Multi-uav operation control method with priority strategy and verification method thereof

Abstract

The application discloses a multi-unmanned aerial vehicle operation control method with priority strategy and a verification method thereof. The control method takes an actual three-dimensional terrain model as a boundary condition, plans an initial path for all unmanned aerial vehicles according to starting points and task target points of the unmanned aerial vehicles through an RRT-star algorithm, and optimizes the initial path. Each unmanned aerial vehicle is controlled to run along the respective optimized path, and position information and priority of each unmanned aerial vehicle are acquired in real time. For unmanned aerial vehicle pairs whose current distance exceeds a safety distance threshold, the unmanned aerial vehicle with a lower priority is controlled to run along the optimized path with a fixed preset time step added to a current time. The verification method generates a three-dimensional terrain and unmanned aerial vehicle positions through noise to verify the control method. The multi-unmanned aerial vehicle is sequentially run according to priorities of the unmanned aerial vehicles to realize cooperative collision avoidance of the multi-unmanned aerial vehicle, and the operation safety and cooperation efficiency of the multi-machine system under complex terrain are significantly improved.

Description

Technical Field

[0001] This invention relates to multi-UAV formation mission planning, specifically to a multi-UAV operation control method that introduces a priority strategy and its verification method. Background Technology

[0002] With the increasing maturity of unmanned aerial vehicle (UAV) technology, its application mode is gradually shifting from single-platform operation to multi-UAV swarm collaboration, a trend that has become an important research direction. UAV swarms simulate the self-organizing characteristics of biological groups, possessing excellent dynamic adaptability and enabling efficient task execution in complex and ever-changing environments through multi-UAV cooperation. The advantages of this type of system lie not only in its ability to achieve parallel task processing and resource sharing, but also in its ability to effectively enhance the overall capacity to respond to emergencies through information exchange and collaborative decision-making among individual UAVs.

[0003] In the field of smart cities, such systems are widely used in traffic monitoring, public security, and infrastructure inspection, enhancing the intelligence level of urban management and the speed of emergency response through multi-drone collaborative operations. Overall, drone swarms, with their high degree of collaboration and environmental adaptability, demonstrate significant value in numerous application scenarios, providing effective support for achieving efficient and flexible task execution.

[0004] During path planning, UAVs need to effectively avoid obstacles in complex 3D terrain while simultaneously handling multi-UAV collision avoidance within the swarm. Before reaching the mission target, UAVs must avoid obstacles in 3D space to ensure safety, while also optimizing their flight paths to shorten distances. Complex mountainous terrain and other obstacle-dense environments place higher demands on path planning algorithms, requiring not only path feasibility but also smoothness. As the swarm size increases, path conflicts and congestion may occur between UAVs, making inter-UAV collision avoidance a critical challenge. Therefore, in addition to handling the interaction between individual UAVs and the external terrain environment, path planning algorithms must also coordinate the behavior within the swarm to achieve efficient collaboration and ensure overall flight safety.

[0005] In conclusion, path planning for UAV swarms is a core element for achieving efficient collaboration and mission execution in complex environments. Path planning must not only address the challenges of complex terrain but also solve collision avoidance problems within the swarm to ensure mission safety and flight efficiency. Summary of the Invention

[0006] Purpose of the invention: To address the above-mentioned shortcomings, this invention provides a multi-UAV operation control method and its verification method that introduces a priority strategy to solve the collision avoidance problem within a cluster.

[0007] Technical Solution: To solve the above problems, this invention adopts a multi-UAV operation control method that introduces a priority strategy, including the following steps:

[0008] (1) Construct a three-dimensional terrain model to obtain the starting point and mission target point of multiple UAVs;

[0009] (2) Using the three-dimensional terrain model as the boundary condition, for all UAVs, the initial path is planned by the RRT-star algorithm based on the starting point of the UAV and the target point of the mission, and the initial path is optimized to obtain the optimized path;

[0010] (3) Control each UAV to run along its own optimized path, and obtain the position information and priority of each UAV in real time. Calculate the current distance between each pair of UAVs. For UAV pairs whose current distance exceeds the safe distance threshold, compare the priority of the UAVs in the UAV pair. For UAVs with lower priority, from the current moment, each UAV will run along the optimized path with a fixed preset time step.

