Formation control method and device for unmanned aerial vehicle cluster to collaboratively realize multi-target task

Abstract

The invention discloses a formation control method and device for an unmanned aerial vehicle cluster to collaboratively realize a multi-target task. The formation control method comprises the following steps: acquiring initial information of a plurality of target tasks and basic information of an unmanned aerial vehicle cluster for executing the plurality of target tasks; according to the initial information, obtaining cluster grouping data when the unmanned aerial vehicle cluster executes a plurality of target tasks; obtaining task allocation data of each cluster group according to the cluster group data and the initial information; obtaining a reference path of each unmanned aerial vehicle cluster according to the task distribution data, the cluster grouping data and the initial information; setting a virtual navigator based on the reference path, and obtaining expected position and expected speed data of each unmanned aerial vehicle for different target tasks according to the reference path and the virtual navigator; and according to the basic information of the unmanned aerial vehicle cluster and the expected position and the expected speed data of each unmanned aerial vehicle, obtaining a formation pattern after each unmanned aerial vehicle arrives at the corresponding expected position at the corresponding expected speed.

Description

Technical Field

[0001] The present disclosure generally relates to the field of drone swarm technology, and more specifically, to a formation control method and device for drone swarms to collaboratively achieve multi-objective tasks. Background Art

[0002] In related fields, with the rapid development of artificial intelligence technology, unmanned swarms represented by drones are increasingly widely used in various industries or fields. Unmanned swarms refer to networked groups composed of multiple autonomous or remote-controlled unmanned systems (for example, drones, unmanned vehicles, unmanned ships, etc.). These unmanned systems can achieve efficient information sharing, collaborative operations and autonomous execution of tasks through advanced communication, perception, collaboration and autonomous decision-making. Supported by swarm intelligence technology, each swarm intelligence unit can change according to different task requirements and application scenarios, and form a human-machine hybrid and autonomous unmanned swarm system through ubiquitous networks, random access, autonomous adaptation, flexible grouping, and dynamic adjustment, thus having diversified application capabilities.

[0003] As an important application aspect of unmanned swarm systems, the configuration generation and formation control issues of UAV swarms have attracted much attention in various practical applications. Summary of the invention

[0004] The embodiments of the present disclosure provide a formation control method and device for drone clusters to collaboratively achieve multi-objective tasks, which can achieve independent control of multiple groups of drone formations in a multi-objective and multi-task scenario.

[0005] In one general aspect, a formation control method for unmanned aerial vehicle clusters to collaboratively achieve multi-target tasks is provided, the formation control method comprising: obtaining initial information of multiple target tasks and basic information of a drone cluster used to perform the multiple target tasks, the initial information comprising location information, value attribute information and area information of multiple target task areas; inputting the initial information into a cluster determination model based on a cuckoo search algorithm to obtain cluster grouping data of the drone cluster when performing the multiple target tasks; inputting the cluster grouping data and the initial information into a task allocation model based on an auction algorithm to obtain task allocation data for each cluster grouping; and inputting the task allocation data, the cluster determination model and the initial information into a task allocation model based on an auction algorithm to obtain task allocation data for each cluster grouping. The grouping data and the initial information are input into a path generation model based on an artificial potential field method to obtain a reference path for each UAV cluster; based on the reference path, a virtual navigator is set for each UAV cluster, and the reference path and the virtual navigator are input into a configuration generation model based on a leader-follower relationship to obtain the expected position and expected speed data of each UAV for different target tasks; the basic information of the UAV cluster, the expected position and expected speed data of each UAV are input into the formation control model based on the bionic swarming algorithm to obtain the formation formation after each UAV arrives at the corresponding expected position at the corresponding expected speed, so as to complete the formation control for each target task.

[0006] Optionally, the formation control method may also include: in response to the communication quality of a part of the drones in the drone cluster dropping below a preset threshold, using the drones in the drone cluster except for the part of the drones as an updated drone cluster for performing the multiple target tasks, executing the formation control method based on the updated drone cluster to obtain an updated formation formation to complete the update of the formation formation, wherein the preset threshold indicates that the communication quality is defined as the lower limit value of the communication quality corresponding to a weak communication situation.

[0007] Optionally, the step of inputting the initial information into a cluster determination model based on a cuckoo search algorithm to obtain cluster grouping data of the drone cluster when performing the multiple target tasks may include: constructing a task area attribute parameter matrix, an optimization space and a task area objective function based on the initial information; performing the following processing for each iteration cycle until a preset convergence condition is met: randomly generating an initial search space including multiple initial solutions, the solutions included in the search space represent the number of drones required to be allocated to each task area; substituting the initial search space, the task area attribute parameter matrix and the optimization space into the task area objective function, and retaining solutions corresponding to higher task area objective function values ​​to the next generation; using a global Lévy flight strategy, randomly updating the search space and removing a portion of solutions, and solutions that are not removed are retained to the next generation; generating new solutions based on a local random walk strategy; calculating the fitness value of the new solution based on an adaptive scaling factor parameter, and retaining solutions corresponding to higher fitness values ​​to the next generation; in response to meeting the preset convergence condition, taking the current solution as the optimal solution, and obtaining cluster grouping data of the drone cluster when performing the multiple target tasks based on the optimal solution.

[0008] Optionally, the step of inputting the cluster grouping data and the initial information into a task allocation model based on an auction algorithm to obtain the task allocation data for each cluster grouping may include: constructing a task allocation optimization function based on the cluster grouping data and the initial information; and based on the task allocation optimization function, through an auction algorithm, obtaining the task allocation data for each cluster grouping while satisfying the maximization of the task allocation optimization function value.

[0009] Optionally, the step of inputting the task assignment data, the cluster grouping data and the initial information into a path generation model based on an artificial potential field method to obtain a reference path for each drone cluster may include: calculating the initial position and target position of each drone cluster for each target task based on the task assignment data, the cluster grouping data and the initial information; and calculating the repulsive force of the obstacle zone and the attractive force of the target point in the scene corresponding to each target task based on the artificial potential field method to obtain a reference path for each drone cluster from the initial position to the target position.

[0010] Optionally, the step of inputting the reference path and the virtual navigator into a configuration generation model based on a pilot-follower relationship to obtain expected position and expected speed data of each UAV for different target tasks may include: based on the reference path and the virtual navigator, according to preset geometric constraint information between the navigator and the follower based on the pilot-follower relationship, obtaining an expected formation topology structure; based on the expected formation topology structure, obtaining expected position and expected speed data of each UAV for different target tasks.

[0011] Optionally, the step of inputting the basic information of the drone cluster, the expected position and expected speed data of each drone into the formation control model based on the bionic swarming algorithm to obtain the formation formation after each drone arrives at the corresponding expected position at the corresponding expected speed may include: initializing model parameters of the formation control model based on the bionic swarming algorithm based on the basic information of the drone cluster, the expected position and expected speed data of each drone, and determining whether each drone is located at the corresponding expected position; in response to determining that all drones are located at the corresponding expected position, obtaining the formation formation after each drone arrives at the corresponding expected position at the corresponding expected speed. The subsequent formation formation; in response to determining that there are UAVs that are not located at the corresponding expected positions, for each UAV that is not located at the corresponding expected position, the following processing is executed in a loop: the communication connection relationship between the UAVs is calculated to obtain the communication connection matrix between the UAVs; based on the communication connection matrix between the UAVs, the attraction of the corresponding expected position to the UAVs and the interaction force between the UAVs are calculated; based on the attraction of the corresponding expected position to the UAVs and the interaction force between the UAVs, the resultant force applied to each UAV is calculated; based on the resultant force applied to each UAV, the acceleration of each UAV is calculated, so as to obtain the speed and position of each UAV at the next moment.

