Multi-path optimization method for low-altitude unmanned aerial vehicle delivery system

Abstract

The invention discloses a multi-path optimization method for a low-altitude unmanned aerial vehicle delivery system, and relates to the technical field of unmanned aerial vehicle logistics, and the method comprises the steps: obtaining N delivery unmanned aerial vehicles, performance parameters of the N unmanned aerial vehicles and delivery tasks of the N unmanned aerial vehicles through the low-altitude unmanned aerial vehicle delivery system, and collecting and obtaining flight data of the N unmanned aerial vehicles through a sensor group; carrying out three-dimensional modeling, and building an unmanned aerial vehicle delivery space model; constructing an unmanned aerial vehicle delivery cost function to obtain N unmanned aerial vehicle delivery paths; and monitoring in real time to obtain unmanned aerial vehicle state data and environment data, performing regulation and control optimization on the delivery paths of the N unmanned aerial vehicles, and determining the optimized delivery paths of the N unmanned aerial vehicles. The technical problems of low path planning efficiency and insufficient safety of low-altitude unmanned aerial vehicle delivery in the prior art are solved, and the technical effects of realizing multi-path efficient optimization of the low-altitude unmanned aerial vehicle delivery system and ensuring path safety are achieved.

Description

Technical Field

[0001] The present invention relates to the field of drone logistics technology, and in particular to a multi-path optimization method for a low-altitude drone delivery system. Background Art

[0002] In the field of low-altitude drone delivery, existing technologies face many challenges in multi-path planning. For example, it is difficult to efficiently integrate drone performance parameters, delivery mission requirements and real-time environmental data, resulting in path planning that does not fully consider the comprehensive optimization of cost factors such as energy consumption, time, and maintenance. It also lacks the ability to dynamically adapt to changes in drone status and environmental risks, which can easily lead to path conflicts, inefficiency or safety hazards.

[0003] Existing technologies for low-altitude drone delivery have technical problems such as low path planning efficiency and insufficient safety. Summary of the Invention

[0004] This application provides a multi-path optimization method for a low-altitude drone delivery system, which is used to solve the technical problems of low-altitude drone delivery in the existing technology, such as low path planning efficiency and insufficient safety.

[0005] In view of the above problems, this application provides a multi-path optimization method for a low-altitude drone delivery system.

[0006] The first aspect of the present application provides a multi-path optimization method for a low-altitude drone delivery system, the method comprising:

[0007] N delivery drones, N drone performance parameters, and N drone delivery tasks are obtained through a low-altitude drone delivery system, and a sensor group is installed on the N delivery drones to collect and obtain N drone flight data through the sensor group; three-dimensional modeling is performed based on delivery map information and the N drone flight data to build a drone delivery space model; a drone delivery cost function is constructed according to the drone delivery target, and multi-path planning is performed within the drone delivery space model based on the N drone performance parameters and the N drone delivery tasks using the drone delivery cost function to obtain N drone delivery paths; drone status data and environmental data are monitored and obtained in real time, and the N drone delivery paths are regulated and optimized based on the drone status data and environmental data to determine N optimized drone delivery paths.

[0008] One or more technical solutions provided in this application have at least the following technical effects or advantages:

[0009] A low-altitude drone delivery system acquires N delivery drones, N drone performance parameters, and N drone delivery tasks. Sensor groups are installed on the N delivery drones to collect flight data from the N drones. Three-dimensional modeling is performed based on the delivery map information and the flight data of the N drones to construct a drone delivery space model. A drone delivery cost function is constructed based on the drone delivery objectives. Multi-path planning is performed within the drone delivery space model using the drone delivery cost function based on the N drone performance parameters and the N drone delivery tasks to obtain N drone delivery paths. Real-time monitoring is performed to acquire drone status and environmental data. Based on the drone status and environmental data, the N drone delivery paths are regulated and optimized to determine the optimal delivery paths for the N drones. This achieves the technical effect of achieving efficient multi-path optimization of the low-altitude drone delivery system and ensuring path safety. BRIEF DESCRIPTION OF THE DRAWINGS

[0010] In order to more clearly illustrate the technical solutions in the embodiments of the present invention, the following briefly introduces the drawings required for use in the description of the embodiments. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without creative work.