[0011] Furthermore, the specific steps for planning the initial path using the RRT-star algorithm based on the UAV's starting point and mission objective point are as follows:

[0012] (21) Initialize the parameters by setting the starting point of the UAV as the root node of the tree and the target point of the mission as the endpoint.

[0013] (22) Randomly generate a sampling point within the search space and find the node closest to the sampling point in the tree;

[0014] (23) Generate a new node by extending one step from the node closest to the sampling point in the direction towards the sampling point. Find all neighboring nodes within the neighborhood radius of the new node. Find the parent node with the minimum path cost among the neighboring nodes. When the distance between the new node and the destination is less than the threshold, backtrack from the destination along the parent node to the starting point to obtain the path set.

[0015] Furthermore, using the 3D terrain model as the boundary condition, obstacle avoidance detection is performed on the generated parent node. If the parent node does not meet the boundary condition of the 3D terrain model, it is deleted.

[0016] Furthermore, the optimization of the initial path includes removing redundant nodes and inflection points in the initial path through a waypoint optimization selection mechanism and smoothing the path.

[0017] Furthermore, the specific steps for deleting redundant nodes and inflection points in the initial path through the waypoint optimization selection mechanism are as follows: taking the starting pathpoint of the initial path as the critical waypoint, the path segments are obtained by connecting the current critical waypoint to any subsequent node of the critical waypoint in the initial path. Starting from the end pathpoint of the initial path, the path segments are traversed back to the initial pathpoint. It is determined whether the path segments meet the boundary conditions of the 3D terrain model. The end pathpoints of the path segments that meet the conditions are added to the critical waypoint set, and the end pathpoints of the path segments that meet the conditions are used as critical waypoints. The selection is repeated until the end pathpoint of the initial path becomes a critical waypoint. If no path segment meets the conditions in the current traversal, the next pathpoint of the critical waypoint on the initial path is used as a new critical waypoint for traversal.

[0018] Furthermore, the path smoothing process specifically involves using a B-spline curve smoothing optimization algorithm to smooth the path. Specifically, based on curvature constraints, a continuous path curve is generated for the path using cubic B-spline basis functions and node vectors. The continuous path curve is then uniformly sampled to generate a set of path points. Each sampled path point in the set is checked to see if it meets the boundary conditions of the three-dimensional terrain model. If it does not meet the conditions, the path curve is regenerated.

[0019] Furthermore, the priority of each UAV is calculated through a priority function, which includes the value of the UAV performing the task, the resource contribution of the UAV to the task, and the distance of the UAV to the task.

[0020] This invention also includes a verification method for the above-mentioned multi-UAV operation control method that introduces a priority strategy, comprising the following steps:

[0021] (1) By using based A noise-based terrain generation method simulates and generates a 3D terrain model; and generates the starting points and mission objective points of multiple UAVs on the 3D terrain model;

[0022] (2) Using the three-dimensional terrain model as the boundary condition, for all UAVs, the initial path is planned by the RRT-star algorithm based on the starting point of the UAV and the target point of the mission, and the initial path is optimized to obtain the optimized path;

[0023] (3) Control each UAV to run along its own optimized path and calculate the running time of the UAV's initial optimized running path;

[0024] Control each drone to run along its own optimized path, and obtain the position information and priority of each drone in real time. Calculate the current distance between each pair of drones. For drone pairs whose current distance exceeds the safe distance threshold, compare the task priorities of the drones in the pair. For drones with lower priority, starting from the current moment, each drone will run along the optimized path with a fixed preset time step. Calculate the running time of the drone running the optimized path after collision avoidance.