[0012] In another general aspect, a formation control device for drone clusters to collaboratively achieve multi-target tasks is provided, the formation control device comprising: a data acquisition module, configured to: acquire initial information of multiple target tasks and basic information of drone clusters used to perform the multiple target tasks, the initial information comprising location information, value attribute information and area information of multiple target task areas; a cluster grouping module, configured to: input the initial information into a cluster determination model based on a cuckoo search algorithm to obtain cluster grouping data of the drone cluster when performing the multiple target tasks; a task allocation module, configured to: input the cluster grouping data and the initial information into a task allocation model based on an auction algorithm to obtain task allocation data for each cluster grouping; a path generation module, configured to: The task allocation data, the cluster grouping data and the initial information are input into a path generation model based on an artificial potential field method to obtain a reference path for each UAV cluster; a configuration generation module is configured to: based on the reference path, set a virtual navigator for each UAV cluster, and input the reference path and the virtual navigator into a configuration generation model based on a leader-follower relationship to obtain the expected position and expected speed data of each UAV for different target tasks; a formation control module is configured to: input the basic information of the UAV cluster, the expected position and expected speed data of each UAV into the formation control model based on the bionic swarming algorithm to obtain the formation formation after each UAV arrives at the corresponding expected position at the corresponding expected speed, so as to complete the formation control for each target task.

[0013] Optionally, the formation control device can also be configured to: in response to the communication quality of a part of the drones in the drone cluster dropping below a preset threshold, use the drones in the drone cluster other than the part of the drones as an updated drone cluster for performing the multiple target tasks, and execute the formation control method as described above based on the updated drone cluster to obtain an updated formation formation to complete the update of the formation formation, wherein the preset threshold indicates that the communication quality is defined as the lower limit value of the communication quality corresponding to the weak communication situation.

[0014] Optionally, the cluster grouping module inputs the initial information into a cluster determination model based on a cuckoo search algorithm, and the operation of obtaining the cluster grouping data of the drone cluster when performing the multiple target tasks may include: based on the initial information, constructing a task area attribute parameter matrix, an optimization space and a task area objective function; for each iteration cycle, performing the following processing until a preset convergence condition is met: randomly generating an initial search space including multiple initial solutions, the solutions included in the search space represent the number of drones required to be allocated to each task area; substituting the initial search space, the task area attribute parameter matrix and the optimization space into the task area objective function, and retaining solutions corresponding to higher task area objective function values ​​to the next generation; using a global Lévy flight strategy, randomly updating the search space and removing a portion of the solutions, and the solutions that are not removed are retained to the next generation; generating a new solution based on a local random walk strategy; calculating the fitness value of the new solution based on an adaptive scaling factor parameter, and retaining the solution corresponding to the higher fitness value to the next generation; in response to meeting the preset convergence condition, taking the current solution as the optimal solution, and obtaining the cluster grouping data of the drone cluster when performing the multiple target tasks based on the optimal solution.

[0015] Optionally, the task allocation module inputs the cluster grouping data and the initial information into a task allocation model based on an auction algorithm, and the operation of obtaining the task allocation data for each cluster grouping may include: constructing a task allocation optimization function based on the cluster grouping data and the initial information; based on the task allocation optimization function, through an auction algorithm, obtaining the task allocation data for each cluster grouping while satisfying the maximization of the task allocation optimization function value.

[0016] Optionally, the path generation module inputs the task assignment data, the cluster grouping data and the initial information into a path generation model based on the artificial potential field method, and the operation of obtaining the reference path of each drone cluster may include: calculating the initial position and target position of each drone cluster for each target task based on the task assignment data, the cluster grouping data and the initial information; based on the artificial potential field method, calculating the repulsive force of the obstacle area and the attractive force of the target point in the scene corresponding to each target task, and obtaining the reference path of each drone cluster from the initial position to the target position.

[0017] Optionally, the configuration generation module inputs the reference path and the virtual navigator into a configuration generation model based on a pilot-follower relationship, and the operation of obtaining expected position and expected speed data of each UAV for different target tasks may include: based on the reference path and the virtual navigator, according to preset geometric constraint information between the navigator and the follower based on the pilot-follower relationship, obtaining an expected formation topology structure; based on the expected formation topology structure, obtaining expected position and expected speed data of each UAV for different target tasks.

[0018] Optionally, the formation control module inputs the basic information of the drone cluster, the expected position and expected speed data of each drone into the formation control model based on the bionic swarming algorithm, and the operation of obtaining the formation formation after each drone arrives at the corresponding expected position at the corresponding expected speed may include: initializing the model parameters of the formation control model based on the bionic swarming algorithm based on the basic information of the drone cluster, the expected position and expected speed data of each drone, and determining whether each drone is located at the corresponding expected position; in response to determining that all drones are located at the corresponding expected position, obtaining the formation formation after each drone arrives at the corresponding expected speed In response to determining that there are UAVs that are not located at the corresponding expected positions, for each UAV that is not located at the corresponding expected positions, the following processing is executed cyclically: the communication connection relationship between the UAVs is calculated to obtain a communication connection matrix between the UAVs; based on the communication connection matrix between the UAVs, the attraction of the corresponding expected position to the UAVs and the interaction forces between the UAVs are calculated; based on the attraction of the corresponding expected position to the UAVs and the interaction forces between the UAVs, the resultant force applied to each UAV is calculated; based on the resultant force applied to each UAV, the acceleration of each UAV is calculated, so as to obtain the speed and position of each UAV at the next moment.

[0019] In another general aspect, a computer program product is provided, the computer program product comprising a computer program / instructions, which when executed by a processor implements the formation control method for drone clusters to collaboratively achieve multi-objective tasks as described above.

[0020] In another general aspect, a computer-readable storage medium is provided, which, when instructions in the computer-readable storage medium are executed by a processor of an electronic device / server, enables the electronic device / server to execute the formation control method for drone clusters to collaboratively achieve multi-objective tasks as described above.

[0021] In another general aspect, an electronic device is provided, comprising: at least one processor; and at least one memory storing computer executable instructions, wherein the computer executable instructions, when executed by the at least one processor, cause the at least one processor to execute the formation control method for drone clusters to collaboratively achieve multi-objective tasks as described above.

[0022] According to the formation control method and device for drone clusters to collaboratively achieve multi-target tasks according to the embodiments of the present disclosure, it is possible to achieve independent control of multiple drone formations in multi-target and multi-task scenarios. In addition, by quickly achieving formation maneuvering control of large-scale drone clusters in different scenarios, it is possible to efficiently complete various tasks such as regional reconnaissance and coverage. BRIEF DESCRIPTION OF THE DRAWINGS

[0023] The above and other objects and features of the embodiments of the present disclosure will become more apparent through the following description in conjunction with the accompanying drawings showing the embodiments, in which:

[0024] Figure 1 is a flow chart showing a formation control method for a drone cluster to collaboratively achieve a multi-objective task according to an embodiment of the present disclosure;

[0025] Figure 2 is a flowchart showing an example of a formation control method for a drone cluster to collaboratively achieve a multi-objective task according to an embodiment of the present disclosure;

[0026] Figure 3 is an example flow chart illustrating a cuckoo search algorithm according to an embodiment of the present disclosure;

[0027] Figure 4 is an example flow chart showing a formation control method based on a bionic swarming algorithm according to an embodiment of the present disclosure;

[0028] FIG. 5A to FIG. 5C is a diagram showing the aggregation-dispersion and communication connection relationship of a drone cluster in a bionic swarming algorithm according to an embodiment of the present disclosure;

