A method and system for judging safety of low-altitude flight of a UAV

Abstract

The application relates to the technical field of low-altitude flight, and discloses a method and system for judging the safety of low-altitude flight of an unmanned aerial vehicle. The method comprises the following steps: performing data preprocessing on flight state data of the unmanned aerial vehicle and three-dimensional environment data to obtain standard flight state data and standard three-dimensional environment data; performing adaptive gridding processing on the standard three-dimensional environment data to obtain a three-dimensional dynamic gridding map; performing multi-dimensional feature extraction on the three-dimensional dynamic gridding map to obtain risk quantification features; performing weight distribution on the risk quantification features, fusing and evaluating the weighted data to obtain a comprehensive risk evaluation value; comparing the comprehensive risk evaluation value with a safety threshold to obtain a flight safety level; and mapping the flight safety level to a preset safety flight instruction library to obtain a flight adjustment control instruction of the unmanned aerial vehicle. The application can improve the efficiency of judging the safety of low-altitude flight of an unmanned aerial vehicle.

Description

Technical Field

[0001] This invention relates to the field of low-altitude flight technology, and in particular to a method and system for judging the safety of low-altitude flight of unmanned aerial vehicles (UAVs). Background Technology

[0002] Existing methods for assessing the safety of low-altitude UAV flight have significant shortcomings in data processing. They fail to perform accurate and comprehensive preprocessing of flight status data and 3D environmental data. They often lack standardized data format parsing and conversion mechanisms, making it difficult to effectively identify and handle outliers and missing segments in the data. Furthermore, they neglect the adaptation and conversion between geographic coordinate systems and UAV navigation coordinate references, and fail to achieve data temporal alignment and sampling rate standardization. This leads to biases in the basic data upon which subsequent analysis relies, directly affecting the accuracy of safety assessments.

[0003] Existing technologies fall short in terms of dynamic adaptability and comprehensiveness in risk assessment. They cannot adjust environmental modeling parameters based on the real-time flight status of UAVs, and their gridding processing often uses fixed granularity and range, making it difficult to match the needs of different flight speeds, altitudes, and mission types. Furthermore, risk feature extraction is limited to a single dimension, failing to fully integrate multi-dimensional information such as airspace type distribution, spatial accessibility, and dynamic environmental changes. Weight allocation also lacks differentiation based on flight phase and mission type, resulting in a comprehensive risk assessment value that cannot objectively reflect actual flight risks. Safety level determination lacks specificity, which in turn affects the effectiveness of flight adjustment and control commands, making it difficult to guarantee the safety and stability of low-altitude flight. Summary of the Invention

[0004] This invention provides a method and system for judging the safety of low-altitude flight of unmanned aerial vehicles (UAVs) to solve the problems mentioned in the background art.

[0005] To achieve the above objectives, the present invention provides a method for determining the safety of low-altitude flight of unmanned aerial vehicles (UAVs), comprising: S1. Perform data preprocessing on the UAV's flight status data and three-dimensional environment data to obtain the UAV's standard flight status data and standard three-dimensional environment data. S2. Based on the standard flight status data, the standard three-dimensional environment data is subjected to adaptive rasterization processing to obtain a three-dimensional dynamic rasterized map of the UAV. S3. Perform multi-dimensional feature extraction on the three-dimensional dynamic rasterized map to obtain the risk quantification features of the UAV; S4. Based on the flight stage and mission type in the flight status data, the risk quantification features are weighted and the weighted data is fused and evaluated to obtain the comprehensive risk assessment value of the UAV. S5. The comprehensive risk assessment value is compared with a safety threshold to obtain the flight safety level of the UAV; S6. Map the flight safety level to a preset safe flight command library to obtain the flight adjustment control command for the UAV.

[0006] In a preferred embodiment, the step of preprocessing the UAV's flight status data and 3D environment data to obtain the UAV's standard flight status data and standard 3D environment data includes: The data formats of the UAV's flight status data and 3D environment data are parsed and converted to obtain the UAV's regular flight status data and regular 3D environment data; Anomaly detection is performed on the regularized flight state data to obtain the abnormal data points and missing data segments of the regularized flight state data. Remove the abnormal data points and fill the missing data segments with linear interpolation to obtain the clean flight status data of the UAV; Based on the preset geographic coordinate system transformation rules, the coordinate system of the regular three-dimensional environment data is transformed to the navigation coordinate reference of the UAV, so as to obtain the corrected three-dimensional environment data of the UAV. The clean flight status data and the three-dimensional environment data are time-aligned, and the sampling rate of the aligned data is standardized and resampled to obtain the standard flight status data and the standard three-dimensional environment data.

