Unmanned aerial vehicle flight path generation method and unmanned aerial vehicle flight path generation device based on road network hub and dynamic safety score

Abstract

The invention discloses an unmanned aerial vehicle flight path generation method and an unmanned aerial vehicle flight path generation device based on a road network hub and a dynamic safety score. The generation method comprises the following steps: constructing a primary road network hub network based on flight infrastructure data and environmental geographic information data; preliminarily screening hub nodes of the primary road network hub network; carrying out the preliminary scoring of the airworthiness capability of the candidate nodes, and obtaining a second-level road network hub network; before a flight task is started, constructing a backbone topology road network; performing fine-grained leg filling on the backbone topology road network, and calculating the total flight consumption value of each leg according to the real-time airworthiness capability score of the key connection point; and forming an undirected graph based on the total flight consumption value, and determining an optimal path and a plurality of redundant paths in the undirected graph. The method has positive and remarkable technical effects in multiple dimensions such as flight safety, real-time adaptability, cost and industry normalization, and solves the problem of pain points existing in the industry for a long time.

Description

Technical Field

[0001] This invention relates to the field of aircraft technology, and in particular to a method and apparatus for generating UAV flight paths based on road network hubs and dynamic safety scores. Background Technology

[0002] With the rapid development of the low-altitude industry and drone technology, drones are increasingly being used in urban inspection, logistics delivery, and emergency response. The number of drones in urban air traffic is expected to experience explosive growth in the future. Against this backdrop, drone flight safety has become a critical issue that urgently needs to be addressed. Especially in medium- and long-range flight missions, drone flight paths must possess high reliability and dynamic adaptability to cope with the complexity and variability of the urban low-altitude environment.

[0003] Currently, drone safety technologies primarily focus on in-flight emergency response and post-flight rescue. For example, by equipping drones with parachutes or thrust reversers, emergency self-rescue measures can be implemented in the event of a malfunction to reduce crash damage. These methods are reactive and cannot proactively avoid potential risks before flight. Furthermore, while existing drone traffic management systems (such as UTM) possess some airspace scheduling and obstacle avoidance capabilities, their risk assessments largely rely on historical data and static rules, resulting in insufficient real-time performance and granularity in the assessment results.

[0004] A particularly lacking mechanism is the path generation system for medium- and long-haul flights. Existing methods typically focus only on point-to-point path planning, lacking a highly adaptable intermediate node system, resulting in low recoverability and overall robustness of paths in unforeseen circumstances. Therefore, there is an urgent need for a proactive safety planning method capable of generating highly reliable paths before flight. On the other hand, existing path planning schemes often aim to generate a single "optimal" path. This approach has significant drawbacks: if a sudden environmental change renders the path unusable, the system must recalculate, leading to decision-making lag and a lack of continuity; furthermore, a single path cannot provide a set of alternative solutions, making dynamic switching and optimization based on real-time safety scores difficult. For complex and highly dynamic urban low-altitude environments, there is a lack of an effective method to generate path networks that combine reliability, redundancy, and real-time adaptability. Summary of the Invention

[0005] The purpose of this application is to provide a method and apparatus for generating UAV flight paths based on road network hubs and dynamic safety scoring, so as to at least solve the aforementioned technical problems. The various technical effects of the optional technical solutions provided by this invention are detailed below.

[0006] To achieve the above objectives, in a first aspect, this application provides a method for generating UAV flight paths based on road network hubs and dynamic safety scoring, comprising: Based on flight infrastructure data and environmental geographic information data, a primary road network hub network is constructed, which includes several hub nodes; A preliminary screening of several hub nodes was conducted to obtain multiple candidate nodes that meet the conditions for hovering and emergency landing. The candidate nodes are initially scored for their airworthiness to obtain a secondary road network hub network including multiple airworthiness nodes; Before the flight mission is launched, based on the flight start point and flight end point, connectivity relationships are searched in the secondary road network hub network to construct a backbone topology road network including multiple key connection points. The main trunk topology network is filled with fine-grained flight segments, and the total flight consumption value of each flight segment from the flight start point to the flight end point is calculated based on the real-time airworthiness score of the key connection points. Based on the total flight consumption value of each flight segment, an undirected graph is formed, and the optimal path and multiple redundant paths are determined in the undirected graph. A comprehensive safety report is generated for the optimal path, allowing the UAV to fly based on the optimal path, and the redundant path is switched when an anomaly occurs on the optimal path.

[0007] In some embodiments, the flight infrastructure data includes apron information, alternate landing point information, charging and battery swapping station information, air safety waypoint information, no-fly zone information, and restricted flight zone information; the environmental geographic information data includes three-dimensional urban building information and terrain surface information.

[0008] In some embodiments, the preliminary screening of a plurality of hub nodes to obtain multiple candidate nodes with hovering and emergency landing capabilities includes: Determine the physical space where several of the hub nodes are located; If the physical space where the hub node is located meets the alternate landing conditions, then the hub node that meets the alternate landing conditions will be used as a candidate node. The alternate landing conditions include compliance with alternate landing location, airspace, and communication.

[0009] In some embodiments, the preliminary airworthiness assessment of the candidate nodes includes: The airworthiness score of each candidate node is calculated by a multi-factor weighted evaluation model. The multi-factor weighted evaluation model is determined based on the weight of the i-th evaluation factor and the normalized score of the i-th evaluation factor. The normalized score of the i-th evaluation factor includes surface stability, real-time meteorological conditions, infrastructure support, historical safety record, and dynamic airspace occupancy rate. The airworthiness scores of the candidate nodes are graded, and candidate nodes whose scores fall within the unairworthiness range are isolated to obtain multiple levels of airworthiness nodes. The node scores are then used as the node weights of the airworthiness nodes.

[0010] In some embodiments, the step of searching for connectivity relationships in the secondary road network hub network based on the flight origin and flight destination, and constructing a backbone topology road network, includes: If the two airworthiness nodes satisfy the adjacent constraint, then a directed edge is established and the directed edge is assigned an edge weight. Using the flight start point as the source point S and the flight destination point as the target point T, a graph theory algorithm is used to search for connectivity in the secondary road network hub network to form a directed topological network containing multiple routes from the flight start point to the flight destination, and this directed topological network is used as the backbone topological network.