[0029] FIG. 6A to FIG. 6E is a schematic diagram showing simulation results of a bionic formation according to an embodiment of the present disclosure;

[0030] FIG. 7A to FIG. 7C is a schematic diagram showing a formation reconstruction process of a bionic formation after encountering interference according to an embodiment of the present disclosure;

[0031] Figure 8 is a block diagram showing a formation control device for drone clusters to collaboratively achieve multi-objective tasks according to an embodiment of the present disclosure;

[0032] Fig. 9is a block diagram illustrating an electronic device according to an embodiment of the present disclosure. DETAILED DESCRIPTION

[0033] The following specific embodiments are provided to help the reader obtain a comprehensive understanding of the methods, devices and / or systems described herein. However, after understanding the disclosure of the present application, various changes, modifications and equivalents of the methods, devices and / or systems described herein will be clear. For example, the order of operations described herein is only an example and is not limited to those orders set forth herein, but can be changed as will be clear after understanding the disclosure of the present application, except for operations that must occur in a specific order. In addition, for greater clarity and simplicity, the description of features known in the art may be omitted.

[0034] Reference will now be made in detail to the embodiments of the present disclosure, examples of which are illustrated in the accompanying drawings, wherein like reference numerals refer to like parts throughout. The embodiments will be described below with reference to the drawings in order to explain the present disclosure.

[0035] As mentioned above, in the related technologies, drone cluster configuration generation and formation control are important application fields of unmanned cluster systems, and have great potential in various practical applications. This disclosure proposes a bionic swarming configuration generation and formation control method for multi-drone coverage reconnaissance, which combines bionic features with drone formations to achieve stable control of large-scale drone formations, and realizes formation changes and formation motion control in weak communication conditions.

[0036] Refer to the following Figures 1 to 9 A formation control method and apparatus for drone clusters to collaboratively achieve multi-objective tasks according to an embodiment of the present disclosure are described in detail.

[0037] Figure 1 is a flow chart showing a formation control method 100 for drone clusters to collaboratively achieve multi-objective tasks according to an embodiment of the present disclosure.

[0038] Reference Figure 1 According to an embodiment of the present disclosure, in step S101, initial information of multiple target tasks and basic information of a drone cluster for executing multiple target tasks are obtained. Here, the initial information includes location information, value attribute information, area information, etc. of multiple target task areas.

[0039] According to an embodiment of the present disclosure, in step S102, initial information is input into a cluster determination model based on a cuckoo search algorithm to obtain cluster grouping data of the drone cluster when performing multiple target tasks.

[0040] As an example, step S102 may further include steps S1021 to S1023:

[0041] In step S1021, based on the initial information, a task area attribute parameter matrix, an optimization space and a task area objective function are constructed.

[0042] In step S1022, for each iteration cycle, the following processing is performed until a preset convergence condition is met:

[0043] In process S21, an initial search space including a plurality of initial solutions is randomly generated, and the solutions included in the search space represent the number of drones required to be allocated to each mission area.

[0044] In processing S22, the initial search space, the task area attribute parameter matrix and the optimization space are substituted into the task area objective function, and the solution corresponding to the higher task area objective function value is retained to the next generation.

[0045] Here, the optimization variables in the mission area objective function include the required number of drones in each mission area, while also considering factors such as the value attributes of the mission area, so as to optimize the number of drones required in each mission area based on the cuckoo search algorithm.

[0046] In processing S23, the global Lévy flight strategy is used to randomly update the search space.

[0047] In processing S24, a new solution is generated based on the local random walk strategy.

[0048] In process S25, based on the adaptive scale factor parameter, the fitness value of the new solution is calculated, and the solution corresponding to the higher fitness value is retained to the next generation.

[0049] In step S1023, in response to satisfying a preset convergence condition, the current solution is taken as the optimal solution, and cluster grouping data of the drone cluster when performing multiple target tasks is obtained based on the optimal solution.

[0050] That is to say, in the cuckoo search algorithm disclosed in the present invention, the solution is obtained through iterative optimization until a preset number of iterations is reached or a preset convergence condition is reached, and the number of drones required for each mission area is obtained when the objective function of the mission area (such as the search profit function) is maximized, thereby completing the division of large-scale drone clusters and completing the drone cluster formation grouping (also known as grouping).

[0051] The following specifically describes the formation control method disclosed in the present invention by taking the reconnaissance missions for multiple mission areas as an example in which multiple missions that the drones need to perform.

[0052] Specifically, first, a task area attribute parameter matrix can be constructed, as shown in the following formula (1):

[0053]

[0054] Among them, λ i is the value attribute of the task area, σ i is the prior intelligence information of the mission area, ε i is the area parameter of the task area, and n is the number of task areas.

[0055] Then, an optimization space can be constructed, S = [T, N, C, R], that is, the optimization space includes the task execution time T, the number of drones required for the task N, the total path of the reconnaissance task C, and the task benefit R. However, the present disclosure is not limited thereto, and other parameters may also be included.

[0056] In addition, the mission area reconnaissance mission benefit function can also be constructed, as shown in the following equations (2) and (3):

[0057]

[0058] n i =t o / ∑t i *∑n i (3)

[0059] Among them, the execution time of each task area is T = [t1t2t i …t n ],v i is the flight speed of the drone, C i =σ i / ∑σ i , n i The number of drones required for each reconnaissance zone.

[0060] Based on the above-mentioned mission area attribute parameter matrix, optimization space and mission area reconnaissance mission benefit function, the present disclosure improves the traditional cuckoo search algorithm for cluster multi-factor optimization problems, and describes the improved cuckoo search algorithm as the following processes (a) to (c):

[0061] (a) Each cuckoo lays only a few eggs at a time.

[0062] (b) Bird nests (solutions) with high quality will be retained to the next generation.

[0063] (c) The number of available nest owners m is fixed, and the probability that a nest owner finds an alien egg is P a , where P a ∈[0,1]; when the nest owner finds an alien egg, the nest owner can choose to discard the egg, or abandon the nest directly, and then build a new nest in a new location. In other words, this hypothesis can be understood as there are P nests among m nests. a ×m bird nests will be replaced by new ones.

[0064] In addition, refer to Figure 3 To understand the cuckoo search algorithm according to an embodiment of the present disclosure, Figure 3 Detailed description of the steps is given in blocks S301 to S308 and will not be repeated here.

[0065] Therefore, the cuckoo search algorithm includes three elements: selecting the best, using local random movement, and performing random selection through global Lévy flight. That is, in the present disclosure, the cuckoo search algorithm is improved for the combinatorial optimization problem of multiple coupled variables to achieve the optimization of large-scale drone resource allocation.

[0066] The improved cuckoo search algorithm can be specifically described below from these three elements.

[0067] First, m solutions are randomly generated, each of which contains m×n variables. The search strategy space for these solutions is shown in the following formula (4):

[0068]

[0069] The optimization variable range can be set to, for example, t=[t min ,t max ]. In addition, the algorithm supports mixed optimization of continuous and discrete variables.

[0070] Secondly, each row in the search strategy space is substituted into the task benefit function R, the strategy benefit value is calculated, and the solution with higher quality will be retained in the next generation.

[0071] Then, the global Lévy flight strategy is used to randomly update the search solution space. Here, a part of the solution may be discarded with probability Pa. In addition, new solutions are generated based on the local random walk strategy. Here, the solutions that are not discarded are retained to the next generation.

[0072] Finally, the fitness value of the new solution is calculated, and based on the elite retention strategy, the solution with a higher fitness value is selected to be retained in the next generation.

