Cooperative target search method for multiple UAVs based on beetle whisker search algorithm

Abstract

The invention discloses a multi-UAV cooperative target search method based on a longicorn beard search algorithm, the method comprising: dividing a mission area into a plurality of grid units; constructing a target existence probability and a target existence probability update mechanism for the grid unit; constructing pheromones and a pheromone update mechanism for the grid unit; constructing a fitness function that can simultaneously consider environmental benefits, target search benefits, avoidance of repeated search benefits and obstacle avoidance benefits; for the UAV, randomly generating a detection main direction; respectively calculating the fitness function values ​​corresponding to the neighborhood grid unit indicated by the detection main direction and the neighborhood grid units on both sides thereof, selecting the grid unit with the highest fitness function value; using the selected grid unit as the next grid unit to be searched by the UAV, and controlling the UAV to move to the selected grid unit. The invention can utilize environmental information, the physical characteristics of the UAV itself and historical search information to realize efficient coverage search and target search in the mission area.

Description

Technical Field

[0001] The present invention relates to the technical field of unmanned aerial vehicle control, and in particular to a multi-unmanned aerial vehicle collaborative target search method based on a longhorn beard search algorithm. Background Art

[0002] Since the collaborative search of multiple UAVs for targets in a designated mission area has very important significance and value in theory and engineering practice, studying the collaborative search mechanism of multiple UAVs plays an important role in improving the efficiency and effectiveness of UAV collaborative search.

[0003] At present, when multiple UAVs conduct collaborative search of the mission area, the mission area is usually divided into grids according to the flight parameters of the UAVs, and then a single UAV is assigned to each grid to perform the task. Alternatively, the UAVs are compared to particles or gene fragments, and the collaborative search of the mission area by multiple UAVs is achieved based on heuristic algorithm mechanisms such as particle swarms or genetic algorithms.

[0004] However, the traditional search method based on regional grid segmentation and assigning drones to search each grid unit individually has no or poor coordination between multiple drones, and the corresponding search efficiency is not high. The multi-UAV collaborative target search method based on traditional heuristic algorithms such as particle swarm requires continuous optimization and iteration using environmental and target information to seek the optimal search track, which has poor timeliness and is only applicable to situations where track planning can be done in advance. It is difficult to achieve fast online track generation. Moreover, the multi-UAV collaborative target search method based on traditional heuristic algorithms such as particle swarm has a sharp increase in the amount of calculation as the number of collaborative drones increases, and it is easy to lose the possibility of practical execution due to the sharp increase in the amount of calculation. Summary of the invention

[0005] In order to solve some or all of the technical problems existing in the above-mentioned prior art, the present invention provides a multi-UAV collaborative target search method based on a longicorn whisker search algorithm.

[0006] The technical solution of the present invention is as follows:

[0007] A multi-UAV cooperative target search method based on a longhorn beard search algorithm is provided, comprising:

[0008] Divide the mission area into a plurality of grid cells, wherein the area of ​​the grid cells is larger than the target size and smaller than or equal to the search area of ​​a single UAV;

[0009] Based on the probability density function and the detection information of the drone sensor, the target existence probability of the grid unit and the target existence probability update mechanism are constructed;

[0010] Based on the familiarity of the drone with the grid unit, the pheromone of the grid unit and the pheromone update mechanism are constructed;

[0011] According to the target existence probability and pheromone of the grid unit, a fitness function is constructed that can simultaneously consider the environmental benefit, target search benefit, avoidance of repeated search benefit and obstacle avoidance benefit;

[0012] For the UAV, a detection main direction is randomly generated, and the direction indicated by the detection main direction has a neighborhood grid unit;

[0013] Calculate the fitness function values ​​of the neighboring grid cells pointed to by the main detection direction and the neighboring grid cells on both sides respectively, and select the grid cell with the highest fitness function value. If there are more than two neighboring grid cells with the same and highest fitness function values, randomly select a grid cell from the grid cells with the highest fitness function value.