[0011] Figure 1 A flowchart of a multi-path optimization method for a low-altitude drone delivery system provided in an embodiment of the present application;

[0012] Figure 2 A schematic diagram of the process of building a drone delivery space model in the multi-path optimization method of the low-altitude drone delivery system provided in an embodiment of the present application. DETAILED DESCRIPTION

[0013] This application provides multi-path optimization for low-altitude drone delivery systems to address the technical problems of low-altitude drone delivery in the existing technology, such as low path planning efficiency and insufficient safety.

[0014] The following will be combined with the accompanying drawings in the embodiments of this application to clearly and completely describe the technical solutions in the embodiments of this application. Obviously, the embodiments described are only part of the embodiments of this application, not all of them. Based on the embodiments in this application, all other embodiments obtained by ordinary technicians in this field without making any creative work are within the scope of protection of this application.

[0015] Examples, such as Figure 1 As shown, the present application provides a multi-path optimization method for a low-altitude drone delivery system, the method comprising:

[0016] Step S100: Obtain N delivery drones, N drone performance parameters, and N drone delivery tasks through a low-altitude drone delivery system, and equip the N delivery drones with a sensor group to collect and obtain N drone flight data through the sensor group.

[0017] Specifically, the low-altitude drone delivery system first acquires N delivery drones and their corresponding performance parameters, including basic attributes such as range, payload capacity, and maximum flight speed. It also acquires information on N delivery missions, covering key elements such as the mission's starting point, destination coordinates, cargo weight, and timeliness requirements. Subsequently, each drone is equipped with a sensor suite consisting of inertial navigation sensors, visual sensors, and radar. This suite collects multidimensional flight data in real time, including the drone's position, flight altitude, attitude angle, ambient wind speed, and obstacle information. This data is then initially synchronized according to the collection sequence, providing raw data input for the subsequent construction of a dynamic delivery space model and path planning.

[0018] Step S200: Perform three-dimensional modeling based on the delivery map information and the flight data of the N drones to build a drone delivery space model.

[0019] Specifically, N drone flight data sets are first preprocessed. Timestamp alignment is used to synchronize multi-source data based on acquisition timing information. Kalman filtering and other algorithms are then used to reduce noise on the raw data, resulting in N standardized drone flight data sets. Model initialization is then performed based on delivery map information (including static geographic features such as topography, building distribution, and no-fly zone boundaries), constructing a basic 3D delivery model containing 3D geographic coordinates. Next, the standardized flight data is mapped into the basic model. By comparing and analyzing the spatial relationships between flight trajectories and static model features, delivery space update data (such as dynamic obstacle locations and real-time airflow distribution areas) is extracted. These updated data are then attributed to clearly define the types of spatial entities (such as temporary no-fly zones and high-altitude obstacles), their location coordinates, and their dimensions. Finally, based on this attributed delivery space attribute data, the basic 3D delivery model is dynamically updated to form a drone delivery space model that integrates static geographic information and real-time flight data, providing accurate 3D spatial scenario support for subsequent multi-path planning.

[0020] Step S300: Constructing a drone delivery cost function according to the drone delivery target, and performing multi-path planning in the drone delivery space model using the drone delivery cost function based on the N drone performance parameters and the N drone delivery tasks to obtain N drone delivery paths.