[0073] The following is a detailed description of the three processing processes involved in the above algorithm (i.e., the process of selecting the optimal solution, the process of global Lévy flight random selection, and the process of local random movement to generate new solutions).

[0074] 1. On the process of selecting the best

[0075] The best solution is selected by retaining the better solution, for example, based on the elite retention strategy, which ensures that the search moves are always within the range of the local optimal solution, so that the optimal solution is retained to the next generation and will not be removed from the population.

[0076] (II) The process of random selection of global Lévy flights

[0077] The purpose of the global Levy flight, which is a unique feature of the cuckoo search algorithm, is to change variables over a large range during the optimization process, effectively avoiding the optimization from falling into the local optimum. The global random search process is performed according to the Lévy flight process, that is, following:

[0078]

[0079] in, represents the position of the current iteration (i.e., generation t) of population i, represents the position of the next update (i.e., the t+1 generation), L(β) represents the random optimization direction of the Levy distribution, and its moving step α follows the Levy distribution, i.e., Levy = t 1-β (0<β≤2).

[0080] In addition, the continuous jumps of the cuckoo here form a random walk process, namely:

[0081]

[0082] Among them, u follows a normal distribution σ v =1,

[0083] In the present disclosure, in order to fully retain high-quality solutions and enable the algorithm to converge faster, the present application proposes to improve the parameter β in the algorithm. That is, in the present disclosure, β can be adaptively adjusted according to the optimization process, so that the solution obtained has a strong diversity in the early iteration process, and the diversity of the solution is gradually reduced in the later iteration process to better perform local search. According to the above analysis, the adaptive scaling factor can be defined as shown in the following formula (7):

[0084]

[0085] Among them, β0 is the initial parameter, is the task benefit of population i in the initial stage, is the task benefit of population i in the t-1th generation, is the initial total profit of the population participating in the optimization, is the total profit of the t-1th generation of the population participating in the optimization.

[0086] 3. On the process of generating new solutions by local random moves

[0087] In the case where the solution after the Lévy flight random selection is discarded, the present disclosure uses local random movement to generate a new solution. The process of generating a new solution by local random movement can be described as the following formula (8):

[0088]

[0089] in, is the ith solution of the t+1th generation, is the ith solution of the tth generation, α is the step size (e.g., usually 1), are two random number sequences, H is the Heaviside function, ε is a random number, and Pa is the probability of being found in the cuckoo search algorithm.

[0090] Through the above two random selection processes, the optimal solution is obtained through cyclic iteration, thereby obtaining the optimal solution for the number of drones required for each mission area and realizing the formation of large-scale drone clusters.

[0091] In the cluster determination method based on the cuckoo search algorithm according to the present disclosure, in order to be closer to the situation in actual application, according to the attributes of the task area for multiple target tasks (also called task scenarios) and the bionic characteristics of the drone, a resource allocation method based on the cuckoo search optimization algorithm is proposed, and the adaptive adjustment mechanism of the optimization parameters of the bionic characteristics is adopted to achieve the rapid convergence of the optimization curve, thereby achieving rapid resource allocation of large-scale drones, while ensuring the maximization of the allocation benefits. In addition, the cluster determination method based on the cuckoo search algorithm according to the present disclosure also effectively avoids the problem of waste of drone resources with different task area attributes.

[0092] According to an embodiment of the present disclosure, in step S103, the cluster grouping data and the initial information are input into a task allocation model based on an auction algorithm to obtain task allocation data for each cluster grouping.

[0093] As an example, step S103 may further include steps S1031 to S1032:

[0094] In step S1031 , a task allocation optimization function is constructed based on the cluster grouping data and the initial information.

[0095] In step S1032, based on the task allocation optimization function, an auction algorithm is used to obtain task allocation data for each cluster grouping under the condition that the value of the task allocation optimization function is maximized.

[0096] Regarding task allocation, according to the number of drones required for each mission area and the grouping of large-scale drones solved above, it is determined that the number of mission areas is greater than the number of drone groups, so one drone group needs to perform reconnaissance tasks in multiple mission areas. Therefore, in this disclosure, the "many-to-many" task allocation problem is mainly aimed at. By constructing a task allocation optimization function, the optimization goal is to minimize the total path of drone movement, that is, the path of the drone cluster moving from one mission area to the next mission area is the shortest, and based on the auction algorithm, task allocation that enables the drone cluster to be conflict-free is achieved while maximizing the task allocation benefits.

[0097] Based on the auction algorithm, the task allocation of UAV formation is carried out. The main goal of this task allocation is to make the movement path of UAV formation between task areas as short as possible.

[0098] Specifically, first, a task allocation optimization function is constructed, as shown in the following formula (9):

[0099]

[0100] Among them, t ij represents the travel time of the UAV formation from mission area i to mission area j, λ1 represents the weight corresponding to the travel time, λ2 represents the weight corresponding to the mission benefit index parameter, and λ1+λ2=1. represents the value allocation parameter of the area j to be covered, V j Represents the task benefit indicator parameter of task area j.

[0101] Based on the task allocation optimization function, the task sequence of each drone group is obtained through the auction algorithm as follows:

[0102]

[0103] Each row is the UAV group number, n id The mission area serial number of the UAV group.

[0104] According to an embodiment of the present disclosure, a task allocation method based on an auction algorithm is used to implement task allocation of large-scale UAVs based on kinematic constraints and biological characteristics.

[0105] According to an embodiment of the present disclosure, in step S104, the task allocation data, cluster grouping data and initial information are input into a path generation model based on an artificial potential field method to obtain a reference path for each drone cluster.

[0106] As an example, step S104 may further include steps S1041 to S1042:

[0107] In step S1041, based on the task allocation data, cluster grouping data and initial information, the initial position and target position of each drone cluster for each target task are calculated.

[0108] In step S1042, based on the artificial potential field method, the repulsive force of obstacles in the scene corresponding to each target task and the attractive force of the target point are calculated to obtain a reference path from the initial position to the target position of each drone cluster.

[0109] Specifically, for example, based on the pilot-follower theory, a virtual pilot is set for each drone cluster grouping, and according to the drone task allocation results, initial position and target position information, based on the artificial potential field method, the repulsive force of the obstacle area in the mission scene and the attractive force of the target point are calculated, so as to obtain the reference path of the drone cluster as the pilot (referred to as the pilot drone) from the initial position to the target position. The specific calculation process is as follows:

[0110] First, the attractive potential field of the target point on the pilot UAV is calculated, as shown in the following formula (11):

[0111]

[0112] Among them, ρ 2 (q,q goal ) is the distance between the target position and the current position of the pilot UAV, and ξ is a preset parameter. The attraction is the derivative of the attraction potential field with respect to the distance, which is expressed as shown in the following formula (12):

[0113]

[0114] Then, the repulsive force field of the edge of the obstacle area on the navigator UAV is calculated, as shown in the following formula (13):

[0115]

[0116] Among them, ρ(q,q obs ) is the distance between the obstacle area and the current position of the navigator UAV, and η is a preset parameter. The repulsive force is the derivative of the repulsive force potential field with respect to the distance, which is expressed as shown in the following formula (14):

[0117]

[0118] Therefore, the resultant force on the pilot UAV can be expressed as shown in the following formula (15):

[0119] F(q)=F att (q)+F rep (q) (15)

[0120] In summary, the combined force acting on the drone is calculated through the attraction and repulsion forces, and the acceleration of the drone is obtained, thereby obtaining the position of the drone, until the leader drone reaches the target position and its path planning is completed.