[0007] In a preferred embodiment, the step of adaptively rasterizing the standard three-dimensional environment data based on the standard flight status data to obtain a three-dimensional dynamic rasterized map of the UAV includes: Based on the real-time flight speed and altitude data in the standard flight status data, the grid granularity parameters and spatial range parameters for spatial division of the standard three-dimensional environment data are determined. Based on the determined grid granularity parameters and spatial range parameters, a three-dimensional cubic grid array with uniform size is established within the three-dimensional space of the standard three-dimensional environmental data. The terrain elevation information, static obstacle location information, and dynamic obstacle prediction location information in the standard three-dimensional environment data are mapped and associated with the three-dimensional cube grid array to obtain the comprehensive attribute information of the three-dimensional cube grid array. The three-dimensional cubic grid array and the comprehensive attribute information are structurally integrated to obtain a three-dimensional dynamic rasterized map of the UAV.

[0008] In a preferred embodiment, determining the raster granularity parameters and spatial range parameters for spatially dividing the standard three-dimensional environmental data based on the real-time flight speed and altitude data in the standard flight status data includes: Extract the real-time flight speed and real-time flight altitude from the standard flight status data; The real-time flight speed is verified and matched with a preset speed threshold range, and the initial raster granularity parameters of the three-dimensional dynamic rasterized map are determined based on the matched speed threshold range. The real-time flight altitude is compared and matched with a preset altitude threshold range, and the initial spatial range parameters of the three-dimensional dynamic raster map are determined based on the matched altitude threshold range. Based on the mission requirements of the UAV in its current flight phase, the initial grid granularity parameters and the initial spatial range parameters are constrained and adjusted to obtain the grid granularity parameters and the spatial range parameters of the standard three-dimensional environment data.

[0009] In a preferred embodiment, the step of extracting multi-dimensional features from the three-dimensional dynamic rasterized map to obtain the risk quantification features of the UAV includes: The three-dimensional dynamic rasterized map is subjected to description category separation to obtain the spatial type attribute and risk identification attribute of the three-dimensional dynamic rasterized map; Based on the airspace type attribute, the distribution ratio of the UAV in different types of airspace within the current flight airspace is statistically analyzed, and the statistical results are integrated with the risk identification attribute to form the environmental risk characteristics of the UAV. In the three-dimensional dynamic rasterized map, with the current position and heading of the UAV as the center direction, a passable space analysis is performed to obtain the spatial accessibility and congestion characteristics of the UAV. A time-series comparative analysis of the raster state change information in the three-dimensional dynamic rasterized map is performed to obtain the environmental dynamic changes and trend characteristics of the three-dimensional dynamic rasterized map; The risk quantification characteristics of the UAV are obtained by fusing the environmental risk characteristics, spatial accessibility, congestion level characteristics, environmental dynamic changes, and trend characteristics through multimodal feature fusion.

[0010] In a preferred embodiment, the step of assigning weights to the risk quantification features based on the flight phase and mission type in the flight status data, and then fusing and evaluating the weighted data to obtain the comprehensive risk assessment value of the UAV, includes: Extract the current flight stage and current mission type of the UAV from the flight status data; Based on a preset flight phase-mission type-weight mapping table, corresponding weight coefficients are assigned to different dimensions of the risk quantification features according to the current flight phase and the current mission type. Based on the weighting coefficients, the risk quantification features are weighted and evaluated to obtain the risk dimension values ​​of the risk quantification features; Based on a preset risk level benchmark, the risk dimension values ​​are fused from multiple dimensions and normalized to obtain the comprehensive risk assessment value of the drone.

