Unmanned aerial vehicle cooperative path planning method and system based on dynamic priority collision avoidance strategy

Abstract

The invention discloses an unmanned aerial vehicle cooperative path planning method and system based on a dynamic priority collision avoidance strategy, and relates to the technical field of unmanned aerial vehicle path planning. The method comprises: acquiring a map model, and initializing a starting point and an ending point of an unmanned plane; when unmanned aerial vehicle path planning is carried out, a global static optimal smooth path is planned from the map model; designing a multi-factor dynamic priority collision avoidance strategy to obtain an influence result of each factor; when it is detected that collision exists in the cluster, the priority of the unmanned aerial vehicles is determined according to the influence result of each factor, the high-priority unmanned aerial vehicles normally plan paths, and the low-priority unmanned aerial vehicles avoid the high-priority unmanned aerial vehicles. According to the multi-factor dynamic priority collision avoidance strategy, when each unmanned aerial vehicle executes path planning, possible resource waste in a traditional method is avoided. Through effective priority distribution, unnecessary calculation and redundant path adjustment are avoided, so that the efficiency of the planning process is improved.

Description

Technical Field

[0001] The present invention relates to the technical field of unmanned aerial vehicle (UAV) path planning, and particularly to a UAV cooperative path planning method and system based on a dynamic priority collision avoidance strategy. Background Art

[0002] With the rapid progress of UAV technology, UAVs have shown extensive application potential in many fields such as logistics distribution, search and rescue, agricultural monitoring, environmental monitoring, and disaster assessment. However, a single UAV has certain limitations in terms of endurance, payload capacity, task complexity, and large-scale reconnaissance and search. Therefore, UAV swarms have gradually become an important mode for performing complex tasks. UAV swarms are usually required to have all UAVs reach the target location simultaneously in the shortest time, and at the same time, the design of the flight route must consider avoiding dangerous areas, meeting various constraints (such as energy consumption, communication limitations, flight altitude limitations, etc.) and correctly handling the relationships between UAVs (such as avoiding collisions, maintaining formations, collaborative decision-making, etc.). Therefore, cooperative path planning technology directly determines whether UAV swarms can complete tasks smoothly, safely, and efficiently.

[0003] Traditional UAV cooperative collision avoidance strategies are often inefficient. For example, in the patent documents with publication numbers CN119440058A and CN119376407A, they mainly rely on simple rules or static path planning methods, lack the ability of dynamic adjustment and real-time response, and do not consider how to complete the path planning task as soon as possible under the premise of safety. Therefore, modern UAV collision avoidance systems must possess the capabilities of efficient collision avoidance and environmental perception to achieve effective cooperative path planning. Summary of the Invention

[0004] The object of the present invention is to propose a UAV cooperative path planning method and system based on a dynamic priority collision avoidance strategy, which can effectively address the collision avoidance problem within the swarm and provide feedback and correction to the planning results, thereby improving the safety and execution efficiency of path planning.

[0005] According to the first aspect of the embodiments of the present disclosure, there is provided a UAV cooperative path planning method based on a dynamic priority collision avoidance strategy, including the following steps:

[0006] Obtain a map model, and initialize the starting point and ending point of the UAV;

[0007] When performing UAV path planning, plan a globally static optimal smooth path from the map model;

[0008] Design a multi-factor dynamic priority collision avoidance strategy to obtain the influence results of each factor; when a collision within the swarm is detected, determine the UAV priorities according to the influence results of each factor, and the UAVs with high priorities plan paths normally, while the UAVs with low priorities avoid the UAVs with high priorities.

[0009] According to a second aspect of an embodiment of the present disclosure, a UAV collaborative path planning system based on a dynamic priority collision avoidance strategy is provided, comprising:

[0010] Initialization module, obtains the map model, and initializes the starting point and end point of the drone;

[0011] The planning module plans the global static optimal smooth path from the map model when planning the UAV path;

[0012] Collision avoidance module, designs a multi-factor dynamic priority collision avoidance strategy to obtain the impact results of each factor; when a collision is detected within the cluster, the priority of the drone is determined according to the impact results of each factor, the high-priority drone plans the path normally, and the low-priority drone avoids the high-priority drone.

[0013] According to a third aspect of an embodiment of the present disclosure, there is provided an electronic device, comprising a memory, a processor, and a computer program stored and running on the memory, wherein when the processor executes the program, the method for collaborative path planning of unmanned aerial vehicles based on a dynamic priority collision avoidance strategy is implemented.

[0014] According to a fourth aspect of an embodiment of the present disclosure, a computer-readable storage medium is provided, on which a computer program is stored. When the program is executed by a processor, the method for collaborative path planning of unmanned aerial vehicles based on a dynamic priority collision avoidance strategy is implemented.

[0015] Compared with the prior art, the above technical solution adopted by the present invention has the following advantages:

[0016] 1. The multi-factor dynamic priority collision avoidance strategy can automatically adjust the priority of each drone to ensure more efficient avoidance of potential conflicts in complex environments. This adaptability enables planning to respond to emergencies in a timely manner, avoids unchanging planning strategies, and enhances the system's resilience.

[0017] 2. The multi-factor dynamic priority collision avoidance strategy enables each UAV to avoid the waste of resources that may occur in traditional methods when performing path planning. Through effective priority allocation, unnecessary calculations and redundant path adjustments are avoided, thereby improving the efficiency of the planning process.

[0018] 3. The multi-factor dynamic priority collision avoidance strategy can respond to changes in the environment in real time, such as the appearance of new obstacles or the path adjustment of other drones. This adaptability and real-time performance can ensure that drones can still complete tasks efficiently and safely in dynamic environments without being constrained by fixed priority strategies.

[0019] 4. In multi-UAV collaborative missions, by dynamically adjusting the priorities of each UAV, it is possible to ensure a timely response to the requirements of collaboration during flight, avoid unnecessary conflicts, and maximize the overall efficiency. Especially under limited resources, the dynamic allocation of priorities can ensure the most efficient execution of tasks for each UAV.