[0121] Through the above-mentioned path generation model based on the artificial potential field method, according to the drone task allocation results, initial position and target position, based on the artificial potential field method, the repulsive force of the obstacle area in the mission scene and the attractive force of the target point are calculated, so that the reference path of each drone cluster from the initial position to the target position can be obtained.

[0122] According to an embodiment of the present disclosure, in step S105, a virtual navigator is set for each drone cluster based on a reference path, and the reference path and the virtual navigator are input into a configuration generation model based on a pilot-follower relationship to obtain the expected position and expected speed data of each drone for different target tasks.

[0123] As an example, step S105 may further include steps S1051 to S1052:

[0124] In step S1051, based on the reference path and the virtual navigator, and according to preset geometric constraint information between the navigator and the follower based on the navigator-follower relationship, an expected formation topology structure is obtained.

[0125] In step S1052, based on the expected formation topology, the expected position and expected speed data of each UAV for different target tasks are obtained.

[0126] Through the above-mentioned configuration generation model based on the pilot-follower relationship, the expected position of the drone cluster formation is constructed according to the reference path of each drone cluster. According to the geometric constraint information between the pilot and the follower, various topological structures for the entire formation can be formed, thereby obtaining the expected position of each drone in the formation.

[0127] According to an embodiment of the present disclosure, in step S106, the basic information of the drone cluster, the expected position and expected speed data of each drone are input into a formation control model based on the bionic swarming algorithm to obtain the formation formation after each drone arrives at the corresponding expected position at the corresponding expected speed, so as to complete the formation control for each target task.

[0128] As an example, step S106 may further include steps S1061 to S1063:

[0129] In step S1061, based on the basic information of the drone cluster, the expected position and expected speed data of each drone, the model parameters of the formation control model based on the bionic swarming algorithm are initialized, and it is determined whether each drone is located at the corresponding expected position.

[0130] In step S1062, in response to determining that all the drones are located at the corresponding expected positions, a formation is obtained after each drone arrives at the corresponding expected position at the corresponding expected speed.

[0131] In step S1063, in response to determining that there is a drone that is not located at the corresponding expected position, for each drone that is not located at the corresponding expected position, the following processes S61 to S64 are executed in a loop:

[0132] In processing S61, the communication connection relationship between the drones is calculated to obtain the communication connection matrix between the drones.

[0133] In processing S62, based on the communication connection matrix between the drones, the attraction of the corresponding expected position to the drones and the interaction force between the drones are calculated.

[0134] In processing S63, the resultant force on each drone is calculated based on the attraction of the corresponding expected position on the drones and the interaction force between the drones.

[0135] In processing S64, the acceleration of each drone is calculated based on the resultant force on each drone, so as to obtain the speed and position of each drone at the next moment.

[0136] Furthermore, the formation control method disclosed in the present invention is used to perform bionic formation control on each drone group, and based on the desired position of the drone cluster formation, large-scale drone formation control is achieved. The specific implementation process can be as follows:

[0137] In general, the cluster behavior in this disclosure is constructed through the following three rule constraints:

[0138] Rule constraint 1-avoid collision: UAVs will maintain a certain distance between each other. When the distance is too close, there will be a corresponding repulsive force between adjacent UAVs; when the distance is too far, there will be an attractive force between adjacent UAVs.

[0139] Rule Constraint 2 - Gathering towards the center: Drones will move closer to the surrounding groups.

[0140] Constraint 3 - Speed ​​consistency: The speed of each drone is the same as the average speed of its neighbors.

[0141] Specifically, refer to Figure 4 , Figure 4 is an example flow chart showing a formation control method based on a bionic swarming algorithm according to an embodiment of the present disclosure.

[0142] First, the parameters of the bionic swarming algorithm are obtained and initialized (S401 and S402), the distances between the drones in the formation are calculated, the communication connection relationship of the drones is obtained according to the communication distances of the drones, and the communication connection matrix of the drone cluster is calculated (S404 and S405). Here, the calculation method of the communication connection matrix is ​​shown in the following equations (16) to (21):

[0143] a ij (p) = Φ z (z)*nrb∈[0,1] (16)

[0144] Among them, Φ z (z) represents a functional function, which is specifically defined as follows:

[0145] Φ z (z) = ρ h (z / r α ) (17)

[0146] Where z = || p j -p i || σ represents the distance between drones (where p j 、p i are the positions of the two drones respectively), r α is a constant, and nrb is the communication relationship between the two UAVs (its expression is shown in formula (20)).

[0147] The definition of σ norm is as shown in the following formula (18):

[0148]

[0149] In addition, h (z) is a uniformly changing bump function, which is specifically defined as shown in the following formula (19):

[0150]

[0151] Then, the distance between the drones is determined, and if they are within the communication range, the two drones are adjacent.

[0152]

[0153] Among them, r_c is the communication distance of the drone.

[0154] Therefore, the communication connection matrix a of the drone cluster is obtained ij (p) is defined as shown in the following formula (21):

[0155] a ij (p) = ρ h (∥pj -p i ∥ σ / r α )*nrb∈[0,1] (21)

[0156] The control input of the bionic formation control algorithm disclosed in the present invention includes the attraction of the target position to the drone (S406) and the interaction force between the drones (S408). According to the communication connection matrix of the drone cluster, if there is a communication relationship between two drones, there is an interaction force, and the target position also has an attraction to the drone. The specific control input variables can be expressed as shown in the following formula (22):

[0157]

[0158] in, is the interaction force between UAVs, is the attraction of the target location to the drone.

[0159] Here, the interaction force between drones consists of two parts: the gradient-based part and the consistent part, which can be specifically expressed as shown in the following formula (23):

[0160]

[0161] Among them, p j 、p i are the positions of the two drones in the drone cluster, n ij represents the number of neighbors (i.e., adjacent drones) of the i-th drone, which is expressed as shown in the following formula (24):

[0162]

[0163] In addition, Φ α (z) is the performance function defined by the following formula (25):

[0164] Φ α (z) = a ij (p)Φ(zd α ) (25)

[0165]

[0166] Among them, a ij (p) is the communication connection matrix between the drone clusters, z is the distance between the drones, d α , a, b, and c are all preset constants.

[0167] In the present disclosure, the attraction control input of the target point to the drone is as shown in the following formula (27):

[0168]

[0169] Among them, c1 and c2 are preset parameters, v i 、v t are the speed of the drone and the speed of the target point respectively.

[0170] The acceleration of the drone is calculated by the resultant force on the drone (S409), and the speed and position of the drone at the next moment are obtained (S412), so as to update the position of the formation. In addition, according to the reference path, the position information of the virtual navigator is updated based on the proposed configuration generation method of the pilot-follower theory. The position of the drone formation is iteratively updated, and the aggregation-dispersion process of the drones in the formation is completed based on the bionic swarming algorithm, so that the drones are driven to reach the target position in turn without collision, and the formation operation is realized, thereby completing the construction of the bionic swarming configuration generation and formation control method.

[0171] Through the above-mentioned formation control model based on the bionic swarming algorithm, the formation control of large-scale drones is realized based on the expected position of the drone cluster formation obtained above.

[0172] The formation control method disclosed in the present invention can solve the defect of the prior art in lacking an overall design for optimizing the task allocation of large-scale drone swarms and performing bionic control of formations, thereby realizing optimization of the task allocation of large-scale drones, bionic swarming formation control, and formation reorganization and control under weak communication conditions.

