Distributed heterogeneous unmanned aerial vehicle time-sensitive task allocation and flight path collaborative planning method

Abstract

The invention relates to a distributed heterogeneous unmanned aerial vehicle time-sensitive task allocation and flight path collaborative planning method, which comprises a fast path cost dynamic estimation module, a distributed task allocation and path planning coupling module and a local reset module for new tasks in a dynamic environment. The method has the advantages that a real-time evaluation mechanism of the path cost is embedded into a task allocation process, deep fusion of task allocation and path reachability is achieved, and decision deviation caused by decoupling processing is effectively avoided. On the basis, a local reset strategy based on a task time window and path cost is introduced, the response capability of the system to dynamic task changes is remarkably improved, and the calculation overhead is reduced. Meanwhile, an improved consensus-based task coordination framework is adopted, and an adaptive communication mechanism is fused, so that the robustness and convergence speed of the algorithm under a time-varying network condition are enhanced.

Description

Technical Field

[0001] This invention relates to the field of unmanned aerial vehicle (UAV) planning methods, specifically to a distributed heterogeneous UAV time-sensitive task allocation and trajectory collaborative planning method. Background Technology

[0002] With the rapid development of drone technology, drones are increasingly widely used in military, civilian, and commercial fields. For example, in disaster relief, drones can perform search and rescue, material delivery, and disaster monitoring tasks; in agriculture, drones can be used for pesticide spraying and crop health monitoring; and in logistics and distribution, drones can achieve fast and efficient cargo transportation. However, with the increasing complexity of application scenarios and the expansion of mission scale, single drones are no longer sufficient to meet the needs of multiple tasks and multiple objectives. Therefore, distributed heterogeneous multi-drone systems have gradually become a research hotspot.

[0003] Distributed heterogeneous multi-UAV systems are characterized by their composition of different types of UAVs (such as fixed-wing, rotary-wing, and hybrid-powered UAVs), each with different performance parameters (such as endurance, payload capacity, and speed range) and mission adaptability. These UAVs collaborate in a distributed manner to solve complex task allocation and trajectory planning problems. The core challenge of such systems lies in how to efficiently optimize task allocation and trajectory planning to ensure maximum task completion efficiency, optimal resource utilization, and overall system performance.

[0004] Currently, research on multi-UAV task allocation and trajectory planning mainly focuses on the following aspects: 1. Centralized Task Allocation Method: This method models the task allocation and trajectory planning of all UAVs as a single global optimization problem and solves it using mathematical programming or heuristic algorithms (such as genetic algorithms and simulated annealing). The advantage of this approach is that it can obtain theoretically optimal global solutions, but its disadvantage is high computational complexity, making it difficult to apply in real-time to large-scale UAV systems.

[0005] 2. Distributed Task Allocation Methods: Distributed task allocation algorithms are typically designed based on game theory, auction mechanisms, or market mechanisms. Each drone makes independent decisions based on local information and negotiates with its neighboring nodes. Typical algorithms include the Contract Net Protocol (CNP) and the Auction Algorithm. The advantages of these methods are distributed computational burden and strong scalability, but they suffer from slow convergence speed and a tendency to get trapped in local optima.

[0006] 3. Path planning algorithms: Path planning algorithms mainly include graph search-based methods (such as A* algorithm and Dijkstra's algorithm), sampling-based methods (such as RRT and PRM), and potential field-based methods (such as artificial potential field method). These algorithms can generate safe paths from the starting point to the destination, but in dynamic environments or multi-aircraft collaborative scenarios, they often lack consideration for real-time performance and global coordination.

[0007] 4. Joint Optimization of Task Assignment and Path Planning: Some existing techniques attempt to combine task assignment and path planning to form a joint optimization framework. For example, by introducing mixed-integer linear programming (MILP) models or reinforcement learning methods, the objective functions of both task assignment and path planning are considered simultaneously. However, the computational complexity of such methods remains high, and their adaptability to heterogeneous UAV systems is insufficient.