[0025] (4) By comparing the running time of the initial optimized path of the UAV and the running time of the optimized path after collision avoidance, the verification results of the multi-UAV operation control method with priority strategy are obtained.

[0026] The present invention also employs a computer device, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to implement the steps of the above method.

[0027] The present invention also employs a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the steps of the above method.

[0028] Beneficial effects: Compared with the prior art, the significant advantages of this invention are:

[0029] (1) By prioritizing the sequential operation of multiple UAVs, the cooperative collision avoidance of multiple UAVs can be achieved, which significantly improves the operational safety and collaborative efficiency of multi-UAV systems in complex terrain.

[0030] (2) By adopting the improved RRT-star algorithm and combining it with the waypoint optimization selection mechanism, the redundant nodes and inflection points of the path are significantly reduced, which improves the efficiency of path planning and the feasibility of UAV flight while ensuring obstacle avoidance safety.

[0031] (3) Using B-spline curves to smooth the initial path effectively improves the path continuity and smoothness, reduces unnecessary turning and speed fluctuations during UAV flight, thereby reducing energy consumption and extending flight time.

[0032] (4) The noise-based terrain generation method constructs a highly realistic and flexibly configurable three-dimensional terrain environment, which can effectively simulate complex mountain undulations and obstacle distribution, and provide experimental and verification conditions for UAV path planning that are closer to actual tasks. Attached Figure Description

[0033] Figure 1 This is a flowchart illustrating the operation method of the present invention.

[0034] Figure 2 This is a schematic diagram of a complex three-dimensional mountainous terrain based on noise in this invention.

[0035] Figure 3 This is a schematic diagram of the initial path generation in this invention.

[0036] Figure 4 This is a schematic diagram of waypoint optimization path generation in this invention.

[0037] Figure 5 This is a schematic diagram of smooth path generation in this invention.

[0038] Figure 6 This is a schematic diagram illustrating the generation of all paths in this invention.

[0039] Figure 7 This is the time increment histogram in this invention.

[0040] Figure 8 This is a schematic diagram comparing the average path time and the maximum path time of the initial path and the path after collision avoidance in this invention. Detailed Implementation

[0041] like Figure 1 As shown, this invention proposes a planning method based on an improved RRT-STAR algorithm for path planning problems in complex 3D environments. This method first introduces a waypoint optimization selection mechanism to filter path nodes and reduce redundant inflection points. Then, B-spline curves are applied to smooth the path, thereby reducing the number of turns and energy consumption. Furthermore, a priority comprehensive evaluation method integrating task value, resource contribution, and initial path length is constructed. Finally, based on the obtained smoothed path and priority allocation, cooperative collision avoidance control of multiple UAVs is achieved, effectively preventing inter-UAV collisions. To illustrate the feasibility of this invention, the following embodiments will be used for further explanation.

[0042] Example 1

[0043] This embodiment presents a multi-UAV operation control method that incorporates a priority strategy, comprising the following steps:

[0044] Step 1: Construct a 3D terrain model to obtain the starting points and mission objective points of multiple UAVs;

[0045] Step 2: Using the 3D terrain model as boundary conditions, for all UAVs, plan the initial path using the RRT-star algorithm based on the UAV's starting point and mission objective point, and optimize the initial path to obtain the optimized path;

[0046] Step 3: Control each drone to run along its own optimized path, and obtain the position information and priority of each drone in real time. Calculate the current distance between each pair of drones. For drone pairs whose current distance exceeds the safe distance threshold, compare the task priorities of the drones in the pair. For drones with lower task priorities, starting from the current moment, each drone will run along the optimized path with a fixed preset time step.

[0047] In this embodiment, an improved Rapidly-exploring RandomTree (RRT-star) algorithm is used for initial path planning, and a waypoint optimization selection mechanism is used to reduce redundant nodes and inflection points in the path. The RRT-star algorithm is a path planning method based on random sampling, which has the ability to efficiently explore free space.

