A method and device for unmanned aerial vehicle path planning based on meteorological environment

Abstract

The application discloses a kind of unmanned vehicle path planning method and device based on meteorological environment, method includes: obtaining system parameter and meteorological data of planning area;Meteorological data is based on fusion processing to multi-source positioning information, and initial track coordinates are obtained;Based on initial track coordinates, task data and performance parameter, global task allocation is carried out using adaptive genetic algorithm, and access sequence data are obtained;Based on access sequence data and meteorological data, search tree optimization is carried out by risk cost, and initial path planning data are obtained;Initial path planning data are carried out conflict coordination and post-processing, and target path planning data are obtained.The application is positioned by fusing rate meteorological data to carry out positioning compensation and path planning, and search tree optimization is guided using risk cost, can effectively reduce the flight risk of unmanned vehicle under adverse weather, can be widely applied in path planning technical field.

Description

Technical Field

[0001] This invention relates to the field of path planning technology, and in particular to a method and apparatus for unmanned aerial vehicle (UAV) path planning based on meteorological environment. Background Technology

[0002] With the increasingly widespread application of drone technology in emergency logistics, urban inspection, and environmental monitoring, the autonomous flight capability of drones in complex meteorological environments, especially after extreme weather events such as typhoons and severe convection, has become a key challenge restricting their further application. In such scenarios, urban near-ground wind fields exhibit high spatiotemporal variability, posing stringent requirements on the positioning accuracy, flight safety, and path planning of drones.

[0003] Currently, the industry has developed a variety of methods for drone path planning.

[0004] One type is graph search-based algorithms, such as Algorithm A. These algorithms typically perform path searches on discretized grid maps, which are computationally efficient, but struggle to handle continuously changing wind field environments and complex dynamic constraints.

[0005] Another type is sampling-based algorithms, such as Rapid Exploratory Random Tree (RRT) and its improved version RRT. These algorithms can effectively perform path search in high-dimensional spaces, possess asymptotic optimality, and are suitable for complex environments. However, existing sampling-based methods often oversimplify the consideration of wind field influence during the planning process. For example, they may only use steady-state wind field models or ignore the real-time impact of wind field on path cost, leading to planned paths that may face significant safety risks or excessive energy consumption in actual flight. Summary of the Invention

[0006] The main objective of this invention is to provide a method, apparatus, electronic device, storage medium, and program product for unmanned aerial vehicle (UAV) path planning based on meteorological environment, aiming to solve at least one problem in the prior art.

[0007] To achieve the above objectives, one aspect of this invention proposes a UAV path planning method based on meteorological environment, the method comprising: Acquire system parameters and meteorological data for the planned area; among which, system parameters include task data, multi-source positioning information and performance parameters of the UAV swarm, and task data includes delivery point coordinates and material requirements; The initial track coordinates are obtained by fusing multi-source positioning information based on meteorological data; Based on the initial track coordinates, task data, and performance parameters, an adaptive genetic algorithm is used for global task allocation to obtain access sequence data; the access sequence data includes the access sequence of each delivery point corresponding to each drone in the drone cluster. Based on access sequence data and meteorological data, search tree optimization is performed through risk cost to obtain initial path planning data; the initial path planning data includes local continuous path planning data between adjacent delivery points for each drone. The initial path planning data is reconciled and post-processed to obtain the target path planning data.

[0008] In some embodiments, the initial track coordinates are obtained by fusing multi-source positioning information based on meteorological data, including the following steps: Based on meteorological data, a layered atmospheric delay correction model is used to compensate for meteorological interference positioning errors in multi-source positioning information, resulting in compensated multi-source positioning information. The initial track coordinates are obtained by fusing and denoising the compensation multi-source positioning information through fuzzy adaptive Kalman filtering.

[0009] In some embodiments, based on initial track coordinates, mission data, and performance parameters, an adaptive genetic algorithm is used to perform global mission allocation to obtain access sequence data, including the following steps: The delivery point numbers of all delivery points are divided into subsequences of each drone in the drone swarm, and an initial population is generated by random arrangement. The initial population includes several individuals, and each individual corresponds to an arrangement scheme of delivery point numbers in the drone swarm. The initial population is taken as the first population; Based on the arrangement scheme corresponding to the first group, the fitness value of each individual is calculated using a preset quantization function based on the initial track coordinates and task data; where the fitness value represents the reciprocal of the total delivery cost. The maximum fitness value in the first population is used as the candidate fitness value, and the corresponding individual is used as the candidate individual; Genetic evolution was performed on the first population based on fitness values ​​to obtain the second population; Among them, genetic evolution includes roulette wheel strategy, crossover operation and mutation operation. The crossover probability applied by the crossover operation is quantified based on the larger fitness value of the two individuals to be crossed and the extreme fitness value and average fitness value of the first population. The mutation probability applied by the mutation operation is quantified based on the fitness value of the individual to be mutated and the extreme fitness value and average fitness value of the first population. The feasibility of the arrangement scheme for each individual in the second population is tested based on the performance parameters. If the feasibility test result is unsuccessful, the corresponding individual is repaired to make it pass the feasibility test. Increment the number of iterations; If the rate of change of the candidate fitness values ​​is less than the improvement threshold after a preset number of consecutive iterations, or if the number of iterations reaches the preset maximum number of iterations, the orchestration scheme corresponding to the candidate individual is used as the access sequence data; otherwise... Treating the second population as the first population, the process returns to the step of calculating the fitness value of each individual based on the initial track coordinates and mission data according to the arrangement scheme corresponding to the first population, until the number of iterations reaches the preset maximum number of iterations, and the arrangement scheme corresponding to the candidate individual is used as the access sequence data.

[0010] In some embodiments, adjacent delivery points include a start point and an end point. Based on access sequence data and meteorological data, initial route planning data is obtained by performing search tree optimization through risk cost, including the following steps: Obtain building data for the configuration space between adjacent delivery points, and extract wind speed and vertical wind shear intensity for the configuration space from meteorological data; Add the starting point to the search tree as the root node; Sampling points are generated in the configuration space through uniform random sampling; Based on wind speed, vertical wind shear intensity, and building data, the risk cost of the sampling point is quantified; whereby the risk cost represents the weighted sum of the high wind speed risk corresponding to wind speed, the wind shear risk corresponding to vertical wind shear intensity, and the building proximity risk corresponding to building data. The risk cost is normalized and mapped to the rejection probability of the sample points; The selection type of the sampling point is determined by comparing a random number within a preset range with the probability of rejection; the selection type includes acceptance and rejection. If the type is selected as Accept, the sampled points will be added as nodes to the search tree; The sampling point is used as the first candidate point, and the node closest to the first candidate point in the search tree is used as the second candidate point. Based on a preset step size, a new node is generated starting from the second candidate point and moving in the direction of the first candidate point; Based on the wind speed, vertical wind shear intensity, and building data corresponding to the new node, constrained dynamic detection is performed on the new node. If the result of the constraint dynamic detection is passed, the new node is added to the search tree; Based on the new node, candidate paths are planned from the search tree by using a fast exploration random tree star schema algorithm with the constraint of minimizing risk cost; If the distance between the new node and the end point is less than the preset step size, the candidate path is used as local continuous path planning data between adjacent delivery points; otherwise, the process returns to the step of generating sampling points in the configuration space through uniform random sampling until the distance between the new node and the end point is less than the preset step size. Initial path planning data is obtained by organizing the local continuous path planning data between all adjacent delivery points corresponding to each drone.