[0011] In a preferred embodiment, the risk dimension value is calculated using the following formula: ; In the formula, The value of the aforementioned risk dimension. For the first The weighting coefficients for each risk dimension, The total number of risk dimensions included in the aforementioned risk quantification features. For the first Normalized eigenvalues ​​for each risk dimension For the first Normalized eigenvalues ​​for each risk dimension For the first The risk dimension for the first The correlation coefficient of each risk dimension.

[0012] In a preferred embodiment, the step of comparing the comprehensive risk assessment value with a safety threshold to obtain the flight safety level of the UAV includes: Based on the current flight stage and current mission type in the flight status data, the comprehensive risk assessment value is mapped to a preset configuration database to obtain the dynamic safety threshold configuration of the UAV. The comprehensive risk assessment value is compared with the dynamic security threshold configuration to obtain a specific risk threshold range for the comprehensive risk assessment value. Discrete safety level calibration is performed on the specific risk threshold range to obtain the flight safety level of the UAV.

[0013] In a preferred embodiment, mapping the flight safety level to a preset safe flight command library to obtain flight adjustment control commands for the UAV includes: Read the flight safety level of the UAV, the current standard flight status data of the UAV, and the current mission objective; Based on the flight safety level, a preset safe flight instruction library is searched and matched to obtain a candidate instruction set for the UAV. The safe flight instruction library stores flight operation instructions that can be executed under different safety levels and their execution conditions. Based on the current standard flight status data and the current mission objective, the candidate instruction set is verified for feasibility and prioritized to obtain the optimized instruction sequence for the UAV. The optimized instruction sequence is encapsulated into flight adjustment control instructions for the UAV.

[0014] To address the above problems, the present invention also provides a safety assessment system for low-altitude flight of unmanned aerial vehicles (UAVs), the system comprising: The data preprocessing module is used to preprocess the flight status data and three-dimensional environment data of the UAV to obtain the standard flight status data and standard three-dimensional environment data of the UAV. An adaptive rasterization processing module is used to perform adaptive rasterization processing on the standard three-dimensional environment data based on the standard flight status data to obtain a three-dimensional dynamic rasterized map of the UAV. A multi-dimensional feature extraction module is used to extract multi-dimensional features from the three-dimensional dynamic rasterized map to obtain the risk quantification features of the UAV. The weighting and fusion evaluation module is used to assign weights to the risk quantification features based on the flight phase and mission type in the flight status data, and to fuse and evaluate the weighted data to obtain the comprehensive risk assessment value of the UAV. The safety threshold comparison module is used to compare the comprehensive risk assessment value with a safety threshold to obtain the flight safety level of the UAV. The command mapping module is used to map the flight safety level to a preset safe flight command library to obtain the flight adjustment control commands of the UAV.

[0015] Compared with the prior art, the present invention has the following beneficial effects: 1. This invention implements a comprehensive and standardized preprocessing process for UAV flight status data and 3D environmental data. This includes data format parsing and conversion, removal of outlier data points, filling of missing data segments, coordinate system adaptation and conversion, time sequence alignment, and sampling rate standardization. This ensures that the standard data possesses high accuracy and consistency, laying a solid data foundation for subsequent safety assessments. Simultaneously, based on real-time flight speed and altitude data, the invention adaptively adjusts the grid granularity and spatial range parameters to construct a 3D dynamic rasterized map that accurately maps environmental information such as terrain elevation and static and dynamic obstacles. This achieves dynamic adaptation between environmental modeling and flight status, significantly improving the real-time performance and relevance of risk perception.

[0016] 2. This invention integrates key information such as environmental risk, spatial accessibility, congestion level, and dynamic changes and trends in the environment through multi-dimensional feature extraction. It assigns differentiated weights to risk features of different dimensions based on flight phase and mission type. After multi-dimensional fusion and normalization, a comprehensive risk assessment value is obtained, making risk assessment more comprehensive and aligned with actual needs. By comparing dynamic safety thresholds, the invention accurately calibrates flight safety levels and performs feasibility verification and priority ranking of candidate commands based on the current flight status and mission objectives, generating optimized flight adjustment control commands. This effectively improves the accuracy and efficiency of low-altitude flight safety assessment for UAVs, ensuring safe and stable flight operations. Attached Figure Description

[0017] Figure 1 This is a flowchart illustrating a method for determining the safety of low-altitude flight of an unmanned aerial vehicle (UAV) according to an embodiment of the present invention. Figure 2 This is a functional block diagram of a low-altitude flight safety assessment system for unmanned aerial vehicles (UAVs) provided in an embodiment of the present invention. The realization of the objective, functional features and advantages of the present invention will be further explained in conjunction with the embodiments and with reference to the accompanying drawings. Detailed Implementation

[0018] It should be understood that the specific embodiments described herein are merely illustrative of the invention and are not intended to limit the invention.

