Multi-unmanned aerial vehicle cooperative target searching method and system based on particle swarm algorithm

Abstract

The present application belongs to the field of swarm intelligence and multi-agent target search technology, and provides a multi-unmanned aerial vehicle cooperative target search method and system based on a particle swarm algorithm. The present application calculates the position of each unmanned aerial vehicle at the next time point through the particle swarm algorithm, and detects the surrounding environment in real time during the actual flight of the unmanned aerial vehicle to the target point at the next time point, and re-plans in time when an obstacle is detected to realize the functions of avoiding obstacles and avoiding inter-aircraft conflicts. In the particle swarm algorithm, an energy factor that decays over time is introduced to simulate the rule that the search ability decreases over time, and a maximum speed constraint is introduced, so that after the size and direction of the speed at the next time point are calculated, the speed of the unmanned aerial vehicle in the calculated direction does not exceed the maximum speed constraint.

Description

Technical Field

[0001] This invention belongs to the field of swarm intelligence and multi-agent target search technology, and particularly relates to a multi-UAV cooperative target search method and system based on particle swarm algorithm. Background Technology

[0002] In recent years, considering the advantages of Particle Swarm Optimization (PSO) such as simple parameters, fast convergence speed, low computational complexity and small data transmission volume, PSO has been widely used in multi-UAV target search. Specifically, each UAV is abstracted as a particle in the particle swarm, and the position of the particle is encoded, so that the UAV can gradually move towards the global optimum and then discover the target location.

[0003] The inventors discovered that current multi-UAV cooperative target search methods based on particle swarm optimization do not fully capture the characteristics of UAVs, and most research is based on idealized assumptions. Specifically, UAVs are treated as point masses, there are no obstacles in the environment, and inter-UAV collisions are not considered. A few algorithms that consider obstacle avoidance only try to avoid collisions by changing speed direction, braking suddenly, or changing altitude when an obstacle is detected. However, in real-world applications, obstacles are ubiquitous. Simply changing speed direction, braking suddenly, or changing altitude is insufficient for obstacle avoidance in multi-obstacle environments, let alone resolving inter-UAV conflicts, severely impacting search success rates. Furthermore, considering the inherent characteristics of UAVs, their speed cannot approach infinity; drastic speed changes are inconsistent with UAV dynamics. In practical applications, drastic speed changes not only cause greater energy loss but also greatly increase the risk of accidents. Summary of the Invention

[0004] To address the aforementioned problems, this invention proposes a multi-UAV cooperative target search method and system based on particle swarm optimization (PSO). This invention considers the actual characteristics of UAVs and introduces an autonomous obstacle avoidance algorithm to optimize the actual operating trajectory of the UAVs, solving the obstacle avoidance problem in multi-obstacle environments and the problem of inter-UAV conflicts. Furthermore, it considers the energy limitations of UAVs by adding a time-decreasing energy factor to the PSO algorithm, simulating the objective law of the gradual decline in UAV search capability over time, and providing a reliable basis for the study of energy loss processes.

[0005] To achieve the above objectives, the present invention is implemented through the following technical solution:

[0006] In a first aspect, the present invention provides a multi-UAV cooperative target search method based on particle swarm optimization, comprising:

[0007] Obtain the location information of the target to be searched;

[0008] Based on a pre-defined particle swarm optimization algorithm, multiple drones are controlled to move towards the location of the target to be searched.

[0009] Among them, an energy factor that decays over time is introduced into the particle swarm algorithm to simulate the law of the decline in search capability over time, and a maximum speed constraint is introduced. After calculating the magnitude and direction of the speed at the next moment, the speed of the UAV in the calculated direction is controlled to not exceed the maximum speed constraint.

[0010] Furthermore, after calculating the target position of each UAV from the current time to the next time using the particle swarm optimization algorithm, and determining the guidance path from the current position to the target position at the next time, a UAV control command to push away the obstacle is generated at all positions on the guidance path where there are obstacles. When determining obstacles on the guidance path, other UAVs within a preset range around the currently controlled UAV and with higher priority are regarded as obstacles.

[0011] Furthermore, inertia weight parameters, individual acceleration factor parameters, and social acceleration factor parameters are added to the particle swarm algorithm, and the weights of each parameter are dynamically adjusted at each stage of control.