[0008] Although existing technologies have solved the problem of multi-UAV task allocation and trajectory planning to some extent, the following major shortcomings still exist: 1. High computational complexity: Although centralized optimization methods and joint optimization methods can theoretically find better solutions, their computational complexity increases exponentially with the number of UAVs and the scale of the mission, making it difficult to meet the real-time requirements in practical applications.

[0009] 2. Lack of support for heterogeneity: Most existing methods assume that UAVs are homogeneous, meaning all UAVs have the same performance parameters and mission capabilities. However, in real-world applications, UAVs are typically heterogeneous, with different types exhibiting significant differences in endurance, payload capacity, and flight speed. Existing methods fail to adequately consider these heterogeneous characteristics, leading to unreasonable task allocation results.

[0010] 3. Insufficient Protocol Optimization: Current task allocation and trajectory planning methods are mostly performed in stages, i.e., task allocation is performed first, and then trajectory planning is optimized separately. This method ignores the coupling relationship between the two, which may lead to suboptimal solutions. For example, some task allocation schemes may fail to generate feasible trajectories, or the generated trajectories may not meet the time constraints of the task.

[0011] 4. Poor adaptability to dynamic environments: Most existing algorithms perform well in static environments, but in dynamic environments (such as sudden obstacles, changes in task priorities, etc.), the robustness and adaptability of the algorithms are poor, and it is difficult to adjust task allocation and trajectory planning strategies in a timely manner.

[0012] 5. High communication overhead: Although distributed algorithms reduce the computational burden of individual nodes, they require frequent communication negotiations, resulting in high communication overhead. This is especially true in large-scale UAV systems, where communication latency and bandwidth limitations can become bottlenecks.

[0013] In summary, existing multi-UAV task allocation and trajectory planning technologies still have many shortcomings in terms of computational efficiency, heterogeneity support, collaborative optimization, dynamic adaptability, and communication overhead. Therefore, there is an urgent need for a distributed heterogeneous UAV time-sensitive task allocation and trajectory collaborative planning method to overcome the above problems and improve the overall performance of the system. Summary of the Invention

[0014] The purpose of this invention is to address the shortcomings and defects of existing technologies by providing a distributed heterogeneous UAV time-sensitive task allocation and trajectory collaborative planning method. This invention achieves deep integration of task allocation and path reachability by embedding a real-time path cost evaluation mechanism into the task allocation process, effectively avoiding decision bias caused by decoupled processing. Furthermore, a local reset strategy based on task time windows and path costs is introduced, significantly improving the system's responsiveness to dynamic task changes and reducing computational overhead. Simultaneously, an improved consensus-based task coordination framework is adopted, incorporating an adaptive communication mechanism to enhance the algorithm's robustness and convergence speed under time-varying network conditions. Moreover, this invention can improve the timeliness of task response and flight safety while maximizing global task benefits, making it particularly suitable for application scenarios with high requirements for timeliness and path feasibility, such as disaster relief, and has promising engineering application prospects.

[0015] To achieve the above objectives, the present invention adopts the following technical solution: a distributed heterogeneous UAV time-sensitive task allocation and trajectory cooperative planning method, which includes a fast path cost dynamic prediction module, a distributed task allocation and path planning coupling module, and a local reset module for new tasks in dynamic environments. The fast path cost dynamic prediction module is used to improve path planning efficiency and reduce computational overhead. It pre-builds an initial visibility graph before task allocation and dynamically expands the graph structure according to the current position of the UAV or the task point during the actual path calculation stage. It combines the shortest path search algorithm to perform path calculation, thereby achieving fast and efficient path generation. Distributed task allocation and path planning coupled module: This module includes three core stages: task addition stage, new task partial reset stage, and task conflict resolution stage. Each stage works together to achieve efficient and reliable task scheduling; for the partial reset module for new tasks in dynamic environments: This includes task compatibility judgment, construction of candidate reset task set, path cost calculation and priority ranking, execution of partial reset operation, and task reallocation and optimization.