[0019] This application provides a method for determining the safety of low-altitude flight of unmanned aerial vehicles (UAVs). The executing entity of this method includes, but is not limited to, at least one of the following electronic devices that can be configured to execute the method provided in this application: a server, a terminal, etc. In other words, the method for determining the safety of low-altitude flight of UAVs can be executed by software or hardware installed on a terminal device or a server device. The server includes, but is not limited to, a single server, a server cluster, a cloud server, or a cloud server cluster. The server can be an independent server or a cloud server that provides basic cloud computing services such as cloud services, cloud databases, cloud computing, cloud functions, cloud storage, network services, cloud communication, middleware services, domain name services, security services, content delivery networks (CDNs), and big data and artificial intelligence platforms.

[0020] Reference Figure 1 The diagram shown is a flowchart illustrating a method for determining the safety of low-altitude flight of a drone according to an embodiment of the present invention. In this embodiment, the method for determining the safety of low-altitude flight of a drone includes: S1. Perform data preprocessing on the UAV's flight status data and three-dimensional environment data to obtain the UAV's standard flight status data and standard three-dimensional environment data. In this embodiment of the invention, the step of preprocessing the UAV's flight status data and three-dimensional environment data to obtain the UAV's standard flight status data and standard three-dimensional environment data includes: The data formats of the UAV's flight status data and 3D environment data are parsed and converted to obtain the UAV's regular flight status data and regular 3D environment data; Anomaly detection is performed on the regularized flight state data to obtain the abnormal data points and missing data segments of the regularized flight state data. Remove the abnormal data points and fill the missing data segments with linear interpolation to obtain the clean flight status data of the UAV; Based on the preset geographic coordinate system transformation rules, the coordinate system of the regular three-dimensional environment data is transformed to the navigation coordinate reference of the UAV, so as to obtain the corrected three-dimensional environment data of the UAV. The clean flight status data and the three-dimensional environment data are time-aligned, and the sampling rate of the aligned data is standardized and resampled to obtain the standard flight status data and the standard three-dimensional environment data.

[0021] When parsing and converting the flight status data and 3D environment data of the UAV, the original storage format of the UAV flight status data is first identified, including binary format, character-delimited format, etc. Then, key information in the flight status data is extracted according to the preset format parsing rules. Key information includes flight altitude, flight speed, flight heading, etc. At the same time, the original storage format of the 3D environment data is identified, including point cloud format, raster format, etc. Key information in the 3D environment data is extracted according to the corresponding format parsing rules. Key information includes environmental terrain elevation, obstacle position, spatial coordinates, etc. Subsequently, the two types of key information are converted into preset structured data formats. The fields of the structured data format include data identifier, data content, timestamp, etc. Finally, the regularized flight status data and regularized 3D environment data of the UAV are obtained.

[0022] When performing anomaly detection on the regularized flight status data, a preset normal range of flight status values ​​is first retrieved. This range is determined based on the performance parameters calibrated by the UAV at the factory and the measured data from routine flight tests. The value of each data point in the regularized flight status data is compared with this normal range one by one. Flight status data points whose values ​​exceed the normal range are identified as abnormal data points. Then, a preset flight status data sampling interval threshold is retrieved. This threshold is determined based on the original acquisition frequency of the UAV flight status data. The timestamp sequence of the regularized flight status data is checked. Time periods where the interval between two adjacent timestamps is greater than the sampling interval threshold are identified as data missing segments. Finally, the abnormal data points and data missing segments of the regularized flight status data are obtained. When removing the abnormal data points and performing linear interpolation to fill the missing data segments, the detected abnormal data points are directly removed from the regular flight status data. Then, for each missing data segment, the value of the previous valid data point corresponding to the start timestamp of the missing segment and the value of the next valid data point corresponding to the end timestamp of the missing segment are determined. According to the number of timestamps in the missing segment, the uniform change between the values ​​of the two valid data points is calculated. The value corresponding to the uniform change is sequentially filled into the position corresponding to each timestamp of the missing segment, and finally the clean flight status data of the UAV is obtained.