[0012] Furthermore, at the beginning of the search task, the inertia weight value is the largest; as the search task is executed, the inertia weight decreases and the social acceleration factor increases; after the social acceleration factor reaches the preset value, the individual acceleration factor gradually increases.

[0013] Furthermore, the energy factor at the next moment is equal to the product of the energy factor at the current moment and the energy decay rate, where the energy decay rate simulates the rate of energy decay during the flight of the UAV.

[0014] Furthermore, each drone is set as a single particle in the particle swarm algorithm; the actual position of the drone is set as the particle position.

[0015] Furthermore, the control process from the current moment to the next moment is divided into a search phase and an optimization phase. In the search phase, the A* algorithm is used to obtain a guidance path that does not consider obstacles. In the optimization phase, the environment around the UAV is detected, the trajectory existing in the obstacle is compared with the collision-free guidance path generated in the search phase, and gradient information is used to pull the trajectory in the obstacle out of the obstacle.

[0016] Secondly, the present invention also provides a multi-UAV cooperative target search system based on particle swarm optimization algorithm, comprising:

[0017] The data acquisition module is configured to: acquire the location information of the target to be searched;

[0018] The search control module is configured to control multiple drones to move towards the location of the target to be searched, based on a preset particle swarm algorithm.

[0019] Among them, an energy factor that decays over time is introduced into the particle swarm algorithm to simulate the law of the decline in search capability over time, and a maximum speed constraint is introduced. After calculating the magnitude and direction of the speed at the next moment, the speed of the UAV in the calculated direction is controlled to not exceed the maximum speed constraint.

[0020] Thirdly, the present invention also provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the steps of the multi-UAV cooperative target search method based on particle swarm optimization described in the first aspect.

[0021] Fourthly, the present invention also provides an electronic device, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the program to implement the steps of the multi-UAV cooperative target search method based on particle swarm optimization described in the first aspect.

[0022] Compared with the prior art, the beneficial effects of the present invention are as follows:

[0023] 1. This invention introduces an energy factor that decays over time into the particle swarm optimization algorithm to simulate the decline in search capability over time. It also introduces a maximum speed constraint. After calculating the magnitude and direction of the velocity at the next moment, the invention controls the UAV's speed in the calculated direction to not exceed the maximum speed constraint. While adhering to the objective law that the UAV's energy level decreases and its search capability weakens over time during mission execution, this invention allows the speed to reach its maximum without exceeding the maximum speed constraint, thus solving the problem of excessive energy loss and accidents caused by drastic speed changes.

[0024] 2. This invention calculates the target position of each UAV from the current time to the next time using the particle swarm optimization algorithm, and determines the guidance path from the current position to the target position. Then, it generates a UAV control command to push away obstacles at all positions on the guidance path where there are obstacles. Obstacle avoidance can be achieved in multi-obstacle environments by using the UAV control command to push away obstacles. Furthermore, other UAVs within a preset range around the currently controlled UAV and with higher priority are regarded as obstacles. Priority is given to ensuring that the higher priority UAVs fly according to the guidance path, while the lower priority UAVs perform obstacle avoidance tasks on the guidance path. This ensures the flight stability of the higher priority UAVs among multiple UAVs, guarantees the control stability and search capability of multiple aircraft, and solves the problem of inter-UAV conflict.

[0025] 3. This invention introduces an adaptive strategy into the particle swarm optimization algorithm. By dynamically adjusting the weights of inertia weight, individual acceleration factor, and social acceleration factor at each stage, it avoids the problem of control getting trapped in local optima and improves search capability. Specifically, at the beginning of the search task, inertia weight dominates to obtain strong global search capability. As the search task progresses, inertia weight begins to decrease, social acceleration factor begins to increase, and the attraction of global optimum to particles begins to increase, thus enabling particles to escape the current local optimum. After the social acceleration factor reaches a preset value, individual acceleration factor begins to gradually increase, and the attraction of individual optimum to particles increases, thus enabling UAVs to gather near the target location in the later stages. Attached Figure Description

[0026] The accompanying drawings, which form part of this embodiment, are used to provide a further understanding of this embodiment. The illustrative embodiments and their descriptions are used to explain this embodiment and do not constitute an improper limitation of this embodiment.