[0014] The selected grid cell is used as the next grid cell to be searched by the UAV, and the UAV is controlled to move toward the selected grid cell.

[0015] In some possible implementations, the initial target existence probability of a grid cell is:

[0016]

[0017] Among them, P represents the initial target existence probability corresponding to the grid cell, f(·) represents the mixed two-dimensional Gaussian probability density function corresponding to the target, s represents the area corresponding to f(·), and D represents the integral area corresponding to the grid cell.

[0018] In some possible implementations, the target existence probability update mechanism of the grid unit includes:

[0019] If the UAV detects a target in a grid cell at time k, the target existence probability of the grid cell is updated using the following formula to obtain the target existence probability of the grid cell at the next moment;

[0020]

[0021] If the UAV does not detect the target in the grid cell at time k, the target existence probability of the grid cell is updated using the following formula to obtain the target existence probability of the grid cell at the next moment;

[0022]

[0023] Among them, P(k+1) represents the probability of the target existing in the grid cell at time k+1, P d represents the detection probability of the drone sensor, P(k) represents the probability of the target existing in the grid cell at time k, and P fIndicates the false alarm rate.

[0024] In some possible implementations, the pheromone update mechanism of the grid cell is defined as:

[0025] τ(k)=(1-p)·τ(k-1)+ρΔτ(k)

[0026] Among them, τ(k) represents the pheromone of the grid unit at time k, ρ represents the pheromone volatility coefficient, τ(k-1) represents the pheromone of the grid unit at time k-1, Δτ(k) represents the update amount of the pheromone of the grid unit from time k-1 to time k, Δτ(k)=num*Q, num represents the number of visits to the grid unit from time k-1 to time k, and Q represents the pheromone strength.

[0027] In some possible implementations, the fitness function is defined as:

[0028] J(k)=(ω1J1(k)+ω2J2(k)+ω3J3(k))&&J4(k)

[0029] Among them, J(k) represents the fitness function corresponding to the grid cell at time k, ω1 represents the environmental benefit weight, J1(k) represents the environmental benefit of the task area corresponding to the grid cell at time k, ω2 represents the target search benefit weight, J2(k) represents the target search benefit corresponding to the grid cell at time k, ω3 represents the avoidance of repeated search benefit weight, J3(k) represents the avoidance of repeated search benefit corresponding to the grid cell at time k, and J4(k) represents the obstacle avoidance benefit at time k.

[0030]

[0031] In some possible implementations, the mission area environmental benefits are defined as:

[0032]

[0033] e(k)=(1-P(k))ln(P(k))

[0034] Among them, β1 represents the environmental benefit coefficient, e(k) represents the uncertainty of the task area environment at time k, and P(k) represents the probability of the target existence in the grid cell at time k.

[0035] In some possible implementations, the target search benefit is defined as:

[0036] J2(k)=b(k)P(k)

[0037] Among them, P(k) represents the probability of the target existing in the grid cell at time k, σ represents the weight coefficient.

[0038] In some possible implementations, the benefit of avoiding duplicate searches is defined as:

[0039]

[0040] Among them, τ(k) represents the pheromone of the grid unit at time k, and β3 represents the benefit coefficient of avoiding repeated search.

[0041] In some possible implementations, the randomly generating a detection main direction includes:

[0042] Determine the neighboring grid cells of the grid cell where the drone is located;

[0043] Taking the flight direction of the drone as the reference direction and the center of the drone as the starting point, for each neighborhood grid cell, determine a direction that points to the corresponding neighborhood grid cell and has the smallest angle with the reference direction;

[0044] A direction is randomly selected from the multiple determined directions as the main detection direction of the UAV.

[0045] In some possible implementations, the method further includes:

[0046] Determine whether the number of steps of the drone exceeds the preset step threshold. If not, continue to randomly generate the main detection direction, determine the final movement direction of the drone based on the main detection direction, and control the drone to continue mobile search. If yes, end the search of the drone and output the search results.