[0011] In some embodiments, conflict coordination and post-processing are performed on the initial path planning data to obtain the target path planning data, including the following steps: The initial path planning data is subjected to multi-UAV path conflict detection and coordination to obtain the second path planning data. The second path planning data is post-processed and smoothed to obtain the target path planning data.

[0012] In some embodiments, multi-UAV path conflict detection and coordination are performed on the initial path planning data to obtain second path planning data, including the following steps: The initial path planning data is discretized over time to obtain the spatiotemporal sequence for each UAV; the spatiotemporal sequence includes several spatiotemporal points. Spatiotemporal conflict detection between UAVs is performed based on spatiotemporal points to obtain the conflict points; The search tree optimization is re-executed on the local continuous path planning data corresponding to the conflict points to eliminate the conflict points, resulting in local continuous path replanning data, which is then summarized to obtain the second path planning data. If there are no conflict points, the initial path planning data will be used as the second path planning data.

[0013] In some embodiments, post-processing and smoothing are performed on the second path planning data to obtain the target path planning data, including at least one of the following steps: The second path planning data is pruned by removing redundant intermediate nodes. Cubic B-spline curves are used to smooth the trajectory of the second path planning data.

[0014] To achieve the above objectives, another aspect of the present invention proposes a UAV path planning device based on meteorological environment, the device comprising: The first module is used to acquire system parameters and meteorological data for the planned area; the system parameters include task data, multi-source positioning information and performance parameters of the UAV cluster, and the task data includes delivery point coordinates and material requirements. The second module is used to fuse multi-source positioning information based on meteorological data to obtain the initial track coordinates; The third module is used to perform global task allocation based on the initial track coordinates, task data and performance parameters using an adaptive genetic algorithm to obtain access sequence data; the access sequence data includes the access sequence of each delivery point corresponding to each drone in the drone cluster; The fourth module is used to perform search tree optimization based on access sequence data and meteorological data, through risk cost, to obtain initial path planning data; wherein, the initial path planning data includes local continuous path planning data between adjacent delivery points for each drone; The fifth module is used to coordinate conflicts and perform post-processing on the initial path planning data to obtain the target path planning data.

[0015] To achieve the above objectives, another aspect of the present invention provides an electronic device, which includes a memory and a processor. The memory stores a computer program, and the processor executes the computer program to implement the aforementioned method.

[0016] To achieve the above objectives, another aspect of the present invention provides a computer-readable storage medium storing a computer program that, when executed by a processor, implements the aforementioned method.

[0017] To achieve the above objectives, another aspect of the present invention provides a computer program product, including a computer program that, when executed by a processor, implements the aforementioned method.

[0018] The embodiments of the present invention include at least the following beneficial effects: The present invention provides a method, apparatus, electronic device, storage medium, and program product for UAV path planning based on meteorological environment. This solution acquires system parameters and meteorological data of the planning area; wherein the system parameters include task data, multi-source positioning information of the UAV swarm, and performance parameters; the task data includes delivery point coordinates and material requirements; based on the meteorological data, the multi-source positioning information is fused to obtain initial trajectory coordinates; based on the initial trajectory coordinates, task data, and performance parameters, an adaptive genetic algorithm is used for global task allocation to obtain access sequence data; wherein the access sequence data includes the access sequence of each UAV in the UAV swarm corresponding to a delivery point; based on the access sequence data and meteorological data, a search tree is optimized through risk cost to obtain initial path planning data; wherein the initial path planning data includes local continuous path planning data between adjacent delivery points for each UAV; conflict coordination and post-processing are performed on the initial path planning data to obtain target path planning data. This invention utilizes fusion meteorological data for positioning compensation and path planning, and employs risk-cost-guided search tree optimization to effectively reduce the flight risks of UAVs in adverse weather conditions. Specifically, this invention employs a two-stage collaborative architecture of "global task allocation followed by local path planning." First, an adaptive genetic algorithm optimizes multi-aircraft task allocation and access order from a global perspective. Then, a refined local path search is performed for each flight segment. This collaborative mechanism overcomes the tendency of single-stage planning to fall into local optima, ensuring both global task efficiency and the local feasibility of each flight segment. Furthermore, by acquiring and processing meteorological data and performing planning based on this data, this invention enables the UAV system to dynamically respond to complex and changing weather environments. Subsequent conflict coordination steps further ensure the spatiotemporal safety of multi-aircraft collaborative operations, significantly improving the system's adaptability and robustness in complex environments. Attached Figure Description

[0019] Figure 1 This is a schematic diagram of an implementation environment for the UAV path planning method based on meteorological environment provided in an embodiment of the present invention; Figure 2 This is a flowchart illustrating the UAV path planning method based on meteorological environment provided in an embodiment of the present invention. Figure 3 This is a schematic diagram of the overall process of the UAV path planning method based on meteorological environment provided in the embodiments of the present invention; Figure 4 This is a schematic diagram of the structure of the UAV path planning device based on meteorological environment provided in an embodiment of the present invention; Figure 5 This is a schematic diagram of the structure of the electronic device provided in an embodiment of the present invention. Detailed Implementation

[0020] To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative of the invention and are not intended to limit the invention. In the following description, when referring to the accompanying drawings, unless otherwise indicated, the same numbers in different drawings represent the same or similar elements. The embodiments described in the following exemplary embodiments do not represent all embodiments consistent with the embodiments of this invention; they are merely examples of apparatuses and methods consistent with some aspects of the embodiments of this invention as detailed in the appended claims.

[0021] It is understood that the terms “first,” “second,” etc., used in this invention may be used herein to describe various concepts, but unless specifically stated otherwise, these concepts are not limited by these terms. These terms are used only to distinguish one concept from another. For example, first information may also be referred to as second information without departing from the scope of embodiments of the invention, and similarly, second information may also be referred to as first information. Depending on the context, the words “if,” “when,” or “in response to determination” as used herein may be interpreted as “when…” or “when…” or “in response to determination.”

[0022] The terms “at least one,” “multiple,” “each,” “any,” etc., used in this invention, “at least one” includes one, two, or more than two; “multiple” includes two or more than two; “each” refers to each of the corresponding multiple; and “any” refers to any one of the multiple.

[0023] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention pertains. The terminology used herein is for the purpose of describing embodiments of the invention only and is not intended to limit the invention.

[0024] To facilitate understanding of the technical solution of this invention, the following explanations are provided for the technical terms that may be involved in the technical solution of this invention: GA (Genetic Algorithm): A global optimization algorithm based on natural selection and genetic mechanisms, used to generate an initial path population.

[0025] RRT* (Rapid Exploratory Random Tree): A sampling-based path planning algorithm with asymptotic optimality, suitable for local path optimization in complex environments.

[0026] High-resolution wind field: refers to a wind field model with fine-grained temporal and spatial data, including information such as wind speed, wind direction, and turbulence.

[0027] Safety corridor: A spatial constraint region generated based on the global path set in path planning, used to limit the local path search range.

[0028] Risk cost model: A mathematical model used to quantify risks such as high wind speed, wind shear, and proximity to buildings during flight.

[0029] B-spline smoothing: A curve fitting method used to smooth paths and improve the executability of flight trajectories.

[0030] Existing technical solutions generally suffer from the following shortcomings: they fail to effectively integrate high-resolution, dynamically changing wind field data; the planning process is simplistic, lacking coordination between global task allocation and local path optimization; and the quantitative modeling of flight risks is relatively simple, failing to systematically consider multiple constraints such as wind shear, proximity of buildings, and the performance limits of UAVs. Therefore, how to plan a safe and efficient flight path for UAVs in complex meteorological environments such as typhoons remains a pressing technical challenge in this field.