[0020] 5. The multi-factor dynamic priority collision avoidance strategy can adjust the trajectory planning of each UAV according to real-time priorities, avoid overcrowded flight routes, reduce path intersections and interference, and ultimately improve the spatial efficiency and safety of path planning. Brief Description of the Drawings

[0021] The accompanying drawings forming a part of this application are used to provide a further understanding of this application. The schematic embodiments of this application and their descriptions are used to explain this application and do not constitute an improper limitation of this application.

[0022] Figure 1 is a schematic diagram of the process of the UAV collaborative path planning method based on the dynamic priority collision avoidance strategy;

[0023] Figure 2 is a schematic diagram of the two-dimensional matrix generation map model shown in the embodiment;

[0024] Figure 3 is the global static path map under the Lazy Theta algorithm shown in the embodiment;

[0025] Figure 4 is a schematic diagram of the process of the UAV path planning link shown in the embodiment;

[0026] Figure 5 is a schematic diagram of the UAV oscillation shown in the embodiment;

[0027] Figure 6 is a schematic diagram of the multi-factor dynamic priority collision avoidance strategy process shown in the embodiment;

[0028] Figure 7 is the global path planning roadmap of the MS-Lazy Theta algorithm shown in the embodiment;

[0029] Figure 8 is the minimum inter-aircraft distance broken line graph shown in the embodiment;

[0030] Figure 9 is the minimum distance from aircraft to obstacle broken line graph shown in the embodiment;

[0031] Figure 10 is the end point fuel dot graph shown in the embodiment;

[0032] Figure 11It is the global path planning roadmap of the Lazy Theta algorithm shown in the embodiment;

[0033] Figure 12 It is the global path planning roadmap of the S-Lazy Theta algorithm shown in the embodiment;

[0034] Figure 13 It is the inter-machine constraint broken line graph under the condition of the predicted collision detection method shown in the embodiment. Detailed implementation manners

[0035] The present disclosure will be further described below in conjunction with the accompanying drawings and embodiments.

[0036] It should be noted that the following detailed descriptions are all illustrative and are intended to provide further descriptions of the present application. Unless otherwise specified, all technical and scientific terms used in the present invention have the same meanings as those commonly understood by those of ordinary skill in the technical field to which the present application belongs.

[0037] It should be noted that the terms used herein are only for describing specific implementation manners and are not intended to limit the exemplary implementation manners according to the present application. As used herein, unless the context clearly indicates otherwise, the singular forms are also intended to include the plural forms. In addition, it should also be understood that when the terms "include" and / or "comprise" are used in this specification, they indicate the presence of features, steps, operations, devices, components, and / or combinations thereof.

[0038] It should be noted that the flowcharts and block diagrams in the accompanying drawings illustrate the possible architectures, functions, and operations of the methods and systems according to various embodiments of the present disclosure. It should be noted that each block in the flowchart or block diagram may represent a module, a program segment, or a part of code, and the module, the program segment, or the part of code may include one or more executable instructions for implementing the logical functions defined in each embodiment. It should also be noted that in some alternative implementations, the functions marked in the blocks may occur in a different order than marked in the accompanying drawings. For example, two consecutive blocks shown may actually be executed substantially in parallel, or they may sometimes be executed in the reverse order, depending on the functions involved. Similarly, it should also be noted that each block in the flowchart and / or block diagram, and the combinations of blocks in the flowchart and / or block diagram, may be implemented by a dedicated hardware-based system for performing the specified functions or operations, or may be implemented by a combination of dedicated hardware and computer instructions.

[0039] Embodiment 1:

[0040] As Figure 1 shown, this embodiment provides a UAV cooperative path planning method based on a dynamic priority collision avoidance strategy, including the following steps:

[0041] S1. Obtain the map model and initialize the starting and ending points of the UAV.

[0042] As Figure 2 shown, the map model of the present invention is generated by a two-dimensional matrix, and the black blocks in the map model are obstacles.

[0043] S2. When planning the UAV path, plan the global static optimal smooth path from the map model.

[0044] As Figure 3 shown, the colored line is the global static path. The Lazy Theta algorithm can effectively remove redundant paths to obtain the initial UAV path. After obtaining the initial path, the global static smooth path is obtained from the map model by the cubic spline interpolation method. To address the drawback of mutations in the cubic spline interpolation method, the moving window method is added for improvement. Finally, the global static optimal smooth path is planned, and its nodes are saved as the input for the multi-factor dynamic priority collision avoidance strategy for reference in intra-cluster collision avoidance.

[0045] When planning the UAV path, first introduce time variables as references, namely the time stay variable the time start variable and the time end variable where the time stay variable represents the time required for the UAV numbered i to complete the j-th grid; the time start variable represents the start time of the j-th grid where the UAV numbered i arrives; the time end variable represents the end time of the j-th grid where the UAV numbered i completes; the above variables are expressed as follows:

[0046]

[0047] In the formula: x i (j), y i (j) represent the coordinates of the starting point of the i-th UAV at the current grid; x i (j + 1), y i (j + 1) represent the coordinates of the starting point of the i-th UAV at the next grid; V represents the running speed of the UAV.

[0048] Based on the above time variables, the grid where the UAV is located and the usage time can be determined while planning the path. Regarding the UAV as a dynamic obstacle, the definition of grid dynamic occupancy is realized:

[0049]

[0050] Where: P(x, y) represents the grid occupancy function, which is used to characterize the occupancy of the grid map.

[0051] To improve the accuracy of path planning, the cubic spline interpolation method is introduced to smooth the path formed by the Lazy Theta algorithm. The cubic spline interpolation method is a method of constructing a smooth curve by using cubic polynomials to interpolate between a series of known data points. The method is as follows:

[0052] Suppose there is a series of known path points The goal is to find a function S(x) that is a cubic polynomial on each interval [x m , x m+1 and satisfies the following conditions:

[0053] (1) Interpolation condition: S(x m ) = y m , that is, the function is equal to the corresponding dependent variable value at the known data points.