[0021] Specifically, the method first extracts a set of delivery cost elements based on the dimensions of energy, time, and maintenance, targeting the delivery objective (e.g., minimizing total cost, minimizing delivery time, or comprehensive optimization). Using the analytic hierarchy process (AHP), weights are assigned to each element (e.g., 40% for energy, 50% for time, and 10% for maintenance). This results in a cost function: f(path) = w1 × E + w2 × T + w3 × M, where E is the energy cost, T is the flight time cost, and M is the maintenance cost. Next, combining the performance constraints of the N drones (e.g., maximum range, payload limit, and flight altitude range) with the delivery mission constraints (e.g., order timeliness requirements, cargo weight thresholds, and no-fly zone avoidance rules), an intelligent optimization algorithm, such as a genetic algorithm or an ant colony algorithm, is used within the constructed drone delivery spatial model to perform multi-path optimization within the feasible flight area for each drone. By iteratively calculating the cost function values ​​of different paths, the optimal path that meets the constraints is screened out, and finally a delivery path including flight trajectory, turning point coordinates, and speed planning is generated for N drones, realizing the initial path allocation of multi-drone tasks.

[0022] Step S400: Real-time monitoring and acquisition of drone status data and environmental data, and optimization and control of the N drone delivery paths based on the drone status data and environmental data to determine the optimized delivery paths of the N drones.

[0023] Specifically, the system uses a suite of drone-mounted sensors and external environmental monitoring equipment to collect real-time drone status data (such as battery charge, motor temperature, flight attitude deviation, and navigation signal strength) and environmental data (such as real-time wind speed, precipitation probability, air traffic flow, and sudden no-fly orders). First, anomaly detection is performed on both types of data. Anomaly parameters (such as battery charge below a critical value or attitude angle exceeding a limit) and environmental risk parameters (such as strong wind areas or coordinates of temporary no-fly zones) are identified using an isolation forest algorithm or a threshold discriminant method. Subsequently, the original delivery routes of N drones are dynamically adjusted based on these risk parameters. For drones affected by anomalies, their paths are prioritized to find nearby safe landing points or switch to backup power modes. For drones encountering environmental risks, a path replanning algorithm (such as the Dijkstra algorithm) is used to generate detours within the feasible flight space to avoid high-risk areas. During this control process, dynamic obstacle information is updated in real time within the drone delivery space model, and a multi-drone collaborative mechanism is employed to mitigate the risk of collisions caused by path adjustments. Ultimately, an optimized delivery path is generated for each drone that adapts to the real-time status and environment to ensure the safety and timeliness of the delivery mission.

[0024] In one possible implementation, Figure 2 As shown, step S200 also includes:

[0025] Step S210: Synchronously align and filter the N UAV flight data according to the acquisition timing information to obtain N standard UAV flight data.

[0026] Step S220: Initialize the model based on the delivery map information and construct a basic delivery 3D model.

[0027] Step S230: Map the N standard drone flight data to the basic delivery three-dimensional model for updating to obtain a drone delivery space model.

[0028] Specifically, the N collected UAV flight data are synchronized and aligned according to the timing information of data collection (such as timestamp) to ensure the consistency of multi-source data in the time dimension. Then, Kalman filtering, median filtering and other algorithms are used to reduce the noise of the original data to eliminate interference signals such as sudden changes in flight attitude and abnormal fluctuations, and obtain standardized N standard UAV flight data, providing a clean data source for subsequent modeling.

[0029] Based on the delivery map information (including static geographic elements such as terrain, building distribution, fixed no-fly zone boundaries, road networks, etc.), the basic delivery 3D model is initialized and constructed through 3D modeling technology (such as BIM modeling and point cloud rendering). The model covers XYZ axis geographic coordinates, spatial entity geometry and attribute labels (such as building height, no-fly zone range), forming a static spatial foundation for drone delivery.

[0030] N processed standard drone flight data (including real-time flight trajectories, dynamic obstacle detection results, airflow parameters, etc.) are mapped to the basic delivery 3D model. Through spatial coordinate matching and data comparison, elements in the model that need to be updated (such as temporary obstacles and real-time changing airflow areas) are identified, and updated delivery space data is generated. The updated data is then attributed to clarify the type of spatial entity (such as mobile obstacles and passable areas), location coordinates, and size parameters. Based on the labeled delivery space attribute data, the basic model is dynamically updated. Ultimately, a drone delivery space model is constructed that integrates static geographic information and real-time flight data, providing dynamic and accurate 3D scene support for multi-path planning.