[0048] During the path generation phase, all newly generated path nodes undergo obstacle avoidance detection to determine whether they are above and above the terrain surface. If the detection finds that a node intersects with the terrain or is below the terrain height, the node is considered invalid and discarded. This mechanism effectively prevents paths from directly traversing the terrain.

[0049] The specific steps of the path planning algorithm based on RRT-star are as follows:

[0050] Step 21-1: Initialize parameters and set the starting point of the drone. As the root node of the tree, define the tree. ; Set the mission objective point Set as endpoint Define the current node list. and initialized to ;

[0051] Step 21-2: Randomly generate a point within the search space. , ;

[0052] Step 21-3: In the tree Find the distance point The nearest node :

[0053] ;

[0054] Step 21-4: From node To point Extend the direction to generate a new node in one step :

[0055] ;

[0056] Step 21-5: In the new node neighborhood radius Find all neighboring nodes and add them to the neighboring node set. :

[0057] ;

[0058] In the neighboring node set Find the parent node with the minimum path cost in the middle. :

[0059] ;

[0060] in, To start from the starting point To the node Path cost; for the set of neighboring nodes Other nodes besides the parent node Check if it can be passed through the new node. Optimized path:

[0061]

[0062] if Then reset the new node. For nodes The parent node;

[0063] Step 21-6: Determine the new node With the mission objective The distance between them, if or reaching the maximum number of iterations. When, then the target point Join the tree, and from the target point Backtracking from the parent node to the starting point Result in a set of paths : ;if Then return to step 21-2.

[0064] In this embodiment, optimizing the initial path includes removing redundant nodes and inflection points from the initial path using a waypoint optimization selection mechanism. During the path optimization phase, the straight path between two points is segmented and sampled, and each sampled point is checked to see if it lies on the terrain surface. If a path segment poses a risk of intersecting with the terrain, that path segment is marked as infeasible and readjusted. This process ensures the overall safety of the path segment.

[0065] The goal of the waypoint optimization selection mechanism is to select a set of key waypoints in the initial path, ensuring that the straight paths between these waypoints do not collide with the terrain surface, thereby simplifying the path structure. Specifically, this mechanism iterates through the initial path, reducing the number of path nodes and optimizing path complexity to ultimately generate a concise path composed of key waypoints. During optimization, the start and end points of the path are first marked as initial optimization waypoints. Then, the algorithm recursively filters from the end of the path towards the start, obtaining path segments based on the connections from the current key waypoint to any subsequent node in the path. For each path segment, a collision detection method is used to verify its safety and feasibility, prioritizing the longest possible path segment. If a path segment does not collide with the terrain surface, the end node of the path is added to the set of key waypoints. Simultaneously, the starting reference point of the entire path traversal is updated to the current node, i.e., the starting index of the path traversal is updated; this neither regenerates the path nor modifies the original path. This process is repeated until the entire path is fully optimized.

[0066] The steps of the waypoint optimization selection mechanism are as follows:

[0067] Step 22-1: Initialize parameters: Set waypoints Define as an empty list; set initial index End of index ;make ,Will Add to waypoint set middle;

[0068] Step 22-2: Traverse the path from the end point to the starting point, and set the index... from Decrease, check from the current starting point Time Does the connecting path segment collide with the terrain? If there is no collision, then the point... Add to waypoint set and update This means setting the current node as the new starting point. If no collision-free path segment is found in the current traversal, then... Increment by 1 (select the nearest next node), Add to waypoint set middle.

[0069] Step 22-3: When The loop terminates when the path ends. Added to waypoint set If not already added, the optimized waypoint set is finally returned. .

[0070] In this embodiment, optimizing the initial path includes smoothing the path and using B-spline curves to smooth and optimize the planned path, thereby reducing unnecessary turning and energy consumption of the UAV.

[0071] Path curve generation: A set of paths obtained through a waypoint optimization selection mechanism. Used as input to the B-spline smoothing algorithm, combined with cubic B-spline basis functions. Combine node vectors to generate continuous path curves. .