[0023] Based on preset geographic coordinate system transformation rules, when transforming the coordinate system of the regularized 3D environment data to the navigation coordinate reference of the UAV, the preset geographic coordinate system transformation rules are first retrieved. These rules include coordinate origin translation parameters and coordinate axis rotation angle parameters. The coordinate origin translation parameter refers to calculating the spatial position difference between the origin of the original coordinate system of the regularized 3D environment data and the target origin, with the origin of the UAV's navigation coordinate reference as the target origin. The coordinate axis rotation angle parameter refers to calculating the angle between the coordinate axis of the original coordinate system of the regularized 3D environment data and the target coordinate axis, with the direction of the coordinate axis of the UAV's navigation coordinate reference as the target direction. Then, following the order of translation followed by rotation, the coordinates of each spatial point in the regularized 3D environment data are adjusted so that the coordinate system of the adjusted 3D environment data is completely consistent with the UAV's navigation coordinate reference, ultimately obtaining the corrected 3D environment data of the UAV.

[0024] When performing time-series alignment of the clean flight status data and the corrected 3D environment data, and standardizing the sampling rate of the aligned data for resampling, the timestamp sequence of the clean flight status data is first used as the reference sequence. The timestamp sequence of the corrected 3D environment data is matched one by one with the reference sequence. Data points in the corrected 3D environment data that have the same timestamp as the reference sequence are retained, while data points in the corrected 3D environment data that exceed the time range of the reference sequence are removed. This completes the time-series alignment of the two types of data. Then, a preset standard sampling rate is retrieved. This sampling rate is determined based on the needs of subsequent UAV data processing. The time-series aligned clean flight status data and corrected 3D environment data are resampled. The original data points corresponding to the timestamps of the standard sampling rate are retained. For positions where there are no corresponding original data points for the timestamps of the standard sampling rate, the average value of the preceding and following valid data points adjacent to that position is used to supplement the data. Finally, the standard flight status data and the standard 3D environment data are obtained.

[0025] The beneficial effects are that by performing a series of processes on UAV flight status data and 3D environment data, including format parsing and conversion, anomaly detection and removal, interpolation and filling, coordinate system transformation, temporal alignment, and standardized resampling, the original two types of data with different formats, coordinate references, temporal sampling rates, and anomaly / missing issues can be transformed into standard flight status data and standard 3D environment data with regular formats, clean data, unified coordinate references, consistent temporal sequences, and standardized sampling rates. This effectively solves the defects of the original data, such as disordered formats, data anomalies / missing information, mismatched coordinate references, asynchronous temporal sequences, and inconsistent sampling rates, providing high-quality and highly consistent data source support for subsequent UAV-related analysis and control work based on these two types of data.

[0026] S2. Based on the standard flight status data, the standard three-dimensional environment data is subjected to adaptive rasterization processing to obtain a three-dimensional dynamic rasterized map of the UAV. In this embodiment of the invention, the step of performing adaptive rasterization processing on the standard three-dimensional environment data based on the standard flight state data to obtain a three-dimensional dynamic rasterized map of the UAV includes: Based on the real-time flight speed and altitude data in the standard flight status data, the grid granularity parameters and spatial range parameters for spatial division of the standard three-dimensional environment data are determined. Based on the determined grid granularity parameters and spatial range parameters, a three-dimensional cubic grid array with uniform size is established within the three-dimensional space of the standard three-dimensional environmental data. The terrain elevation information, static obstacle location information, and dynamic obstacle prediction location information in the standard three-dimensional environment data are mapped and associated with the three-dimensional cube grid array to obtain the comprehensive attribute information of the three-dimensional cube grid array. The three-dimensional cubic grid array and the comprehensive attribute information are structurally integrated to obtain a three-dimensional dynamic rasterized map of the UAV.