[0011] In some embodiments, the step of using the flight origin as the source point S and the flight destination as the target T, employing graph theory algorithms to search for connectivity relationships in the secondary road network hub network, forming a directed topological network containing multiple routes from the flight origin to the flight destination, and using this directed topological network as the backbone topological network, includes: Set the edge weights and the number of hubs n; Based on the edge weights and the number of hubs n, a bidirectional breadth-first search is performed on the airworthiness nodes, starting from the starting set {S_i} and ending set {T_j} respectively. Compare all airworthiness nodes found in the starting direction with those found in the ending direction, and use the airworthiness nodes that exist in both the starting set and the ending set as key connection points. Based on the key connection points, the backbone topology network is established.

[0012] In some embodiments, the fine-grained segment filling of the backbone topology network includes: Collision detection and airspace compliance verification are performed based on flight infrastructure data and environmental geographic information data to determine whether there is a straight flight path between two key connection points. If so, fill the straight segment between the two key connection points; Otherwise, activate the detour algorithm to avoid obstacles and controlled airspace, filling the detour segment between the two key connection points.

[0013] In some embodiments, calculating the total flight consumption value for each segment from the flight origin to the flight destination based on the real-time airworthiness score of the key connection point includes: Calculate the initial flight cost for each segment; The total flight cost for each segment is calculated based on the initial flight cost for each segment and the real-time airworthiness scores of the key connection points at both ends of the segment.

[0014] In some embodiments, generating a comprehensive security report for the optimal path includes: The comprehensive safety report is presented in a three-dimensional visualization, including the overlay of the optimal path with a 3D real-world map, airspace rules, and signal heatmap.

[0015] Secondly, this application also provides a drone flight path generation apparatus, comprising: one or more processors; a memory for storing one or more computer programs, wherein the one or more processors are configured to execute the one or more computer programs stored in the memory, so that the one or more processors perform a drone flight path generation method based on road network hubs and dynamic safety scoring as described in any of the first aspects.

[0016] Thirdly, this application provides an aircraft, which includes the UAV flight path generation device described in the second aspect.

[0017] Implementing one of the above-described technical solutions of this application has the following advantages or beneficial effects: This application discloses a method and apparatus for generating UAV flight paths based on road network hubs and dynamic safety scoring. First, a primary road network hub network is constructed using multi-source data, including flight infrastructure data and environmental geographic information data. Preliminary node screening reveals multiple candidate nodes with excellent hovering and emergency landing conditions, providing a safe haven for the UAV throughout its flight path and significantly enhancing its ability to handle sudden malfunctions such as power failure or brief signal interruptions. Second, a hierarchical and quantifiable dynamic safety scoring mechanism is introduced, transforming various risks into specific and calculable values ​​to obtain a secondary road network hub network containing multiple airworthiness nodes. After constructing the backbone topology road network, fine-grained flight segment filling is performed, along with real-time airworthiness scoring. This proactively avoids dangerous areas, achieving a leap from "collision avoidance" to "optimal safety." The redundant path design ensures that if a node on the optimal path fails, a route can be quickly planned, achieving seamless switching, avoiding mission interruption, and improving overall reliability.

[0018] This application, through precise risk assessment and optimal path optimization, eliminates the need for drones to frequently perform large-scale emergency maneuvers, reducing overall energy consumption and saving energy and operating costs. It achieves the best balance between computing efficiency and response speed, providing a foundation for large-scale applications. It is suitable for large-scale, standardized low-altitude operations and has produced positive and significant technical effects in multiple dimensions such as flight safety, real-time adaptability, operational economy, and industry standardization, solving long-standing pain points in the industry. Attached Figure Description

[0019] To more clearly illustrate the technical solutions of the embodiments of the present invention, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort. In the drawings: Figure 1 This is a flowchart illustrating a method for generating UAV flight paths based on road network hubs and dynamic safety scoring, according to an embodiment of this application. Figure 2 This is a schematic block diagram of a UAV flight path generation system according to an embodiment of this application; Figure 3 This is a schematic diagram of the backbone topology network of this application; Figure 4 This is a schematic diagram of the dynamically weighted route score for the nodes and segments of this application; Figure 5 This is another schematic diagram of the UAV flight path generation method based on road network hubs and dynamic safety scoring in this application embodiment. Detailed Implementation

[0020] To make the objectives, technical solutions, and advantages of this application clearer, various exemplary embodiments described below can be referenced to the corresponding drawings, which constitute a part of the exemplary embodiments and illustrate various exemplary embodiments that may be adopted to implement this application. 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 this disclosure. It should be understood that they are merely examples of processes, methods, and apparatuses consistent with some aspects of this application disclosed as detailed in the appended claims, and other embodiments may be used, or structural and functional modifications may be made to the embodiments listed herein without departing from the scope and spirit of this application.

[0021] In the description of this application, it should be understood that the terms "first," "second," etc., are used for descriptive purposes only and should not be construed as indicating or implying relative importance or implicitly specifying the number of technical features indicated. The term "multiple" means two or more. The terms "connected" and "linked" should be interpreted broadly, for example, they can refer to fixed connections, detachable connections, integral connections, mechanical connections, electrical connections, communication connections, direct connections, indirect connections through an intermediate medium, and can refer to the internal connection of two elements or the interaction relationship between two elements. For those skilled in the art, the specific meaning of the above terms in this application can be understood according to the specific circumstances.

[0022] To illustrate the technical solutions described in this application, specific embodiments are provided below, showing only the parts related to the embodiments of this application.

[0023] like Figure 1-5 As shown, this application provides a method and apparatus for generating UAV flight paths based on road network hubs and dynamic safety scoring, which are applied to UAV flight path generation systems, such as... Figure 2 As shown, the system includes a firewall / security group enabling HTTPS, load balancing / reverse proxy, a Docker container cluster, and an external data source API. The Docker container cluster includes a web application container (with the core business of the security scoring system), a database container, a cache container, a message queue container, and other auxiliary containers such as file services and scheduled tasks. The system is deployed using Docker containerization on an Ubuntu server. The deployment process typically begins with an environment pre-check to confirm the system version, hardware resources, and Docker environment are ready. The application image building process is defined by writing a Dockerfile, and multiple service containers are defined and connected using a Docker Compose orchestration file (docker-compose.yml), enabling one-click deployment and collaborative service startup.