[0031] In view of this, this invention provides a method and apparatus for UAV path planning based on meteorological environment. This method acquires system parameters and meteorological data of the planning area. The system parameters include task data, multi-source positioning information of the UAV swarm, and performance parameters. The task data includes delivery point coordinates and material requirements. Based on the meteorological data, the multi-source positioning information is fused to obtain initial trajectory coordinates. Based on the initial trajectory coordinates, task data, and performance parameters, an adaptive genetic algorithm is used for global task allocation to obtain access sequence data. The access sequence data includes the access sequence of each UAV in the UAV swarm corresponding to a delivery point. Based on the access sequence data and meteorological data, a search tree is optimized using risk cost to obtain initial path planning data. The initial path planning data includes local continuous path planning data between adjacent delivery points for each UAV. The initial path planning data undergoes conflict coordination and post-processing to obtain target path planning data. This invention utilizes fusion meteorological data for positioning compensation and path planning, and employs risk-cost-guided search tree optimization to effectively reduce the flight risks of UAVs in adverse weather conditions. Specifically, this invention employs a two-stage collaborative architecture of "global task allocation followed by local path planning." First, an adaptive genetic algorithm optimizes multi-aircraft task allocation and access order from a global perspective. Then, a refined local path search is performed for each flight segment. This collaborative mechanism overcomes the tendency of single-stage planning to fall into local optima, ensuring both global task efficiency and the local feasibility of each flight segment. Furthermore, by acquiring and processing meteorological data and performing planning based on this data, this invention enables the UAV system to dynamically respond to complex and changing weather environments. Subsequent conflict coordination steps further ensure the spatiotemporal safety of multi-aircraft collaborative operations, significantly improving the system's adaptability and robustness in complex environments.

[0032] It is understood that the UAV path planning method based on meteorological environment provided by this invention can be applied to any computer device with data processing and computing capabilities, and this computer device can be various terminals or servers. When the computer device in the embodiment is a server, the server is an independent physical server, or a server cluster or distributed system composed of multiple physical servers, or a cloud server that provides basic cloud computing services such as cloud services, cloud databases, cloud computing, cloud functions, cloud storage, network services, cloud communication, middleware services, domain name services, security services, CDN (Content Delivery Network), and big data and artificial intelligence platforms. Optionally, the terminal can be a smartphone, tablet, laptop, or desktop computer, but it is not limited to these.

[0033] like Figure 1The diagram shown is a schematic representation of an implementation environment provided by an embodiment of the present invention. (Refer to...) Figure 1 The implementation environment includes at least one terminal 102 and a server 101. The terminal 102 and the server 101 can be connected via a network, either wirelessly or via a wired connection, to complete data transmission and exchange.

[0034] Server 101 can be a standalone physical server, a server cluster or distributed system consisting of multiple physical servers, or a cloud server that provides basic cloud computing services such as cloud services, cloud databases, cloud computing, cloud functions, cloud storage, network services, cloud communication, middleware services, domain name services, security services, CDN (Content Delivery Network), and big data and artificial intelligence platforms.

[0035] Additionally, server 101 can also be a node server in a blockchain network. Blockchain is a novel application model of computer technologies such as distributed data storage, peer-to-peer transmission, consensus mechanisms, and encryption algorithms.

[0036] Terminal 102 can be a smartphone, tablet computer, laptop computer, desktop computer, smart speaker, smartwatch, etc., but is not limited to these. Terminal 102 and server 101 can be directly or indirectly connected via wired or wireless communication, and this embodiment of the invention does not impose any limitations.

[0037] For example, based on Figure 1 The implementation environment shown in this embodiment of the invention provides a UAV path planning method based on meteorological environment. The following description uses the application of this UAV path planning method based on meteorological environment in server 101 as an example. It can be understood that this UAV path planning method based on meteorological environment can also be applied to terminal 102.

[0038] Reference Figure 2 , Figure 2 This is an optional flowchart of the UAV path planning method based on meteorological environment provided in the embodiments of the present invention. The execution subject of the UAV path planning method based on meteorological environment can be any of the aforementioned computer devices (including servers or terminals). Figure 2 The method may include, but is not limited to, steps S100 to S500.

[0039] Step S100: Obtain system parameters and meteorological data for the planned area; The system parameters include task data, multi-source positioning information of the drone swarm, and performance parameters. The task data includes delivery point coordinates and material requirements. Step S200: Based on meteorological data, multi-source positioning information is fused to obtain initial track coordinates; It should be noted that in some embodiments, step S200 may include the following steps: based on meteorological data, the multi-source positioning information is compensated for meteorological interference positioning errors by using a layered atmospheric delay correction model to obtain compensated multi-source positioning information; the compensated multi-source positioning information is fused and denoised by using fuzzy adaptive Kalman filtering to obtain initial track coordinates.

[0040] For example, in some specific implementations, the UAV can be equipped with multi-source positioning sensors, including a GNSS receiver, an inertial measurement unit (IMU), and a barometric altimeter. Simultaneously, the UAV can be configured to receive high-resolution wind field data of the planned area in real time from a ground-based meteorological monitoring network via a 4G / 5G communication link. This data includes three-dimensional wind speed vectors (u, v, w components), temperature, and air pressure, etc. Specifically, this can be achieved through the following process: Step 1: Weather interference positioning error compensation processing; First, the positioning fusion computing unit receives raw multi-source positioning information (such as GNSS pseudorange, carrier phase, IMU acceleration and angular velocity, and barometric altimeter atmospheric pressure data) and high-resolution meteorological data.

[0041] Using the atmospheric temperature and pressure information from the meteorological data, the computing unit initiates a layered atmospheric delay correction model. This model divides the troposphere at the UAV's current location into multiple thin layers based on altitude and humidity variations. Specifically, for each layer, the model accurately calculates the atmospheric refractive index based on real-time meteorological data and uses this to correct the path curvature and propagation delay of the GNSS signal as it passes through each layer. For example, in the low-altitude region after a typhoon, the GNSS signal delay will significantly increase due to the extremely high water vapor content.

[0042] Through the hierarchical model, the embodiments of the present invention can accurately calculate this delay error and deduct it from the original GNSS positioning results, thereby obtaining compensated positioning information that is less affected by meteorological interference, i.e., compensated multi-source positioning information.

[0043] Step 2: Fusion and noise reduction processing; The multi-source positioning information (including corrected GNSS position, IMU data, and barometric altitude data) after compensation in step one is input into the fuzzy adaptive Kalman filter module.

[0044] It should be noted that during UAV flight, the noise characteristics of each sensor change dynamically due to variations in wind turbulence and maneuvers. For example, when the UAV traverses a region of strong turbulence, the noise of the IMU's accelerometer increases dramatically. Using fixed noise parameters with a traditional Kalman filter will lead to distortion in the fusion results.

[0045] In this embodiment of the invention, the fuzzy adaptive Kalman filter module first calculates the statistical information (such as mean and variance) of the observation residuals (i.e., the difference between the actual observed value and the filter's predicted value) in real time. Then, a pre-designed fuzzy logic system uses this statistical information as input and performs inference through preset fuzzy rules (e.g., "if the residual mean is large and the variance is large, then the observation noise is large") to dynamically adjust the noise covariance matrices Q and R of each sensor in the filter. After adaptive adjustment, the filter fuses data from GNSS, IMU, and barometric altimeter in an optimal way, which can fully utilize high-precision GNSS information, smoothly transition to an IMU-based inference mode when the GNSS signal is interfered with, and effectively suppress high-frequency noise from the sensors.