[0047] The main advantages of the technical solution of the present invention are as follows:

[0048] The multi-UAV collaborative target search method based on the beetle whisker search algorithm of the present invention designs the target existence probability and the target existence probability update mechanism, pheromones and the pheromone update mechanism for the divided grid units, and constructs a fitness function that can simultaneously consider environmental benefits, target search benefits, avoidance of repeated search benefits and obstacle avoidance benefits based on the target existence probability and pheromones of the grid units. During the collaborative target search of the UAVs, the UAVs are moved to the grid units with larger fitness function values, and the environmental information in the search process, the physical characteristics of the UAVs themselves and the historical search information can be utilized to the maximum extent, so as to achieve efficient search coverage of the mission area, obstacle avoidance search and avoidance of repeated search among multiple UAVs, and improve the collaborative coverage search efficiency and target search efficiency of the UAVs. BRIEF DESCRIPTION OF THE DRAWINGS

[0049] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0050] Figure 1 This is a flow chart of a multi-UAV cooperative target search method based on a longicorn beard search algorithm according to an embodiment of the present invention;

[0051] Figure 2 A schematic diagram of a grid layout of a task area according to an embodiment of the present invention;

[0052] Figure 3 The present invention is a flowchart of another multi-UAV collaborative target search method based on the beetle whisker search algorithm according to an embodiment of the present invention. DETAILED DESCRIPTION

[0053] In order to make the purpose, technical solution and advantages of the present invention clearer, the technical solution of the present invention will be clearly and completely described below in conjunction with the specific embodiments of the present invention and the corresponding drawings. 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 making creative work are within the scope of protection of the present invention.

[0054] The technical solution provided by the embodiments of the present invention is described in detail below with reference to the accompanying drawings.

[0055] See also Figure 1 An embodiment of the present invention provides a multi-UAV cooperative target search method based on a longhorn beard search algorithm, the method comprising the following steps:

[0056] Step S1, dividing the mission area into a plurality of grid cells, wherein the area of ​​the grid cell is larger than the target size and smaller than or equal to the search area of ​​a single UAV;

[0057] Step S2, constructing the target existence probability of the grid unit and the target existence probability update mechanism based on the probability density function and the detection information of the drone sensor;

[0058] Step S3, constructing the pheromone of the grid unit and the pheromone update mechanism based on the familiarity of the drone with the grid unit;

[0059] Step S4, constructing a fitness function that can simultaneously consider environmental benefits, target search benefits, avoidance of repeated search benefits, and obstacle avoidance benefits based on the target existence probability and pheromone of the grid unit;

[0060] Step S5, for the UAV, randomly generate a main detection direction, and the detection main direction indicates a direction where there is a neighborhood grid cell;

[0061] Step S6, respectively calculating the fitness function values ​​corresponding to the neighborhood grid cell indicated by the main detection direction and the neighborhood grid cells on both sides thereof, selecting the grid cell with the highest fitness function value, and if there are more than two neighborhood grid cells with the same and highest fitness function values ​​corresponding to them, randomly selecting a grid cell from the grid cells with the highest fitness function value;

[0062] Step S7, taking the selected grid cell as the next grid cell to be searched by the drone, and controlling the drone to move toward the selected grid cell.

[0063] A multi-UAV collaborative target search method based on a longicorn beetle whisker search algorithm provided in an embodiment of the present invention designs target existence probability and target existence probability update mechanism, pheromones and pheromone update mechanism for divided grid cells, and constructs a fitness function based on the target existence probability and pheromones of the grid cells that can simultaneously consider environmental benefits, target search benefits, avoiding repeated search benefits and obstacle avoidance benefits. During the collaborative target search of the UAVs, the UAVs are moved to the grid cells with larger fitness function values, and the environmental information, the physical characteristics of the UAVs themselves and the historical search information in the search process can be maximized to achieve efficient search coverage of the mission area, and realize obstacle avoidance search and avoidance of repeated search among multiple UAVs, thereby improving the collaborative coverage search efficiency and target search efficiency of the UAVs.

[0064] The following is a detailed description of the steps and principles of a multi-UAV collaborative target search method based on a longhorn beard search algorithm provided by an embodiment of the present invention:

[0065] Step S1, dividing the task area into a plurality of grid units.