[0024] like Figure 1 As shown, the method for generating UAV flight paths based on road network hubs and dynamic safety scores includes the following steps S10 to S70.

[0025] S10: Based on flight infrastructure data and environmental geographic information data, construct a primary road network hub network, which includes several hub nodes.

[0026] Flight infrastructure data and environmental geographic information data are combined to form road network data, which is the core computational foundation of the UAV flight route safety scoring system. It provides a basic spatial framework for UAV path planning, real-time navigation, and risk gap detection. Road network data not only includes traditional road network information, but also geographic information of various infrastructures necessary for low-altitude flight activities. In some implementations, the flight infrastructure data includes information on helipads, alternate landing points, charging and battery swapping stations, safe waypoints, no-fly zones, and restricted flight zones. The environmental geographic information data includes three-dimensional urban building information and terrain surface information.

[0027] Specifically, flight infrastructure data is usually collected and entered manually, obtained through on-site surveying and mapping, official planning documents, or defined by the operating company and input into the UAV flight route safety scoring system. The characteristics of flight infrastructure data are: high accuracy requirements, and update frequency is directly related to changes in infrastructure. It serves as the benchmark for high-precision route management by the UAV flight route safety scoring system.

[0028] Environmental geographic information data is used to construct the flight environment, including basic elevation and topographic information, water system surface information, and vegetation surface information, as well as 3D urban building information. This information depicts the outline, height, and shape of buildings and is crucial for assessing potential collision risks, electromagnetic signal blockage, and complex airflow fields such as inter-building winds. The main source of environmental geographic information data is professional data models and services provided by third parties, such as the mainstream domestic map service provider Tianditu and maps provided by various enterprises. These provide application programming interfaces (APIs) to obtain basic geographic information and points of interest. For the establishment of 3D urban building information, oblique photogrammetry or LiDAR technology can be used to acquire point cloud data and generate high-precision real-scene 3D models and digital surface models. For areas with special needs, such as key inspection areas or complex building environments, refined modeling can be performed based on UAV aerial photography data to supplement and improve the level of detail in third-party data, ensuring the comprehensiveness and accuracy of risk assessment.

[0029] Flight infrastructure data and environmental geographic information data form multi-source static and dynamic data. After obtaining the flight infrastructure data and environmental geographic information data, a primary road network hub network is constructed based on this data. The obtained primary road network hub network... Figure 3 The data includes only the hub nodes of node AM, but does not include the scores for each flight segment.

[0030] S20: Perform preliminary screening on several hub nodes to obtain multiple candidate nodes that have hovering and emergency landing capabilities.

[0031] In some implementations, a preliminary automated screening of several hub nodes can be performed, and step S20 may include: Determine the physical space where several of the hub nodes are located; If the physical space where the hub node is located meets the alternate landing conditions, then the hub node that meets the alternate landing conditions will be used as a candidate node. The alternate landing conditions include compliance with alternate landing location, airspace, and communication.

[0032] Specifically, physical space refers to the geographical space where the candidate node is located. Compliance of the alternate landing location in the alternate landing conditions means that the candidate node has sufficient physical space around it, and is located in an open area, such as a rooftop, plaza, park, open green space, or standard take-off and landing pad. In other words, the space where the candidate node is located is larger than the predetermined size, meeting the basic requirements for safe hovering, take-off, landing, and circling for the drone. Compliance of the alternate landing location can be quantitatively determined by analyzing the Digital Surface Model (DSM) and satellite imagery to calculate the building coverage and average height within a certain radius around the candidate node.

[0033] The compliance of the alternate landing location in the alternate landing conditions can also include superior alternate landing address, which means that there must be one or more certified and available emergency alternate landing points, such as aprons or alternate landing points, near the candidate node. That is, the candidate node has compliant aprons and / or alternate landing points within a preset range.

[0034] Airspace compliance means that candidate nodes must be absolutely far away from permanent and temporary no-fly zones and restricted areas, such as airport airspace, military control zones, nuclear facilities, etc., and comply with all preset air corridor rules. Their coordinates must be compared and verified in real time with the airspace control data released by the authorities.

[0035] Communication compliance refers to the requirement that the communication signal strength of candidate nodes remains consistently and stably above the minimum threshold set by the system, ensuring uninterrupted data transmission and command interaction between the drone and the control platform. Communication signals include 4G / 5G cellular network signals, dedicated data link signals, etc.

[0036] Furthermore, for the candidate nodes that have passed the initial screening, a manual screening step can be added. For example, domain experts or system administrators can manually confirm and calibrate the nodes to give the system the necessary flexibility and eliminate some candidate nodes that are not identified by experience knowledge or potential risk points that the automated rules have failed to identify.

[0037] After initial screening of hub nodes and elimination of potential risk points, the remaining candidate nodes all meet the mandatory conditions.

[0038] S30: Perform a preliminary airworthiness assessment on the candidate nodes to obtain a secondary road network hub network including multiple airworthiness nodes.

[0039] After screening the hub nodes, each candidate node is assigned a quantitative airworthiness capability score (ACS), which is a real-time reflection of its overall safety status.

[0040] In some implementations, step S30 may include: The airworthiness score of each candidate node is calculated by a multi-factor weighted evaluation model. The multi-factor weighted evaluation model is determined based on the weight of the i-th evaluation factor and the normalized score of the i-th evaluation factor. The normalized score of the i-th evaluation factor includes surface stability, real-time meteorological conditions, infrastructure support, historical safety record, and dynamic airspace occupancy rate. The airworthiness scores of the candidate nodes are graded, and candidate nodes whose scores fall within the unairworthiness range are isolated to obtain multiple levels of airworthiness nodes. The node scores are then used as the node weights of the airworthiness nodes.

[0041] Specifically, the airworthiness score can be expressed by the following formula: ; Here, ACS represents the airworthiness rating. This represents the weight of the i-th evaluation factor. This represents the normalized score of the i-th evaluation factor.