[0027] The determination of raster granularity parameters and spatial range parameters for spatial division of the standard three-dimensional environment data based on real-time flight speed and altitude data from the standard flight status data includes: Extract the real-time flight speed and real-time flight altitude from the standard flight status data; The real-time flight speed is verified and matched with a preset speed threshold range, and the initial raster granularity parameters of the three-dimensional dynamic rasterized map are determined based on the matched speed threshold range. The real-time flight altitude is compared and matched with a preset altitude threshold range, and the initial spatial range parameters of the three-dimensional dynamic raster map are determined based on the matched altitude threshold range. Based on the mission requirements of the UAV in its current flight phase, the initial grid granularity parameters and the initial spatial range parameters are constrained and adjusted to obtain the grid granularity parameters and the spatial range parameters of the standard three-dimensional environment data.

[0028] Based on the real-time flight speed and altitude data in the standard flight status data, when determining the grid granularity parameters and spatial range parameters for spatial division of the standard three-dimensional environment data, the preset correspondence between flight speed and grid granularity, and the correspondence between flight altitude and spatial range are first retrieved. This correspondence is determined based on the environmental perception accuracy requirements of the UAV at different flight speeds and the environmental coverage requirements at different flight altitudes. The real-time flight speed in the standard flight status data is matched with the speed range in the correspondence. After successful matching, the corresponding grid granularity parameters are extracted. Then, the real-time flight altitude in the standard flight status data is matched with the altitude range in the correspondence. After successful matching, the corresponding spatial range parameters are extracted. The spatial range parameters include the longitudinal start and end coordinates, the lateral start and end coordinates, and the vertical start and end coordinates of the grid array.

[0029] Based on the determined grid granularity parameters and spatial range parameters, when establishing a three-dimensional cubic grid array with uniform size within the three-dimensional space of the standard three-dimensional environment data, the side length of a single three-dimensional cubic grid is determined by the grid granularity parameters to ensure that the length, width, and height of all three-dimensional cubic grids are consistent. Then, using the longitudinal start and end coordinates, lateral start and end coordinates, and vertical start and end coordinates determined by the spatial range parameters as boundaries, the three-dimensional space is divided sequentially according to the side length of a single three-dimensional cubic grid within the three-dimensional space corresponding to the standard three-dimensional environment data, starting from the boundary start position, until the entire three-dimensional space determined by the spatial range parameters is covered, ultimately forming a three-dimensional cubic grid array with uniform size.

[0030] Mapping and associating the terrain elevation information, static obstacle location information, and dynamic obstacle predicted location information from the standard 3D environmental data to the 3D cube grid array to obtain the comprehensive attribute information of the 3D cube grid array involves first extracting the terrain elevation information from the standard 3D environmental data, matching the spatial position of each 3D cube grid with the coordinates of the terrain elevation information, and assigning the corresponding terrain elevation value to the 3D cube grid if a match is successful. Then, the static obstacle location information from the standard 3D environmental data is extracted, and the spatial position of each 3D cube grid is compared with the coordinate range of the static obstacle. If the 3D cube... If the spatial location of a grid cell is within the coordinate range of a static obstacle, an attribute identifier indicating the presence of a static obstacle is added to the 3D cube grid cell. Then, the predicted location information of dynamic obstacles is extracted from the standard 3D environment data. The spatial location corresponding to each 3D cube grid cell is compared with the predicted coordinate range of the dynamic obstacle. If the spatial location of the 3D cube grid cell is within the predicted coordinate range of the dynamic obstacle, an attribute identifier indicating the predicted location of the dynamic obstacle is added to the 3D cube grid cell. The terrain elevation value, the static obstacle attribute identifier, and the dynamic obstacle predicted location attribute identifier are integrated to finally obtain the comprehensive attribute information of the 3D cube grid array.

[0031] When the three-dimensional cubic grid array and the comprehensive attribute information are structurally integrated to obtain the three-dimensional dynamic rasterized map of the UAV, a preset raster map structured format is retrieved. This format contains the spatial index information of the three-dimensional cubic grid and the corresponding attribute information entries. The spatial index information of each grid in the three-dimensional cubic grid array is bound one by one with the comprehensive attribute information corresponding to that grid, so that the spatial position of each grid and the attribute information form a unique corresponding relationship. Then, all the bound grid information is arranged in an orderly manner according to the preset structured format, and finally the three-dimensional dynamic rasterized map of the UAV is obtained.