[0027] Figure 1 This is a flowchart of Embodiment 1 of the present invention;

[0028] Figure 2 This is a schematic diagram of the basic particle swarm optimization algorithm speed update principle in Embodiment 2 of the present invention;

[0029] Figure 3 This is a simulated implementation scenario diagram of Embodiment 2 of the present invention;

[0030] Figure 4 This is a simulation diagram of the control after 1 second in Embodiment 2 of the present invention;

[0031] Figure 5 This is a simulation diagram of the control process for 30 seconds in Embodiment 2 of the present invention;

[0032] Figure 6 This is a simulation diagram of the control process for 60 seconds in Embodiment 2 of the present invention;

[0033] Figure 7 This is a simulation diagram of the control 90s in Embodiment 2 of the present invention;

[0034] Figure 8 This is a schematic diagram illustrating conflict avoidance in Embodiment 2 of the present invention. Detailed Implementation

[0035] The present invention will be further described below with reference to the accompanying drawings and embodiments.

[0036] It should be noted that the following detailed descriptions are exemplary and intended to provide further explanation of this application. Unless otherwise specified, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this application pertains.

[0037] Example 1:

[0038] Currently, when using particle algorithms to control UAVs for collaborative target search, collisions are avoided by simply changing speed direction, emergency braking, or changing flight altitude when obstacles are present. However, in real-world applications, obstacles are ubiquitous in the environment, and relying solely on changing speed direction, emergency braking, or changing flight altitude is insufficient for obstacle avoidance in multi-obstacle environments. Furthermore, traditional control methods do not consider conflicts between UAVs and cannot resolve inter-UAV conflicts during flight, severely impacting the search success rate.

[0039] To address the obstacle avoidance problem and inter-drone conflict in multi-drone control environments, this embodiment provides a multi-drone cooperative target search method based on particle swarm optimization, including:

[0040] Obtain the location information of the target to be searched; it is understood that the target to be searched can be the same target to be searched by multiple aircraft, or different targets to be searched can be determined for one or more different aircraft.

[0041] Based on a pre-defined particle swarm optimization algorithm, multiple drones are controlled to move towards the location of the target to be searched.

[0042] Specifically, the particle swarm optimization algorithm is used to calculate the target position of each drone from the current time to the next time, and to determine the guidance path from the current position to the target position in the next time. Then, a drone control command to push away the obstacle is generated at all locations on the guidance path where there are obstacles. When determining obstacles on the guidance path, other drones within a preset range around the currently controlled drone and with higher priority are regarded as obstacles.

[0043] Specifically, the particle swarm optimization algorithm is used to calculate the target position of each UAV from the current time to the next time, and to determine the guidance path from the current position to the target position in the next time. Then, a UAV control command to push away obstacles is generated at all positions on the guidance path where there are obstacles. Obstacle avoidance can be achieved in multi-obstacle environments by using the UAV control command to push away obstacles. Furthermore, other UAVs within a preset range around the currently controlled UAV and with higher priority are regarded as obstacles. Priority is given to ensuring that the higher priority UAVs fly according to the guidance path, while the lower priority UAVs are controlled to perform obstacle avoidance tasks on the guidance path. This ensures the flight stability of the higher priority UAVs among multiple UAVs, guarantees the control stability and search capability of multiple aircraft, and solves the problem of inter-UAV conflict.

[0044] To explain the multi-UAV cooperative target search method based on particle swarm optimization in this embodiment, the steps of the method may include:

[0045] S1. Model the search environment and the multiple controlled drones;

[0046] S2. Initialize the parameters in the particle swarm algorithm and initialize the position and velocity of all drones;

[0047] S3. When each controlled drone flies from the current moment to the target position in the next moment, it applies an obstacle avoidance algorithm to avoid obstacles and prevent the controlled drone from colliding with other drones.

[0048] S4. The UAV reaches the target position at a certain moment and calculates the fitness of that position; the UAVs share group information and calculate the target position at the next moment.

[0049] In step S1, each drone can be set as a single particle in the particle swarm algorithm; the actual position of the drone is set as the particle position.