[0046] Step S300: Based on the initial track coordinates, mission data, and performance parameters, an adaptive genetic algorithm is used to perform global mission allocation to obtain access sequence data; The access sequence data includes the access sequence of each delivery point corresponding to each drone in the drone cluster; It should be noted that in some embodiments, step S300 may include the following steps: dividing the delivery point numbers of all delivery points into subsequences of each drone in the drone swarm, and generating an initial population through a random arrangement; wherein, the initial population includes several individuals, each individual corresponding to an arrangement scheme of delivery point numbers in the drone swarm; taking the initial population as the first population; calculating the fitness value of each individual based on the arrangement scheme corresponding to the first population, based on the initial track coordinates and task data, through a preset quantization function; wherein, the fitness value represents the reciprocal of the total delivery cost; taking the maximum fitness value in the first population as the candidate fitness value, and taking the corresponding individual as the candidate individual; performing genetic evolution on the first population based on the fitness value to obtain a second population; wherein, the genetic evolution includes a roulette wheel strategy, crossover operation, and mutation operation, and the crossover probability applied by the crossover operation is based on the larger fitness value of the two individuals to be crossovered and the fitness of the first population. The extreme values ​​and average fitness values ​​are quantized. The mutation probability applied in the mutation operation is based on the fitness value of the individual to be mutated and the extreme values ​​and average fitness values ​​of the first population. A feasibility test is performed on the arrangement scheme of each individual in the second population based on performance parameters. If the feasibility test result is unsuccessful, a repair operation is performed on the corresponding individual to make it pass the feasibility test. The number of iterations is incremented. If the rate of change of the candidate fitness values ​​for a consecutive preset number of iterations is less than the improvement threshold, or if the number of iterations reaches the preset maximum number of iterations, the arrangement scheme corresponding to the candidate individual is used as the access sequence data. Otherwise, the second population is treated as the first population, and the process returns to the step of calculating the fitness value of each individual based on the arrangement scheme corresponding to the first population, based on the initial track coordinates and mission data, using a preset quantization function, until the number of iterations reaches the preset maximum number of iterations, at which point the arrangement scheme corresponding to the candidate individual is used as the access sequence data.

[0047] For example, in some specific implementations, global task allocation based on an adaptive genetic algorithm (GA) (this step aims to assign delivery points to multiple drones and determine the order of access) can be implemented as follows: 1. Task parameter input: Input the coordinates of the delivery point, material requirements, and drone performance parameters (payload, battery level, maximum range, etc.).

[0048] 2. Chromosome Encoding and Population Initialization: A permutation-based encoding method is adopted. One chromosome represents a complete delivery plan. The gene sequence consists of all delivery point numbers, and a special separator is introduced to divide the subsequences that each drone is responsible for. The initial population is generated randomly.

[0049] 3. Fitness Assessment: Calculate the fitness value for each individual (scheme). The fitness function is defined as the reciprocal of the total delivery cost.

[0050] The formula for calculating total delivery cost is:

[0051] in, , and These represent fixed costs ( ), risk cost ( ) and time cost ( The weighting coefficients are used to balance the scale of drone scheduling, flight safety, and emergency delivery efficiency, and can be set according to task preferences. It should be noted that this embodiment of the invention does not limit the specific quantification methods for various costs. In some optional implementations, for example, the delivery distance can be determined by the distance between the initial flight path coordinates and the delivery point coordinates, and then various costs can be determined proportionally according to the data size and a preset mapping table based on the task data.

[0052] 4. Genetic evolution: A roulette wheel strategy is used to select individuals with high fitness to enter the next generation. Crossover probability. and mutation probability Dynamic adjustments are made based on the variance of population fitness and the number of generations to avoid premature convergence. The specific adjustment strategy is as follows:

[0053]

[0054] in, This represents the maximum value of the current population. This represents the average value of the current population. The value of the individual with the larger fitness function value among the two crossover objects. The values ​​for the individuals to be mutated are given. k1, k2, k3, and k4 are constants, typically k1=k3=1.0 and k2=k4=0.5. When the population fitness variance is small (tending to convergence), the crossover and mutation probabilities are increased to increase diversity; when the population fitness variance is large (good diversity), the crossover and mutation probabilities are decreased to protect superior individuals.

[0055] 5. Constraint Embedding: In each generation, a feasibility test is performed on each individual drone. This checks whether the total payload of each drone exceeds limits, whether the estimated range exceeds battery capacity, and whether the mission sequence traverses known areas of sustained high wind speeds. Individuals violating constraints are either repaired or assigned extremely low fitness.

[0056] 6. Iteration and Output: Set the maximum number of iterations. When the iteration limit is reached or the optimal solution remains unimproved, the algorithm terminates and outputs the optimal task allocation scheme, i.e., the specific delivery point visit sequence for each drone.

[0057] Step S400: Based on the access sequence data and meteorological data, the search tree is optimized through risk cost to obtain the initial path planning data; The initial path planning data includes local continuous path planning data between adjacent delivery points for each drone; It should be noted that in some embodiments, step S400 may include the following steps: acquiring building data of the configuration space between adjacent delivery points, extracting wind speed and vertical wind shear intensity of the configuration space from meteorological data; adding the starting point to the search tree as the root node; generating sampling points in the configuration space through uniform random sampling; quantifying the risk cost of the sampling points based on wind speed, vertical wind shear intensity, and building data; wherein, the risk cost represents the weighted sum of the high wind speed risk corresponding to wind speed, the wind shear risk corresponding to vertical wind shear intensity, and the building proximity risk corresponding to building data; normalizing the risk cost and mapping it to the rejection probability of the sampling point; comparing the rejection probability with a random number in a preset interval, and determining the selection type of the sampling point based on the comparison result; wherein, the selection type includes acceptance and rejection; if the selection type is acceptance, adding the sampling point as a node to the search tree; and adding the sampling point... As the first candidate point, the node closest to the first candidate point in the search tree is found as the second candidate point. Starting from the second candidate point, a new node is generated in the direction of the first candidate point based on a preset step size. Based on the wind speed, vertical wind shear intensity, and building data corresponding to the new node, a constraint dynamic detection is performed on the new node. If the constraint dynamic detection result is passed, the new node is added to the search tree. Based on the new node, a candidate path is planned from the search tree by using a fast exploration random tree star schema algorithm with the constraint of minimizing risk cost. If the distance between the new node and the end point is less than the preset step size, the candidate path is used as the local continuous path planning data between adjacent delivery points. Otherwise, the step of generating sampling points in the configuration space by uniform random sampling is returned until the distance between the new node and the end point is less than the preset step size. The initial path planning data is obtained by organizing the local continuous path planning data between all adjacent delivery points corresponding to each drone.

[0058] For example, in some specific implementations, local continuous path planning based on the improved RRT* (Fast Exploratory Random Tree Star Schema) can be implemented as follows: For each drone's access sequence, continuous spatial path planning is performed sequentially for the flight segments between every two adjacent delivery points.

[0059] 1. Parameter input: GA output task sequence, high-resolution wind field data, and multiple constraints.

[0060] 2. Risk Cost Calculation: A uniform random sampling point x is generated in the configuration space. Based on high-resolution wind field data, the comprehensive risk cost R(x) at x is calculated. The risk cost is obtained by weighted summation of the following three components: R(x) = α1·R wind (x)+α2·R shear (x)+α3·R building (x) Among them, R wind (x) represents a high wind speed risk, which is directly proportional to the degree to which the current wind speed exceeds the safe threshold (14 m / s); R shear (x) represents wind shear risk, which is directly proportional to the vertical wind shear intensity (the rate of change of wind speed per unit height, such as greater than 0.5 m / s / m); R building (x) represents the building proximity risk, which is inversely proportional to the distance to the nearest building (the closer the distance, the greater the risk). α1, α2, and α3 are weighting coefficients that can be set according to the task's safety preferences.