[0031] In one possible implementation, step S230 further includes:

[0032] Step S231: Map the N standard drone flight data to the basic delivery three-dimensional model for comparison to obtain delivery space update data.

[0033] Step S232: Attribute tagging is performed on the delivery space update data to obtain delivery space attribute data, where the delivery space attribute data includes the type, location, and size of the space entity.

[0034] Step S233: Update the basic delivery three-dimensional model based on the delivery space attribute data to obtain the drone delivery space model.

[0035] Specifically, N standard drone flight data (including real-time flight trajectories, obstacle detection results, environmental parameters, etc.) that have undergone synchronous alignment and filtering noise reduction processing are loaded into the basic delivery 3D model through spatial coordinate mapping technology. The spatial position and geometric form are compared with the static geographic elements within the model (such as terrain, buildings, and fixed no-fly zones). By analyzing the differences between the flight data and model elements (such as flight trajectories deviating from the preset route and detecting obstacles not marked in the model), dynamically changing spatial elements are identified, such as the coordinates of temporary aerial obstacles and the real-time updated boundaries of airflow disturbance areas. Ultimately, delivery space update data containing information such as the location and range of the newly added dynamic elements is generated, providing an update basis for the dynamic iteration of the model.

[0036] The acquired delivery space update data is structured and attributed according to pre-set classification rules. First, each spatial entity in the update data is assigned a unique type label. For example, dynamic obstacles are marked as mobile obstacles, temporarily restricted areas are marked as temporary no-fly zones, and areas with abnormal airflow are marked as high-wind speed zones. Second, the precise location coordinates of each entity are extracted, including three-dimensional spatial parameters such as latitude, longitude, and altitude to ensure the entity's accurate positioning in the drone delivery space model. Finally, the geometric dimensions of the entity are measured and recorded, such as the radius or length, width, and height of the obstacle, the coverage area or boundary range of the no-fly zone, and the influence radius of the airflow area. Through the above operations, the original update data is converted into standardized delivery space attribute data containing the key attributes of type, location, and size, providing structured information support for subsequent basic model updates and path planning.

[0037] Based on the generated delivery space attribute data (including the type, location, and size of spatial entities), the basic delivery 3D model is dynamically updated. For newly added spatial entities marked in the attribute data (such as temporary no-fly zones and mobile obstacles), 3D geometric objects are created at the corresponding locations in the basic model. For example, a sphere is used to represent a circular obstacle, and a cube is used to represent a no-fly zone. These objects are assigned corresponding type labels. For dimensional parameters involved in the attribute data (such as the obstacle radius and the range of the no-fly zone), visualization is achieved by adjusting the vertex coordinates or boundary range of the geometric objects in the model. Furthermore, if the attribute data contains a state change to the original model entity (such as the removal of a fixed obstacle), the corresponding object in the basic model is deleted or hidden, and environmental parameters (such as wind speed and airflow direction) are updated in the model's dynamic layer. Through these operations, the basic delivery 3D model is iterated into a drone delivery space model that integrates static geographic information with real-time dynamic data. This provides a 3D scene with precise spatial constraints and real-time environmental characteristics for subsequent multi-path planning, ensuring that the path planning results meet actual flight conditions.

[0038] In one possible implementation, step S300 further includes:

[0039] Step S310: extracting cost elements from the drone delivery target to obtain a delivery cost element set, wherein the delivery cost element set includes energy consumption cost, time cost, and maintenance cost.

[0040] Step S320: using the analytic hierarchy process to assign weights to the cost elements in the delivery cost element set, and determining the cost element weight factors.

[0041] Step S330: Fitting the delivery cost element set based on the cost element weight factor to construct the drone delivery cost function.