[0050] In step S3, an obstacle avoidance and conflict avoidance mechanism between the autonomous obstacle avoidance drones is introduced into the particle swarm optimization (PSO) algorithm. When the drone calculates and proceeds to the target position x(t+1) at the next moment using the PSO algorithm, the traditional PSO control method of moving along a straight line is changed. In this embodiment, a path planning mechanism is triggered. The entire planning process from the current moment to the next moment is divided into two processes: front-end search and back-end optimization. In the front-end search phase, each drone searches for a path from the current moment to the next moment's target position using dynamics A*, without considering obstacles. In the back-end trajectory optimization phase, it checks whether the trajectory generated in the previous stage will collide with obstacles and generates a force based on the gradient to push the trajectory away from the obstacle, serving as the drone control command to push away the obstacle. Furthermore, if the generated trajectory violates the dynamic constraints due to unreasonable time allocation, a time reallocation process is executed. To save limited onboard resources, a back-to-the-horizon strategy is introduced into the planning process, ensuring that path planning is only performed within the drone's perception range, and environments outside the drone's sensor detection range are not considered. In addition, this embodiment also considers inter-drone conflict avoidance by introducing a priority strategy into the drone. Priority can be understood as determining the order in which multiple drones execute their plans when multiple drones plan to pass through a narrow space at the same time. Drones with higher priority can plan and pass first, while lower-priority drones will treat higher-priority drones within a certain sensing range as obstacles and perform obstacle avoidance.

[0051] In step S4, inertia weights are introduced into the particle swarm optimization algorithm to eliminate the influence of the maximum velocity V in each dimension of the particle. max The requirements. The formulas for particle velocity and position are expressed as follows:

[0052] υ id =ω*υ id +c1*rand()*(p best -x id )+c2*rand()*(g best -x id )

[0053] x id =x id +υ id

[0054] Where ω represents the inertia weight; p best Represents the best position that each particle has found so far; g best This represents the optimal position found so far for the entire particle swarm; c1 and c2 are called the individual acceleration factor and the social acceleration factor, respectively, representing the position that pulls the particle towards p. best or g bestThe weights of the random acceleration term; υ id x represents the velocity of the i-th particle in the d-th iteration; id This represents the position of the i-th particle in the d-th iteration.

[0055] This real-time particle swarm optimization algorithm introduces an adaptive strategy, dynamically adjusting the weights of inertia weight ω, individual acceleration factor c1, and social acceleration factor c2 at each stage to avoid getting trapped in local optima and thus improve search capability. At least one search method in this embodiment is for single-target search, therefore, getting trapped in local optima should be avoided. Specifically, at the beginning of the search task, the inertia weight is assigned its maximum value and should dominate to obtain strong global search capability. As the search task progresses, the inertia weight begins to decrease, the social acceleration factor begins to increase, and the attraction of the global optimum to the particles begins to strengthen, enabling the particles to escape the current local optimum. After the social acceleration factor reaches its preset optimal value, the individual acceleration factor gradually increases, and the attraction of the individual optimum to the particles increases, allowing the UAV to achieve aggregation near the target location in the later stages of the experiment.

[0056] In this embodiment, energy and velocity constraints can be added to the particle swarm optimization algorithm. Regarding energy, considering that the battery life of commonly used drones is extremely limited in practical applications—that is, the energy level inevitably decreases over time during actual flight—an energy constraint is introduced into the traditional particle swarm optimization algorithm. This is achieved by introducing an energy factor that decays over time to simulate the decline in search capability. Optionally, the energy factor at the next moment is equal to the product of the energy factor at the current moment and the energy decay rate, where the energy decay rate simulates the rate of energy decay during drone flight. Regarding velocity, considering that the speed of a drone in actual flight has an upper limit and cannot arbitrarily approach infinity, a maximum speed constraint is introduced. When the magnitude of the velocity vector calculated by the particle swarm optimization algorithm at the next moment exceeds the maximum safe speed that the drone can reach, the direction of the vector is maintained, but its magnitude is limited to the upper limit.

[0057] Specifically, an energy factor that decays over time is introduced into the particle swarm optimization algorithm to simulate the decline in search capability over time. A maximum speed constraint is also introduced. After calculating the magnitude and direction of the speed at the next moment, the speed of the UAV in the calculated direction is controlled to not exceed the maximum speed constraint. Based on the objective law that the energy level of the UAV decreases and the search capability weakens over time during mission execution, the speed can reach the maximum without exceeding the maximum speed constraint and causing safety issues. This solves the problem of excessive energy loss and accidents caused by drastic speed changes.