[0032] When extracting real-time flight speed and real-time flight altitude from the standard flight status data, a preset structured field is retrieved from the standard flight status data. This structured field includes a flight speed field and a flight altitude field that correspond one-to-one with a timestamp. A timestamp consistent with the current UAV flight time is selected, and the flight speed field value corresponding to this timestamp is extracted as the real-time flight speed, and the flight altitude field value corresponding to this timestamp is extracted as the real-time flight altitude. The real-time flight speed is verified and matched with a preset speed threshold interval. When determining the initial raster granularity parameters of the 3D dynamic rasterized map based on the matched speed threshold interval, a correspondence table between the preset speed threshold interval and the initial raster granularity parameters is retrieved. This correspondence table is established based on the environmental perception resolution requirements of the UAV at different flight speeds. The preset speed threshold interval is divided into multiple continuous and non-overlapping intervals, and each interval corresponds to a unique initial raster granularity parameter. The extracted real-time flight speed is compared with each speed threshold interval one by one to determine the speed threshold interval to which the real-time flight speed belongs, and the initial raster granularity parameters corresponding to that speed threshold interval are extracted.

[0033] The real-time flight altitude is compared and matched with a preset altitude threshold range. When determining the initial spatial range parameters of the three-dimensional dynamic raster map based on the matched altitude threshold range, a correspondence table between the preset altitude threshold range and the initial spatial range parameters is retrieved. This correspondence table is established based on the environmental coverage requirements of the UAV at different flight altitudes. The preset altitude threshold range is divided into multiple continuous and non-overlapping intervals. Each interval corresponds to a unique initial spatial range parameter, which includes longitudinal start and end coordinates, lateral start and end coordinates, and vertical start and end coordinates. The extracted real-time flight altitude is compared with each altitude threshold range one by one to determine the altitude threshold range to which the real-time flight altitude belongs, and the initial spatial range parameters corresponding to that altitude threshold range are extracted.

[0034] Based on the mission requirements of the current flight phase of the UAV, the initial grid granularity parameters and the initial spatial range parameters are constrained and adjusted to obtain the grid granularity parameters and spatial range parameters of the standard three-dimensional environment data. Then, a correspondence table between the UAV flight phase mission requirements and parameter adjustment rules is retrieved. This table covers common flight phases such as takeoff, cruise, and landing, with each flight phase corresponding to specific parameter adjustment rules. The current flight phase of the UAV is determined, and the corresponding parameter adjustment rules are extracted. The initial grid granularity parameters are refined or coarsened according to these rules, and the start and end coordinates of the longitudinal, lateral, and vertical directions of the initial spatial range parameters are enlarged or reduced. The adjusted parameters are the grid granularity parameters and spatial range parameters of the standard three-dimensional environment data.

[0035] The beneficial effects are as follows: by accurately extracting real-time flight speed and altitude from standard flight status data, and combining this with a preset correspondence table to match speed and altitude with threshold ranges, the initial grid granularity parameters and initial spatial range parameters are determined. Then, the initial parameters are constrained and adjusted according to the mission requirements of the current flight phase of the UAV, ensuring that the grid granularity and spatial range parameters are adapted to the actual flight status and mission objectives. On this basis, a three-dimensional cubic grid array of uniform size is established, and terrain elevation, static obstacle position, and dynamic obstacle prediction position information are accurately mapped and associated to the grid array to form comprehensive attribute information. After structured integration, a three-dimensional dynamic gridded map is obtained, realizing deep adaptation between gridded processing and UAV flight status, mission requirements, and environmental data. This ensures that the three-dimensional dynamic gridded map can accurately reflect comprehensive environmental information, providing reliable environmental data support for subsequent UAV flight planning, obstacle avoidance, and other operations, thereby improving the safety and environmental adaptability of UAV flight.