[0042] Specifically, focusing on the drone delivery goal, the resource consumption and costs of the delivery process are analyzed, core cost elements are extracted, and a delivery cost element set is constructed. Energy consumption costs focus on the energy loss of the drone during flight, such as battery power consumption or fuel usage; time costs cover the entire process from receiving the delivery mission to completing the delivery of the goods, including takeoff, flight, landing, and possible waiting time; maintenance costs involve equipment loss incurred by the drone during the delivery mission, such as motor wear, sensor aging, and other subsequent maintenance costs. By clarifying these three types of cost elements, a structured set of cost elements is formed, providing a quantitative analysis basis for the subsequent cost function construction and multi-path planning.

[0043] The analytic hierarchy process (AHP) is used to assign weights to the set of delivery cost elements (energy cost, time cost, and maintenance cost). First, a hierarchical model is constructed, with the overall drone delivery goal set as the top layer and each cost element as the middle layer. A judgment matrix is ​​constructed using expert experience or historical data to compare cost elements pairwise (for example, comparing the relative importance of energy cost and time cost to the delivery goal). Next, the eigenvectors and maximum eigenvalues ​​of the judgment matrix are calculated, and a consistency test is performed to ensure the rationality of the weight assignment. Finally, the weight factors for each cost element are determined. For example, if the judgment matrix shows that time cost is more important than energy cost, a weight of 0.5 might be assigned to time cost, 0.4 to energy cost, and 0.1 to maintenance cost. This allows the weight factors to reflect the priority differences of different cost elements in the delivery goal, providing a quantitative basis for constructing the cost function.

[0044] Based on the determined cost factor weight factors (such as a weight of 0.4 for energy cost, 0.5 for time cost, and 0.1 for maintenance cost), the delivery cost factor set (energy cost, time cost, and maintenance cost) is mathematically fitted to construct a drone delivery cost function using a linear weighted approach. This is achieved specifically through the formula f(path) = w1×E+w2×T+w3×M, where w1, w2, and w3 are the weight factors for each cost factor, E is the quantified value of energy cost, T is the quantified value of time cost, and M is the quantified value of maintenance cost. This function integrates multi-dimensional cost factors into a single quantitative indicator, providing a unified optimization objective function for subsequent multi-path planning within the drone delivery space model. This enables the path planning results to balance different cost factors according to preset weights, achieving comprehensive cost optimization for the delivery task.

[0045] In one possible implementation, step S300 further includes:

[0046] Step S340: Determine N drone performance constraint parameters and N delivery task constraint parameters based on the N drone performance parameters and the N drone delivery tasks.

[0047] Step S350: Based on the N drone performance constraint parameters and the N delivery task constraint parameters, the drone delivery cost function is used to perform multi-path planning in the drone delivery space model to obtain N drone delivery paths.

[0048] Specifically, based on the acquired N UAV performance parameters (including endurance, payload capacity, maximum flight speed, safe flight altitude range, etc.) and N UAV delivery mission information (such as mission start and destination coordinates, cargo weight, delivery time requirements, and no-fly zone distribution), constraint parameters are extracted and quantified. For each UAV, performance constraint parameters include the maximum flight distance threshold determined by the range, the cargo weight limit limited by the payload capacity, and the flight speed range. For each delivery mission, constraint parameters include the maximum allowable flight time based on time requirements, the coordinate range of the no-fly zone, and the matching requirements between cargo weight and UAV payload. Through this process, a corresponding set of constraint parameters is established for each UAV and each delivery mission, forming the feasibility boundary conditions for the subsequent multi-path planning, ensuring that the path planning results meet the UAV's physical performance and mission execution requirements.