[0058] In this embodiment, the particle swarm optimization (PSO) algorithm is applied to a multi-UAV target search task. Each UAV in the swarm is treated as a single particle in the swarm, and the position of each particle is encoded. This allows the particles to gradually move towards the global optimum as iterations progress, thereby effectively discovering the target location. An adaptive strategy is introduced into the PSO algorithm, dynamically adjusting the weights of various parameters throughout the search process. This gives the UAVs strong global search capabilities in the early stages of the search task, effectively avoids getting trapped in local minima, and achieves rapid convergence to the target location in the later stages of the search. Considering the characteristics that UAVs should possess in actual flight, such as energy limitations during flight, an energy factor that decays over time is introduced into the traditional PSO algorithm. This respects the objective law that the energy level of UAVs decreases and their search capabilities weaken over time during task execution. In addition, a maximum speed limit is added to the UAVs, constraining their speed within a safe range to prevent it from approaching infinity. Considering the possibility of environmental obstacles and inter-drone collisions during actual mission execution, an autonomous obstacle avoidance algorithm is introduced into the traditional particle swarm optimization algorithm to optimize the actual flight trajectory of UAVs. A priority strategy is also introduced so that high-priority UAVs are regarded as obstacles by low-priority UAVs and obstacle avoidance is implemented to achieve the conflict avoidance function in complex obstacle environments. This significantly improves the practical application value of the multi-UAV cooperative particle swarm search algorithm.

[0059] Example 2:

[0060] To further explain and verify the multi-UAV cooperative target search method based on particle swarm optimization in Example 1, this example provides a specific simulation case, as follows:

[0061] like Figure 3 As shown, optionally, the search task is conducted in a rectangular space of 12m × 13m. To ensure the universality of the proposed method, at the start of each search task, the environment can be filled with obstacles of relatively random number, size, and position. These obstacles can be circular, cylindrical, or other shapes. In one embodiment, circular obstacles allow the drone to pass through or fly over them, while cylindrical obstacles allow the drone to bypass them from the left or right sides. The target position is randomly selected in the search space and will not conflict with any obstacles.

[0062] Any number of drones can be selected. In one embodiment, four drones are selected. Specifically, four drones of the same model and configuration are arranged side-by-side at the edge of the search scene, spaced 1 meter apart. Each drone is equipped with hardware modules such as a binocular camera, depth camera, GPS, and communication module, and all modules maintain good performance during flight. Throughout the search process, the four drones can communicate with each other via wireless communication to exchange information. The specific implementation steps are as follows:

[0063] S1. Model the search environment. Optionally, for a given rectangular search area, divide it into square grids, for example, setting the actual area of ​​each grid to 1m × 1m. Targets are distributed in the search environment according to probability. Model the drones. Optionally, treat the drone swarm as a particle swarm, with each drone modeled as a single particle in the swarm, and the actual position of the drone considered as the particle position. Treat the sensor return values ​​on the drones as the environmental fitness value f. The specific formula is as follows:

[0064] Search area:

[0065] S: {x: 0 ≤ x ≤ x} max}, {y: 0 ≤ y ≤ y max}

[0066] Assume d is the return value of the drone sensor:

[0067] The search objective of f(x, y) = d is to find (x*, y*) ∈ S such that

[0068] S2. Initialize particle swarm parameters. Optional parameters: number of particles n = 4, maximum velocity υ. max = 2 m / s, the maximum number of iterations is set to 10,000, the termination condition is set to reaching the maximum number of iterations or no new global optimum is generated after 20 iterations, c1 = c2 = 2, R1 and R2 are set to random numbers with equal probability between 0 and 1, and the energy decay rate λ is set to 0.95. Initialize the speed and position of the 4 UAVs respectively.

[0069] In practice, the initial parameters used for particle swarm optimization can be assigned based on references, experimental parameters, or empirical parameters. However, the initial parameters often fail to achieve optimal results. During debugging, a specific parameter can be adjusted while keeping other parameter values ​​unchanged. The improvement in search performance can be observed. If the performance improves, the modification can be retained. This process of adjusting each parameter continues until the overall search performance reaches its optimal level.