[0016] Furthermore, the fast path cost dynamic prediction module specifically employs a visibility graph method that avoids the computational resource consumption associated with high-resolution map modeling. While ensuring path safety, it significantly improves path planning efficiency, making it particularly suitable for multi-UAV systems requiring rapid response and efficient collaboration. The visibility graph is composed of the visibility relationships between the vertices of polygonal obstacles. The prerequisite for visibility between two vertices is that the straight line segment connecting these two vertices does not intersect with the boundaries of other obstacles. To determine whether a vertex can "see" other vertices, a visibility determination method based on a scanline strategy is used. Specifically, the visibility determination method based on a scanline strategy involves using a vertex... Starting from a point, emit a ray in a horizontal direction. , as the main scan line; define the set Store all vertices Mutually visible points, initially empty; define an ordered list. Stores the obstacle edge information that currently intersects with the scan line; initially, this is empty. During the initialization phase, all obstacle vertices are first processed according to their intersection with the scan line. Distance and relative to The vertices are sorted by the angles of the axes to form an ordered sequence. Then, it is determined whether each obstacle edge is aligned with the initial scan line. If edges intersect, add them to the ordered edge list. and press with vertex The distances are sorted in ascending order; during the traversal of each vertex, the scan line rotates sequentially to the next target vertex. And based on the current edge list Determine if there are obstacles obstructing the view, if the list Empty indicates from the vertex arrive If the connection is not blocked by any obstacle, both are visible; otherwise, it is necessary to further determine whether the first edge intersects with the connecting line to determine the visibility status. In addition, after processing a new vertex each time, the system will dynamically update the edge list E: remove the edges located in the clockwise direction of the scan line and add new edges located in the counterclockwise direction to ensure the accuracy and efficiency of subsequent visibility judgments.

[0017] Furthermore, the distributed task allocation and path planning coupling module specifically comprises three core stages: a task addition stage, a new task partial reset stage, and a task conflict resolution stage. These stages work collaboratively to achieve efficient and reliable task scheduling. Each UAV maintains two key data structures: a task list. : indicates drone The currently assigned set of tasks; execution order This indicates the order in which these tasks are performed during flight. The above structure is continuously iterated and updated during algorithm execution and is subject to the maximum mission carrying capacity of the drone. Limitations; during the task addition phase, the drone determines whether to accept a new task by evaluating the task score, and Sipi is defined as the drone's... The total task score in the current execution order is calculated using the following formula: ,in: This is the time decay coefficient; For the task The estimated start time under the current execution order; For the task The earliest executable time; For the task Value rating for the task; Inserted into the execution order The first in At each position, the system will recalculate the score. The details are as follows: If task Already exists or If the task is selected correctly, no additional score will be added; otherwise, the system will select the insertion position that maximizes the increase in total score for task addition. Furthermore, before inserting a task, the system will use a visibility graph model to calculate the visibility of the task from the current position. To the new mission The shortest path, and based on path length and drone speed Calculate flight time = Thus, the task can be accurately estimated. Start time: in For the previous task Execution time.

[0018] Furthermore, the distributed task allocation and path planning coupling module also includes a task conflict resolution mechanism: during the task conflict resolution phase, each UAV exchanges task allocation information through communication, including: the winning task set. : Indicates that the current drone The task to be performed; the task offer : indicates drone Bid value for each task; timestamp Record the latest update time of information; after communicating with other drones, the drone updates its local mission status based on the received information. If a mission has been won by another drone with a higher bid, the drone originally assigned the mission will release the mission and its subsequent missions, and update the execution order synchronously. The process mainly includes three processing methods: Update: Synchronize the other party's task winning information and bid; Reset: Clear the drone's task winning information; Leave: Retain the original information unchanged; When all drones reach a consensus, that is, when the winners and bids of all tasks no longer change, the task allocation process ends.