[0042] Surface stability is used to assess the surface material of candidate nodes, such as concrete, grass, soil, and its flatness and slope, which has a direct impact on landing safety.

[0043] Real-time weather conditions integrate data such as real-time wind speed, rainfall, visibility, and temperature. Among these, wind speed is a key factor affecting the hovering stability and energy consumption of drones.

[0044] Infrastructure support assesses the availability and distance of charging / battery swapping stations and maintenance points near the candidate node, which affects the drone's continuous operational capability.

[0045] Historical security records refer to the frequency of failures, accidents, or signal interruptions that have occurred in the candidate node's history.

[0046] Dynamic airspace occupancy rate is an assessment of the density of other aircraft and the risk of conflict in the airspace surrounding a candidate node by UAVs based on real-time perception data.

[0047] The airworthiness score of each candidate node is a dynamic value, and the update strategy can be flexibly set according to the characteristics of different data sources and business needs.

[0048] For high-frequency changing data in flight infrastructure data and environmental geographic information data, such as weather, airspace occupancy, and signal strength, the updated score can be updated in real time, such as at a frequency of minutes or seconds.

[0049] For low- to medium-frequency change data in flight infrastructure data and environmental geographic information data, such as geographic information and infrastructure information, the update score is based on a preset cycle or event-driven. The preset cycle can be an hourly or daily frequency, while the event-driven update can be triggered when the infrastructure data is updated.

[0050] It should be noted that even if a data source is missing or outdated for an extended period, its corresponding scoring factor will continue to decay due to "information obsolescence." The system will set a confidence decay curve for this factor, and its scoring contribution will gradually decrease over time. When the overall ACS of a candidate node consistently falls below the system-set disable threshold (20 points) due to data obsolescence or deteriorating status, the node will be automatically marked as "unavailable" and temporarily removed from the currently effective path planning network until its score recovers to a safe level.

[0051] When scoring candidate nodes, the scores are graded, normalized to a percentage, and divided into the following risk levels: 0-19 points: Inoperative zone. Candidate nodes in this category are of extremely high risk, and the system prohibits any route from planning them as a waypoint and isolates them from the real-time road network.

[0052] 20-39 points: High Risk range. Candidate nodes in this category are only recommended for temporary use in extreme situations such as emergency avoidance. In the path planning algorithm, paths passing through this candidate node will be subject to extremely high penalty weights, and the algorithm will prioritize avoiding them.

[0053] 40-69 points: Low Risk range. Candidate nodes in this category have satisfactory airworthiness, are stable components of the road network, and can participate normally in path planning.

[0054] 70-100 points: Optimal airworthiness range. Candidate nodes in this category have excellent conditions in all aspects and are high-safety, high-reliability nodes that are preferred for route planning. The algorithm will tend to guide the route through these candidate nodes.

[0055] Understandably, candidate nodes located in the medium-to-high risk zone, low risk zone, and optimal airworthiness zone are all airworthy nodes, while candidate nodes located in the unairworthiness zone are directly isolated.

[0056] It should be emphasized that the above-mentioned score range, factor weights and specific algorithms are only one feasible way to achieve the purpose of this application. This application is not limited to this specific mathematical expression. Any method that uses multi-source data to dynamically quantify and score road network hubs to achieve the purpose of safe path planning falls within the protection scope of this application.

[0057] By scoring the airworthiness of candidate nodes, multiple airworthiness nodes are obtained, and the combinations of these airworthiness nodes form a secondary road network hub network.

[0058] S40: Before the flight mission is launched, based on the flight start point and flight end point, search for connectivity in the secondary road network hub network and construct a backbone topology road network including multiple key connection points.

[0059] The backbone topology network provides a highly reliable and multi-choice path backbone framework for UAVs during long-distance flights, and the construction of this backbone topology network is a dynamic process of continuous evolution.

[0060] The construction of the backbone topology network is triggered by the system scheduler. The triggering strategy can be flexibly configured according to actual business needs, including two methods: regular construction and on-demand construction. Regular construction is for fixed routes or routine inspection tasks, and is refreshed according to a preset cycle, such as daily or hourly. On-demand construction is triggered by the user setting the start and end points in the user interface to generate flight tasks.

[0061] Each time a task is built, unseaworthy nodes are isolated to ensure that subsequent computations utilize currently available safe resources.

[0062] In some implementations, step S40 may include: If the two airworthiness nodes satisfy the adjacent constraint, then a directed edge is established and the directed edge is assigned an edge weight. Using the flight start point as the source point S and the flight destination point as the target point T, a graph theory algorithm is used to search for connectivity in the secondary road network hub network to form a directed topological network containing multiple routes from the flight start point to the flight destination, and this directed topological network is used as the backbone topological network.

[0063] Specifically, the "adjacent" constraint in the adjacent constraint is not purely geographical proximity, but rather a combination of spatial distance and safe accessibility in airspace. The adjacent constraint can include distance constraints, airspace visibility constraints, and signal continuity constraints.

[0064] Distance constraints are used to maximize path redundancy and safety margin, ensuring that drones can frequently reach safe nodes to deal with emergencies. The straight-line distance between two airworthy nodes is controlled within a preset range, which is relatively small, such as within 3 kilometers. This preset range is set based on flight safety and emergency response efficiency in complex urban environments, with the aim of forming a dense and highly available safety network.

[0065] Airspace visibility means that there must be no permanent or temporary no-fly zones or restricted-fly zones on the line connecting two airworthiness nodes. At the same time, obstacle shielding analysis is conducted based on a three-dimensional city model to ensure that the planned flight path can avoid static obstacles such as tall buildings and mountains and maintain basic airspace visibility.

[0066] Signal continuity is based on the communication signal spectrum. The signal strength along the flight path of the two airworthiness nodes should be higher than the minimum operating requirements throughout the entire flight to avoid large-scale signal blind spots.

[0067] When two airworthiness nodes meet the aforementioned adjacency constraints, a directed edge is established, resulting in multiple directed edges. Each directed edge is then assigned a weight. Edge weights can be determined by considering factors such as distance, average flight time, and historical success rate.