[0173] In addition, according to an embodiment of the present disclosure, the formation control method may further include the following processing: in response to the communication quality of some drones in the drone cluster falling below a preset threshold, the drones other than some drones in the drone cluster are used as an updated drone cluster for performing multiple target tasks, and the above steps of the formation control method are performed based on the updated drone cluster to obtain an updated formation formation to complete the update of the formation formation. Here, the preset threshold indicating the communication quality is defined as the lower limit value of the communication quality corresponding to the weak communication situation.

[0174] That is to say, when the communication of the drone is disturbed, the method disclosed in the present invention can realize the formation reorganization of the drone formation and the formation maneuvering control under weak communication conditions. Specifically, when the communication of the drone cluster is disturbed, based on the formation control method, the drone cluster can adjust the formation in time and complete the formation reorganization under weak communication. In addition, the drone cluster can change different formations at any time to achieve stable formation control.

[0175] For example, when a drone passes through an interference area, the communication network of the drone cluster is destroyed, some drones lose communication (the communication quality is so poor that communication cannot be achieved), and the original formation is destroyed. At this time, with the existing number of drones (including drones that can communicate normally and drones that can achieve weak communication), the configuration generation method based on the leader-follower relationship can obtain the expected position of the new formation, so that the original drone formation can change the formation structure through formation reorganization, thereby achieving the damage self-healing of the drone cluster formation.

[0176] Below, refer to Figure 2 An example of a formation control method for drone swarms to collaboratively achieve multi-objective tasks according to the present disclosure is described. Figure 2 is a flowchart showing an example of a formation control method for drone clusters to collaboratively achieve multi-objective tasks according to an embodiment of the present disclosure.

[0177] Reference Figure 2 ,In step S201, according to the initial position and the task attribute value, based on the cuckoo search method, the drone cluster is divided into groups to complete the drone cluster formation grouping.

[0178] In step S202, according to the grouping situation, when the task allocation benefit is maximized, the task allocation of the drone cluster grouping is performed based on the auction algorithm.

[0179] In step S203, after completing the drone cluster grouping and task allocation, a reference path is generated for each drone cluster based on the artificial potential field method.

[0180] In step S204, based on the reference path, a virtual navigator is set for each drone cluster formation, and based on the pilot-follower configuration generation method, the expected position and expected speed of each drone in the formation are obtained.

[0181] In step S205, based on the bionic swarming algorithm, the aggregation-dispersion process of the UAVs in the formation is completed, and the UAVs are driven to reach the target positions in sequence without collision, forming a formation operation to realize formation maneuver control.

[0182] In step S206, based on the formation control algorithm, when the drone cluster communication encounters interference, the drone cluster can adjust the formation in time and complete the formation reorganization under weak communication. At the same time, the drone cluster can change different formations at any time to achieve stable formation control.

[0183] Here, it should be noted that the above Figure 2 The description of each step is merely exemplary, and the steps in the method according to the present disclosure are not limited thereto.

[0184] The effect of the formation control method disclosed in the present invention is described below from the perspective of simulation results. FIG. 5A to FIG. 5C is a diagram showing the aggregation-dispersion and communication connection relationship of drone clusters in the bionic swarming algorithm according to an embodiment of the present disclosure, FIG. 6A to FIG. 6E is a schematic diagram showing simulation results of a bionic formation according to an embodiment of the present disclosure, and FIG. 7A to FIG. 7C It is a schematic diagram showing the formation reconstruction process of the bionic formation after encountering interference according to an embodiment of the present disclosure.

[0185] Reference FIG. 5A to FIG. 5C , Figure 5A It shows the aggregation status of drone cluster. Figure 5B It shows the state where the cluster starts to separate. Figure 5C It shows the status of the drone cluster separation completion.

[0186] Reference FIG. 6A to FIG. 6E , Fig. 6A The schematic diagram of the construction of the communication relationship of the drone cluster is shown. Figure 6B The schematic diagram of the construction of the target formation of the drone cluster is shown. Figure 6C A schematic diagram showing the target formation of a swarm of drones maintaining motion. Fig.6D A schematic diagram showing an example of constructing and maintaining the motion of a UAV cluster in a "human" formation. Fig. 6E A schematic diagram showing another example of constructing and maintaining motion of a swarm of drones in a V-shaped formation.

[0187] Reference FIG. 7A to FIG. 7C , Fig. 7A The diagram shows a drone swarm in a V-shaped formation being broken up due to communication interference. Figure 7B A schematic diagram showing the process of a drone cluster gradually reconnecting after communications are disrupted. Figure 7C A schematic diagram showing the change and maintenance of the drone cluster formation.

[0188] In the simulation verification of the present disclosure, it is assumed that the electromagnetic reconnaissance equipment and communication equipment carried by all drones can be used normally. From the above simulation results, it can be concluded that the formation control method proposed in the present disclosure can realize the independent control of multiple groups of drone formations in multi-target and multi-task scenarios. By adopting the multiple optimization control algorithms as described above, an effective solution framework for bionic swarming configuration generation and formation control for multi-drone coverage reconnaissance is proposed.

[0189] Figure 8 is a block diagram showing a formation control device 800 for drone clusters to collaboratively achieve multi-objective tasks according to an embodiment of the present disclosure.

[0190] Reference Figure 8According to an embodiment of the present disclosure, a formation control device 800 for drone clusters to collaboratively achieve multi-objective tasks may include a data acquisition module 810, a cluster grouping module 820, a task allocation module 830, a path generation module 840, a configuration generation module 850 and a formation control module 860.

[0191] According to an embodiment of the present disclosure, the data acquisition module 810 may execute: acquiring initial information of multiple target tasks and basic information of a drone cluster for performing multiple target tasks. Here, the initial information includes location information, value attribute information, and area information of multiple target task areas.

[0192] According to an embodiment of the present disclosure, the cluster grouping module 820 may execute: inputting initial information into a cluster determination model based on a cuckoo search algorithm to obtain cluster grouping data of the drone cluster when performing multiple target tasks.

[0193] As an example, the cluster grouping module 820 inputs the initial information into the cluster determination model based on the cuckoo search algorithm, and the operation of obtaining the cluster grouping data of the drone cluster when performing multiple target tasks may include operations 821) to 823):

[0194] In operation 821), based on the above initial information, a task area attribute parameter matrix, an optimization space and a task area objective function are constructed.

[0195] In operation 822), for each iteration cycle, perform the following processes 1) to 5) until the preset convergence conditions are met: 1) Randomly generate an initial search space including multiple initial solutions, and the solutions included in the search space represent the number of drones required to be allocated to each mission area; 2) Substitute the initial search space, the mission area attribute parameter matrix and the optimization space into the mission area objective function, and retain the solutions corresponding to the higher mission area objective function values ​​to the next generation; 3) Utilize the global Lévy flight strategy to randomly update the search space and remove a part of the solutions, and the solutions that are not removed are retained to the next generation; 4) Generate new solutions based on the local random walk strategy; 5) Calculate the fitness value of the new solution based on the adaptive scaling factor parameter, and retain the solutions corresponding to the higher fitness value to the next generation.

[0196] In operation 823), in response to satisfying a preset convergence condition, the current solution is taken as the optimal solution, and cluster grouping data of the drone cluster when performing multiple target tasks is obtained based on the optimal solution.

[0197] According to an embodiment of the present disclosure, the task allocation module 830 may execute: inputting the cluster grouping data and the initial information into a task allocation model based on an auction algorithm to obtain task allocation data for each cluster grouping.

[0198] As an example, the task allocation module 830 inputs the cluster grouping data and the initial information into the task allocation model based on the auction algorithm, and the operation of obtaining the task allocation data of each cluster grouping may include: constructing a task allocation optimization function based on the cluster grouping data and the initial information; based on the task allocation optimization function, through the auction algorithm, obtaining the task allocation data of each cluster grouping while satisfying the maximization of the task allocation optimization function value.