[0049] The N determined UAV performance constraints (such as maximum flight distance, payload limit, and safe altitude range) and N delivery mission constraints (such as timeliness requirements, no-fly zone coordinates, and cargo weight thresholds) are combined with a constructed UAV delivery space model. The constraints are then used to filter the three-dimensional space, defining a feasible flight delivery space for each UAV (i.e., excluding areas that exceed performance limits or violate mission rules). Subsequently, using the constructed UAV delivery cost function (a weighted function integrating energy consumption, time, and maintenance costs) as the optimization objective, intelligent optimization algorithms such as genetic algorithms and ant colony algorithms are used to optimize multiple paths within each feasible space. The algorithm iteratively calculates the cost function values ​​for different paths while simultaneously verifying that the paths satisfy all constraints (e.g., total path length ≤ UAV range, flight trajectory avoids no-fly zones). Ultimately, a cost-optimal, yet consistent delivery path is generated for each UAV, enabling differentiated path planning for the N UAVs within the three-dimensional space and ensuring the efficiency and feasibility of multi-drone mission execution.

[0050] In one possible implementation, step S350 further includes:

[0051] Step S351: Based on the N drone performance constraint parameters and the N delivery task constraint parameters, constraint planning is performed on the drone delivery space model to obtain N feasible flight delivery spaces.

[0052] Step S352: Utilize the drone delivery cost function to perform multi-path planning and optimization within the N feasible flight delivery spaces to determine the N drone delivery paths.

[0053] Specifically, N drone performance constraint parameters (such as the maximum flight distance corresponding to the range, the upper limit of cargo weight limited by the payload capacity, and the safe flight altitude range) and N delivery mission constraint parameters (such as the maximum flight time determined by timeliness requirements, the geographic coordinates of the no-fly zone, and the matching rules between cargo weight and drone payload) are input into the drone delivery space model as spatial constraints. Through three-dimensional spatial geometric operations (such as convex polygon clipping and spatial region segmentation), based on the unique constraint combination of each drone, areas that exceed performance limits or violate mission rules are eliminated from the global model. For example, long-distance areas with insufficient range, no-fly zones above the safe altitude, or obstacle-dense areas below the minimum flight altitude are excluded. Ultimately, an independent feasible flight delivery space is generated for each drone. All paths within this space meet the physical performance limitations and mission execution requirements of the corresponding drone, thus defining an effective search range for subsequent path optimization.

[0054] For each of the N feasible delivery flight spaces, a multi-path optimization algorithm (such as a genetic algorithm or an ant colony algorithm) is employed to optimize the drone delivery cost function (which integrates energy, time, and maintenance costs, with weights determined using the analytic hierarchy process) within each space. The algorithm iteratively generates candidate paths and calculates the cost function for each path (e.g., total energy consumption = flight distance × energy consumption per unit distance; time cost = estimated flight duration; maintenance cost = estimated value from a modeled equipment wear and tear analysis). The algorithm also verifies that the paths meet the corresponding drone's performance constraints (e.g., range and payload limits) and mission constraints (e.g., time requirements and no-fly zone avoidance). Paths are continuously optimized through operations such as crossover and mutation, ultimately selecting the lowest-cost delivery path for each drone that meets all constraints. This results in N drone delivery paths with specific flight trajectories, turning point coordinates, and speed plans, enabling efficient path allocation for multi-drone missions in a dynamic three-dimensional space.

[0055] In one possible implementation, step S400 further includes:

[0056] Step S410: Perform anomaly detection on the drone state data and environmental data to obtain drone abnormal state parameters and environmental risk parameters.

[0057] Step S420: Based on the abnormal state parameters of the drone and the environmental risk parameters, the N drone delivery paths are regulated and optimized to obtain N drone optimized delivery paths.