[0054] (2) Smoothness condition: The function S(x) is smooth on the entire interval [x 0 , x n , that is, its first derivative and second derivative are continuous at the connection points of adjacent intervals.

[0055] To find the cubic polynomial that satisfies these conditions, a cubic polynomial S m , x m+1 (x) is defined for each interval [x m and can be expressed as follows:

[0056] S m (x) = a m + b m (x - x m ) + c m (x - x m ) 2 + d m (x - x m ) 3 (5)

[0057] The values of a m , b m , c m , d m are solved by establishing the following system of equations:

[0058] (1) From the interpolation condition, the interpolation equation can be obtained as follows:

[0059]

[0060] (2) From the smoothness condition, it can be obtained that S m (x) and Sm+1 (x) The first derivative and the second derivative at x = x i+1 are equal, and the smoothing equation can be obtained as follows:

[0061]

[0062] (3) The slope of the curve at the endpoints is zero, and the boundary equation can be obtained as follows:

[0063]

[0064] This method can flexibly adjust the path shape by controlling the positions and derivatives of the nodes, thereby improving the efficiency and feasibility of the planned path. Although the traditional cubic spline interpolation method can ensure global smoothness, due to its global constraints, there may be unnecessary fluctuations or mutations in some regions. The moving window method limits the control points on which the interpolation calculation depends by defining a window, so that each interpolation considers the current and nearby nodes, which can erase potential mutations and eliminate short-term fluctuations. The key concept of the moving average method is the "moving window". Given the path points after interpolation Select a window size w and calculate the average value of the path points within the window as the smoothed path points The larger the window, the more obvious the smoothing effect, but it will lead to an increase in the amount of calculation, especially when the number of path points is large. If the window is too small, it may not be able to eliminate the short-term fluctuations in the path. The acquisition method is as follows:

[0065]

[0066]

[0067] In the formula: k represents the index of the points within the window, represents the coordinate value of the path points within the window, and m represents the interval after interpolation.

[0068] The moving window method reduces unnecessary fluctuations or mutations through local interpolation, thereby improving the path smoothness. While retaining the overall shape of the path, this method significantly reduces the impact of short-term fluctuations.

[0069] As Figure 4 shown, by completing the process of the UAV path planning link, a globally static optimal smooth path can be obtained, which serves as a foundation for path planning and subsequent multi-factor dynamic priority collision avoidance strategies.

[0070] S3. Design a multi-factor dynamic priority collision avoidance strategy to obtain the influence results of each factor; when a collision is detected within the cluster, determine the UAV priorities according to the influence results of each factor. The UAVs with high priorities plan their paths normally, and the UAVs with low priorities avoid the UAVs with high priorities.

[0071] When designing a multi-factor dynamic priority collision avoidance strategy, the detection method of UAV collisions is directly related to the success or failure of UAV cooperative path planning. Based on the prediction collision detection method with a time window, the setting of buffering is added. According to the prediction collision detection with a time window, the UAVs that do not meet the cooperative collision avoidance constraints are initially obtained. Then, the positions of UAV A and UAV B at the current moment and the previous moment are extracted, and are set as (x A (t), y A (t)), (x A (t - 1), y A (t - 1)); (x B (t), y B (t)), (x B (t - 1), y B (t - 1)); The direction vectors representing UAV A and UAV B can be obtained:

[0072]

[0073] Then, the path direction angle is obtained:

[0074]

[0075] Based on the path direction angle, the running direction information of the UAV is obtained, and then the formula for the buffer collision detection method based on the time window is obtained as follows:

[0076]

[0077] In the formula: D buf represents the buffer distance, and σ represents the path direction angle.

[0078] The buffer collision method based on the time window can effectively avoid potential collisions by dynamically adjusting the buffer distance, thereby improving the task safety and efficiency of UAV cooperative path planning.

[0079] When a collision situation is detected within the cluster, the multi-factor dynamic priority collision avoidance strategy combines the following indicators:

[0080] (1) The remaining fuel quantity f l , and the remaining fuel of the UAV should be at least sufficient to support it to reach the task end point;

[0081] (2) The total distance S, which is the cumulative distance of the UAV flight path;

[0082] (3) The remaining distance S c , that is, the remaining distance of the UAV flight path;

[0083] When the minimum remaining fuel quantity fs Less than the fuel constraint f min At this time, the UAV obtains a higher priority to avoid mission failure caused by not meeting the fuel constraint after collision avoidance; the method for obtaining the minimum remaining fuel quantity is as follows:

[0084]

[0085] In the formula: η represents the fuel consumption in time sequence, which refers to the constant fuel consumption of the UAV within 1 time unit in the non-low-cost hovering state; V represents the flight speed of the UAV.

[0086] When the minimum remaining fuel quantity f s is greater than the fuel constraint f min , considering the total distance S and the remaining distance ratio S c , a priority model is constructed as shown in the following formula; the larger the comprehensive index P i of the UAV, the higher its priority;

[0087] P i =α*S + γ*S c , i = 1, 2..., n (15)

[0088] In the formula: α and γ are the influence weights of the corresponding indicators.

[0089] After the priority is determined, as Figure 5 shown, in view of the fact that if both conflicting parties adopt collision avoidance methods to eliminate collisions, it may lead to continuous collision avoidance of the UAV and eventually oscillation; therefore, a single UAV collision avoidance form is adopted. The present invention proposes a hybrid avoidance method, which combines the path replanning method and the in-situ waiting method to ensure that the high-priority UAV can plan the path according to the original plan, while the low-priority UAV selects the optimal collision avoidance strategy according to the specific situation.

[0090] For low-priority UAVs, the hybrid avoidance method is selected through the following logic:

[0091] (1) Obtain the collision vector angle:

[0092] Let the current position of the i-th UAV be The target end point is The direction vector representing the highest path planning efficiency is obtained as The direction point after path replanning behavior is set as The path vector representing path replanning is obtained as The vector angle θ is:

[0093]

[0094] (2) Establish an avoidance strategy as follows:

[0095]

[0096] In the formula: λ represents the conflict tolerance value, which can be given artificially and determines the proportion of the two collision avoidance methods.