[0072] Collision detection: in generating path curves Then, a set of path points is generated through uniform sampling, and each sampled path point is checked to see if it meets the requirement of no collision with the terrain surface. If a collision risk is detected, the control points are adjusted. Adjustments were made and the curves were regenerated until all path points met safety requirements. This ensured that the smoothed B-spline path always flew above the 3D terrain surface, avoiding collisions with the ground or obstacles.

[0073] Curvature constraint optimization: A curvature constraint mechanism is introduced to limit the maximum curvature of the path. The curvature function is defined as:

[0074]

[0075] in, and These are the velocity and acceleration vectors of the path curve, respectively; by limiting the curvature The maximum value;

[0076] Path Sampling and Output: While path curve equations provide a continuous and smooth path description, they have significant limitations in practical inter-drone collision avoidance applications. Since collision avoidance decisions need to be completed rapidly within high-frequency update cycles, directly using curve equations introduces a complex real-time computational burden, making it difficult to meet time response requirements. Furthermore, cooperative collision avoidance between UAVs relies on position comparisons at a unified discrete time step, while continuous curves lack natural discrete-time alignment properties. Therefore, after path optimization, the path curves are processed according to mission requirements. Dynamic sampling is performed to generate a dense set of path points. Since it is necessary to design a cluster collision avoidance algorithm based on the initial path of each drone and in combination with a priority mechanism, the purpose of path sampling is to provide initial path data for drone cluster collision avoidance in the scenario.

[0077] In this embodiment, a priority calculation method that integrates task value, resource contribution, and initial path length is designed to assign dynamic priorities to UAVs, coordinate paths in real time during multi-UAV cooperative flight, and implement swarm collision avoidance strategies.

[0078] The path priority of a drone is a crucial factor in collision avoidance decision-making, and its level directly impacts the path adjustment strategy. The formula for calculating the priority function is:

[0079]

[0080] in, It is a task value, It is a drone For the task The degree of resource contribution Drones generated based on the improved RRT-star algorithm To the mission Path length, It is a constant value used to balance the impact of resource contribution factors on the flight priority of drones. , , The global weights of each factor;

[0081] The priority value calculated by this function will be used in subsequent collision avoidance decisions. The higher the priority of a drone, the higher its path adjustment priority. This mechanism can coordinate the action order of a swarm of drones in conflict situations; the generated initial path serves as the data foundation, and combined with the priority function and collision avoidance adjustment strategy, it gradually resolves potential collision problems between drones. The algorithm mainly includes the following steps:

[0082] 1. Path expansion: For each path Need to find the current time The last point of the path ; after that from Start, press Add waypoints at intervals until... The spatial location and other parameters (such as priority) of all newly added points remain consistent with those of the last point;

[0083] 2. Collision threat detection: calculation Location information of drones at all times The distance matrix between the drones is constructed as follows:

[0084]

[0085] Detect all that meet the requirements Collision ;

[0086] 3. Collision Threat Detection and Decision-Making: For each collision pair It is necessary to compare current drones and drones path and Priority, if drones Pause, otherwise, drone pause.

[0087] 4. Time Advancement: Proceed to the next timestamp Recalculate ,like If the condition is met, the algorithm terminates; otherwise, path expansion is performed again. , The path with the longest numerical value among all initial paths. The speed of each drone. The final output is the path set after applying a cluster collision avoidance algorithm. .