[0068] In some implementations, the step of using the flight origin as the source point S and the flight destination as the target T, and employing graph theory algorithms to search for connectivity in the secondary road network hub network to form a directed topological network containing multiple routes from the flight origin to the flight destination, and using this directed topological network as the backbone topological network, may include: Set the edge weights and the number of hubs n; Based on the edge weights and the number of hubs n, a bidirectional breadth-first search is performed on the airworthiness nodes, starting from the starting set {S_i} and ending set {T_j} respectively. Compare all airworthiness nodes found in the starting direction with those found in the ending direction, and use the airworthiness nodes that exist in both the starting set and the ending set as key connection points. Based on the key connection points, the backbone topology network is established.

[0069] Specifically, when constructing the backbone topology network, the flight start point and flight destination have been determined. The edge weight can be set to 1, and the number of hubs to be n (where n is a positive integer). Then, using the edge weight and the number of hubs n as references, a bidirectional breadth-first search is performed on the airworthiness nodes, starting from the start point set {S_i} and ending point set {T_j} respectively. That is, n layers are searched from the start point set {S_i} to the ending point set {T_j}, and simultaneously n layers are searched from the ending point set {T_j} to the start point set {S_i}. Next, all airworthiness nodes found in the start direction are compared with all airworthiness nodes found in the ending direction. Airworthiness nodes appearing in both sets are considered key connection points. Finally, paths from the start point set to key connection points, and from key connection points to the ending point set, are merged to form a connected directed acyclic graph, i.e., the backbone topology network.

[0070] The backbone topology network generated by bidirectional search ensures that the number of hubs traversed by any feasible path from the flight start point to the flight end point is within a reasonable range, such as n to 2n. This avoids overly tortuous and lengthy paths while providing sufficient redundancy options. In subsequent optimization, the system can quickly calculate specific paths that balance safety, efficiency, and other factors in this backbone topology network based on real-time scores and edge weights.

[0071] After the backbone topology network is constructed, it will be persistently stored and used as the input framework for the next step of fine-grained segment filling. The backbone topology network will be updated in real time as the airworthiness scores of the network hubs change dynamically. When the score of a node on a certain path drops sharply, the system can quickly remove the failed node from the redundant network and quickly replan the path based on the remaining nodes, which greatly improves the system's real-time response capability and robustness.

[0072] S50: Perform fine-grained segment filling on the main trunk topology network, and calculate the total flight consumption value of each segment from the flight start point to the flight end point based on the real-time airworthiness score of the key connection point.

[0073] The backbone topology network defines the "skeleton" of flight, but it does not provide specific safe and flyable flight segments. Therefore, it is necessary to calculate the globally optimal path based on the backbone topology network.

[0074] Generating one or more flyable detailed paths for any two adjacent critical connection points in the backbone network topology is called fine-grained segment filling. Fine-grained segment filling is computationally intensive, therefore it is asynchronously completed by a separate, continuously running geographic information processing service. This service monitors events such as hub status, no-fly zone updates, and 3D model updates in the backbone network topology. Once it detects data changes affecting segment feasibility, it triggers a recalculation of the affected segments.

[0075] When performing fine-grained segment filling on the backbone road network, the filling process deeply integrates multi-source data. The specific filling process is as follows: Collision detection and airspace compliance verification are performed based on flight infrastructure data and environmental geographic information data to determine whether there is a straight flight path between two key connection points. If so, fill the straight segment between the two key connection points; Otherwise, activate the detour algorithm to avoid obstacles and controlled airspace, filling the detour segment between the two key connection points.

[0076] For example, to calculate a straight flight path from node A to node B, it is necessary to perform collision checks and airspace compliance checks based on information such as 3D urban building information in environmental geographic information data and real-time airspace data in flight infrastructure data. If the checks pass, it is determined that a direct flight segment can be generated between node A and node B, and a straight flight segment is filled between node A and node B. If the checks fail, it means that the straight flight segment will be blocked by obstacles, such as tall buildings, mountains, or no-fly zones. In this case, a detour algorithm is activated, starting from node A and ending at node B, avoiding static obstacles and controlled airspace as hard constraints, and using factors such as signal strength, ground risks, and public privacy as optimization objectives to automatically generate a smooth and safe detour segment.

[0077] The fine-grained flight segments are filled with a series of ordered latitude and longitude coordinates, altitude coordinates, length, estimated flight time and other metadata. This data is published to a shared storage and notified to the path planning service in the form of notification messages, achieving decoupling and efficient updates.

[0078] Each fine-grained flight segment is assigned a quantified Flight Cost Value (FCV), which comprehensively reflects all the costs and risks of flying that segment. In some implementations, calculating the total flight cost value for each segment from the flight origin to the flight destination based on the real-time airworthiness score of the key connection point may include: Calculate the initial flight cost for each segment; The total flight cost for each segment is calculated based on the initial flight cost for each segment and the real-time airworthiness scores of the key connection points at both ends of the segment.

[0079] Specifically, the Initial Flight Cost (IFC) is the basic cost of a flight segment's inherent attributes, calculated using the weighted statistical model of the aforementioned multi-source data. Its calculation factors include distance cost, signal cost, terrain risk cost, and energy cost. Distance cost refers to the actual flight distance of the segment; signal cost is based on the communication map and predicts the average signal quality on the segment, with higher costs for weaker signals; terrain risk cost refers to the risk weights of the land surface types below the segment, such as densely populated areas, highways, and waterways; and energy cost is an estimated energy consumption figure that takes into account changes in altitude and predicted wind direction and speed.

[0080] To transmit the dynamic safety status of the backbone network to path selection, this application introduces a key dynamic penalty mechanism. The final flight cost of a segment is dynamically adjusted based on the safety status of its two critical connection points. Taking the segment between critical connection point A and critical connection point B as an example, the total flight cost of the segment between critical connection point A and critical connection point B is expressed by the following formula: Final FCV = Initial FCV × (100 / min(ACS_A, ACS_B)); Here, ACS_A and ACS_B are the real-time airworthiness scores of critical connection points A and B at the two ends of the flight segment, respectively, and the initial FCV represents the initial flight consumption value.

