D s (t) = C s (t)+M s (t);

[0100] Where: D s (t) is the distance the sooty tern moves from its current position to the optimal position.

[0101] Attack behavior: During migration, sooty terns can use their wings to increase their flight altitude, and can also adjust their speed and attack angle. When attacking prey, their hovering behavior in the air can be defined as the following mathematical model:

[0102]

[0103] Where r is the radius of each spiral; θ is a random angle value in the range [0, 2π]; u and v are related constants that define the shape of the spiral, both of which can be set to 1; e is the base of the natural logarithm.

[0104] In the original sooty tern algorithm, only the optimal sooty tern position is used for guidance to update the sooty tern position. In order to more effectively improve the global search ability of the sooty tern, the position update mechanism of the reptile search algorithm and the satin bowerbird optimization algorithm is introduced to improve the sooty tern position update method. The sooty tern position is updated by comprehensively considering factors such as switching different position update modes with different iteration times, the optimal position of the sooty tern in this iteration, the average value of each dimension of the current solution, the difference percentage between each dimension of the current solution and the optimal solution, the position of other sooty terns within the population, and the fitness value of the courtship nest, so as to avoid the local optimum in each iteration, thereby improving the global search ability of the sooty tern algorithm.

[0105] By introducing the reptile search algorithm and the satin bowerbird optimization algorithm, the improved sooty tern position update formula is as follows:

[0106]

[0107] Where:

[0108] η (j)(t) = P bs(j) (t)×P (j) (t);

[0109]

[0110]

[0111] Among them, P s(j) (t+1) is the sooty tern P after the t+1th iteration s The position of the jth dimension of (t+1); P s(j) (t) is the number of sooty terns after the tth iteration s The position of the jth dimension of (t); P bs(j) (t) represents the optimal position P after the current tth iteration bs The jth dimension of (t); D s(j) (t) is the distance D that the sooty tern moves to its new location s (t) is the jth dimension; γ is a random number between [0,1]; rand is a random number between [0,1]; η (j) (t) represents the sooty tern P at the tth iteration s (t) is the j-th dimension hunting operator; ε is a small positive number; R (j) (t) is a reduction function used to reduce the search area; M(P s (t)) indicates Sooty Tern P s (t) is the average position of all j-th dimensions; n represents the dimension of the planning problem to be solved; P (j) (t) represents the best solution P of the sooty tern after the tth iteration bs (t) and the current solution P s The percentage difference of the j-th dimension position of (t); P s (t) is the position of the sooty tern after the tth iteration; r1 and r2 are random integers between [1,N], represents the r1th sooty tern after the tth iteration The j-th dimension position of represents the r2th sooty tern after the tth iteration The j-th dimension position of (j) and LB (j) denote the upper and lower bounds of the j-th dimension position of the sooty tern respectively; ES(t) is the evolution factor, which randomly decreases between 2 and -2 during the entire iteration process; r3 denotes a random integer between [-1,1]; α is a sensitive parameter used to control the search accuracy of hunting cooperation during the iteration process; β is a sensitive parameter that controls the exploration accuracy of the encirclement phase during the iteration process; F f (·) is the fitness function when calculating the fitness value; F tis the fitness value of the courtship bird nest.

[0112] The present invention improves the position update method of the sooty tern, introduces the position update mechanism of the reptile search algorithm and the satin bowerbird optimization algorithm to improve the position update method of the sooty tern, comprehensively considers different position update modes switched with different iteration times, the optimal position of the sooty tern in this iteration, the average value of each dimension of the current solution, the difference percentage between each dimension of the current solution and the optimal solution, the position of other sooty terns within the population, the fitness value of the courtship bird nest and other factors to update the position of the sooty tern, thereby achieving an increase in the search range of the algorithm and enhancing the adaptability of the algorithm.

[0113] S6: Calculate the fitness value.

[0114] fitness(t)=F f (P s (t+1));

[0115] In the formula, F f (·) is the fitness function when calculating the fitness value;

[0116] S7: Record information, record the best sooty tern in the iteration.

[0117] S8: For the optimal sooty tern, bidirectional sine mutation is performed dimension by dimension, realizing the ability of the algorithm to jump out of the local optimal solution in the later stage. For dimension j, the sine chaos value is first calculated according to the current number of iterations, and the positive and negative directions are switched with equal probability.

[0118] sinValue = sin(πx0);

[0119]

[0120] In the formula, rand is a random number between 0 and 1; x0 is the iteration sequence value;

[0121] Perform mutation perturbation on the optimal position:

[0122] P bs(j) (t+1)′=P bs(j) (t+1)+SinValue×P bs(j) (t+1);

[0123] Where: P bs(j) (t+1) represents the optimal position P of the t+1th iteration bs The jth dimension of (t+1);

[0124] Greedy Update:

[0125]

[0126] After each dimension has been mutated, stop mutation.

[0127] S9: Record information, record the best sooty tern in the iteration.