[0061] 3. Probabilistic Acceptance Sampling: The calculated risk cost R(x) is normalized and mapped to the probability of a sample point being rejected. For a randomly generated candidate sample point, a random number p in the interval [0,1] is first generated. If p is greater than the normalized risk value of the point, the sample point is accepted; otherwise, it is rejected. Through this mechanism, high-risk points will be selected with a lower probability, thus guiding the search to avoid high-risk areas overall. At the same time, due to the existence of a certain acceptance probability, the algorithm still has the ability to explore necessary high-risk areas, ensuring the completeness of path search under complex constraints.

[0062] 4. Node Expansion and Dynamic Constraint Detection: Find the node closest to x in the search tree and attempt to expand a new node x_new in the x-direction by a step size. During the expansion process and immediately after generating x_new, multiple safety checks are performed: check whether the path segment intersects with the 3D model of the building, check whether the wind speed at the location of x_new exceeds the wind resistance limit of the drone (14m / s), and check whether the vertical wind shear exceeds the safety threshold (0.5 m / s / m). If any condition is violated, x_new is discarded.

[0063] 5. Search Tree Optimization and Path Extraction: Add valid `x_new` nodes to the search tree and perform the path optimization steps of RRT* (check if there are existing nodes near the new node that offer a lower cost and update their parent nodes) to optimize the tree structure and reduce path costs. When the search tree first connects to the target point, extract an initial feasible path. Through continuous iteration, the path asymptotically approaches optimality.

[0064] Step S500: Perform conflict coordination and post-processing on the initial path planning data to obtain the target path planning data; It should be noted that in some embodiments, step S500 may include the following steps: step S510, performing multi-UAV path conflict detection and coordination on the initial path planning data to obtain second path planning data; step S520, performing path post-processing and smoothing output on the second path planning data to obtain target path planning data.

[0065] It should be noted that in some embodiments, step S510 may include the following steps: discretizing the initial path planning data into a time series to obtain a spatiotemporal sequence for each UAV; wherein the spatiotemporal sequence includes several spatiotemporal points; performing spatiotemporal conflict detection between UAVs based on the spatiotemporal points to obtain conflict points; re-performing search tree optimization on the local continuous path planning data corresponding to the conflict points to perform local replanning to eliminate the conflict points, obtaining local continuous path replanning data, and then summarizing to obtain second path planning data; if there are no conflict points, the initial path planning data is used as the second path planning data.

[0066] For example, in some specific implementations, multi-UAV path conflict detection and coordination can be achieved as follows: 1. Spatiotemporal Conflict Detection: After completing the initial path planning for all UAVs, all paths are discretized according to time sequence to form the time-position sequence of each UAV. The system checks for spatiotemporal points where two or more UAVs are expected to enter the same airspace within the same time period, and the three-dimensional spatial distance between them is less than the set minimum safety interval (10 meters). If such a point exists, it is determined to be a path conflict.

[0067] 2. Conflict Resolution and Iterative Replanning: Conflict detection and replanning is an iterative process, with the following specific steps: a. Check for conflicts in all paths. If no conflicts are found, proceed to the next step.

[0068] b. If a conflict exists, select one of the conflicts and perform local replanning for one or more UAVs involved (the principle is the same as step S300, and will not be repeated here). During replanning, additional constraints or costs can be introduced near the conflict area to guide new paths to avoid that spatiotemporal region.

[0069] c. Repeat conflict detection until all conflicts are resolved, or the maximum number of iterations is reached (e.g., 10 times to avoid infinite loops). Since replanning may change the original path, it may introduce new conflicts, which can be gradually eliminated through iterative detection.

[0070] It should be noted that in some embodiments, step S520 may include at least one of the following steps: performing path pruning on the second path planning data by deleting redundant intermediate nodes; and performing trajectory smoothing on the second path planning data using cubic B-spline curves.

[0071] For example, in some specific implementations, path post-processing and smoothing output can be implemented as follows: Post-processing optimization is performed on each drone path for which conflicts have been resolved: 1. Path pruning: Traverse the path nodes and delete redundant intermediate nodes that do not change obstacle avoidance and safety properties to shorten the path length.

[0072] 2. Smooth trajectory: Cubic B-spline curves are used to fit the pruned path nodes to generate a continuous flight trajectory with smooth curvature changes, ensuring that it conforms to the dynamic constraints of the UAV.

[0073] 3. Final output: Output a complete, smoothed 3D flight path that meets all weather, environmental, maneuver, and cooperative constraints, completing the UAV delivery mission planning.

[0074] To explain in detail the principle of the technical solution of the present invention, the overall process of the present invention will be described below with reference to some specific embodiments. It is easy to understand that the following is an explanation of the technical principle of the present invention and should not be regarded as a limitation of the present invention.

[0075] First, it should be noted that the existing technology has the following main drawbacks: (1) Insufficient integration of high-resolution wind field data: Existing methods mostly use low-resolution or steady-state wind fields, which cannot reflect the real spatiotemporal changes of urban wind fields after typhoons.

[0076] (2) Single planning stage and lack of coordination: Most existing solutions are single-stage planning and fail to effectively combine global task allocation with local continuous path optimization.

[0077] (3) Incomplete consideration of constraints: Most methods only consider obstacle avoidance and distance, without systematically integrating multiple safety and operation constraints such as wind shear, high wind speed, power consumption, and load.

[0078] (4) Insufficient risk quantification: Existing research has relatively simple modeling of flight risks and has not established a comprehensive risk cost function to guide the generation of safe paths.

[0079] (5) Poor adaptability: Traditional algorithms are prone to getting stuck in local optima in dynamic wind fields, resulting in poor path smoothness and actual flyability.

[0080] Therefore, the object of the present invention is: This paper presents a two-stage UAV path planning method and system based on high-resolution wind field and multiple constraints. Through a collaborative architecture of "global task allocation + local continuous planning", it integrates high-precision wind field data and multiple risk constraints to achieve safe, efficient and adaptive flight of UAVs in complex weather environments.

[0081] like Figure 3 As shown in the figure, the UAV path planning method based on meteorological environment provided in this embodiment of the invention can be implemented through the following process: Step 1: System initialization and environment data input; 1. Real-time data such as three-dimensional wind speed vector, temperature, humidity, and horizontal pressure gradient field within the planned area are collected through airborne or ground-based meteorological sensing units.

[0082] 2. The positioning fusion computing unit uses a hierarchical atmospheric delay correction model and fuzzy adaptive Kalman filtering to fuse multi-source positioning information and output high-precision track coordinates.

[0083] Step 2: Global task allocation based on adaptive genetic algorithm (GA); This step aims to assign delivery points to multiple drones and determine the order of access.

[0084] 1. Task parameter input: Input the coordinates of the delivery point, material requirements, and drone performance parameters (payload, battery level, maximum range, etc.).

[0085] 2. Chromosome Encoding and Population Initialization: A permutation-based encoding method is adopted. One chromosome represents a complete delivery plan. The gene sequence consists of all delivery point numbers, and a special separator is introduced to divide the subsequences that each drone is responsible for. The initial population is generated randomly.

[0086] 3. Fitness Assessment: Calculate the fitness value for each individual (scheme). The fitness function is defined as the reciprocal of the total delivery cost.

[0087] The formula for calculating total delivery cost is:

[0088] in, , and These represent the weighting coefficients for fixed costs, risk costs, and time costs, respectively, used to balance the scale of drone dispatch, flight safety, and emergency delivery efficiency. They can be set according to mission preferences.