[0019] Furthermore, the partial reset module for new tasks in a dynamic environment specifically includes: task compatibility judgment: when a new task... Upon arrival, the system first determines whether each drone is capable of performing the task. If a drone lacks the necessary sensors, payloads, or functional modules for the corresponding task type, it is skipped, and its original task sequence remains unchanged. Only drones with the capability to perform the task participate in the subsequent partial reset operation. A candidate reset task set is then constructed: For drones with the capability, the system evaluates the time window and path relationship between the new task and the original task, filtering out tasks that may be affected by the new task to form a candidate reset task set. Specifically, only tasks whose execution time windows overlap with the new task are selected for reset processing to avoid unnecessary global task updates; path cost calculation and priority ranking: after determining candidate tasks, the system calls the visibility graph model and shortest path algorithm to calculate the flight path cost between the candidate task and the new task, and sorts the candidate tasks according to the path cost, prioritizing the task with the minimum path cost to the new task as the insertion point to improve the success rate of task insertion and the overall task score; perform local reset operation: for the selected task The system will clear this task from the current task list. All tasks in and after this point, and update the execution order synchronously. This operation ensures that the principle of diminishing marginal returns still holds throughout the entire task sequence, thereby maintaining the monotonicity of the task scoring function and the convergence of the algorithm; Task reallocation and optimization: After completing the local reset, the system calls the distributed task allocation algorithm again to reallocate the released tasks to suitable drones, thereby achieving rebalancing and optimization of task resources.

[0020] The beneficial effects of this invention, achieved by embedding a real-time path cost evaluation mechanism into the task allocation process, are as follows: This invention deeply integrates task allocation and path reachability, effectively avoiding decision-making biases caused by decoupling. Furthermore, the introduction of a local reset strategy based on task time windows and path costs significantly improves the system's responsiveness to dynamic task changes and reduces computational overhead. Simultaneously, the adoption of an improved consensus-based task coordination framework, combined with an adaptive communication mechanism, enhances the algorithm's robustness and convergence speed under time-varying network conditions. Moreover, this invention can improve the timeliness of task response and flight safety while maximizing global task benefits, making it particularly suitable for application scenarios with high requirements for timeliness and path feasibility, such as post-disaster relief, and possesses promising engineering application prospects. Attached Figure Description

[0021] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0022] Figure 1 This is a schematic diagram of the overall scheme of the present invention.

[0023] Figure 2 This is a schematic diagram of obstacle vertex visibility determination in this invention.

[0024] Figure 3 This is a comparison chart of the scores of the algorithm in this invention with the mainstream CBBA algorithm under different drones and number of tasks.

[0025] Figure 4 This is a comparison chart of the flight distance of the algorithm in this invention with the mainstream CBBA algorithm under different drones and mission numbers.

[0026] Figure 5 This is a comparison chart of the scores and convergence times of the algorithm and the new task reset algorithm in this invention under the conditions of 4 drones and 25 tasks.

[0027] Figure 6 Table 1 shows a comparison of the scores and convergence times of the algorithm and the new task reset algorithm under different drone and task numbers in this invention.

[0028] Figure 7 Table 2 shows a comparison of the scores and convergence times of the algorithm and the new task reset algorithm under different drone and task numbers in this invention. Detailed Implementation

[0029] (1) Problem definition and background: The research object of this invention is the task allocation and path planning of heterogeneous multiple unmanned aerial vehicles (UAVs, such as rescue and search and rescue UAVs) for heterogeneous multiple tasks (such as rescue and search and rescue tasks) in a dynamic, time-sensitive, and constrained rescue environment. Here, several assumptions are introduced to help define the research problem: 1) Given the invention's context of a dynamic, time-sensitive, and constrained rescue environment, survivor waiting time is particularly critical, as prolonged delays can severely adversely affect their health. To ensure survivors' well-being, both rescue and search missions are assigned time windows, and all missions must be completed within these windows; otherwise, the mission is considered a failure.