[0199] According to an embodiment of the present disclosure, the path generation module 840 may execute: inputting the task allocation data, cluster grouping data and initial information into a path generation model based on an artificial potential field method to obtain a reference path for each drone cluster.

[0200] As an example, the path generation module 840 inputs the task assignment data, cluster grouping data and initial information into the path generation model based on the artificial potential field method, and the operation of obtaining the reference path of each drone cluster may include: calculating the initial position and target position of each drone cluster for each target task based on the task assignment data, cluster grouping data and initial information; calculating the repulsive force of the obstacle area and the attractive force of the target point in the scene corresponding to each target task based on the artificial potential field method, and obtaining the reference path of each drone cluster from the initial position to the target position.

[0201] According to an embodiment of the present disclosure, the configuration generation module 850 can execute: based on the reference path, set a virtual navigator for each drone cluster, and input the reference path and the virtual navigator into the configuration generation model based on the leader-follower relationship to obtain the expected position and expected speed data of each drone for different target tasks.

[0202] As an example, the configuration generation module 850 inputs the reference path and the virtual navigator into the configuration generation model based on the pilot-follower relationship, and the operation of obtaining the expected position and expected speed data of each UAV for different target tasks may include: based on the reference path and the virtual navigator, according to the preset geometric constraint information between the navigator and the follower based on the pilot-follower relationship, obtaining the expected formation topology structure; based on the expected formation topology structure, obtaining the expected position and expected speed data of each UAV for different target tasks.

[0203] According to an embodiment of the present disclosure, the formation control module 860 may execute: inputting the basic information of the drone cluster, the expected position and expected speed data of each drone into a formation control model based on the bionic swarming algorithm, and obtaining the formation formation after each drone arrives at the corresponding expected position at the corresponding expected speed, so as to complete the formation control for each target task.

[0204] As an example, the formation control module 860 inputs the basic information of the drone cluster, the expected position and expected speed data of each drone into the formation control model based on the bionic swarming algorithm, and the operation of obtaining the formation formation after each drone arrives at the corresponding expected position at the corresponding expected speed may include operations 861) to 863):

[0205] In operation 861), based on the basic information of the drone cluster, the expected position and expected speed data of each drone, the model parameters of the formation control model based on the bionic swarming algorithm are initialized, and it is determined whether each drone is located at the corresponding expected position.

[0206] In operation 862), in response to determining that all the drones are located at the corresponding expected positions, a formation is obtained after each drone arrives at the corresponding expected position at the corresponding expected speed.

[0207] In operation 863), in response to determining that there are drones that are not located at the corresponding expected positions, for each drone that is not located at the corresponding expected position, the following processes 1) to 4) are executed in a loop: 1) the communication connection relationship between the drones is calculated to obtain a communication connection matrix between the drones; 2) based on the communication connection matrix between the drones, the attraction of the corresponding expected position to the drones and the interaction forces between the drones are calculated; 3) based on the attraction of the corresponding expected position to the drones and the interaction forces between the drones, the resultant force acting on each drone is calculated; 4) based on the resultant force acting on each drone, the acceleration of each drone is calculated to obtain the speed and position of each drone at the next moment.

[0208] In addition, according to an embodiment of the present disclosure, the formation control device may also be configured as: in response to the communication quality of some drones in the drone cluster dropping below a preset threshold, the drones other than some drones in the drone cluster are used as an updated drone cluster for performing multiple target tasks, and the above formation control method is executed based on the updated drone cluster to obtain an updated formation formation to complete the update of the formation formation.

[0209] Here, the preset threshold value indicates that the communication quality is defined as a lower limit value of the communication quality corresponding to a weak communication situation.

[0210] It should be noted that the operations performed on the above-mentioned structural frames can be referred to in Figure 1 The related contents described are similar and will not be repeated here.

[0211] Fig. 9 is a block diagram illustrating an electronic device 900 according to an embodiment of the present disclosure.

[0212] Reference Fig. 9According to an embodiment of the present disclosure, the electronic device 900 may include a processor 910 and a memory 920. The processor 910 may include (but not limited to) a central processing unit (CPU), a digital signal processor (DSP), a microcomputer, a field programmable gate array (FPGA), a system on a chip (SoC), a microprocessor, an application-specific integrated circuit (ASIC), etc. The memory 920 may store computer-executable instructions to be executed by the processor 910. The memory 920 includes a high-speed random access memory and / or a non-volatile computer-readable storage medium. When the processor 910 executes the computer-executable instructions stored in the memory 920, the formation control method for the drone cluster to collaboratively achieve multi-objective tasks as described above may be implemented.

[0213] According to an embodiment of the present disclosure, the formation control method for unmanned aerial vehicle clusters to collaboratively achieve multi-objective tasks can be written as a computer program / instruction to form a computer program product and stored on a computer-readable storage medium. When the computer program / instruction is executed by a processor, the formation control method for unmanned aerial vehicle clusters to collaboratively achieve multi-objective tasks as described above can be implemented. When the instructions in the computer-readable storage medium are executed by a processor of an electronic device / server, the electronic device / server is enabled to execute the formation control method for unmanned aerial vehicle clusters to collaboratively achieve multi-objective tasks as described above. Examples of computer-readable storage media include: read-only memory (ROM), random access programmable read-only memory (PROM), electrically erasable programmable read-only memory (EEPROM), random access memory (RAM), dynamic random access memory (DRAM), static random access memory (SRAM), flash memory, non-volatile memory, CD-ROM, CD-R, CD+R, CD-RW, CD+RW, DVD-ROM, DVD-R, DVD+R, DVD-RW, DVD+RW, DVD-RAM, BD-ROM, BD-R, BD-RLTH, B D-RE, Blu-ray or optical disk storage, hard disk drive (HDD), solid state drive (SSD), card storage (such as, multimedia card, secure digital (SD) card or extreme digital (XD) card), magnetic tape, floppy disk, magneto-optical data storage device, optical data storage device, hard disk, solid state disk and any other device, any other device is configured to store computer program and any associated data, data files and data structures in a non-transitory manner and provide the computer program and any associated data, data files and data structures to a processor or computer so that the processor or computer can execute the computer program. In one example, the computer program and any associated data, data files and data structures are distributed on a networked computer system so that the computer program and any associated data, data files and data structures are stored, accessed and executed in a distributed manner by one or more processors or computers.

[0214] According to the formation control method for drone clusters to collaboratively achieve multi-objective tasks according to the embodiments of the present disclosure, it is possible to achieve independent control of multiple groups of drone formations in a multi-objective and multi-task scenario.

[0215] On the other hand, according to the formation control method and device for drone clusters to collaboratively achieve multi-target tasks according to the embodiments of the present disclosure, by quickly realizing formation maneuver control of large-scale drone clusters in different scenarios, it is possible to efficiently complete various tasks such as regional reconnaissance and coverage.

[0216] On the other hand, through the formation control method and device for drone clusters to collaboratively achieve multi-objective tasks according to the embodiments of the present invention, drone cluster formation resource optimization, task allocation, path planning and independent formation control can be achieved in multi-objective and multi-task scenarios.

[0217] On the other hand, through the formation control method and device for drone swarm collaboration to achieve multi-target tasks according to the embodiments of the present disclosure, it is possible to generate bionic configurations for different scenarios and achieve formation motion maintenance. In addition, in the case of weak communication, it is possible to quickly complete formation reorganization and configuration transformation, achieve the effect of intelligent emergence of single weak drone swarm behavior, and achieve a breakthrough in the emergence and evolution technology of unmanned swarm intelligent behavior in multi-domain operations.