[0089] 4. Genetic evolution: A roulette wheel strategy is used to select individuals with high fitness to enter the next generation. Crossover probability. and mutation probability Dynamic adjustments are made based on the variance of population fitness and the number of generations to avoid premature convergence. The specific adjustment strategy is as follows:

[0090]

[0091] in, This represents the maximum value of the current population. This represents the average value of the current population. The value of the individual with the larger fitness function value among the two crossover objects. The values ​​for the individuals to be mutated are given. k1, k2, k3, and k4 are constants, typically k1=k3=1.0 and k2=k4=0.5. When the population fitness variance is small (tending to convergence), the crossover and mutation probabilities are increased to increase diversity; when the population fitness variance is large (good diversity), the crossover and mutation probabilities are decreased to protect superior individuals.

[0092] 5. Constraint Embedding: In each generation, a feasibility test is performed on each individual drone. This checks whether the total payload of each drone exceeds limits, whether the estimated range exceeds battery capacity, and whether the mission sequence traverses known areas of sustained high wind speeds. Individuals violating constraints are either repaired or assigned extremely low fitness.

[0093] 6. Iteration and Output: Set the maximum number of iterations. When the iteration limit is reached or the optimal solution remains unimproved, the algorithm terminates and outputs the optimal task allocation scheme, i.e., the specific delivery point visit sequence for each drone.

[0094] Step 3: Local continuous path planning based on improved RRT*; For each drone's access sequence output in step two, continuous spatial path planning is performed sequentially for the flight segments between every two adjacent delivery points.

[0095] 1. Parameter input: GA output task sequence, high-resolution wind field data, and multiple constraints.

[0096] 2. Risk Cost Calculation: A uniform random sampling point x is generated in the configuration space. Based on high-resolution wind field data, the comprehensive risk cost R(x) at x is calculated. The risk cost is obtained by weighted summation of the following three components: R(x) = α1·R wind (x)+α2·R shear (x)+α3·R building (x) Among them, R wind (x) represents a high wind speed risk, which is directly proportional to the degree to which the current wind speed exceeds the safe threshold (14 m / s); R shear(x) represents wind shear risk, which is directly proportional to the vertical wind shear intensity (the rate of change of wind speed per unit height, such as greater than 0.5 m / s / m); R building (x) represents the building proximity risk, which is inversely proportional to the distance to the nearest building (the closer the distance, the greater the risk). α1, α2, and α3 are weighting coefficients that can be set according to the task's safety preferences.

[0097] 3. Probabilistic Acceptance Sampling: The calculated risk cost R(x) is normalized and mapped to the probability of a sample point being rejected. For a randomly generated candidate sample point, a random number p in the interval [0,1] is first generated. If p is greater than the normalized risk value of the point, the sample point is accepted; otherwise, it is rejected. Through this mechanism, high-risk points will be selected with a lower probability, thus guiding the search to avoid high-risk areas overall. At the same time, due to the existence of a certain acceptance probability, the algorithm still has the ability to explore necessary high-risk areas, ensuring the completeness of path search under complex constraints.

[0098] 4. Node Expansion and Dynamic Constraint Detection: Find the node closest to x in the search tree and attempt to expand a new node x_new in the x-direction by a step size. During the expansion process and immediately after generating x_new, multiple safety checks are performed: check whether the path segment intersects with the 3D model of the building, check whether the wind speed at the location of x_new exceeds the wind resistance limit of the drone (14m / s), and check whether the vertical wind shear exceeds the safety threshold (0.5 m / s / m). If any condition is violated, x_new is discarded.

[0099] 5. Search Tree Optimization and Path Extraction: Add valid `x_new` nodes to the search tree and perform the path optimization steps of RRT* (check if there are existing nodes near the new node that offer a lower cost and update their parent nodes) to optimize the tree structure and reduce path costs. When the search tree first connects to the target point, extract an initial feasible path. Through continuous iteration, the path asymptotically approaches optimality.

[0100] Step 4: Detection and coordination of path conflicts among multiple UAVs; 1. Spatiotemporal Conflict Detection: After completing the initial path planning for all UAVs, all paths are discretized according to time sequence to form the time-position sequence of each UAV. The system checks for spatiotemporal points where two or more UAVs are expected to enter the same airspace within the same time period, and the three-dimensional spatial distance between them is less than the set minimum safety interval (10 meters). If such a point exists, it is determined to be a path conflict.

[0101] 2. Conflict Resolution and Iterative Replanning: Conflict detection and replanning is an iterative process, with the following specific steps: a. Check for conflicts in all paths. If no conflicts are found, proceed to step five.

[0102] b. If a conflict exists, select one of the conflicts and perform local replanning for one or more drones involved (return to step three). During replanning, introduce additional constraints or costs near the conflict area to guide a new path to avoid that spatiotemporal region.

[0103] c. Repeat conflict detection until all conflicts are resolved, or the maximum number of iterations is reached (e.g., 10 times to avoid infinite loops). Since replanning may change the original path, it may introduce new conflicts, which can be gradually eliminated through iterative detection.

[0104] Step 5: Post-processing of the path and smoothing of the output; Post-processing optimization is performed on each drone path for which conflicts have been resolved: 1. Path pruning: Traverse the path nodes and delete redundant intermediate nodes that do not change obstacle avoidance and safety properties to shorten the path length.

[0105] 2. Smooth trajectory: Cubic B-spline curves are used to fit the pruned path nodes to generate a continuous flight trajectory with smooth curvature changes, ensuring that it conforms to the dynamic constraints of the UAV.

[0106] 3. Final output: Output a complete, smoothed 3D flight path that meets all weather, environmental, maneuver, and cooperative constraints, completing the UAV delivery mission planning.

[0107] In specific application scenarios, taking the emergency logistics scenario in a city after a typhoon as an example, the Fengyi Ark ARK40 UAV was used for testing in a simulation environment built with MATLAB. The system successfully planned a multi-UAV cooperative path that satisfies all constraints. Compared with the traditional RRT* method, the risk cost was reduced by about 25%, and the task completion time was shortened by about 18%.

[0108] In some alternative implementations, GA can be replaced by Particle Swarm Optimization (PSO) or Ant Colony Optimization (ACO) for global task allocation; RRT can be replaced by Informed RRT or PRM* for local path search; the risk model can be further incorporating machine learning to predict risks, such as risk map generation based on historical data; the smoothing algorithm can be replaced by Bézier curves or polynomial interpolation; and the positioning unit can use an RTK-GNSS and inertial navigation fusion scheme to replace pure filtering positioning.

[0109] In summary, this invention, through a two-stage planning architecture of "global optimization + local refinement," effectively overcomes the problem of poor planning performance in complex wind fields caused by traditional single-stage methods. Specific advantages include: (1) Enhanced safety: The system integrates multiple constraints such as high wind speed, wind shear, and building risk, and establishes a risk cost function to guide planning.

[0110] (2) Better adaptability: Adaptive GA and risk perception RRT* are adopted, which can dynamically respond to wind field changes and mission requirements.

[0111] (3) Higher efficiency: By coordinating task allocation and path search, invalid searches are reduced, and planning speed and path quality are improved.

[0112] (4) Better scalability: It supports collaborative planning of multiple UAVs and can easily access actual meteorological data and flight control systems.

[0113] Compared with the prior art, the embodiments of the present invention have at least the following beneficial effects: (1) Path security is significantly improved and risk exposure is reduced; (2) High planning efficiency, adaptable to dynamic wind field changes; (3) The system has strong scalability and is suitable for multi-machine collaboration and complex task scenarios.