[0030] 2) It is assumed that the search and rescue team already has a familiar understanding of the site environment using satellite maps before the drones carry out the search and rescue missions. Therefore, the obstacle and danger zones within the map are known, and this paper does not consider exploration issues. This assumption is considered reasonable for the scope of our research. Therefore, the main focus is on improving the overall score of drone mission allocation, that is, completing as many search and rescue missions as possible before the survivors' health deteriorates.

[0031] 3) Assuming a dynamic rescue environment, the search and rescue team may receive distress calls from survivors at any time. Therefore, the drone swarm needs to respond promptly to new tasks and assign them to drones capable of performing them, ensuring survivors receive timely assistance.

[0032] Under the above assumptions, the main problems that this work needs to solve are summarized as follows: This invention addresses the distributed task allocation problem in search and rescue missions for heterogeneous unmanned aerial vehicle (UAV) systems. Existing algorithms exhibit significant limitations when dealing with real-world rescue scenarios, primarily in three aspects: insufficient modeling of environmental constraints, lack of collaborative optimization between task allocation and path planning, and delayed response to dynamic tasks.

[0033] (2) Problem modeling: In the distributed heterogeneous multi-UAV task allocation and trajectory planning collaborative optimization algorithm described in this invention, a UAV set is defined. This refers to an aircraft formation consisting of multiple heterogeneous drones, performing various types of missions including rescue and search; defining a mission set. This represents the set of tasks to be assigned. Due to the limited carrying capacity of each drone, a single drone can be assigned a maximum of [number missing]. One task; among them, drones The cruising speed is expressed as .

[0034] Depending on the task attributes, each task has a specific time window constraint, which includes: Task The earliest time it can be executed Latest completion time and the execution time required. All tasks must be uniquely assigned to a single drone with the corresponding execution capability, and it must be ensured that the task can be executed within its time window.

[0035] For distributed task allocation modeling, let... Indicates initial allocation to drones Task sequence list, This indicates the sequential path of the actual mission performed by the drone, by Generated after sorting. (If it's a drone...) In execution order The Middle The task to be executed next is the task. Then it is written as = At the same time, set Indicates drone According to the path order The planned start time sequence for each task; if the task It is a drone No. The next task to be executed, This indicates the expected start time of the task, and it should meet the following requirements: .

[0036] In a constrained search and rescue environment, it is assumed that all obstacles in the environment can be modeled using geometric polygons, and the set of obstacles is denoted as . Based on this, we define drones. The complete flight path is From the drone's starting point to the first mission point The path segment is denoted as To ensure the safety and feasibility of the mission, the planned path must avoid collisions with any obstacles. The specific mathematical model is shown in the following formula: See Figures 1-7 As shown, the technical solution adopted in this specific implementation is as follows: it includes a fast path cost dynamic prediction module, a distributed task allocation and path planning coupling module, and a local reset module for new tasks in dynamic environments. Rapid path cost dynamic prediction module: In dynamic multi-UAV collaborative search and rescue missions, the task sequence is frequently adjusted, which leads to the need for continuous updates to path planning, bringing a large computational burden to the distributed task allocation algorithm. Although traditional environment modeling based on fine-grained grid maps can improve the safety and reliability of the path, its high modeling complexity is in obvious contradiction with the time-sensitive requirements of dynamic rescue scenarios. Therefore, a fast path cost dynamic prediction module is adopted to improve path planning efficiency and reduce computational overhead. This method pre-constructs an initial visibility graph before task allocation and dynamically expands the graph structure according to the current position of the UAV or the task point during the actual path calculation stage. It combines the shortest path search algorithm (such as Dijkstra's algorithm) to perform path calculation, thereby achieving fast and efficient path generation.