[0058] Specifically, the drone's onboard sensors (such as IMU, barometer, battery monitoring module) and external environmental monitoring equipment (such as weather radar, traffic management system) are used to collect drone status data (including battery power, motor temperature, flight attitude angle, navigation signal strength, etc.) and environmental data (such as wind speed, wind direction, precipitation probability, no-fly zone dynamic information, etc.) in real time. Subsequently, the data is processed using preset anomaly detection algorithms (such as threshold detection, Kalman filter residual analysis, isolation forest algorithm, etc.): For drone status data, abnormal states are identified by comparing historical data distribution or physical limit thresholds (such as battery power below the safety threshold, motor temperature exceeding the critical value), and drone abnormal state parameters are generated; for environmental data, risk factors that may affect flight safety (such as strong wind areas, temporary no-fly zone coordinates) are extracted by analyzing real-time weather forecasts or traffic control information to form environmental risk parameters. Finally, through data fusion and verification, structured parameters containing information such as anomaly type, location, and severity are output to provide a basis for subsequent path control.

[0059] Based on acquired drone abnormality parameters (such as low battery charge and motor failure warnings) and environmental risk parameters (such as the coordinates of strong wind areas and temporary no-fly zones), dynamic control and optimization are implemented for N drone delivery routes. First, in response to drone abnormalities, a path replanning mechanism is prioritized. For example, a drone with low battery is rerouted to the nearest charging point, or a forced landing path is generated for a drone with out-of-control attitude. Second, for environmental risks, three-dimensional spatial modeling techniques are used to identify risk area boundaries. An A* algorithm or a dynamic windowing algorithm (DWA) is then used to generate detours within the feasible flight space, avoiding high-risk areas. During the control process, a dynamic obstacle layer is updated in the drone delivery space model in real time, and path adjustment information is synchronized via a multi-drone communication protocol to avoid collisions. Finally, through local path correction and global optimization strategies, an optimized delivery path is generated for each drone, adapting to the real-time status and environment, ensuring efficient and safe delivery.

[0060] In one possible implementation, step S420 further includes:

[0061] Step S421: Based on the abnormal state parameters and environmental risk parameters of the drones, the N drone delivery paths are locally marked as abnormal and the path control is optimized to determine the optimized delivery paths of the N drones.

[0062] Specifically, based on the output of drone anomaly parameters (e.g., a drone's battery charge is below 15% or its motor temperature exceeds a certain limit) and environmental risk parameters (e.g., the coordinates of a sudden no-fly zone or the extent of a high-wind zone), N drone delivery routes are first spatially matched. The affected route segments are marked in a three-dimensional model using a geographic information system (GIS). For infeasible areas of a route caused by drone anomalies, a local path reconstruction algorithm is automatically triggered. For example, a nearby safe alternative route is searched based on the original route to ensure that the new route meets the drone's remaining battery or payload limits. For environmental risk areas, an expansion algorithm is used to expand the risk area boundaries, and the Dijkstra algorithm is used to recalculate the shortest path that avoids the risk. During the control process, the impact of route adjustments on overall delivery timeliness is evaluated in real time, and the option with the lowest incremental cost is prioritized. A multi-drone coordination mechanism is then used to synchronously update the path plans of adjacent drones to avoid global conflicts caused by local adjustments. Finally, through drone-by-drone and segment-by-segment anomaly marking and path optimization, N optimized delivery routes are generated that take into account real-time status and environmental adaptability, ensuring the safety and efficiency of the delivery mission.

[0063] It should be noted that the order in which the embodiments of the present application are presented is for illustrative purposes only and does not necessarily represent the superiority or inferiority of the embodiments. Furthermore, the foregoing descriptions of specific embodiments of this specification are provided. Furthermore, the processes depicted in the accompanying drawings do not necessarily require the specific order or sequential sequence shown to achieve the desired results. In certain embodiments, multitasking and parallel processing are also possible or may be advantageous.

[0064] The above description is only a preferred embodiment of the present application and is not intended to limit the present application. Any modifications, equivalent replacements, improvements, etc. made within the spirit and principles of the present application should be included in the scope of protection of the present application.

[0065] This specification and drawings are merely illustrative of the present application and are intended to cover any and all modifications, variations, combinations, or equivalents within the scope of this application. Obviously, those skilled in the art may make various modifications and variations to this application without departing from the scope of this application. Thus, this application is intended to include such modifications and variations as fall within the scope of this application and its equivalents.