[0128] S10: Repeat steps S5 to S9. When the maximum number of iterations Miter is reached, the algorithm stops and the optimal path result is output.

[0129] In this embodiment:

[0130] Using MATLAB as the simulation platform, assuming that a 20×20 grid map is constructed, and taking the shortest moving distance as the goal, the STOA method and the ISTOA method are analyzed. The parameters in the STOA algorithm are: N=50, Maxiter=200, LB=1, UB=20; the parameters in the ISTOA algorithm are: N=50, Maxiter=200, LB=1, UB=20. The simulation environment and the moving paths obtained by the two methods are shown in Figure 2. Figure 2 As shown, Figure 3 is the iterative process curve. Table 1 is the comparison of the data results of the two algorithms.

[0131] Table 1. Comparison of algorithm path results;

[0132] algorithm Path length STOA 40.6274 ISTOA 32.6274

[0133] from Figure 2 It can be intuitively found that the moving path obtained by STOA is longer and more circuitous than that of ISTOA, while the path obtained by ISTOA is more reasonable. Figure 2 and Figure 3 From the results in , we can see that when the STOA algorithm is used, the algorithm converges slowly; when the ISTOA algorithm is used, the convergence speed is faster and a better path can be found faster. It can be seen that the ISTOA algorithm designed in this paper has a faster convergence speed and convergence accuracy. The simulation results show that under a variety of identical environments, the ISTOA algorithm has a stronger search capability and obtains a better mobile path, verifying the effectiveness of the algorithm.

[0134] The above are only preferred embodiments of the present invention and are not intended to limit the present invention. Any modifications, equivalent substitutions and improvements made within the spirit and principles of the present invention should be included in the protection scope of the present invention.

Claims

1. A path planning method for unmanned vehicles based on air-ground collaborative perception, It is characterized in that The following steps are involved: A grid map of the ground environment is established based on the perception system and positioning navigation system carried by the drone. The unmanned vehicle obtains the grid map established by the drone and corrects and supplements the grid map based on the environmental information perceived by itself. According to the modified and supplemented grid map, the starting point is determined, and an objective function with the shortest moving path, the shortest time consumption or the least energy consumption as constraints is established; According to the objective function, the optimal position is updated in the grid map by using the improved sooty tern optimization algorithm to determine the optimal sooty tern position; wherein, the improved sooty tern optimization algorithm introduces the position update mechanism of the reptile search algorithm and the satin bowerbird optimization algorithm into the position update method of the original sooty tern optimization algorithm, and comprehensively considers the factors of switching different position update modes with different iteration times, the optimal position of the sooty tern in this iteration, the average value of each dimension of the current solution, the difference percentage between each dimension of the current solution and the optimal solution, the positions of other sooty terns within the population, and the fitness value of the courtship nest to update the sooty tern position; The optimal path planning result is determined by sequentially updating the optimal sooty tern position according to the preset maximum number of iterations.

2. According to claim 1, a method for unmanned vehicle path planning based on air-ground collaborative perception, It is characterized in that The optimal position update by the improved Sooty Tern optimization algorithm comprises the following steps: Set the size of the sooty tern population, the maximum number of iterations, and the upper and lower bounds of the sooty tern optimization; Introduce Gaussian mapping to initialize the population location of sooty terns; The migration and attack behaviors of sooty terns were simulated, and the position update mechanism of reptile search algorithm and satin bowerbird optimization algorithm was introduced to improve the position update method of sooty terns, perform position update, and record the optimal position of sooty terns in the iteration; The optimal position of the sooty tern is further updated through bidirectional sine mutation in each dimension, and the optimal position in the iteration is recorded; The optimal position is updated in sequence according to the preset number of iterations, and the optimal path result is output.