[0114] like Figure 4 As shown, this embodiment of the invention also provides a UAV path planning device 900 based on meteorological environment, which can implement the above-described method. This device may include: The first module 910 is used to acquire system parameters and meteorological data of the planned area; among them, the system parameters include task data, multi-source positioning information and performance parameters of the UAV cluster, and the task data includes delivery point coordinates and material requirements; The second module 920 is used to fuse multi-source positioning information based on meteorological data to obtain the initial track coordinates; The third module 930 is used to perform global task allocation based on the initial track coordinates, task data and performance parameters using an adaptive genetic algorithm to obtain access sequence data; wherein, the access sequence data includes the access sequence of each delivery point corresponding to each drone in the drone cluster; The fourth module 940 is used to perform search tree optimization based on access sequence data and meteorological data through risk cost to obtain initial path planning data; wherein, the initial path planning data includes local continuous path planning data between adjacent delivery points for each drone; The fifth module 950 is used to perform conflict coordination and post-processing on the initial path planning data to obtain the target path planning data.

[0115] It is understood that the content of the above method embodiments is applicable to the present device embodiments. The specific functions implemented by the present device embodiments are the same as those of the above method embodiments, and the beneficial effects achieved are also the same as those achieved by the above method embodiments.

[0116] This invention also provides an electronic device, which includes a memory and a processor. The memory stores a computer program, and the processor executes the computer program to implement the method described above. This electronic device can be any smart terminal, including tablet computers, in-vehicle computers, etc.

[0117] It is understood that the content of the above method embodiments is applicable to this device embodiment. The specific functions implemented by this device embodiment are the same as those of the above method embodiments, and the beneficial effects achieved are also the same as those achieved by the above method embodiments.

[0118] like Figure 5 As shown, Figure 5 The hardware structure of an electronic device 1000 according to another embodiment is illustrated. The electronic device 1000 includes: The processor 1001 can be implemented using a general-purpose CPU (Central Processing Unit), microprocessor, application-specific integrated circuit (aSIC), or one or more integrated circuits, and is used to execute relevant programs to implement the technical solutions provided in the embodiments of the present invention. The memory 1002 can be implemented as a read-only memory (ROM), a static storage device, a dynamic storage device, or a random access memory (RaM). The memory 1002 can store the operating system and other application programs. When the technical solutions provided in the embodiments of this specification are implemented through software or firmware, the relevant program code is stored in the memory 1002 and is called and executed by the processor 1001. Input / output interface 1003 is used to implement information input and output; The communication interface 1004 is used to enable communication and interaction between this device and other devices. Communication can be achieved through wired means (such as USB, network cable, etc.) or wireless means (such as mobile network, WIFI, Bluetooth, etc.). Bus 1005 transmits information between various components of the device (e.g., processor 1001, memory 1002, input / output interface 1003, and communication interface 1004); The processor 1001, memory 1002, input / output interface 1003 and communication interface 1004 are connected to each other within the device via bus 1005.

[0119] The electronic device embodiments described above are merely illustrative. The units described as separate components may or may not be physically separate; that is, they may be located in one place or distributed across multiple network units. Some or all of the modules can be selected to achieve the purpose of this embodiment according to actual needs.

[0120] This invention also provides a computer-readable storage medium storing a computer program that, when executed by a processor, implements the above-described method.

[0121] It is understood that the content of the above method embodiments is applicable to this storage medium embodiment. The specific functions implemented in this storage medium embodiment are the same as those in the above method embodiments, and the beneficial effects achieved are also the same as those achieved in the above method embodiments.

[0122] This invention also provides a computer program product, including a computer program that, when executed by a processor, implements the above-described method.

[0123] It is understood that the content of the above method embodiments is applicable to the embodiments of this program product. The specific functions implemented by the embodiments of this program product are the same as those of the above method embodiments, and the beneficial effects achieved are also the same as those achieved by the above method embodiments.

[0124] Memory, as a non-transitory computer-readable storage medium, can be used to store non-transitory software programs and non-transitory computer-executable programs. Furthermore, memory may include high-speed random access memory, and may also include non-transitory memory, such as at least one disk storage device, flash memory device, or other non-transitory solid-state storage device. In some embodiments, memory may optionally include memory remotely located relative to the processor, and these remote memories can be connected to the processor via a network. Examples of such networks include, but are not limited to, the Internet, intranets, local area networks, mobile communication networks, and combinations thereof.

[0125] The present invention provides a method, apparatus, electronic device, storage medium, and program product for UAV path planning based on meteorological environment. This method acquires system parameters and meteorological data of the planning area. The system parameters include task data, multi-source positioning information of the UAV swarm, and performance parameters. The task data includes delivery point coordinates and material requirements. Based on the meteorological data, the multi-source positioning information is fused to obtain initial trajectory coordinates. Based on the initial trajectory coordinates, task data, and performance parameters, an adaptive genetic algorithm is used for global task allocation to obtain access sequence data. The access sequence data includes the access sequence of each UAV in the UAV swarm corresponding to a delivery point. Based on the access sequence data and meteorological data, a search tree is optimized using risk cost to obtain initial path planning data. The initial path planning data includes local continuous path planning data between adjacent delivery points for each UAV. The initial path planning data undergoes conflict coordination and post-processing to obtain target path planning data. This invention utilizes fusion meteorological data for positioning compensation and path planning, and employs risk-cost-guided search tree optimization to effectively reduce the flight risks of UAVs in adverse weather conditions. Specifically, this invention employs a two-stage collaborative architecture of "global task allocation followed by local path planning." First, an adaptive genetic algorithm optimizes multi-aircraft task allocation and access order from a global perspective. Then, a refined local path search is performed for each flight segment. This collaborative mechanism overcomes the tendency of single-stage planning to fall into local optima, ensuring both global task efficiency and the local feasibility of each flight segment. Furthermore, by acquiring and processing meteorological data and performing planning based on this data, this invention enables the UAV system to dynamically respond to complex and changing weather environments. Subsequent conflict coordination steps further ensure the spatiotemporal safety of multi-aircraft collaborative operations, significantly improving the system's adaptability and robustness in complex environments.

[0126] The embodiments described in this invention are for the purpose of more clearly illustrating the technical solutions of the embodiments of this invention, and do not constitute a limitation on the technical solutions provided by the embodiments of this invention. As those skilled in the art will know, with the evolution of technology and the emergence of new application scenarios, the technical solutions provided by the embodiments of this invention are also applicable to similar technical problems.

[0127] Those skilled in the art will understand that the technical solutions shown in the figures do not constitute a limitation on the embodiments of the present invention, and may include more or fewer steps than shown, or combine certain steps, or different steps.

[0128] The device embodiments described above are merely illustrative. The units described as separate components may or may not be physically separate; that is, they may be located in one place or distributed across multiple network units. Some or all of the modules can be selected to achieve the purpose of this embodiment according to actual needs.

[0129] Those skilled in the art will understand that all or some of the steps in the methods disclosed above, as well as the functional modules / units in the systems and devices, can be implemented as software, firmware, hardware, or suitable combinations thereof.

[0130] The preferred embodiments of the present invention have been described above with reference to the accompanying drawings, but this does not limit the scope of the claims of the present invention. Any modifications, equivalent substitutions, and improvements made by those skilled in the art without departing from the scope and spirit of the present invention should be within the scope of the claims of the present invention.