[0066] In one embodiment of the present invention, the mission area is rasterized according to the specific structural size of the mission area to divide the mission area into multiple grid units; at the same time, when performing grid division, the area of ​​each grid unit obtained is controlled to be larger than the target size and smaller than or equal to the search area of ​​a single UAV, ensuring that the target can exist completely in a certain grid unit and that a single UAV is capable of completely searching a single grid unit.

[0067] See also Figure 2 For example, when the task area is a rectangular area, the task area can be divided into two parts including L m ×L n Grid cells A={(m,n)|1≤m≤L m ,1≤n≤L n}, and the grid cells can be sorted from left to right and from bottom to top, then the grid cell (m,n) represents the mnth grid cell. Figure 2 In the figure, the black area indicates that there is a target in the grid cell.

[0068] Step S2, based on the probability density function and the detection information of the drone sensor, construct the target existence probability of the grid unit and the target existence probability update mechanism.

[0069] In one embodiment of the present invention, a mixed two-dimensional Gaussian probability density function is used to describe the target distribution in the task area. The mixed two-dimensional Gaussian probability density function used to describe the target distribution is specifically expressed as:

[0070]

[0071] Among them, μ1 and μ2 represent the mean of a certain prior target point distribution in the set x-direction and y-direction, σ1 and σ2 represent the standard deviation of the target distribution in the x-direction and y-direction, and ρ represents the covariance, ρ = cov(x,y). Among them, the x-direction and y-direction of the task area can be set in advance.

[0072] When the mission area contains N targets, the target existence probability density function of the mission area is:

[0073]

[0074] Among them, f i (x, y) represents the mixed two-dimensional Gaussian probability density function corresponding to the i-th target, α i represents the weight corresponding to the i-th target, α i The specific value can be set according to the importance of different goals.

[0075] Based on the above settings, the initial target existence probability corresponding to the grid cell is expressed as:

[0076]

[0077] Among them, P represents the initial target existence probability corresponding to the grid cell, f(·) represents the mixed two-dimensional Gaussian probability density function corresponding to the target, s represents the area corresponding to f(·), and D represents the integral area corresponding to the grid cell, which is the area contained in the grid cell.

[0078] Specifically, taking the mnth grid cell (m,n) as an example, the corresponding initial target existence probability can be expressed as:

[0079]

[0080] Among them, Pmn represents the initial target existence probability corresponding to the mnth grid cell (m, n), and Dmn represents the integral area corresponding to the grid cell (m, n).

[0081] Furthermore, the target existence probability update mechanism of the grid unit includes:

[0082] If the UAV detects a target in a grid cell at time k, the target existence probability of the grid cell is updated using the following formula to obtain the target existence probability of the grid cell at the next moment;

[0083]

[0084] If the UAV does not detect the target in the grid cell at time k, the target existence probability of the grid cell is updated using the following formula to obtain the target existence probability of the grid cell at the next moment;

[0085]

[0086] Among them, P(k+1) represents the probability of the target existing in the grid cell at time k+1, P d represents the detection probability of the drone sensor, P(k) represents the probability of the target existing in the grid cell at time k, and P f Indicates the false alarm rate.

[0087] The detection probability of the drone sensor is the probability that there is a target in the grid cell and it is detected by the drone sensor. Its value is between [0,1]. The specific value is determined according to the actual performance of the drone sensor. The false alarm rate is the probability that there is no target in the grid cell but the drone sensor detects the target. Its value is between [0,1]. The specific value is also determined according to the actual performance of the drone sensor.

[0088] In one embodiment of the present invention, the larger the value of the target existence probability of the grid cell, the greater the possibility that the target exists in the grid cell. For this purpose, a threshold θ can be set. When the target existence probability of the grid cell is greater than the set threshold θ, it is considered that the target exists in the grid cell.

[0089] Step S3, based on the drone's familiarity with the grid unit, construct the pheromone of the grid unit and the pheromone update mechanism.