[0088] Example 2

[0089] This embodiment presents a verification method for a multi-UAV operation control method that incorporates a priority strategy, comprising the following steps:

[0090] Step 1: By using based A noise-based terrain generation method simulates and generates a 3D terrain model; and generates the starting points and mission objective points of multiple UAVs on the 3D terrain model;

[0091] Step 2: Using the 3D terrain model as boundary conditions, for all UAVs, plan the initial path using the RRT-star algorithm based on the UAV's starting point and mission objective point, and optimize the initial path to obtain the optimized path;

[0092] Step 3: Control each drone to run along its own optimized path and calculate the running time of the drone's initial optimized path;

[0093] Control each drone to run along its own optimized path, and obtain the position information and priority of each drone in real time. Calculate the current distance between each pair of drones. For drone pairs whose current distance exceeds the safe distance threshold, compare the task priorities of the drones in the pair. For drones with lower task priorities, starting from the current moment, each drone will run along the optimized path with a fixed preset time step. Calculate the running time of the drone running the optimized path after collision avoidance.

[0094] Step 4: Compare the running time of the initial optimized path of the UAV with the running time of the optimized path after collision avoidance to obtain the verification results of the multi-UAV operation control method that introduces the priority strategy.

[0095] In this embodiment, a noise-based terrain generation method is used to construct a three-dimensional terrain model, simulating the undulation characteristics of natural terrain and the distribution of obstacles, and generating the UAV starting point, mission target point and environmental elements according to the requirements of the UAV mission scenario.

[0096] The three-dimensional features of the terrain are represented by a discrete mesh, where the height value of each mesh point is determined by... The noise generation function is given. Specifically, for terrain grid points... height for:

[0097]

[0098] The terrain range height matrix is ​​as follows: , For terrain at grid points height, The number of layers stacked when generating noise. For the first The amplitude of the layer, For the first The frequency of the layer.

[0099] The noise value is obtained by smooth interpolating the contribution values ​​of the mesh vertices, using the following formula:

[0100]

[0101] Linear interpolation function Defined as:

[0102]

[0103] in, and Let the two values ​​be the values ​​to be interpolated. These are the weighting coefficients of the linear interpolation function, used to control the interpolation points at the two endpoints. and The position between them, the range of values ​​is .

[0104] exist In noise, smooth interpolation weighting coefficients and , specifically by The function is calculated as follows:

[0105]

[0106] in, and For point The formula for calculating the relative position in the y-direction of the current mesh element is:

[0107]

[0108]

[0109] in, The integer part of the absolute coordinate corresponds to the starting point of the grid. It is the integer part of the absolute coordinate, corresponding to the starting point of the grid.

[0110] The generated height value matrix is ​​normalized using the following formula:

[0111]

[0112] in, and These are the minimum and maximum values ​​of the initial terrain height matrix, respectively. This represents the maximum height of the terrain after normalization.

[0113] The location of the drone and the mission is determined by three-dimensional coordinates. It means that among them For terrain grid coordinates, For height coordinates, satisfy the following constraints:

[0114]

[0115] in The elevation of the terrain surface. The safe height between the location and the terrain follows a uniform distribution. The initial position generation algorithm for drones and missions follows these steps: randomly selecting terrain grid points. Calculate terrain height Randomly generated based on the safe height range Finally ensure Afterwards, record As a valid position.

[0116] To ensure the efficiency and reproducibility of the simulation experiments, this embodiment implemented all algorithms and performed simulation analysis using Python, and presented the experimental results intuitively using its scientific computing and visualization capabilities. The parameter settings for terrain, UAV, obstacle avoidance, and collision avoidance algorithms are shown in the table below.

[0117]

[0118] To support experimental reproduction and subsequent analysis, terrain data and generated location information are stored in .npy and .mat formats, respectively. At the same time, supporting data loading and verification interfaces are provided to ensure the reliability and reusability of the data. Figure 2 The visualization results showcasing the terrain, UAV, and mission location clearly present their distribution in three-dimensional space. The mountainous terrain environment constructed by this modeling method not only reproduces the undulating features of real landforms but also flexibly adapts to different experimental scenario settings, providing an effective simulation foundation for UAV path planning research in complex environments.