Claims

1. A method for unmanned aerial vehicle (UAV) path planning based on meteorological environment, characterized in that, The method includes the following steps: Acquire system parameters and meteorological data for the planned area; wherein, the system parameters include task data, multi-source positioning information and performance parameters of the UAV cluster, and the task data includes delivery point coordinates and material requirements; The initial track coordinates are obtained by fusing the multi-source positioning information based on the meteorological data. Based on the initial track coordinates, the task data, and the performance parameters, an adaptive genetic algorithm is used to perform global task allocation to obtain access sequence data; wherein, the access sequence data includes the access sequence of each delivery point corresponding to each drone in the drone cluster; Based on the access sequence data and the meteorological data, search tree optimization is performed through risk cost to obtain initial path planning data; wherein, the initial path planning data includes local continuous path planning data between adjacent delivery points for each drone; The initial path planning data is then subjected to conflict coordination and post-processing to obtain the target path planning data.

2. The method according to claim 1, characterized in that, The process of fusing the multi-source positioning information based on the meteorological data to obtain the initial track coordinates includes the following steps: Based on the meteorological data, the multi-source positioning information is compensated for meteorological interference positioning errors by using a layered atmospheric delay correction model to obtain compensated multi-source positioning information. The initial track coordinates are obtained by fusing and denoising the compensated multi-source positioning information using fuzzy adaptive Kalman filtering.

3. The method according to claim 1, characterized in that, The step of using an adaptive genetic algorithm to perform global task allocation based on the initial track coordinates, the task data, and the performance parameters to obtain access sequence data includes the following steps: The delivery point numbers of all the delivery points are divided into subsequences of each drone in the drone cluster, and an initial population is generated by random arrangement; wherein, the initial population includes several individuals, and each individual corresponds to an arrangement scheme of the delivery point numbers of the drone cluster; The initial population is taken as the first population; According to the arrangement scheme corresponding to the first population, based on the initial track coordinates and the task data, the fitness value of each individual is calculated by a preset quantization function; wherein, the fitness value represents the reciprocal of the total delivery cost; The maximum fitness value in the first population is used as the candidate fitness value, and the corresponding individual is used as the candidate individual; Based on the fitness value, the first population is subjected to genetic evolution to obtain the second population; The genetic evolution includes a roulette wheel strategy, a crossover operation, and a mutation operation. The crossover probability applied in the crossover operation is quantified based on the larger fitness value of the two individuals to be crossed, as well as the extreme fitness value and the average fitness value of the first population. The mutation probability applied in the mutation operation is quantified based on the fitness value of the individual to be mutated, as well as the extreme fitness value and the average fitness value of the first population. Based on the performance parameters, the arrangement scheme of each individual in the second population is tested for feasibility. If the result of the feasibility test is not passed, the corresponding individual is repaired so that it passes the feasibility test. Increment the number of iterations; If the rate of change of the candidate fitness value is less than the improvement threshold after a preset number of consecutive iterations, or if the number of iterations reaches a preset maximum number of iterations, the orchestration scheme corresponding to the candidate individual is used as the access sequence data; otherwise... Using the second population as the first population, return to the step of calculating the fitness value of each individual based on the initial track coordinates and the task data using a preset quantization function according to the orchestration scheme corresponding to the first population, until the number of iterations reaches the preset maximum number of iterations, and use the orchestration scheme corresponding to the candidate individual as the access sequence data.

4. The method according to claim 1, characterized in that, The adjacent delivery points include a start point and an end point. The initial path planning data is obtained by performing search tree optimization based on the access sequence data and the meteorological data, using risk cost, and includes the following steps: Obtain building data of the configuration space between adjacent delivery points, and extract the wind speed and vertical wind shear intensity of the configuration space from the meteorological data; Add the starting point to the search tree as the root node; Sampling points are generated in the configuration space through uniform random sampling; Based on the wind speed, the vertical wind shear intensity, and the building data, the risk cost of the sampling point is quantified; wherein, the risk cost represents the weighted sum of the high wind speed risk corresponding to the wind speed, the wind shear risk corresponding to the vertical wind shear intensity, and the building proximity risk corresponding to the building data; The risk cost is normalized and mapped to the rejection probability of the sampling point; The selection type of the sampling point is determined by comparing a random number within a preset range with the rejection probability; wherein the selection type includes acceptance and rejection. If the selection type is "accept", the sampling point is added as a node to the search tree; The sampling point is used as the first candidate point, and the node closest to the first candidate point in the search tree is used as the second candidate point. Based on a preset step size, a new node is generated from the second candidate point in the direction of the first candidate point; Based on the wind speed, vertical wind shear intensity, and building data corresponding to the new node, constrained dynamic detection is performed on the new node. If the result of the constraint dynamic detection is passed, the new node is added to the search tree; Based on the new node, candidate paths are planned from the search tree by using a fast-exploration random tree star schema algorithm with the constraint of minimizing risk cost; If the distance between the new node and the end point is less than the preset step size, the candidate path is used as local continuous path planning data between adjacent delivery points; otherwise, the process returns to the step of generating sampling points in the configuration space by uniform random sampling until the distance between the new node and the end point is less than the preset step size. The initial path planning data is obtained by organizing the local continuous path planning data between all adjacent delivery points corresponding to each drone.

5. The method according to claim 1, characterized in that, The process of resolving conflicts and post-processing the initial path planning data to obtain the target path planning data includes the following steps: The initial path planning data is subjected to multi-UAV path conflict detection and coordination to obtain the second path planning data; The second path planning data is post-processed and smoothed to obtain the target path planning data.

6. The method according to claim 5, characterized in that, The process of performing multi-UAV path conflict detection and coordination on the initial path planning data to obtain the second path planning data includes the following steps: The initial path planning data is discretized over time to obtain a spatiotemporal sequence for each UAV; wherein the spatiotemporal sequence includes several spatiotemporal points; Based on the aforementioned spatiotemporal points, spatiotemporal conflict detection is performed between the UAVs to obtain the conflict points; The search tree optimization is re-executed on the local continuous path planning data corresponding to the conflict point to perform local replanning in order to eliminate the conflict point, thereby obtaining local continuous path replanning data, which is then summarized to obtain the second path planning data. If the conflict point is empty, the initial path planning data will be used as the second path planning data.

7. The method according to claim 5, characterized in that, The step of performing path post-processing and smoothing output on the second path planning data to obtain the target path planning data includes at least one of the following steps: The second path planning data is pruned by removing redundant intermediate nodes. The second path planning data is smoothed using cubic B-spline curves.

8. A UAV path planning device based on meteorological environment, characterized in that, The device includes: The first module is used to acquire system parameters and meteorological data of the planned area; wherein, the system parameters include task data, multi-source positioning information and performance parameters of the UAV cluster, and the task data includes delivery point coordinates and material requirements; The second module is used to fuse the multi-source positioning information based on the meteorological data to obtain the initial track coordinates; The third module is used to perform global task allocation using an adaptive genetic algorithm based on the initial track coordinates, the task data, and the performance parameters to obtain access sequence data; wherein, the access sequence data includes the access sequence of each delivery point corresponding to each drone in the drone cluster; The fourth module is used to perform search tree optimization based on the access sequence data and the meteorological data, through risk cost, to obtain initial path planning data; wherein, the initial path planning data includes local continuous path planning data between adjacent delivery points for each UAV; The fifth module is used to perform conflict coordination and post-processing on the initial path planning data to obtain the target path planning data.

9. An electronic device, characterized in that, The electronic device includes a memory and a processor, the memory storing a computer program, and the processor executing the computer program to perform the method according to any one of claims 1 to 7.

10. A computer program product, characterized in that, The computer program product includes a computer program that, when executed by a processor, implements the method according to any one of claims 1 to 7.

Images