[0090] In one embodiment of the present invention, digital pheromones are used to describe the familiarity of the drone with the corresponding grid unit, and the pheromones of the grid unit and the pheromone update mechanism are established.

[0091] Specifically, the pheromone of the grid unit and the pheromone update mechanism are defined as:

[0092] τ(k)=(1-ρ)·τ(k-1)+ρΔτ(k)

[0093] Among them, τ(k) represents the pheromone of the grid unit at time k, ρ represents the pheromone volatility coefficient, τ(k-1) represents the pheromone of the grid unit at time k-1, Δτ(k) represents the update amount of the pheromone of the grid unit from time k-1 to time k, Δτ(k)=num*Q, num represents the number of visits to the grid unit from time k-1 to time k, and Q represents the pheromone strength.

[0094] Among them, 0<ρ≤1, Q is a constant, and the specific values ​​of ρ and Q are set according to actual conditions.

[0095] The specific value of the pheromone of the grid unit at the initial moment is set according to the actual situation. In one embodiment of the present invention, the pheromone of the grid unit at the initial moment is set to 0.

[0096] Step S4, based on the target existence probability and pheromone of the grid unit, construct a fitness function that can simultaneously consider the environmental benefit, the target search benefit, the benefit of avoiding repeated search and the benefit of avoiding obstacles.

[0097] In one embodiment of the present invention, information entropy is used to describe the uncertainty of the task area environment, and the uncertainty of the task area environment is used to determine the task area environment benefit.

[0098] Specifically, the uncertainty of the mission area environment is expressed as:

[0099] e(k)=(1-P(k))ln(P(k))

[0100] The environmental benefits of the mission area are expressed as:

[0101]

[0102] Among them, e(k) represents the uncertainty of the task area environment at time k, P(k) represents the probability of target existence in the grid cell at time k, J1(k) represents the environmental benefit of the task area corresponding to the grid cell at time k, β1 represents the environmental benefit coefficient, and the specific value of β1 can be set according to actual conditions.

[0103] In one embodiment of the present invention, the target search benefit is defined as:

[0104] J2(k)=b(k)P(k)

[0105] Among them, J2(k) represents the target search benefit corresponding to the grid cell at time k, P(k) represents the target existence probability of the grid cell at time k, σ represents a weight coefficient, which is used to adjust the target perception weight. The specific value of the weight coefficient can be set according to actual conditions.

[0106] In one embodiment of the present invention, the benefit of avoiding repeated searches is defined as:

[0107]

[0108] Among them, J3(k) represents the benefit of avoiding repeated search corresponding to the grid unit at time k, τ(k) represents the pheromone of the grid unit at time k, β3 represents the benefit coefficient of avoiding repeated search, and the specific value of β3 can be set according to actual conditions.

[0109] Since the distance between the UAV at time k and other UAVs at the next time is smaller than the grid size through the perception link, a collision may occur. Therefore, in one embodiment of the present invention, the obstacle avoidance benefit is defined as:

[0110]

[0111] Among them, J4(k) represents the obstacle avoidance benefit at time k.

[0112] Based on the above-set mission area environment benefits, target search benefits, avoidance of repeated search benefits and obstacle avoidance benefits, the fitness function is defined as:

[0113] J(k)=(ω1J1(k)+ω2J2(k)+ω3J3(k))&&J4(k)

[0114] Among them, J(k) represents the fitness function corresponding to the grid unit at time k, ω1 represents the environmental benefit weight, ω2 represents the target search benefit weight, ω3 represents the avoidance of repeated search benefit weight, and the specific values ​​of ω1, ω2 and ω3 can be set according to actual conditions.

[0115] The above fitness function can maximize the use of environmental information, the UAV's own physical characteristics and historical search information during the search process. The larger the value of the fitness function corresponding to the grid unit, the greater the benefit of the UAV moving to the grid unit. By using the above fitness function to calculate and determine the next movement direction of the UAV, the target search efficiency can be improved and the environmental uncertainty can be reduced.