[0097] The hybrid avoidance method combines direction judgment and effectively improves the cooperative flight ability of the UAV swarm in complex environments. This strategy enables the low-priority UAVs to approach the target point more efficiently by dynamically selecting the collision avoidance method.

[0098] Such as Figure 6 shown, the process of completing the multi-factor dynamic priority collision avoidance strategy can obtain a globally optimal smooth path that meets the inter-aircraft constraints.

[0099] To verify the feasibility and advantages of the present invention, simulation verification and comparative experiments on UAV cooperative path planning were carried out. In the experiment, the size of the UAV flight area was 60m×40m, the grid scale was set to 120×80, that is, the grid division accuracy r was 0.5m, and the flight speed V of each UAV was the same and constant, which was 0.4m / s. The conflict tolerance value λ = 90°, the minimum safe fuel f min was set to 10%, the UAV wheelbase l was 0.3m, the flight safety distance s was set to 0.2m, the buffer distance D buf was set to 0.1m, the fuel consumption per time sequence η was set to 0.75%, the window size w was 10, and the index weights of the multi-factor dynamic priority calculation link were α = 0.283 and γ = 0.717.

[0100] To simulate the fuel consumption during the operation of the UAV, the remaining fuel f at time t l (t) is expressed as:

[0101]

[0102] In the formula: f F is the maximum fuel amount, which is set to 100%.

[0103] To verify the effectiveness and practicality of the path planning method proposed by the present invention, this experiment conducted cooperative path planning tests on UAV swarms of different scales. Figure 7 Shows the results of path planning for UAV swarms of 5, 10, and 15 scales when the task start point and task end point are known. This experiment considered various flight scenarios, including the presence of obstacles and the mutual collision avoidance behavior between UAVs, aiming to comprehensively evaluate the performance of this path planning method in UAV swarms of different scales.

[0104] By Figure 7It can be seen that by comparing the path planning results for different cluster sizes, it is possible to clearly observe that as the cluster size increases, the challenges faced by path planning also increase. At a cluster size of 5, path planning is relatively simple, with relatively few collisions and path redundancies. Therefore, the path planning time is short, and the overall path is relatively concise. However, at cluster sizes of 10 and 15, as the number of UAVs increases, the collision risk and path redundancy in the path planning process also increase significantly, and the complexity and computational workload of path planning increase accordingly. These changes indicate that as the cluster size increases, the robustness and computational efficiency of the path planning method become particularly important, and the path planning method proposed in the present invention demonstrates good adaptability and efficiency in this regard.

[0105] To further verify the performance of the path planning method in a complex flight environment, the minimum distance between UAVs at each moment, the minimum distance between UAVs and obstacles, and the remaining fuel amount when the UAVs reach the end point are calculated. These indicators evaluate from multiple perspectives whether the results of path planning meet the constraint conditions.

[0106] Figure 8 It is a line graph showing the change in the minimum distance between UAVs for the above three cluster sizes. In Figure 8 , the red line represents the safe flight distance s between UAVs, and the blue line represents the broken line of the minimum distance between UAVs at the same moment. This line graph can effectively reflect the collision avoidance situation between UAVs in path planning. The blue line below the red line means that the inter-aircraft constraint conditions are not met, and the blue line at and above the red line means that the inter-aircraft constraint conditions are met. In the present invention, as the cluster size increases, the change in the minimum distance between UAVs is relatively obvious, but the blue line is always above the red line, indicating that the path planning method keeps the distance between UAVs within a reasonable range and fully meets the cooperative collision avoidance constraints throughout the process.

[0107] Figure 9 It is a line graph showing the change in the minimum distance between UAVs and obstacles for the above three cluster sizes. In Figure 9 , the red line represents the obstacle safety distance D ob between UAVs and obstacles, and the blue line represents the broken line of the minimum distance between UAVs and obstacles at the same moment. Figure 9 It can clearly reveal the relative position relationship between UAVs and static or dynamic obstacles during flight, and reflect whether the path planning can effectively avoid obstacles. During the experiment, as the cluster size increases, the requirements for avoiding obstacles become more complex. By comparing the changes in the minimum distance for different cluster sizes, it can be seen that the path planning method proposed in the present invention can maintain a relatively stable minimum distance when facing obstacles, ensuring that UAVs avoid collision accidents during flight.

[0108] Figure 10It is a dot plot of the remaining fuel of the drones when they reach the end point under the above three cluster scales. In Figure 10 , the red line represents the fuel constraint f of the drone min . The black dots represent the remaining fuel of different drones when they reach the end point. Under the three cluster scales of the drones, the remaining fuel of all drones when they reach the end point is above the red line, which means that all drones meet the fuel constraint. Figure 10 It reflects the satisfaction of the fuel constraint of the path planning method. By comparing the remaining fuel under the cluster scales of 5, 10, and 15 drones, it can be found that as the cluster scale increases, the fuel consumption points are always above the red line of the fuel constraint, indicating that the path planning method proposed by the present invention can complete the task before the fuel runs out.

[0109] It can be concluded that the path planning method proposed by the present invention shows high planning quality and calculation efficiency in drone clusters of different scales. The algorithm can not only ensure flight safety and avoid collisions, but also effectively optimize the path length. In addition, as the cluster scale increases, the algorithm shows strong adaptability and robustness in dealing with complex flight environments and multi-drone cooperative operations. Therefore, the path planning method proposed by the present invention can provide an efficient and safe solution in practical applications and is applicable to various complex cooperative flight tasks.