[0070] S3 and four drones each travel to their respective locations; the entire process can be divided into two parts: front-end search and back-end optimization. In the front-end search phase, the A* algorithm is used to obtain a guidance path that does not consider obstacles. In the back-end optimization phase, the drones use sensors to detect the surrounding environment, compare the trajectories within obstacles with the collision-free guidance paths generated in the search phase, and use gradient information to pull the trajectories within obstacles out of them, thereby generating and optimizing multiple trajectories across multiple threads, while simultaneously selecting the trajectory with the lowest execution cost. The nonlinear optimization problem of a uniform B-spline with Q control points is given by the following equation:

[0071]

[0072] Where J is the penalty function to be optimized, which is determined by the weighted penalty J. r The composition is defined as follows: r = {s, c, d, t}, representing smoothness, collision, dynamic feasibility, and terminal progress, respectively; λ represents the corresponding weight J. r The desired optimization penalty function is defined by penalties. Since particle swarm optimization is a multi-drone search, collision avoidance between drones must also be considered. To avoid drone conflicts, a drone priority mechanism can be introduced, where lower-priority drones must consider the trajectories of higher-priority drones, treating them as obstacles. To conserve resources, only drones within a certain range need to be considered; if the distance to a higher-priority drone exceeds a threshold, its impact is disregarded. Therefore, the optimization problem becomes J′:

[0073]

[0074] Where, λ w J represents the conflict weight corresponding to the weighted conflict penalty term; ω For the weighted conflict penalty term, λ ω This represents the corresponding penalty weight.

[0075] When the dynamics of a certain trajectory segment are not feasible, time reallocation is used to increase the time allocated to that trajectory segment, thereby increasing its dynamic feasibility. Furthermore, to conserve the limited onboard resources of the UAV, a back-to-the-horizon strategy is introduced into the planning process, ensuring that path planning is only performed within the UAV's perception range.

[0076] S4. The drone completes obstacle avoidance and reaches the previously designated particle position. It uses its onboard sensors to detect the environment, determining if any targets exist in the current area, and uses this as the particle's fitness at that position. Assume there is a broadcast network between the drones, enabling them to share information. The particle swarm optimization algorithm updates the velocity and position of the particle corresponding to each drone, as shown in the following formula:

[0077] υid =ε*(ω*υ id +c1*rand()*(p best -x id )+c2*rand()*(g best -x id ))

[0078] x id =x id +υ′ id

[0079] Where ε is the energy factor; ω is the inertia weight; p best Represents the best position that each particle has found so far; g best c1 and c2 represent the optimal position found so far by the entire particle swarm; c1 and c2 are called the individual acceleration factor and the social acceleration factor, respectively; representing the pull of the particle towards p. best or g best The weights of the random acceleration term; υ id x represents the velocity of the i-th particle in the d-th iteration; id v′ represents the position of the i-th particle in the d-th iteration; id It is the actual execution speed of the drone under speed constraints.

[0080] This embodiment introduces an adaptive strategy into the particle swarm optimization algorithm. By dynamically adjusting the weights of parameters such as inertia weight ω, individual acceleration factor c1, and social acceleration factor c2, the search capability is improved and local optima are avoided. At the start of the search, the inertia weight dominates with a value of 1 to achieve strong global search capability. As the search progresses, the inertia weight decreases from its maximum value, while the social acceleration factor increases from 0, strengthening the attraction of the global optimum to particles and preventing the drone from getting trapped in local optima. After the social acceleration factor reaches its optimal value, the individual acceleration factor increases from 0, further increasing the attraction of the individual optimum to particles. Considering the energy limitations of the drone, a time-decreasing energy constraint is added to the particle swarm optimization algorithm to simulate the gradual decline in the drone's search capability over time. The energy factor ε follows the following decay law:

[0081] ε(t+1)=λ*ε(t)

[0082] Wherein, λ is the energy decay rate, used to simulate how fast the energy decays during the flight of the drone, and the value of λ can be 0.95.

[0083] Furthermore, considering that the speed cannot reach infinity during actual flight, a speed constraint is introduced into the UAV to ensure that while maintaining the direction of the velocity vector, the magnitude of the speed remains within an upper limit, i.e., does not exceed the maximum speed V. max The speed constraints are as follows:

[0084]

[0085] S5. If the termination condition is reached early or the number of iterations exceeds the set maximum number of iterations, proceed to step S6; otherwise, return to step S3 for the next iteration.

[0086] S6. This search task has ended.

[0087] To verify the performance of the proposed algorithm, simulation experiments were conducted on a 2.30GHz Intel i7-11800H processor with 16.0GB of memory and 512GB of SSD. The simulation environment was Linux Ubuntu 18.04. In addition, Rviz was used to visualize the entire search process.