[0119] To evaluate the performance of the proposed obstacle avoidance path planning algorithm, simulation experiments were conducted to compare three results: initial path, waypoint optimized path, and smooth path. The analysis focused on path shape and total length. Simulation results are as follows: Figures 3 to 6 As shown. Initial path (e.g.) Figure 3 The path (as shown) was generated by the improved RRT-star algorithm. It features a dense distribution of pathpoints and exhibits some backtracking and irregular characteristics, demonstrating the algorithm's ability to explore feasible solutions in complex 3D terrain. The optimized path (e.g.) Figure 4 (As shown) The waypoint selection mechanism filters out redundant nodes, significantly improving path simplicity. Smooth paths (such as...) Figure 5 As shown, the optimized path is fitted with a B-spline curve to obtain a continuous and smooth trajectory, which better matches the actual flight requirements of the UAV. A comprehensive comparison chart (as shown) is also available. Figure 6 As shown, the three types of paths are placed in the same view, intuitively presenting their evolution and differences. Simulation results show that the present invention can generate flight paths that are feasible, smooth, and collision-free in complex terrain, meeting all the requirements of obstacle avoidance planning.

[0120] The time increment plot is used to statistically analyze the time increment between the initial path and the path of all UAVs after collision avoidance, in order to analyze the time impact range and stability of the collision avoidance algorithm on different mission paths. This is based on the histogram of time increments (…). Figure 7 The data shows that the time increment for the vast majority of tasks is close to 0, indicating that the path time for most tasks does not change significantly before and after collision avoidance. This proves that the collision avoidance algorithm resolves potential conflicts with minimal impact on path time. This distribution of increments demonstrates the high adaptability of the collision avoidance algorithm, which can rationally allocate path time when collision avoidance is needed, while minimizing interference with other tasks. The comparison chart of average time and maximum time (…) Figure 8Further analysis shows that the average time after collision avoidance increased from 6.65 seconds to 7.48 seconds, and the maximum time increased from 14.10 seconds to 16.65 seconds. This change is mainly due to the collision avoidance algorithm adjusting the path for a few tasks with a higher risk of conflict, while having a smaller impact on most tasks. The results demonstrate that the proposed method can effectively control time costs while ensuring cluster security, achieving a good balance between security and efficiency, and showcasing its practicality and superiority in complex task environments.

[0121] In this embodiment, a real-time path planning method for multiple UAVs incorporating a dynamic priority strategy is proposed. By combining the RRT-star path generation algorithm, waypoint optimization selection mechanism, and B-spline curve smoothing method, the problems of path generation, waypoint optimization, and path smoothing in complex 3D terrain environments are effectively solved. Simultaneously, to address the inter-UAV collision problem that may occur during UAV swarm mission execution, a swarm collision avoidance algorithm based on an initial path and priority mechanism is proposed. This algorithm solves the path conflict problem that UAV swarms may encounter during mission flight through task priority sorting, collision detection, and path adjustment mechanisms. Those skilled in the art will readily understand that the above descriptions are merely preferred embodiments of the present invention and are not intended to limit the invention. Any modifications, equivalent substitutions, and improvements made within the spirit and principles of the present invention should be included within the scope of protection of the present invention.

Claims

1. A multi-UAV operation control method incorporating a priority strategy, characterized in that, Includes the following steps: (1) Construct a three-dimensional terrain model to obtain the starting point and mission target point of multiple UAVs; (2) Using the three-dimensional terrain model as boundary conditions, for all UAVs, an initial path is planned using the RRT-STAR algorithm based on the UAV's starting point and the mission objective point, and the initial path is optimized to obtain the optimized path; the specific steps of planning the initial path using the RRT-STAR algorithm based on the UAV's starting point and the mission objective point are as follows: (21) Initialize the parameters by setting the starting point of the UAV as the root node of the tree and the target point of the mission as the endpoint. (22) Randomly generate a sampling point within the search space and find the node closest to the sampling point in the tree; (23) Generate a new node by extending one step from the node closest to the sampling point in the direction towards the sampling point. Find all neighboring nodes within the neighborhood radius of the new node. Find the parent node with the minimum path cost among the neighboring nodes. When the distance between the new node and the destination is less than the threshold, backtrack from the destination along the parent node to the starting point to obtain the path set. (3) Control each UAV to run along its own optimized path, and obtain the position information and priority of each UAV in real time. Calculate the current distance between each pair of UAVs. For UAV pairs whose current distance exceeds the safe distance threshold, compare the priority of the UAVs in the UAV pair. For UAVs with lower priority, from the current moment, each UAV will run along the optimized path with a fixed preset time step.