[0110] The path planning method proposed by the present invention includes four key improved parts, namely the MS-Lazy Theta algorithm, the buffer collision detection method based on the time window in the multi-factor dynamic priority collision avoidance strategy, the multi-factor dynamic priority calculation method, and the hybrid avoidance strategy. In order to verify the improvement effect of each part, the present invention conducted a series of comparative experiments and focused on evaluating the improvement of these improvements on the path planning performance. Specifically, the experiment is divided into four main parts: the MS-LazyTheta algorithm experiment, the buffer collision detection method experiment based on the time window, the multi-factor dynamic priority calculation method experiment, and the hybrid avoidance strategy experiment.

[0111] As Figures 11 to 12 shown, in order to evaluate the actual performance of the MS-Lazy Theta algorithm, the present invention conducted experiments in three different scenarios, corresponding to drones with cluster scales of 5, 10, and 15 respectively, and compared the path planning results under the standard LazyTheta algorithm and the Lazy Theta algorithm with cubic spline interpolation (S-Lazy Theta), aiming to test the performance of the MS-Lazy Theta algorithm in different environments.

[0112] Through the comparison of path diagrams, it is found that although the Lazy Theta algorithm generates a relatively concise path, there are still a large number of redundant parts in the path turning section, which may lead to a decrease in the efficiency of path planning; for the S-Lazy Theta algorithm, the cubic spline interpolation algorithm is added to smooth the path. Although the problem of trajectory characteristics is solved, due to the existence of mutations, the path planning results oscillate, and thus the efficiency may be lower than that of the Lazy Theta algorithm; the MS-LazyTheta algorithm generates a path without angular mutations by smoothing and mutating the moving window method, and can slightly reduce the total path length and planning time of the UAV operation. Through the visual comparison and analysis of path diagrams, the MS-Lazy Theta algorithm can better complete the UAV cooperative path planning task compared with other comparison algorithms.

[0113] Table 1 Path planning data information under three path planning methods

[0114]

[0115] Table 1 shows the specific data of the total path length and planning time of path planning obtained by using the Lazy Theta algorithm, S-LazyTheta algorithm, and MS-Lazy Theta algorithm respectively under different cluster scales of 5, 10, and 15. It can be seen from the data analysis in the table that the performances of the three algorithms are different under different cluster scales.

[0116] The Lazy Theta algorithm adds a path simplification function to reduce the complexity of the path by optimizing path nodes. The total path lengths generated under the three cluster scales are 140.01m, 288.56m, and 451.07m respectively, and the planning times are 101.55s, 92.85s, and 92.85s respectively. Although this algorithm can effectively generate a shorter path, the path it generates has obvious corners and mutations, which may not meet the actual needs for some application scenarios.

[0117] Compared with the Lazy Theta algorithm, the S-Lazy Theta algorithm incorporates a cubic spline interpolation algorithm to smooth the path. This improvement helps to eliminate sharp corners and mutations in the path, making the path smoother. Under different cluster scales, the total path length of the path generated by the S-Lazy Theta algorithm increased by 0.35m, 0.8m, and 0.98m respectively under the three cluster scales, and the planning time increased by 0.23s, 0.22s, and 0.22s accordingly. The reason for this phenomenon is that the cubic spline interpolation algorithm may cause mutations and oscillations during the path calculation of some collision avoidance behaviors, especially in complex scenarios with multiple collision avoidance behaviors, the mutation problem is more prominent. Therefore, although the S-Lazy Theta improves the smoothness of the path, the increase in its total path length and planning time, especially in cases with frequent collisions, may affect its overall efficiency and effectiveness.

[0118] Compared with the S-Lazy Theta algorithm, the MS-Lazy Theta algorithm further incorporates a moving window method to eliminate mutations in the path. Through this method, the mutation problem in the path calculation of the S-Lazy Theta algorithm is effectively solved, thereby optimizing the overall effect of path planning. Compared with the S-Lazy Theta algorithm, the MS-Lazy Theta significantly reduces both the total path length and the planning time. Specifically, under the three cluster scales, the total path length is shortened by 1.17m, 1.52m, and 1.79m respectively, and the planning time is shortened by 0.26s, 1.62s, and 1.62s respectively. The reason for this improvement is that the moving window method can smooth the fluctuations in the cubic spline interpolation results, thus effectively avoiding the path mutation problem. The analysis also found that as the cluster scale increases, the MS-Lazy Theta algorithm shows stronger advantages. Especially in the case of a large UAV cluster scale and severe mutation phenomena in path planning, the MS-Lazy Theta can better handle complex collision avoidance scenarios.

[0119] The MS-Lazy Theta algorithm demonstrates superior performance in this invention. By combining the cubic spline interpolation method and the moving window method, the MS-Lazy Theta can effectively solve the problems of sharp corners and mutations in traditional path planning methods, generating a smoother and more efficient path. Therefore, the MS-Lazy Theta algorithm can effectively improve the path planning quality of the overall UAV system and is a more ideal choice for complex flight tasks.

[0120] To comprehensively analyze the computational effect of the buffer collision detection method based on time window proposed by the present invention, the present invention compares it with the predictive collision detection method based on time window. Through this comparison, it aims to investigate the applicability and performance differences of the method of the present invention in different scenarios, especially for the collision detection and collision avoidance capabilities of multi-UAV clusters in complex flight environments.

[0121] Table 2 Path planning information under two detection methods

[0122] Table 2 shows that when the UAV scale is 5, 10, and 15 respectively, the predictive collision method and the buffer collision method are used

[0123]

[0124] to obtain the total path length, planning time consumption, and algorithm time of path planning.

[0125] It can be seen from the table that when the UAV scale is 5, the total path length of the buffer collision detection method is shortened by 0.55 m compared with the predictive collision detection method, the planning time consumption is shortened by 0.01 s, and in terms of the algorithm running time, it is shortened by 34%; when the UAV scale is 10 and 15, the total path length of the buffer collision detection method increases by 0.44 m compared with the collision detection method, the planning time consumption is the same, and in terms of the algorithm running time, it is shortened by 24.9% and 18.8%. The reason is that due to the setting of a farther collision avoidance distance in the buffer collision method, the overall path length increases slightly, while the flight path of the longest UAV is not affected by the collision detection method, so the planning time consumption is the same. The buffer collision detection uses a more excellent detection mechanism and can reduce the number of UAV collisions to a certain extent, so the algorithm time is greatly shortened.