[0116] Step S5: For the UAV, a main detection direction is randomly generated, and the direction indicated by the main detection direction has a neighborhood grid unit.

[0117] Specifically, the main detection direction is randomly generated, including:

[0118] Determine the neighboring grid cells of the grid cell where the drone is located;

[0119] Taking the flight direction of the drone as the reference direction and the center of the drone as the starting point, for each neighborhood grid cell, determine a direction that points to the corresponding neighborhood grid cell and has the smallest angle with the reference direction;

[0120] A direction is randomly selected from the multiple determined directions as the main detection direction of the UAV.

[0121] In one embodiment of the present invention, the following formula may be used to randomly select a direction:

[0122]

[0123] in, Represents the selected direction, rad(W,1) represents a random function, and W represents the dimension of the selectable direction.

[0124] Furthermore, considering the physical constraints of the UAV's motion during flight and to facilitate the flight control of the UAV, if there is a neighborhood grid cell in the direction indicated by the UAV's flight direction, the UAV's flight direction is used as the main detection direction of the UAV.

[0125] Step S6, respectively calculating the fitness function values ​​corresponding to the neighborhood grid cell pointed to by the main detection direction and the neighborhood grid cells on both sides thereof, and selecting the grid cell with the highest fitness function value.

[0126] Specifically, according to the fitness function calculation formula, the fitness function values ​​corresponding to the neighborhood grid cell indicated by the main detection direction and the neighborhood grid cells on both sides of the neighborhood grid cell are calculated respectively to determine the grid cell with the highest fitness function value.

[0127] If there are more than two neighboring grid cells with the same and highest fitness function values, a grid cell is randomly selected from the grid cells with the highest fitness function value.

[0128] Optionally, if one of the two or more grid cells with the highest corresponding fitness function values ​​is the grid cell indicated by the flight direction of the drone, the grid cell specified by the flight direction of the drone is selected. This can make the flight control of the drone more convenient.

[0129] Step S7, taking the selected grid cell as the next grid cell to be searched by the drone, and controlling the drone to move toward the selected grid cell.

[0130] Specifically, according to the grid unit selected in step S6, the grid unit is used as the next grid unit to be searched by the drone, and the drone is controlled to move and search toward the grid unit. During the mobile search process, the target existence probability and pheromone of all grid cells are updated in real time.

[0131] Further, see Figure 3 In one embodiment of the present invention, the method may further include:

[0132] Step S8, determining whether the number of steps of the drone exceeds a preset step threshold, if not, returning to step S5, if yes, ending the search for the drone and outputting the search results.

[0133] It should be noted that, in this article, relational terms such as "first" and "second" are only used to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Moreover, the terms "include", "comprise" or any other variants thereof are intended to cover non-exclusive inclusion, so that a process, method, article or device including a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, method, article or device. In addition, "front", "back", "left", "right", "upper" and "lower" in this article are all referenced to the placement state shown in the accompanying drawings.