[0088] The simulation results of this invention are as follows: Figure 4 , Figure 5 , Figure 6 and Figure 7 As shown in the diagram. The experimental results show that at the beginning of the experiment, due to the significant influence of inertial weights, the four drones flew along essentially straight trajectories without encountering obstacles, and the search range could be expanded to the entire search space in a short time. As the experiment progressed, the drones began to exhibit autonomous turning behavior without obstacle interference, indicating that they had entered the second stage, where the attraction of the globally optimal position began to increase. In the final stage of the experiment, after 90 seconds of searching, it was observed that the drones were attracted by the local optimal position and were able to basically cluster near the target location. The simulation experiment ended when the termination condition was met. Throughout the process, it was observed that the search step size of the drones gradually decreased in each iteration, effectively simulating the actual law that the drones' energy level decreases and their search capability gradually weakens over time.

[0089] While implementing the basic search function, the simulation process also effectively demonstrated the collision avoidance and conflict prevention capabilities of the designed mechanism. During flight, it can be observed that the relative target position of each UAV frequently changes with each iteration. Once a new target is calculated, a new trajectory is quickly planned, generated, and executed. If the generated path overlaps with a detected obstacle, the algorithm's replanning mechanism is triggered within a very short time. Furthermore, it can be observed that the replanning mechanism can be triggered multiple times on the way to the target until a relatively smooth, collision-free trajectory is generated leading to the target position in the current iteration. Regarding conflict prevention, such as... Figure 8 As shown, when multiple drones are performing search tasks by traversing relatively dense gaps between obstacles, the inter-drone conflict avoidance mechanism is triggered, allowing multiple drones to pass through in an orderly manner according to priority.

[0090] To eliminate randomness and prove the universality of the proposed particle swarm search mechanism, this embodiment repeated the above experiment 100 times. Each experiment was conducted until the termination condition was met or the UAV collided with the obstacle. The experiment was considered successful when the UAV converged to the vicinity of the target position without any collisions. The final conclusion was that the success rate of the algorithm was higher than 90%.

[0091] The simulation experiments above show that the method proposed in this invention has a good search effect when applied to UAVs, and can locate the target position in a short time. It also verifies that the introduced obstacle avoidance and conflict avoidance functions are also suitable for group task execution in complex environments, which significantly improves the practicality of particle swarm algorithm.

[0092] Example 3:

[0093] This embodiment provides a multi-UAV cooperative target search system based on particle swarm optimization algorithm, including:

[0094] The data acquisition module is configured to: acquire the location information of the target to be searched;

[0095] The search control module is configured to control multiple drones to move towards the location of the target to be searched, based on a preset particle swarm algorithm.

[0096] Among them, an energy factor that decays over time is introduced into the particle swarm algorithm to simulate the law of the decline in search capability over time, and a maximum speed constraint is introduced. After calculating the magnitude and direction of the speed at the next moment, the speed of the UAV in the calculated direction is controlled to not exceed the maximum speed constraint.

[0097] The working method of the system is the same as that of the multi-UAV cooperative target search method based on particle swarm optimization algorithm in Embodiment 1, and will not be repeated here.

[0098] Example 4:

[0099] This embodiment provides a computer-readable storage medium storing a computer program that, when executed by a processor, implements the steps of the multi-UAV cooperative target search method based on particle swarm optimization described in Embodiment 1.

[0100] Example 5:

[0101] This embodiment provides an electronic device, including a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the program, it implements the steps of the multi-UAV cooperative target search method based on particle swarm optimization described in Embodiment 1.

[0102] The above description is merely a preferred embodiment of this practice and is not intended to limit the scope of this practice. Various modifications and variations can be made to this practice by those skilled in the art. Any modifications, equivalent substitutions, or improvements made within the spirit and principles of this practice should be included within the protection scope of this practice.

Claims

1. A multi-UAV cooperative target search method based on particle swarm optimization algorithm, characterized in that, include: Obtain the location information of the target to be searched; Based on a pre-defined particle swarm optimization algorithm, multiple drones are controlled to move towards the location of the target to be searched. Among them, an energy factor that decays over time is introduced into the particle swarm algorithm to simulate the law of the decline in search capability over time, and a maximum speed constraint is introduced. After calculating the magnitude and direction of the speed at the next moment, the speed of the UAV in the calculated direction is controlled to not exceed the maximum speed constraint. The drone completes obstacle avoidance and reaches the previously designated particle position. It uses its onboard sensors to detect the environment, determining if any targets exist in the current area, and uses this as the particle's fitness at the designated position. Assuming a broadcast network exists between the drones, enabling group information sharing, the velocity and position of the particle corresponding to each drone are updated using a particle swarm optimization algorithm, as shown in the following formula: in, ε Energy factor; It is inertial weight; This represents the best position that each particle has found so far. This represents the optimal position found so far in the entire particle swarm. and These are respectively referred to as the individual acceleration factor and the social acceleration factor; Indicates the first During the nth iteration The velocity of each particle; Indicates the first During the nth iteration The location of each particle; It is the actual execution speed of the drone under speed constraints; Among them, energy factor The following decay law applies: in, It is the energy decay rate, used to simulate how quickly a drone's energy decays during flight.