[0126] The main purpose of the collision detection method is to increase the distance between aircraft, facilitate safe collision avoidance and reduce repeated collision avoidance behaviors. Therefore, the minimum distance curve between UAVs of the predictive detection method is introduced again for further analysis, as Figure 13 shown.

[0127] Through Figure 13 the visual analysis in it, it can be clearly seen that due to the preset safety distance being too close, some UAVs fail to meet the predetermined distance constraint during the collision avoidance process. This indicates that simply presetting the safety distance may lead to potential risks in the collision avoidance process and cannot effectively avoid the occurrence of close collisions. In contrast, the experimental results of using the predictive detection method show that the number of collisions between UAVs increases significantly. This result fully demonstrates that the predictive detection method has certain deficiencies in dealing with dynamic obstacles or complex environments and cannot effectively avoid collisions in practical applications. Under the collision detection method, the number of collisions of UAVs is significantly lower than that of the predictive detection method, verifying the feasibility and advantages of this method in different scenarios.

[0128] Overall, compared with the prediction detection method, the collision detection method can more accurately monitor the state of the UAV in real time and take collision avoidance measures in a timely manner. Especially in a relatively complex flight environment, it shows more stable and reliable collision avoidance performance.

[0129] By comparing and analyzing the actual effects of the multi-factor dynamic priority calculation method in UAV path planning, especially its applicability under different cluster scales. The experiments were carried out under three cluster scales of 5, 10, and 15 respectively. When the UAV cluster used the random priority calculation method, the total path length and planning time of each sample were recorded, so as to provide data support for evaluating the superiority of this method. It should be noted that when the cluster scale is 5, due to the small number of UAVs and the low probability of collision, in the process of multiple samplings, the situation of sample repetition occurred relatively frequently. To avoid this repetitive impact on the experimental results, the sampling times of the 5-UAV cluster were limited to 5 times. In contrast, when the cluster scales are 10 and 15, the samples are sufficient and the collision frequency is high, and the sampling times can meet the data diversity, so 15 samplings were carried out respectively.

[0130] By comparing these data, it aims to explore the performance differences between the random priority calculation method and the multi-factor dynamic priority calculation method under different cluster scales, and further verify the advantages of the multi-factor dynamic priority calculation method in larger-scale clusters. The total path length, planning time of each sample, and the average data under each cluster scale provide the basic data support required for in-depth analysis.

[0131] Table 3 Sample Data Table

[0132]

[0133]

[0134] From the data in Table 3, when the cluster scale is 5, the path lengths and planning times of each sample are relatively close. The path length of Sample 3 is the shortest, which is 139.15m, and the total path length of Sample 5 is the longest, which is 139.49m. And the planning time of each sample is the same as that under the multi-factor dynamic priority calculation method. Due to the low collision risk in the flight path, the path planning results of the random priority calculation method have little difference. It can be calculated that the average value of the total path lengths of the 5 samples is 139.30m, which is slightly higher than the result under the multi-factor dynamic priority calculation method.

[0135] The planned time for each sample is the same as that under the multi-factor dynamic priority calculation method. This phenomenon can be attributed to the fact that the time-consuming of path planning mainly depends on the flight time of the UAV with the longest remaining path and total path. When there is no collision and no collision avoidance behavior occurs during the flight of this UAV, the time-consuming of path planning for each sample will tend to be the same in this case. This makes all samples under this cluster scale perform the same in terms of planned time-consuming. At this time, the total path length can also be related to the flight efficiency of the UAV. By comparing and analyzing the total path lengths of the samples, the efficiency of different collision avoidance algorithms can be better reflected.

[0136] When the cluster scale increases to 10, the complexity of path planning increases significantly, and the differences between samples also begin to become more obvious. The total path length of sample 4 is the shortest, which is 287.64m, and the total path length of sample 5 is the longest, which is 288.83m; the planned time-consuming of sample 4 is the shortest, which is 90.89s, and the planned time-consuming of sample 5 is the longest, which is 92.60s; it can be calculated that the average value of the total path lengths of 15 samples is 288.10m, and the average value of the planned time-consuming is 91.58s, which is higher than the result under the multi-factor dynamic priority calculation method.

[0137] When the cluster scale is 15, with the increase in the number of UAVs, the conflicts and collision avoidance requirements between paths increase significantly, and the limitations of the random priority calculation method gradually become apparent. At this time, the efficiency of path planning is significantly affected by collision avoidance behavior, the planned time-consuming is generally long, and the differences in path lengths are large. The total path length of sample 1 is the shortest, which is 449.38m, and the total path length of sample 12 is the longest, which is 452.43m; the planned time-consuming of sample 4 is the shortest, which is 91.20s, and the planned time-consuming of sample 5 is the longest, which is 94.06s; it can be calculated that the average value of the total path lengths of 15 samples is 450.85m, and the average value of the planned time-consuming is 92.14s, which is higher than the result under the multi-factor dynamic priority calculation method. This shows that under a large cluster scale, although the random priority method can solve the path planning problem through simple priority settings, its lack of priority rationality and optimization ability leads to longer path lengths and lower calculation efficiency during the planning process.

[0138] In some samples, it is occasionally observed that the performance of the random priority calculation method is better than that of the multi-factor dynamic priority calculation method. The reason behind this can be attributed to the chain effect of UAV collision avoidance behavior. Specifically, the collision avoidance decision of each UAV will affect subsequent path planning. Especially during multiple collision avoidance processes, certain previous different collision avoidance operations may cause subsequent UAVs to no longer avoid other UAVs. The occurrence of this situation makes the results of individual path planning perform better. Therefore, when evaluating the overall efficiency of path planning, relying solely on the performance of individual samples may not be sufficient to reflect the true performance of the system. Considering the differences of each sample under different flight conditions, using the average value of the data to comprehensively evaluate the overall efficiency of path planning under each cluster size can provide a more scientific evaluation result.