[0134] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit it. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or make equivalent replacements for some of the technical features therein. However, these modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A multi-UAV cooperative target search method based on the beetle whisker search algorithm, characterized in that: include: Divide the mission area into a plurality of grid cells, wherein the area of ​​the grid cells is larger than the target size and smaller than or equal to the search area of ​​a single UAV; Based on the probability density function and the detection information of the drone sensor, the target existence probability of the grid unit and the target existence probability update mechanism are constructed; Based on the familiarity of the drone with the grid unit, the pheromone of the grid unit and the pheromone update mechanism are constructed; According to the target existence probability and pheromone of the grid unit, a fitness function is constructed that can simultaneously consider the environmental benefit, target search benefit, avoidance of repeated search benefit and obstacle avoidance benefit; For the UAV, a detection main direction is randomly generated, and the direction indicated by the detection main direction has a neighborhood grid unit; Calculate the fitness function values ​​of the neighboring grid cells pointed to by the main detection direction and the neighboring grid cells on both sides respectively, and select the grid cell with the highest fitness function value. If there are more than two neighboring grid cells with the same and highest fitness function values, randomly select a grid cell from the grid cells with the highest fitness function value. The selected grid cell is used as the next grid cell to be searched by the UAV, and the UAV is controlled to move toward the selected grid cell; The fitness function is defined as: J(k)=(ω1J1(k)+ω2J2(k)+ω3J3(k))&&J4(k) Among them, J(k) represents the fitness function corresponding to the grid cell at time k, ω1 represents the environmental benefit weight, J1(k) represents the environmental benefit of the task area corresponding to the grid cell at time k, ω w represents the target search benefit weight, J2(k) represents the target search benefit corresponding to the grid cell at time k, ω3 represents the avoidance of repeated search benefit weight, J3(k) represents the avoidance of repeated search benefit corresponding to the grid cell at time k, and J4(k) represents the obstacle avoidance benefit at time k. Mission area environmental benefits are defined as: e(k)=(1-P(k))ln(P(k)) Among them, β1 represents the environmental benefit coefficient, e(k) represents the uncertainty of the task area environment at time k, and P(k) represents the probability of the target existence in the grid cell at time k; The target search benefit is defined as: J2(k)=b(k)P(k) Among them, P(k) represents the probability of the target existing in the grid cell at time k, σ represents the weight coefficient; The avoidance of duplicate searches benefit is defined as: Among them, τ(k) represents the pheromone of the grid unit at time k, and β3 represents the benefit coefficient of avoiding repeated search.

2. The multi-UAV cooperative target search method based on the beetle whisker search algorithm according to claim 1 is characterized in that: The initial target existence probability of the grid cell is: Among them, P represents the initial target existence probability corresponding to the grid cell, f(·) represents the mixed two-dimensional Gaussian probability density function corresponding to the target, s represents the area corresponding to f(·), and D represents the integral area corresponding to the grid cell.

3. The multi-UAV cooperative target search method based on the beetle whisker search algorithm according to claim 2 is characterized in that: The target existence probability update mechanism of the grid cell includes: If the UAV detects a target in a grid cell at time k, the target existence probability of the grid cell is updated using the following formula to obtain the target existence probability of the grid cell at the next moment; If the UAV does not detect the target in the grid cell at time k, the target existence probability of the grid cell is updated using the following formula according to the target existence probability of the grid cell at time k to obtain the target existence probability of the grid cell at the next time; Among them, P(k+1) represents the probability of the target existing in the grid cell at time k+1, P d represents the detection probability of the drone sensor, P(k) represents the probability of the target existing in the grid cell at time k, and P f Indicates the false alarm rate.

4. The multi-UAV cooperative target search method based on the beetle whisker search algorithm according to claim 1 or 3, characterized in that: The pheromone update mechanism of the grid cell is defined as: τ(k)=(1-ρ)·τ(k-1)+ρΔτ(k) Among them, τ(k) represents the pheromone of the grid unit at time k, ρ represents the pheromone volatility coefficient, τ(k-1) represents the pheromone of the grid unit at time k-1, Δτ(k) represents the update amount of the pheromone of the grid unit from time k-1 to time k, Δτ(k)=num*Q, num represents the number of visits to the grid unit from time k-1 to time k, and Q represents the pheromone strength.

5. The multi-UAV cooperative target search method based on the beetle whisker search algorithm according to claim 1 is characterized in that: The randomly generated detection main direction comprises: Determine the neighboring grid cells of the grid cell where the drone is located; Taking the flight direction of the drone as the reference direction and the center of the drone as the starting point, for each neighborhood grid cell, determine a direction that points to the corresponding neighborhood grid cell and has the smallest angle with the reference direction; A direction is randomly selected from the multiple determined directions as the main detection direction of the UAV.

6. The multi-UAV cooperative target search method based on the beetle whisker search algorithm according to claim 1 is characterized in that: The method further comprises: Determine whether the number of steps of the drone exceeds the preset step threshold. If not, continue to randomly generate the main detection direction, determine the final movement direction of the drone based on the main detection direction, and control the drone to continue mobile search. If yes, end the search of the drone and output the search results.

Images