2. The multi-UAV cooperative target search method based on particle swarm optimization as described in claim 1, characterized in that, After calculating the target position of each drone from the current time to the next time using the particle swarm optimization algorithm, and determining the guidance path from the current position to the target position in the next time, a drone control command to push away the obstacle is generated at all locations on the guidance path where there are obstacles. When identifying obstacles on the guidance path, other drones within a preset range around the currently controlled drone, and with a higher priority than the currently controlled drone, are considered obstacles.

3. The multi-UAV cooperative target search method based on particle swarm optimization as described in claim 1, characterized in that, In the particle swarm optimization algorithm, inertia weight parameters, individual acceleration factor parameters, and social acceleration factor parameters are added, and the weights of each parameter are dynamically adjusted at each stage of control.

4. The multi-UAV cooperative target search method based on particle swarm optimization as described in claim 3, characterized in that, At the beginning of the search task, the inertia weight is at its maximum; as the search task is executed, the inertia weight decreases and the social acceleration factor increases; after the social acceleration factor reaches a preset value, the individual acceleration factor gradually increases.

5. The multi-UAV cooperative target search method based on particle swarm optimization as described in claim 1, characterized in that, The energy factor at the next moment is equal to the product of the energy factor at the current moment and the energy decay rate, where the energy decay rate simulates the rate of energy decay during the flight of the UAV.

6. The multi-UAV cooperative target search method based on particle swarm optimization as described in claim 1, characterized in that, Each drone is set as a single particle in the particle swarm optimization algorithm; the actual position of the drone is set as the particle position.

7. The multi-UAV cooperative target search method based on particle swarm optimization as described in claim 1, characterized in that, The control process from the current moment to the next moment is divided into a search phase and an optimization phase. During the search phase, through The algorithm obtains a guidance path that does not consider obstacles; During the optimization phase, the environment around the drone is detected, the trajectory existing in the obstacles is compared with the collision-free guidance path generated in the search phase, and gradient information is used to pull the trajectory in the obstacles out of the obstacles.

8. A multi-UAV cooperative target search system based on particle swarm optimization algorithm, characterized in that, include: The data acquisition module is configured to: acquire the location information of the target to be searched; The search control module is configured to control multiple drones to move towards the location of the target to be searched, based on a preset particle swarm algorithm. Among them, an energy factor that decays over time is introduced into the particle swarm algorithm to simulate the law of the decline in search capability over time, and a maximum speed constraint is introduced. After calculating the magnitude and direction of the speed at the next moment, the speed of the UAV in the calculated direction is controlled to not exceed the maximum speed constraint. The drone completes obstacle avoidance and reaches the previously designated particle position. It uses its onboard sensors to detect the environment, determining if any targets exist in the current area, and uses this as the particle's fitness at the designated position. Assuming a broadcast network exists between the drones, enabling group information sharing, the velocity and position of the particle corresponding to each drone are updated using a particle swarm optimization algorithm, as shown in the following formula: in, ε Energy factor; It is inertial weight; This represents the best position that each particle has found so far. This represents the optimal position found so far in the entire particle swarm. and These are respectively referred to as the individual acceleration factor and the social acceleration factor; Indicates the first During the nth iteration The velocity of each particle; Indicates the first During the nth iteration The location of each particle; It is the actual execution speed of the drone under speed constraints; Among them, energy factor The following decay law applies: in, It is the energy decay rate, used to simulate how quickly a drone's energy decays during flight.

9. A computer-readable storage medium having a computer program stored thereon, characterized in that, When executed by the processor, the program implements the steps of the multi-UAV cooperative target search method based on the particle swarm algorithm as described in any one of claims 1-7.

10. An electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that, When the processor executes the program, it implements the steps of the multi-UAV cooperative target search method based on particle swarm optimization as described in any one of claims 1-7.

Images