[0139] Through the comparative analysis of experiments, it can be clearly seen that the multi-factor dynamic priority calculation method demonstrates its superiority. Especially in large-scale clusters (such as 10 UAVs, 15 UAVs), the multi-factor dynamic priority calculation method can effectively reduce the path length, optimize the planning time-consuming, and ensure the efficient completion of tasks. Therefore, the multi-factor dynamic priority calculation method has stronger applicability and higher efficiency. Especially when facing large-scale UAV clusters, it shows higher stability and better path planning results.

[0140] By comparing and analyzing the replanning method, the in-situ waiting method, and the hybrid avoidance strategy, the actual application effect of the hybrid avoidance strategy is verified.

[0141] Table 4 Path planning information under three collision avoidance methods

[0142]

[0143] As can be seen from Table 4, in the case of a 5-UAV cluster size, the type of collision avoidance method has an impact on the total path length of path planning. The total path length of the in-situ waiting method is the shortest, which is 138.67m, while the total path length of the replanning method is the longest, which is 139.35m. This indicates that the in-situ waiting method can reduce the unnecessary path length through a simple waiting strategy in a smaller-scale cluster.

[0144] When the cluster size is 10, the hybrid collision avoidance strategy outperforms the replanning method in both the total path length and planning time, with obvious advantages. Among them, the total path length of the hybrid collision avoidance strategy is shortened by 0.74 m compared with the replanning method, and the planning time is shortened by 0.45 s. This shows that when dealing with collision risks, the hybrid collision avoidance strategy can effectively balance path optimization and calculation time by flexibly combining the strategies of waiting in place and replanning, thus improving the planning efficiency. Although the total path length under the hybrid collision avoidance strategy is 0.87 m longer than that of the waiting-in-place method, its planning time is better than that of the waiting-in-place method, reducing by 1.08 s. This shows that the advantage of the waiting-in-place method is only reflected in the path length, and the planning time is relatively long. Especially when facing multiple collision avoidance operations, the waiting process may lead to an increase in flight time. In contrast, although the replanning method has a certain redundancy in the path length, it can respond to sudden collision situations by real-time replanning the path to ensure the safe execution of the task. The hybrid collision avoidance strategy can ensure the effectiveness of collision avoidance while reducing the planning time, with strong adaptability.

[0145] When the cluster size is 15, due to the more complex collision situations among UAVs, especially the head-on collision types within the cluster, the waiting-in-place method fails to effectively avoid these collisions, resulting in planning failure. The waiting-in-place method cannot adjust the path in real time in this case and can only rely on simple waiting operations, which makes its adaptability poor when facing large-scale clusters. Especially when multiple collisions occur, it is easy to cause task interruption or failure. In contrast, when facing a 15-UAV cluster, the hybrid collision avoidance strategy can more flexibly handle the complex flight environment. Compared with the replanning method, it shortens the path length by 1 m and reduces the planning time by 0.82 s. The hybrid collision avoidance strategy can flexibly choose a suitable collision avoidance method while ensuring the collision avoidance effect, thus avoiding the deficiencies of the waiting-in-place method.

[0146] Generally speaking, in large-scale UAV cluster cooperation tasks, although the waiting-in-place method can provide a shorter path length, it has deficiencies in planning time and dealing with complex collision situations. Especially when encountering complex situations such as head-on collisions, the waiting-in-place method often cannot respond in time, resulting in planning failure and unable to ensure the smooth completion of the task. The replanning method can handle relatively complex collision situations, but when collisions occur frequently, it will lead to a long path, increase the calculation amount, and slow down the planning speed. Therefore, although the replanning method is relatively reliable in dealing with large-scale cluster tasks, its path efficiency is low and the planning time is long. In contrast, when facing large-scale cluster tasks, the universality of the hybrid collision avoidance strategy makes it a more advantageous choice, which can provide higher adaptability and planning efficiency in a changing flight environment.

[0147] Example 2:

[0148] This embodiment provides a UAV cooperative path planning system based on a dynamic priority collision avoidance strategy, including:

[0149] An initialization module, which obtains a map model and initializes the starting and ending points of the UAVs;

[0150] A planning module, which plans a globally static optimal smooth path from the map model when performing UAV path planning;

[0151] A collision avoidance module, which designs a multi-factor dynamic priority collision avoidance strategy to obtain the influence results of each factor; when a collision is detected within the cluster, it determines the UAV priorities according to the influence results of each factor. The high-priority UAVs plan their paths normally, and the low-priority UAVs avoid the high-priority UAVs.

[0152] Embodiment 3:

[0153] An electronic device, including a memory, a processor, and a computer program running on the memory. When the processor executes the program, it implements the above-mentioned UAV cooperative path planning method based on a dynamic priority collision avoidance strategy, including:

[0154] Obtaining a map model and initializing the starting and ending points of the UAVs;

[0155] When performing UAV path planning, planning a globally static optimal smooth path from the map model;

[0156] Designing a multi-factor dynamic priority collision avoidance strategy to obtain the influence results of each factor; when a collision is detected within the cluster, determining the UAV priorities according to the influence results of each factor. The high-priority UAVs plan their paths normally, and the low-priority UAVs avoid the high-priority UAVs.

[0157] Embodiment 4:

[0158] A computer-readable storage medium, on which a computer program is stored. When the program is executed by a processor, it implements the above-mentioned UAV cooperative path planning method based on a dynamic priority collision avoidance strategy, including:

[0159] Obtaining a map model and initializing the starting and ending points of the UAVs;

[0160] When performing UAV path planning, planning a globally static optimal smooth path from the map model;

[0161] Designing a multi-factor dynamic priority collision avoidance strategy to obtain the influence results of each factor; when a collision is detected within the cluster, determining the UAV priorities according to the influence results of each factor. The high-priority UAVs plan their paths normally, and the low-priority UAVs avoid the high-priority UAVs.