[0081] The formula above shows that the actual cost of a flight segment is negatively impacted by the weakest hub it connects to. For example... Figure 4 As shown, for example, even if a flight segment itself is ideal (low initial FCV), if it connects to a low-scoring hub (critical connection point), such as ACS=20, its final FCV will increase dramatically to 5 times its original value (calculated as 100 / 20=5). This will have a huge penalty effect on the path optimization algorithm, causing the system to automatically tend to avoid hub nodes (critical connection points) with poor safety status, thereby achieving safety-oriented automatic routing.

[0082] S60: Based on the total flight consumption value of each flight segment, form an undirected graph, and determine the optimal path and multiple redundant paths in the undirected graph.

[0083] After calculating the total flight consumption value for all segments, the entire flight environment from the flight start point S to the flight end point T is transformed into a weighted graph model, and each segment is a bidirectional symmetrical undirected edge with equal weight. Of course, there are explicit airspace rules, such as one-way air corridors, which will enforce directionality.

[0084] The non-directional nature of each flight segment greatly enhances the flexibility of path planning, allowing the algorithm to freely explore movement in any direction, thereby discovering more potential safe and low-cost paths in complex urban airspace and avoiding the possibility of missing better solution paths due to initial directional assumptions.

[0085] When determining the optimal path, a classic graph search algorithm, such as Dijkstra's algorithm, can be used to find the path with the minimum total flight cost from the starting point set {S_i} to the destination set {T_j} in the undirected graph. This path is the globally optimal path under the current environment, considering multiple objectives such as safety and efficiency. Meanwhile, other paths are considered redundant.

[0086] After calculating the optimal and redundant paths, the total flight consumption value of each path is output. Operators can set consumption thresholds for different task types. If the total flight consumption value of a path exceeds the consumption threshold, an alarm is triggered, and it is recommended to delay the task or re-evaluate the plan.

[0087] S70: Generate a comprehensive safety report for the optimal path so that the UAV can fly based on the optimal path, and switch to the redundant path when the optimal path is abnormal.

[0088] In some implementations, generating a comprehensive security report for the optimal path may include: The comprehensive safety report is presented in a three-dimensional visualization, including the overlay of the optimal path with a 3D real-world map, airspace rules, and signal heatmap.

[0089] Specifically, the comprehensive safety report includes total flight costs, airworthiness scores for each node along the optimal route, and key risk sources for each segment, such as weak signal areas and densely populated areas, along with corresponding recommendations. This provides operations personnel with a basis for rapid approval and decision-making. The planned routes, including optimal and redundant routes, are overlaid with data such as 3D reality maps, airspace rules, and signal heatmaps, and displayed through a visual interface. Operations personnel can intuitively review the overall route and zoom in to view details, greatly improving situational awareness and management efficiency.

[0090] Throughout the entire process of the drone flying along the optimal path, it will undergo full-cycle dynamic monitoring and adaptive adjustments, including dynamic risk warning and seamless path replanning.

[0091] Among them, dynamic risk warning continuously compares real-time environmental data with the plan. Once a significant deterioration in the safety status of a node or flight segment is detected, such as a sudden drop in node score, a sudden no-fly zone, or an intrusion of a severe weather system, different levels of alarms will be triggered immediately.

[0092] When a high-risk alarm is triggered, the system will not wait passively, but will select the suboptimal path from the redundant paths to guide the drone to seamlessly switch flight routes, ensuring the continuity and safety of the flight mission.

[0093] In extreme cases, such as when alternative routes are also impassable, the system evolves its route planning into a collaborative decision-making center, capable of issuing advanced commands to the drone, such as: 1. Flight route interruption and emergency landing: Instruct the drone to immediately abandon its original mission and fly to the safest available road network hub for hovering or emergency landing.

[0094] 2. Autonomous obstacle avoidance and emergency response: In the final segment of the flight, the drone is guided to activate its onboard sensors for local real-time obstacle avoidance and autonomously execute emergency landing procedures to maximize the safety of people, the drone, and the environment.

[0095] While the drone is in flight, all input data sources are continuously monitored, including weather, airspace, signals and hub status. Any changes will trigger the recalculation of the FCV for the affected flight segment and the update of the hub landing score.

[0096] If the total flight cost of the drone's current flight path increases significantly due to environmental degradation, it can immediately switch to a redundant path within milliseconds based on the latest undirected graph. Therefore, even for two identical start and end points, the recommended paths obtained at different times or by different drones may be completely different, achieving optimal path planning based on a real-time digital twin environment and full-cycle dynamic adaptation.

[0097] Furthermore, the detailed telemetry data, path compliance, and incident records collected during each flight mission are retrieved and used for the following purposes: 1. Model iteration is used to verify and optimize the accuracy of the weight allocation and segment cost calculation formula of the safety scoring model.

[0098] 2. Hub network optimization: This analyzes frequently used hubs and flight segments to provide data support for the deployment of future infrastructure, such as refueling points and communication base stations.

[0099] 3. Generate flight logs and analysis reports. The logs and reports comply with regulatory requirements, providing an immutable data chain for accident investigation, operational review, and insurance claims.

[0100] Through seamless switching during anomalies, combined with the expansion of the drone's autopilot capabilities, commands can be issued to trigger alarms, autonomous avoidance, and emergency landing, thereby collaboratively ensuring the safety and robustness of the entire flight.

[0101] The UAV flight path generation method and device based on road network hubs and dynamic safety scoring in this application first construct a primary road network hub network using multi-source data of flight infrastructure data and environmental geographic information data, and perform preliminary node screening. The obtained candidate nodes all have excellent hovering and emergency landing conditions, providing a safe haven for the UAV throughout the entire flight path, greatly enhancing the ability to handle sudden failures such as power failure and brief signal interruption. Second, a hierarchical and quantitative dynamic safety scoring mechanism is introduced to transform various risks into specific and calculable values, obtaining a secondary road network hub network containing multiple airworthiness nodes. After constructing the backbone topology road network, fine-grained flight segment filling is performed, and real-time airworthiness capability scoring is conducted. This enables proactive avoidance of dangerous areas, achieving a leap from "collision avoidance" to "pursuing optimal safety." The design of redundant paths ensures that if a node on the optimal path fails, a route can be quickly planned to achieve seamless switching, avoid mission interruption, and improve overall reliability.