[0037] Compared with traditional grid map-based modeling methods, the visibility graph method used in this algorithm avoids the computational resource consumption caused by high-resolution map modeling, and significantly improves path planning efficiency while ensuring path safety. It is particularly suitable for multi-UAV systems that require fast response and efficient collaboration. The visibility map is composed of the visibility relationships between the vertices of polygonal obstacles. The premise for two vertices to have a visibility relationship is that the straight line segment connecting the two vertices does not intersect the boundaries of other obstacles. In order to determine whether a vertex can "see" other vertices, a visibility determination method based on a scan line strategy is adopted.

[0038] See Figure 2 As shown, the visibility determination method based on the scan line strategy is as follows: With a certain vertex Starting from a point, emit a ray in a horizontal direction. , as the main scan line; define the set Store all vertices Mutually visible points, initially empty; define an ordered list. Stores information about obstacles that currently intersect with the scan line; initially, this field is empty. During the initialization phase, all obstacle vertices are first configured according to the vertex... Distance and relative to The vertices are sorted by the angles of the axes to form an ordered sequence. Then, it is determined whether each obstacle edge is aligned with the initial scan line. If edges intersect, add them to the ordered edge list. and press with vertex Sort by distance in ascending order; During the traversal of each vertex, the scan line rotates sequentially to the next target vertex. And based on the current edge list Determine if there are obstacles obstructing the view, if the list Empty indicates from the vertex arrive If the line connecting the two edges is not blocked by any obstacles, both edges are visible; otherwise, it is necessary to further determine whether the first edge intersects with the connecting line to determine the visibility status. In addition, after processing each new vertex, the system dynamically updates the edge list E: removing edges located in the clockwise direction of the scan line and adding new edges located in the counterclockwise direction to ensure the accuracy and efficiency of subsequent visibility determination; Distributed task allocation and path planning coupled module: This module includes three core stages: task addition stage, new task partial reset stage, and task conflict resolution stage. Each stage works together to achieve efficient and reliable task scheduling.

[0039] Each drone maintains two key data structures: Task List : indicates drone The currently assigned set of tasks; Execution order This indicates the order in which these tasks are performed during the flight.

[0040] The aforementioned structure is continuously iterated and updated during algorithm execution, and is subject to the maximum mission carrying capacity of the drone. Restrictions; During the task addition phase, the drone determines whether to accept a new task by evaluating the task score. Sipi is defined as the drone's... The total task score in the current execution order is calculated using the following formula: in: This is the time decay coefficient; For the task The estimated start time under the current execution order; For the task The earliest executable time; For the task Value rating for the task; Inserted into the execution order The first in At each position, the system will recalculate the score. The details are as follows: If task Already exists or If the score is correct, no additional points will be added; otherwise, the system will select the insertion position that maximizes the increase in total score to add the task.

[0041] In addition, before inserting a task, the system uses a visibility graph model to calculate the visibility from the current task. To the new mission The shortest path, and based on path length and drone speed Calculate flight time = Thus, the task can be accurately estimated. Start time: in, For the previous task Execution time.

[0042] The distributed task allocation and path planning coupling module also includes a task conflict resolution mechanism: During the mission conflict resolution phase, the drones exchange mission allocation information via communication, including: Successful Bid Task Set : Indicates that the current drone The task to be performed; Task Bidding : indicates drone Bid value for each task; Timestamp Record the latest update time of the information; After communicating with other drones, the drone updates its local mission status based on the received information. If a mission has been won by another drone with a higher bid, the drone that originally assigned the mission will release the mission and its subsequent missions, and update the execution order accordingly. .