2. The multi-UAV operation control method incorporating a priority strategy according to claim 1, characterized in that, Using the 3D terrain model as the boundary condition, obstacle avoidance detection is performed on the generated parent node. If the parent node does not meet the boundary condition of the 3D terrain model, it is deleted.

3. The multi-UAV operation control method incorporating a priority strategy according to claim 1, characterized in that, The optimization of the initial path includes removing redundant nodes and inflection points in the initial path through a waypoint optimization selection mechanism and smoothing the path.

4. The multi-UAV operation control method incorporating a priority strategy according to claim 3, characterized in that, The specific steps for deleting redundant nodes and inflection points in the initial path through the waypoint optimization selection mechanism are as follows: taking the starting pathpoint of the initial path as the critical waypoint, the path segment is obtained by connecting the current critical waypoint to any subsequent node of the critical waypoint in the initial path. Starting from the end pathpoint of the initial path, the path segment is traversed back to the initial pathpoint. It is determined whether the path segment meets the boundary conditions of the three-dimensional terrain model. The end pathpoint of the path segment that meets the conditions is added to the critical waypoint set. The end pathpoint of the path segment that meets the conditions is used as the critical waypoint, and the selection is carried out again until the end pathpoint of the initial path becomes the critical waypoint. If no path segment meets the conditions in the current traversal, the next path point of the critical point on the initial path will be used as the new critical point for traversal.

5. The multi-UAV operation control method incorporating a priority strategy according to claim 3, characterized in that, The path smoothing process specifically involves using a B-spline curve smoothing optimization algorithm. Specifically, based on curvature constraints, a continuous path curve is generated using cubic B-spline basis functions and node vectors. The continuous path curve is then uniformly sampled to generate a set of path points. Each sampled path point in the set is checked to see if it meets the boundary conditions of the 3D terrain model. If it does not meet the conditions, the path curve is regenerated.

6. The multi-UAV operation control method incorporating a priority strategy according to claim 1, characterized in that, The priority of each UAV is calculated by a priority function, which includes the value of the UAV performing the task, the resource contribution of the UAV to the task, and the distance of the UAV from the task.

7. A verification method for the multi-UAV operation control method incorporating a priority strategy as described in any one of claims 1 to 6, characterized in that, Includes the following steps: (1) By using based A noise-based terrain generation method simulates and generates a 3D terrain model; The starting points and mission objective points of multiple UAVs are generated on the 3D terrain model. (2) Using the three-dimensional terrain model as the boundary condition, for all UAVs, the initial path is planned by the RRT-star algorithm based on the starting point of the UAV and the target point of the mission, and the initial path is optimized to obtain the optimized path; (3) Control each UAV to run along its own optimized path and calculate the running time of the UAV's initial optimized running path; Control each drone to run along its own optimized path, and obtain the position information and priority of each drone in real time. Calculate the current distance between each pair of drones. For drone pairs whose current distance exceeds the safe distance threshold, compare the task priorities of the drones in the pair. For drones with lower priority, starting from the current moment, each drone will run along the optimized path with a fixed preset time step. Calculate the running time of the drone running the optimized path after collision avoidance. (4) By comparing the running time of the initial optimized path of the UAV and the running time of the optimized path after collision avoidance, the verification results of the multi-UAV operation control method with priority strategy are obtained.

8. A computer 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 computer program, it implements the steps of the method according to any one of claims 1 to 6.

9. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by a processor, it implements the steps of the method according to any one of claims 1 to 6.

Images