Claims

1. A multi-path optimization method for a low-altitude UAV delivery system, characterized in that: The method comprises: Obtain N delivery drones, N drone performance parameters, and N drone delivery tasks through a low-altitude drone delivery system, and equip the N delivery drones with a sensor group to collect and obtain flight data of the N drones through the sensor group; Perform three-dimensional modeling based on the delivery map information and the flight data of the N drones to build a drone delivery space model; Constructing a drone delivery cost function according to the drone delivery target, and performing multi-path planning within the drone delivery space model using the drone delivery cost function based on the N drone performance parameters and the N drone delivery tasks to obtain N drone delivery paths; Real-time monitoring is performed to obtain drone status data and environmental data, and the N drone delivery paths are regulated and optimized based on the drone status data and environmental data to determine the optimized delivery paths of the N drones.

2. The multi-path optimization method for a low-altitude UAV delivery system according to claim 1, characterized in that: The construction of the drone delivery space model includes: Synchronously aligning and filtering the N UAV flight data according to the acquisition timing information to obtain N standard UAV flight data; Initialize the model based on the delivery map information and build a basic delivery 3D model; The N standard drone flight data are mapped to the basic delivery three-dimensional model for updating to obtain a drone delivery space model.

3. The multi-path optimization method for a low-altitude UAV delivery system according to claim 2, characterized in that: The obtaining of the drone delivery space model includes: Mapping the N standard drone flight data to the basic delivery 3D model for comparison to obtain delivery space update data; Performing attribute tagging on the delivery space update data to obtain delivery space attribute data, wherein the delivery space attribute data includes the type, location, and size of the space entity; The basic delivery three-dimensional model is updated based on the delivery space attribute data to obtain the drone delivery space model.

4. The multi-path optimization method for a low-altitude UAV delivery system according to claim 1, wherein: The drone delivery cost function is constructed according to the drone delivery target, including: Extracting cost elements from the drone delivery target to obtain a delivery cost element set, wherein the delivery cost element set includes energy consumption cost, time cost, and maintenance cost; Using the analytic hierarchy process to assign weights to the cost elements in the delivery cost element set, and determine the cost element weight factors; The delivery cost element set is fitted based on the cost element weight factors to construct the drone delivery cost function.

5. The multi-path optimization method for a low-altitude UAV delivery system according to claim 1, characterized in that: The N drone delivery paths are obtained, including: Determining N drone performance constraint parameters and N delivery task constraint parameters based on the N drone performance parameters and the N drone delivery tasks; Based on the N drone performance constraint parameters and the N delivery task constraint parameters, the drone delivery cost function is used to perform multi-path planning in the drone delivery space model to obtain N drone delivery paths.

6. The multi-path optimization method for a low-altitude UAV delivery system according to claim 5, characterized in that: The step of obtaining N drone delivery routes includes: Based on the N UAV performance constraint parameters and the N delivery task constraint parameters, constraint programming is performed on the UAV delivery space model to obtain N feasible flight delivery spaces; The drone delivery cost function is used to perform multi-path planning and optimization in the N feasible flight delivery spaces to determine the N drone delivery paths.

7. The multi-path optimization method for a low-altitude UAV delivery system according to claim 1, wherein: Determining the optimized delivery paths of N drones includes: Performing anomaly detection on the drone state data and environmental data to obtain drone abnormal state parameters and environmental risk parameters; The N drone delivery paths are regulated and optimized based on the drone abnormal state parameters and the environmental risk parameters to obtain N drone optimized delivery paths.

8. The multi-path optimization method for a low-altitude UAV delivery system according to claim 7, characterized in that: The N drone optimized delivery paths are obtained, including: Based on the abnormal state parameters and environmental risk parameters of the drones, local abnormality marking and path control optimization are performed on the N drone delivery paths to determine the optimized delivery paths of the N drones.

Images