[0162] Those skilled in the art should understand that each module or step of the above disclosure can be implemented by a general-purpose computer device. Optionally, they can be implemented by program codes executable by a computing device, so that they can be stored in a storage device and executed by the computing device, or they can be separately fabricated into individual integrated circuit modules, or multiple modules or steps among them can be fabricated into a single integrated circuit module for implementation. The present disclosure is not limited to any specific combination of hardware and software.

[0163] The above are only the preferred embodiments of the present application and are not used to limit the present application. For those skilled in the art, various changes and modifications can be made to the present application. Any modifications, equivalent replacements, improvements, etc. made within the spirit and principle of the present application shall be included within the protection scope of the present application.

[0164] Although the specific implementation manners of the present disclosure have been described above in conjunction with the accompanying drawings, it is not a limitation on the protection scope of the present disclosure. Those skilled in the art should understand that, based on the technical solutions of the present disclosure, various modifications or deformations that can be made without creative efforts by those skilled in the art are still within the protection scope of the present disclosure.

Claims

1. A UAV collaborative path planning method based on a dynamic priority collision avoidance strategy, characterized in that: The following steps are involved: Get the map model and initialize the starting and ending points of the drone; When planning the UAV path, a global static optimal smooth path is planned from the map model; A multi-factor dynamic priority collision avoidance strategy is designed to obtain the impact results of each factor; when a collision is detected within the cluster, the priority of the drone is determined according to the impact results of each factor. High-priority drones plan their paths normally, and low-priority drones avoid high-priority drones.

2. The UAV collaborative path planning method based on dynamic priority collision avoidance strategy according to claim 1 is characterized in that: When planning the UAV path, time variables are introduced as references, namely time stay variables Time start variable and time termination variables The time stay variable Indicates the time required for the drone numbered i to complete the jth grid; time starting variable Indicates the starting time when the UAV numbered i reaches the jth grid; Time end variable Indicates the end time when the drone numbered i completes the jth grid; Based on the above time variables, the drone is regarded as a dynamic obstacle to achieve the definition of grid dynamic occupancy: Where: P(x,y) represents the grid occupancy function, which is used to characterize the occupancy of the grid map.

3. The UAV collaborative path planning method based on dynamic priority collision avoidance strategy according to claim 1 is characterized in that: The global static smooth path is obtained from the map model through the cubic spline interpolation method, and the global static optimal smooth path is planned by adding the moving window method, and its nodes are saved as the input of the multi-factor dynamic priority collision avoidance strategy.

4. The method for cooperative path planning of unmanned aerial vehicles based on dynamic priority collision avoidance strategy according to claim 3 is characterized in that: Given the interpolated path point Select a window size w and get the average value of the path points in the window as the smoothed path points as follows: Where: k represents the index of the point in the window, Represents the coordinate value of the path point in the window, and m represents the interpolated interval.

5. The method for cooperative path planning of unmanned aerial vehicles based on dynamic priority collision avoidance strategy according to claim 1 is characterized in that: The multi-factor dynamic priority collision avoidance strategy includes a buffer collision detection method based on a time window, which is implemented by extracting the current position and the previous position of UAV A and UAV B, respectively set as (x A (t),y A (t))、(x A (t-1),y A (t-1)); (x B (t),y B (t))、(x B (t-1),y B (t-1)); get the direction vector representing drone A and drone B: Then get the path direction angle: The running direction information of the UAV is obtained according to the path direction angle, and then the formula of the buffer collision detection method based on the time window is obtained as follows: Where: D buf represents the buffer distance, and σ represents the path direction angle.

6. The method for cooperative path planning of unmanned aerial vehicles based on dynamic priority collision avoidance strategy according to claim 1 is characterized in that: The multi-factor dynamic priority collision avoidance strategy combines the following indicators: (1) Remaining fuel quantity f l , the remaining fuel of the drone must be at least enough to support it to reach the end of the mission; (2) Total distance S, the cumulative distance of the UAV flight path; (3) Remaining distance S c , which is the remaining distance of the UAV’s flight path; When the minimum remaining fuel quantity f s Less than fuel limit f min , the drone gets a higher priority to avoid mission failure due to failure to meet the fuel constraint after collision avoidance. The minimum remaining fuel is obtained as follows: Where: η represents the fuel consumption in time series; V represents the flight speed of the UAV; When the minimum remaining fuel quantity f s Greater than the fuel limit f min When considering the total distance S and the remaining distance ratio S c , build a priority model, as shown in the following formula; the comprehensive index P of the drone i The larger it is, the higher its priority; P i =α*S+γ*S c ,i=1,2...,n Where: α, γ are the influence weights of the corresponding indicators.

7. The method for cooperative path planning of unmanned aerial vehicles based on dynamic priority collision avoidance strategy according to claim 1 is characterized in that: The way low-priority drones avoid high-priority drones is: First get the collision vector angle: Assume the current position of the i-th drone is The target end point is The direction vector with the highest efficiency in path planning is The direction point after the path replanning behavior is set to The path vector representing the path replanning is The vector angle θ is: Establish an avoidance strategy as follows: Where: λ represents the conflict tolerance value.

8. UAV collaborative path planning system based on dynamic priority collision avoidance strategy, characterized by: include: Initialization module, obtains the map model, and initializes the starting point and end point of the drone; The planning module plans the global static optimal smooth path from the map model when planning the UAV path; Collision avoidance module, designs a multi-factor dynamic priority collision avoidance strategy to obtain the impact results of each factor; when a collision is detected within the cluster, the priority of the drone is determined according to the impact results of each factor, the high-priority drone plans the path normally, and the low-priority drone avoids the high-priority drone.

9. An electronic device comprising a memory, a processor and a computer program stored and running on the memory, characterized in that: When the processor executes the program, the UAV collaborative path planning method based on the dynamic priority collision avoidance strategy is implemented.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the program is executed by the processor, the UAV collaborative path planning method based on the dynamic priority collision avoidance strategy is implemented.

Images