[0043] The communication rules are shown in the table, and mainly include three processing methods: Update: Synchronize the other party's task winning information and bids; Reset: Clear the drone's award information for this mission; Leaving: Retain existing information unchanged; Table 1. UAV Communication Rules The task allocation process ends when all drones reach a consensus that the winners and bids for all tasks will no longer change. For the partial reset module of new tasks in dynamic environments: Task compatibility judgment: When new tasks Upon arrival, the system first determines whether each drone is capable of performing the task. If a drone lacks the necessary sensors, payloads, or functional modules for the corresponding task type, it is skipped, and its original task sequence remains unchanged. Only drones with the capability to perform the task will participate in subsequent partial reset operations. A candidate reset task set is then constructed. For drones capable of performing tasks, the system assesses the time window and path relationship between the new task and the original task, filters out tasks that may be affected by the new task, and forms a candidate set of reset tasks. Specifically, only tasks whose execution time windows overlap with the new task are selected for reset processing to avoid unnecessary global task updates; Path cost calculation and priority ranking: After determining the candidate tasks, the system calls the visibility graph model and the shortest path algorithm (such as Dijkstra's algorithm) to calculate the flight path cost between the candidate tasks and the new task. The candidate tasks are sorted according to the path cost, and the task with the minimum path cost to the new task is selected as the insertion point to improve the success rate of task insertion and the overall task score. Perform a partial reset: For the selected task The system will clear this task from the current task list. All tasks in and after this point, and update the execution order synchronously. This operation ensures that the principle of diminishing marginal gains (DMG) still holds throughout the task sequence, thereby maintaining the monotonicity of the task scoring function and the convergence of the algorithm. Task reallocation and optimization: After completing the partial reset, the system calls the distributed task allocation algorithm (such as the two-stage collaborative optimization algorithm described in this invention) again to reassign the released tasks to suitable drones, thereby achieving rebalancing and optimization of task resources.

[0044] The above description is only used to illustrate the technical solution of the present invention and is not intended to limit it. Any other modifications or equivalent substitutions made by those skilled in the art to the technical solution of the present invention, as long as they do not depart from the spirit and scope of the technical solution of the present invention, should be covered within the scope of the claims of the present invention.

Claims

1. A method for distributed heterogeneous UAV time-critical task allocation and trajectory cooperative planning, characterized in that: It includes a fast path cost dynamic estimation module, a distributed task allocation and path planning coupling module, and a local reset module for new tasks in a dynamic environment: The fast path cost dynamic estimation module is used to improve path planning efficiency and reduce computational overhead. It pre-constructs an initial visibility graph before task allocation and dynamically expands the graph structure according to the current position of the UAV or the task point during actual path calculation. It combines the shortest path search algorithm to calculate the path, thereby achieving fast and efficient path generation. The distributed task allocation and path planning coupling module includes three core stages: task addition stage, new task local reset stage, and task conflict resolution stage. Each stage works together to achieve efficient and reliable task scheduling. The local reset module for new tasks in a dynamic environment includes task compatibility judgment, candidate reset task set construction, path cost calculation and priority sorting, local reset operation execution, and task re-allocation and optimization.

2. The method of claim 1, wherein: The fast path cost dynamic estimation module specifically includes: The visibility graph method avoids the computational resource consumption caused by high-resolution map modeling, significantly improving path planning efficiency while ensuring path safety, especially suitable for multi-UAV systems that require fast response and efficient collaboration. The visibility graph is composed of the visibility relationship between the vertices of the polygonal obstacle. The prerequisite for two vertices to have a visibility relationship is that the straight line segment connecting the two vertices does not intersect with other obstacle boundaries. To determine whether a vertex can "see" another vertex, a visibility judgment method based on the scan line strategy is used. The visibility determination method based on the scanline strategy is as follows: Specifically, taking a certain vertex... Starting from a point, emit a ray in a horizontal direction. , as the main scan line; define the set Store all vertices Mutually visible points, initially empty; define an ordered list. Stores information about obstacles that currently intersect with the scan line; initially, this field is empty. In the initialization phase, first, all obstacle vertices are sorted according to the distance from the vertex and the angle relative to the axis , forming an ordered vertex sequence, then it is determined whether each obstacle edge intersects the initial scan line , if it intersects, it is added to the ordered edge list , and arranged in ascending order of the distance from the vertex . In the process of traversing each vertex, the scan line is rotated to the next target vertex in turn , and according to the current edge list , it is determined whether there is an obstacle blocking, if the list is empty, it indicates that the line from the vertex to is not blocked by any obstacle, and both can be seen; Otherwise, further judgment is needed to determine whether the first edge intersects with the connecting line to determine the visibility state. In addition, after processing a new vertex each time, the system dynamically updates the edge list E: removes the edges located in the clockwise direction of the scan line and adds new edges located in the counterclockwise direction, ensuring the accuracy and efficiency of subsequent visibility judgment.