[0102] This invention significantly enhances the real-time adaptability and decision-making efficiency of route planning. Specifically, it continuously updates real-time scores for flight segments and key connection points through asynchronous services, decoupling time-consuming fine-grained calculations from real-time route optimization. This allows the system to respond to environmental changes such as sudden weather events and temporary flight restrictions within milliseconds. Secondly, by building global route planning on an undirected graph model, the algorithm gains greater flexibility, enabling it to discover more potential safe routes in complex urban airspace and avoiding missing optimal solutions due to model limitations. Finally, the intelligent management process automatically generates visual reports and safety alerts, freeing operators from tedious data analysis and allowing them to focus on high-order decision-making, significantly lowering the operational threshold and improving decision-making efficiency.

[0103] This invention also optimizes operational economics, saving energy and operating costs. On one hand, through precise risk assessment and optimal path optimization, drones no longer need to frequently perform large-scale emergency maneuvers such as emergency detours or return flights, thereby reducing overall energy consumption and extending single flight time and mission radius. On the other hand, high reliability and high mission completion rate mean a lower risk of drone damage and maintenance frequency, directly saving hardware and maintenance costs. Simultaneously, automated workflows reduce reliance on experienced pilots, lowering labor costs.

[0104] This invention also achieves an optimal balance between computational efficiency and response speed, providing a foundation for large-scale applications. Specifically, facing the computational challenges of massive amounts of data in city-level low-altitude road networks, this invention, through innovative system architecture design, provides an ideal theoretical basis and practical solution for rapid computation and real-time response while ensuring planning quality. First, it adopts the core idea of ​​"layered computation and dynamic updating." Time-consuming tasks such as fine-grained segment filling and hub score updates are executed asynchronously by background services, decoupling them from lightweight global path planning. This eliminates the need for large-scale environmental modeling for each request, greatly reducing the pressure on real-time computation. Second, the dynamic local update mechanism means that when the environment changes, the system only needs to recalculate the scores and costs of affected hubs and segments, rather than reconstructing the entire road network, thus achieving millisecond-level path replanning response. Finally, pre-computation based on redundant topological road networks and undirected graph models provides a structured foundation for rapid path search. Solving for the optimal path is transformed into a pathfinding problem on a relatively sparse but information-complete graph, allowing classical path algorithms to achieve maximum efficiency. This design enables the invention to meet the demanding requirements of complex urban environments without relying on massive computing resources, making it feasible for implementation on both high-end servers and ordinary edge computing devices.

[0105] This invention also provides a technological foundation for large-scale, standardized low-altitude operations. It not only enhances the performance of individual drones but also empowers the entire system. Its standardized safety scores (ACS, FCV) provide drone operators with unified quantitative standards for safety, facilitating compliance reviews, insurance pricing, and performance evaluation. The constructed digital twin model of the low-altitude road network forms valuable spatial data assets, providing solid data support and insights for future urban air traffic (UAM) airspace management, infrastructure planning, and regulatory development, effectively promoting the maturity and development of the entire low-altitude economy.

[0106] In summary, this invention, through its innovative path generation method and systematic innovation, has produced positive and significant technical effects in multiple dimensions, including flight safety, real-time adaptability, operational economy, and industry standardization, solving long-standing pain points in the industry.

[0107] The present invention has found that current solutions for achieving safe path planning for UAVs in complex low-altitude flights over medium and long distances include fixed infrastructure heavy-duty solutions, high-authority embodied intelligence solutions, and pure real-time dynamic planning solutions.

[0108] One common alternative to the fixed infrastructure heavy-duty solution is to build fully fixed drone hubs, modeled after ground transportation networks or logistics transit stations, with complete operational and maintenance functions such as automatic charging, cargo transshipment, and in-depth maintenance. Drones complete long-distance missions through multiple takeoffs and landings and hopping-style "station-to-station" flights.

[0109] The high-authority embodied intelligence solution greatly enhances the autonomous intelligence of the UAV itself, namely "embodied intelligence," by granting it extremely high local decision-making authority, enabling it to fully autonomously perceive, calculate, and respond to all emergencies in real time during flight.

[0110] The pure real-time dynamic programming scheme abandons the concept of any pre-built network and relies entirely on a powerful central calculator. Each time the UAV requests a mission, it calculates an optimal path from the starting point to the destination from scratch based on the latest global real-time data.

[0111] In comparison, the advantages of this application are shown in Table 1 below: Table 1 Those skilled in the art will understand that all or part of the features / steps of the above-described method embodiments can be implemented by methods, data processing systems, or computer programs. These features may be implemented without hardware, entirely in software, or in a combination of hardware and software. The aforementioned computer program may be stored in one or more computer-readable storage media. When the computer program is executed (e.g., by a processor), it performs the above-described method for generating UAV flight paths based on road network hubs and dynamic safety scores.

[0112] The aforementioned storage media capable of storing program code include: static hard disks, solid-state hard disks, random access memory (SRAM), electrically erasable programmable read-only memory (EEPROM), erasable programmable read-only memory (EPROM), programmable read-only memory (PROM), read-only memory (ROM), optical storage devices, magnetic storage devices, flash memory, magnetic disks or optical disks, and / or combinations of the above devices, that is, they can be implemented by any type of volatile or non-volatile storage devices or combinations thereof.

[0113] This application also provides a drone flight path generation apparatus, comprising: one or more processors; a memory for storing one or more computer programs, wherein the one or more processors are configured to execute the one or more computer programs stored in the memory, so that the one or more processors execute a drone flight path generation method based on road network hubs and dynamic safety scoring as described in any one of Embodiment 1.

[0114] The UAV flight path generation device of the present invention can be a chip. When the UAV flight path generation device is a chip, each module in the chip can be implemented entirely or partially through software, hardware, or a combination thereof. The above-mentioned modules can be embedded in the processor of the computing device in hardware form or independent of it, or they can be stored in the memory of the computing device in software form, so that the processor can call and execute the operations corresponding to the above-mentioned modules.

[0115] The drone described in this application may include a drone flight path generation device. The drone may be a fixed-wing drone, a multi-rotor drone, etc.