3. According to claim 2, a method for unmanned vehicle path planning based on air-ground collaborative perception, It is characterized in that The method of introducing Gaussian mapping to initialize the position of the sooty tern population includes the following steps: Determine the population size N, the lower boundary LB and the upper boundary UB of the sooty tern; Generate random number x through Gaussian mapping t : In the formula, mod(·) is the remainder function, x t+1 is the next random number; Use the generated Gaussian random numbers to initialize the positions of the sooty terns: P s (t)=(UB-LB)×x t +LB。 4. According to claim 2, a method for unmanned vehicle path planning based on air-ground collaborative perception, It is characterized in that The method of introducing the position update mechanism of the reptile search algorithm and the satin bowerbird optimization algorithm to improve the position update method of the sooty tern and perform position update includes the following steps: Simulating the migration behavior of sooty terns, including conflict avoidance, aggregation, and regeneration; Conflict avoidance: The conflict avoidance behavior process of the sooty tern is simulated and expressed as follows: C s (t)=S A ×P s (t); Where: P s (t) represents the position of the sooty tern at the current iteration t; C s (t) represents the new position of the sooty tern without colliding with other sooty terns; S A Represents a variable factor to avoid collision, which is used to calculate the position after avoiding collision. Its constraint formula is as follows: S A =C f -(t×C f / Miter); Where: C f To adjust S A The control variable; t represents the current iteration number; S A As the number of iterations increases, from C f Gradually decreases to 0; if we assume that C f =2, S A will gradually decrease from 2 to 0; Miter is the number of iterations; Aggregation: Aggregation means that the current sooty tern moves closer to the best position among the adjacent sooty terns under the premise of avoiding conflict, that is, moves closer to the optimal position. Its mathematical expression is as follows: M s (t)=C B ×(P bs (t)-P s (t)); Where: P bs (t) is the optimal position of the sooty tern in the tth iteration; M s (t) represents the position P at different locations s (t) Towards the optimal position P bs (t) The process of moving; C B is a random variable that makes exploration more comprehensive, and varies according to the following formula: C B =0.5×rand; Where: rand is a random number in the range of [0,1]; Update: Update means that the current sooty tern moves towards the direction of the optimal position and updates the position. Its mathematical expression is: D s (t)=C s (t)+M s (t); Where: D s (t) is the distance the sooty tern moves from the current position to the optimal position; Simulating the attack behavior of sooty terns: During migration, sooty terns can increase their flight altitude by using their wings, and can also adjust their speed and attack angle. When attacking prey, their hovering behavior in the air can be defined as the following mathematical model: Where: r is the radius of each spiral; θ is a random angle value in the range [0, 2π]; u and v are related constants that define the shape of the spiral, both of which can be set to 1; e is the base of the natural logarithm; By introducing the reptile search algorithm and the satin bowerbird optimization algorithm, the improved sooty tern position update formula is as follows: Where: η (j) (t)=P bs(j) (t)×P (j) (t); Among them, P s(j) (t+1) is the sooty tern P after the t+1th iteration s The position of the jth dimension of (t+1); P s(j) (t) is the number of sooty terns P after the tth iteration s The position of the jth dimension of (t); P bs(j) (t) represents the optimal position P after the current tth iteration bs The jth dimension of (t); D s(j) (t) is the distance D that the sooty tern moves to its new location s (t) is the jth dimension; γ is a random number between [0, 1]; rand is a random number between [0, 1]; η (j) (t) represents the sooty tern P at the tth iteration s (t) is the j-th dimension hunting operator; ε is a small positive number; R (j) (t) is a reduction function used to reduce the search area; M(P s (t)) indicates Sooty Tern P s (t) is the average position of all the ,th dimension; n represents the dimension of the planning problem to be solved; P (j) (t) represents the best solution P of the sooty tern after the tth iteration bs (t) and the current solution P s The percentage difference of the t-th dimension position; P s (t) is the position of the sooty tern after the tth iteration; r 1 and r 2 is a random integer between [1, N], represents the rth time after the tth iteration 1 Sooty Tern The j-th dimension position of represents the rth time after the tth iteration 2 Sooty Tern The j-th dimension position of (j) and LB (j) denote the upper and lower bounds of the j-th dimension position of the sooty tern, respectively; ES(t) is the evolution factor, which randomly decreases between 2 and -2 during the entire iteration process; r 3 represents a random integer between [-1, 1]; α is a sensitive parameter used to control the search accuracy of hunting cooperation during the iteration process; β is a sensitive parameter that controls the exploration accuracy of the encirclement phase during the iteration process; F f (·) is the fitness function when calculating the fitness value; F t is the fitness value of the courtship nest; Calculate the fitness value: fitness(t)=F f (P s (t+1)); In the formula, F f (·) is the fitness function when calculating the fitness value; Record the best sooty tern in the iteration.

5. According to claim 4, a method for unmanned vehicle path planning based on air-ground collaborative perception, It is characterized in that The method of further updating the optimal position by bidirectional sine mutation in each dimension includes the following steps: For dimension j, the sine chaos value is calculated according to the current number of iterations, and the positive and negative directions are switched with equal probability: sin Value=sin(πx 0 ) In the formula, rand is a random number between 0 and 1; x 0 is the iteration sequence value; Perform mutation perturbation on the optimal position: P bs(j) (t+1)′=P bs(j )(t+1)+SinValue×P bs(j) (t+1); Where: P bs(j) (t+1) represents the optimal position P of the t+1th iteration bs The jth dimension of (t+1); Greedy Update: After each dimension has been mutated, stop mutation.

6. An unmanned vehicle path planning system based on air-ground collaborative perception, It is characterized in that The system comprises: processor; a memory having stored thereon a computer program executable on the processor; Wherein, when the computer program is executed by the processor, the steps of the unmanned vehicle path planning method based on air-ground collaborative perception as described in any one of claims 1 to 5 are implemented.

7. A computer-readable storage medium, It is characterized in that The computer-readable storage medium stores a data processing program, which, when executed by a processor, implements the steps of the unmanned vehicle path planning method based on air-ground collaborative perception as described in any one of claims 1 to 5.

Images