3. The method of claim 1, wherein: The distributed task allocation and path planning coupling module specifically includes: The module includes three core stages: task addition stage, new task local reset stage, and task conflict resolution stage. Each stage works together to achieve efficient and reliable task scheduling. Each UAV maintains two key data structures: Task list : represents a drone a set of tasks currently assigned; Execution order : indicates the order in which these tasks are executed during the flight; The above structure is constantly updated during the algorithm running process, and is limited by the maximum task carrying capacity of the unmanned aerial vehicle ; In the task adding phase, the UAV decides whether to receive a new task by evaluating the task score, defining Sipi represents the UAV The total task score under the current execution sequence is calculated as follows: wherein: is the time decay coefficient; To task Expected start time under current execution order; For the task The earliest executable time; To task a value score; When the task Inserted into the execution order The first in At each position, the system will recalculate the score. The details are as follows: if task has already existed in or , no additional score is added; otherwise, the system will select the insertion position that maximizes the total score increment for task addition; In addition, before inserting a task, the system uses a visibility graph model to calculate the visibility from the current task. To the new mission The shortest path, and based on path length and drone speed Calculate flight time = Thus, the task can be accurately estimated. Start time: wherein is the execution duration of the previous task .

4. The method of claim 1, wherein: The distributed task allocation and path planning coupling module also includes a task conflict resolution mechanism: In the task conflict resolution stage, each UAV exchanges task allocation information through communication, including: winning task set : indicates a task currently executed by the unmanned aerial vehicle ; Task bidding : represents a UAV bid values for each task; Timestamp : the latest update time of the recorded information; After the UAV communicates with other UAVs, the local task state is updated according to the received information, and if a task has been awarded to other UAVs with a higher bid, the original UAV assigned with the task releases the task and subsequent tasks and synchronously updates the execution sequence ; It mainly includes three processing methods: Update: synchronize the task bid information and bid of the opposite party; Reset: clear the bid information of the UAV for the task; Leave: keep the original information unchanged; When all UAVs reach a consensus, i.e., the bid winner and bid of all tasks no longer change, the task allocation process ends.

5. The method of claim 1, wherein: The local reset module for new tasks in a dynamic environment specifically includes: Task compatibility check: When a new task... Upon arrival, it is first determined whether each drone has the capability to perform the task. If a drone does not have the sensors, payload, or functional modules required for the corresponding task type, the drone is skipped and its original task sequence remains unchanged. Only drones with the capability to perform the task will participate in the subsequent partial reset operation. Constructing a candidate reset task set: for the unmanned aerial vehicle with execution ability, the system evaluates the time window and path relationship between the new task and the original task, screens the tasks that may be affected by the new task, and constitutes a candidate reset task set Specifically, only those tasks whose execution time window overlaps with the new task are selected for reset processing to avoid unnecessary global task updating. Path cost calculation and priority sorting: after determining the candidate tasks, the system calls the visibility graph model and the shortest path algorithm to calculate the flight path cost between the candidate tasks and the new task, sorts the candidate tasks according to the path cost, and preferentially selects the task with the minimum path cost as the insertion point to improve the success rate of task insertion and the overall task score; Performing a partial reset operation: for a selected task The system clears all tasks from the current task list and after it, and updates the execution order synchronously This operation ensures that the diminishing returns principle still holds for the entire task sequence, thus maintaining the monotonicity of the task scoring function and the convergence of the algorithm; Task reallocation and optimization: after completing the local reset, the system calls the distributed task allocation algorithm again to reallocate the released tasks to suitable UAVs, achieving task resource rebalancing and optimization.

Images