[0116] Since the implementation of the aforementioned drones is described in detail in Embodiment 1, it will not be repeated here.

[0117] The above description is merely a preferred embodiment of the present invention. Those skilled in the art will understand that various changes or equivalent substitutions can be made to these features and embodiments without departing from the spirit and scope of the present invention. Furthermore, under the teachings of the present invention, these features and embodiments can be modified to adapt to specific situations and materials without departing from the spirit and scope of the present invention. Therefore, the present invention is not limited to the specific embodiments disclosed herein, and all embodiments falling within the scope of the claims of this application are within the protection scope of the present invention.

Claims

1. A method for generating UAV flight paths based on road network hubs and dynamic safety scoring, characterized in that, include: Based on flight infrastructure data and environmental geographic information data, a primary road network hub network is constructed, which includes several hub nodes; A preliminary screening of several hub nodes was conducted to obtain multiple candidate nodes that meet the conditions for hovering and emergency landing. The candidate nodes are initially scored for their airworthiness to obtain a secondary road network hub network including multiple airworthiness nodes; Before the flight mission is launched, based on the flight start point and flight end point, connectivity relationships are searched in the secondary road network hub network to construct a backbone topology road network including multiple key connection points. The main trunk topology network is filled with fine-grained flight segments, and the total flight consumption value of each flight segment from the flight start point to the flight end point is calculated based on the real-time airworthiness score of the key connection points. Based on the total flight consumption value of each flight segment, an undirected graph is formed, and the optimal path and multiple redundant paths are determined in the undirected graph. A comprehensive safety report is generated for the optimal path, allowing the UAV to fly based on the optimal path, and the redundant path is switched when an anomaly occurs on the optimal path.

2. The method for generating UAV flight paths based on road network hubs and dynamic safety scoring according to claim 1, characterized in that, The flight infrastructure data includes information on aprons, alternate landing points, charging and battery swapping stations, safe waypoints, no-fly zones, and restricted flight zones; the environmental geographic information data includes three-dimensional urban building information and topographic surface information.

3. The method for generating UAV flight paths based on road network hubs and dynamic safety scoring according to claim 2, characterized in that, The preliminary screening of several hub nodes yields multiple candidate nodes possessing hovering and emergency landing capabilities, including: Determine the physical space where several of the hub nodes are located; If the physical space where the hub node is located meets the alternate landing conditions, then the hub node that meets the alternate landing conditions will be used as a candidate node. The alternate landing conditions include compliance with alternate landing location, airspace, and communication.

4. The method for generating UAV flight paths based on road network hubs and dynamic safety scoring according to claim 1, characterized in that, The preliminary airworthiness assessment of the candidate nodes includes: The airworthiness score of each candidate node is calculated by a multi-factor weighted evaluation model. The multi-factor weighted evaluation model is determined based on the weight of the i-th evaluation factor and the normalized score of the i-th evaluation factor. The normalized score of the i-th evaluation factor includes surface stability, real-time meteorological conditions, infrastructure support, historical safety record, and dynamic airspace occupancy rate. The airworthiness scores of the candidate nodes are graded, and candidate nodes whose scores fall within the unairworthiness range are isolated to obtain multiple levels of airworthiness nodes. The node scores are then used as the node weights of the airworthiness nodes.

5. The method for generating UAV flight paths based on road network hubs and dynamic safety scoring according to claim 1, characterized in that, The process of searching for connectivity relationships in the secondary road network hub network based on the flight origin and destination, and constructing the backbone topology road network, includes: If the two airworthiness nodes satisfy the adjacent constraint, then a directed edge is established and the directed edge is assigned an edge weight. Using the flight start point as the source point S and the flight destination point as the target point T, a graph theory algorithm is used to search for connectivity in the secondary road network hub network to form a directed topological network containing multiple routes from the flight start point to the flight destination, and this directed topological network is used as the backbone topological network.

6. The method for generating UAV flight paths based on road network hubs and dynamic safety scoring according to claim 5, characterized in that, The process involves using the flight origin as the source point S and the flight destination as the target T, employing graph theory algorithms to search for connectivity relationships within the secondary road network hub network. This results in a directed topological network containing multiple routes from the flight origin to the flight destination, which is then used as the backbone topological network. Set the edge weights and the number of hubs n; Based on the edge weights and the number of hubs n, a bidirectional breadth-first search is performed on the airworthiness nodes, starting from the starting set {S_i} and ending set {T_j} respectively. Compare all airworthiness nodes found in the starting direction with those found in the ending direction, and use the airworthiness nodes that exist in both the starting set and the ending set as key connection points. Based on the key connection points, the backbone topology network is established.

7. The method for generating UAV flight paths based on road network hubs and dynamic safety scoring according to claim 5, characterized in that, The fine-grained segment filling of the backbone topology network includes: Collision detection and airspace compliance verification are performed based on flight infrastructure data and environmental geographic information data to determine whether there is a straight flight path between two key connection points. If so, fill the straight segment between the two key connection points; Otherwise, activate the detour algorithm to avoid obstacles and controlled airspace, filling the detour segment between the two key connection points.

8. The method for generating UAV flight paths based on road network hubs and dynamic safety scoring according to claim 5, characterized in that, The calculation of the total flight consumption value for each segment from the flight origin to the flight destination based on the real-time airworthiness score of the key connection point includes: Calculate the initial flight cost for each segment; The total flight cost for each segment is calculated based on the initial flight cost for each segment and the real-time airworthiness scores of the key connection points at both ends of the segment.

9. The method for generating UAV flight paths based on road network hubs and dynamic safety scoring according to claim 1, characterized in that, The generation of a comprehensive security report for the optimal path includes: The comprehensive safety report is presented in a three-dimensional visualization, including the overlay of the optimal path with a 3D real-world map, airspace rules, and signal heatmap.

10. A drone flight path generation device, characterized in that, include: One or more processors; A memory for storing one or more computer programs, and one or more processors for executing the one or more computer programs stored in the memory to cause the one or more processors to perform a method for generating UAV flight paths based on road network hubs and dynamic safety scores as described in any one of claims 1-9.

Images