[0218] Although some embodiments of the present disclosure have been disclosed and described, it will be appreciated by those skilled in the art that modifications and variations may be made to the embodiments without departing from the concept and spirit of the present disclosure, the scope of which is defined by the claims and their equivalents.

Claims

1. A formation control method for drone swarms to collaboratively achieve multi-objective tasks, characterized in that: The formation control method comprises: Acquire initial information of multiple target tasks and basic information of a drone cluster used to execute the multiple target tasks, wherein the initial information includes location information, value attribute information, and area information of multiple target task areas; Inputting the initial information into a cluster determination model based on a cuckoo search algorithm to obtain cluster grouping data of the drone cluster when performing the multiple target tasks; Inputting the cluster grouping data and the initial information into a task allocation model based on an auction algorithm to obtain task allocation data for each cluster grouping; Inputting the task allocation data, the cluster grouping data and the initial information into a path generation model based on an artificial potential field method to obtain a reference path for each drone cluster; Based on the reference path, a virtual navigator is set for each drone cluster, and the reference path and the virtual navigator are input into a configuration generation model based on a pilot-follower relationship to obtain expected position and expected speed data of each drone for different target tasks; The basic information of the drone cluster, the expected position and expected speed data of each drone are input into the formation control model based on the bionic swarming algorithm, and the formation formation after each drone arrives at the corresponding expected position at the corresponding expected speed is obtained to complete the formation control for each target task.

2. The formation control method according to claim 1, characterized in that: The formation control method further includes: In response to the communication quality of some drones in the drone cluster falling below a preset threshold, the drones in the drone cluster other than the part of the drones are used as an updated drone cluster for performing the multiple target tasks, and the formation control method is executed based on the updated drone cluster to obtain an updated formation formation, so as to complete the update of the formation formation. The preset threshold indicating the communication quality is defined as a lower limit value of the communication quality corresponding to a weak communication situation.

3. The formation control method according to claim 1, characterized in that: The step of inputting the initial information into a cluster determination model based on a cuckoo search algorithm to obtain cluster grouping data of the drone cluster when performing the multiple target tasks includes: Based on the initial information, construct a task area attribute parameter matrix, an optimization space, and a task area objective function; For each iteration cycle, the following processing is performed until the preset convergence condition is met: An initial search space including a plurality of initial solutions is randomly generated, wherein the solutions included in the search space represent the number of drones required to be assigned to each mission area; Substituting the initial search space, the task area attribute parameter matrix and the optimization space into the task area objective function, and retaining the solution corresponding to the higher task area objective function value to the next generation; Using the global Lévy flight strategy, the search space is randomly updated and some solutions are removed, and the solutions that are not removed are retained to the next generation; Generate new solutions based on local random walk strategy; Based on the adaptive scale factor parameter, the fitness value of the new solution is calculated, and the solution corresponding to the higher fitness value is retained to the next generation; In response to satisfying a preset convergence condition, the current solution is taken as the optimal solution, and cluster grouping data of the drone cluster when performing the multiple target tasks is obtained based on the optimal solution.

4. The formation control method according to claim 1, characterized in that: The step of inputting the cluster grouping data and the initial information into the task allocation model based on the auction algorithm to obtain the task allocation data for each cluster grouping comprises: Based on the cluster grouping data and the initial information, construct a task allocation optimization function; Based on the task allocation optimization function, through the auction algorithm, the task allocation data of each cluster grouping is obtained under the condition of maximizing the value of the task allocation optimization function.

5. The formation control method according to claim 1, characterized in that: The step of inputting the task allocation data, the cluster grouping data and the initial information into a path generation model based on an artificial potential field method to obtain a reference path for each drone cluster comprises: Calculate the initial position and target position of each drone cluster for each target task based on the task assignment data, the cluster grouping data and the initial information; Based on the artificial potential field method, the repulsive force of the obstacle area and the attractive force of the target point in the scene corresponding to each target task are calculated to obtain the reference path of each drone cluster from the initial position to the target position.

6. The formation control method according to claim 1, characterized in that: The step of inputting the reference path and the virtual navigator into the configuration generation model based on the pilot-follower relationship to obtain the expected position and expected speed data of each UAV for different target tasks includes: Based on the reference path and the virtual navigator, and according to preset geometric constraint information between the navigator and the follower based on a navigator-follower relationship, an expected formation topology structure is obtained; Based on the expected formation topology, expected position and expected speed data of each UAV for different target tasks are obtained.

7. The formation control method according to claim 1, characterized in that: The step of inputting the basic information of the drone cluster, the expected position and expected speed data of each drone into the formation control model based on the bionic swarming algorithm to obtain the formation after each drone arrives at the corresponding expected position at the corresponding expected speed includes: Based on the basic information of the UAV cluster, the expected position and expected speed data of each UAV, the model parameters of the formation control model based on the bionic swarming algorithm are initialized, and it is determined whether each UAV is located at the corresponding expected position; In response to determining that all the drones are located at the corresponding expected positions, obtaining a formation after each drone arrives at the corresponding expected position at the corresponding expected speed; In response to determining that there is a drone that is not located at the corresponding expected position, for each drone that is not located at the corresponding expected position, the following process is performed in a loop: Calculate the communication connection relationship between the UAVs and obtain the communication connection matrix between the UAVs; Based on the communication connection matrix between the UAVs, the attraction of the corresponding expected position to the UAVs and the interaction force between the UAVs are calculated; Calculate the resultant force on each drone based on the attraction of the corresponding expected position on the drones and the interaction forces between the drones; Based on the resultant force acting on each drone, the acceleration of each drone is calculated to obtain the speed and position of each drone at the next moment.

8. A formation control device for drone swarms to collaboratively achieve multi-objective tasks, characterized in that: The formation control device comprises: The data acquisition module is configured to: acquire initial information of multiple target tasks and basic information of a drone cluster for executing the multiple target tasks, wherein the initial information includes location information, value attribute information, and area information of multiple target task areas; The cluster grouping module is configured to: input the initial information into a cluster determination model based on a cuckoo search algorithm to obtain cluster grouping data of the drone cluster when performing the multiple target tasks; A task allocation module is configured to: input the cluster grouping data and the initial information into a task allocation model based on an auction algorithm to obtain task allocation data for each cluster grouping; A path generation module is configured to: input the task allocation data, the cluster grouping data and the initial information into a path generation model based on an artificial potential field method to obtain a reference path for each drone cluster; The configuration generation module is configured to: set a virtual navigator for each drone cluster based on the reference path, and input the reference path and the virtual navigator into a configuration generation model based on a pilot-follower relationship to obtain expected position and expected speed data of each drone for different target tasks; The formation control module is configured to: input the basic information of the drone cluster, the expected position and expected speed data of each drone into the formation control model based on the bionic swarming algorithm, and obtain the formation formation after each drone arrives at the corresponding expected position at the corresponding expected speed, so as to complete the formation control for each target task.

9. A computer program product, characterized in that The computer program product includes a computer program / instruction, which, when executed by a processor, implements a formation control method for drone clusters to collaboratively achieve multi-objective tasks as described in any one of claims 1 to 7.

10. An electronic device, characterized in that: The electronic device includes: at least one processor; at least one memory storing computer-executable instructions, wherein the computer-executable instructions, when executed by the at least one processor, prompt the at least one processor to execute the formation control method for drone clusters to collaboratively achieve multi-objective tasks as described in any one of claims 1 to 7.

Images