[0036] S3. Perform multi-dimensional feature extraction on the three-dimensional dynamic rasterized map to obtain the risk quantification features of the UAV; In this embodiment of the invention, the step of extracting multi-dimensional features from the three-dimensional dynamic rasterized map to obtain the risk quantification features of the UAV includes: The three-dimensional dynamic rasterized map is subjected to description category separation to obtain the spatial type attribute and risk identification attribute of the three-dimensional dynamic rasterized map; Based on the airspace type attribute, the distribution ratio of the UAV in different types of airspace within the current flight airspace is statistically analyzed, and the statistical results are integrated with the risk identification attribute to form the environmental risk characteristics of the UAV. In the three-dimensional dynamic rasterized map, with the current position and heading of the UAV as the center direction, a passable space analysis is performed to obtain the spatial accessibility and congestion characteristics of the UAV. A time-series comparative analysis of the raster state change information in the three-dimensional dynamic rasterized map is performed to obtain the environmental dynamic changes and trend characteristics of the three-dimensional dynamic rasterized map; The risk quantification characteristics of the UAV are obtained by fusing the environmental risk characteristics, spatial accessibility, congestion level characteristics, environmental dynamic changes, and trend characteristics through multimodal feature fusion.

[0037] When performing descriptive category separation on the three-dimensional dynamic raster map to obtain the airspace type attribute and risk identification attribute of the three-dimensional dynamic raster map, a preset attribute classification rule is retrieved. This rule is established based on the functional attributes of airspace and the risk level classification standard. The comprehensive attribute information of all three-dimensional cube grids in the three-dimensional dynamic raster map is traversed, and the attribute content related to airspace function is filtered out and classified and integrated to form the airspace type attribute. The identification content related to risk level is filtered out and classified and integrated to form the risk identification attribute. Based on the airspace type attribute, when statistically analyzing the distribution ratio of different airspace types within the current flight airspace of the UAV, and integrating the statistical results with the risk identification attribute to form the environmental risk characteristics of the UAV, the boundary range of the current flight airspace of the UAV is first determined. This range is determined by the current position of the UAV and the preset airspace analysis radius. According to the classification standard of the airspace type attribute, the three-dimensional cubic grid within the current flight airspace is divided into different airspace type groups. The proportion of the number of grids corresponding to each airspace type group to the total number of grids in the current flight airspace is statistically analyzed. The statistical results of this proportion are associated and bound one-to-one with the risk identification attribute, so that the airspace distribution ratio information and the risk level information form a corresponding relationship, and finally the environmental risk characteristics are obtained.

[0038] In the three-dimensional dynamic rasterized map, with the current position and heading of the UAV as the center direction, a passable space analysis is performed to obtain the spatial accessibility and congestion characteristics of the UAV. A fan-shaped analysis area with a preset angle range and a preset distance range is defined with the current position of the UAV as the origin and the current heading as the central axis. All three-dimensional cubic grids within this fan-shaped analysis area are traversed. Based on the comprehensive attribute information of the grids, it is determined whether each grid is a passable grid. The proportion of the number of passable grids in the fan-shaped analysis area to the total number of grids in the area is counted; this proportion represents the spatial accessibility. The distribution density of passable grids within the fan-shaped analysis area is also counted; this density is reflected by the number of passable grids per unit volume; this distribution density represents the congestion characteristics.

[0039] To obtain the environmental dynamic changes and trend characteristics of the 3D dynamic rasterized map, a time-series comparative analysis of the raster state change information in the rasterized map is performed. This involves retrieving multiple frames of 3D dynamic rasterized map data within a preset time interval, determined based on the UAV's environmental perception frequency. The comprehensive attribute information of 3D cube rasteres at the same spatial location is compared one by one in different frames of map data. The locations and content of raster attribute changes are recorded, and the proportion of changed rasteres to the total number of rasteres is calculated. Based on the direction and frequency of raster attribute changes in consecutive frames of map data, the development trend of environmental changes is determined. The raster attribute change content, change ratio, and trend judgment results are integrated to finally obtain the environmental dynamic changes and trend characteristics.

[0040] When fusing the environmental risk characteristics, spatial accessibility, congestion level characteristics, environmental dynamic changes, and trend characteristics into multimodal features to obtain the risk quantification characteristics of the UAV, a preset feature fusion rule is invoked. This rule is established based on the influence weight of each feature on the flight risk of the UAV. The influence weight is determined through historical data of UAV flight tests. According to this rule, the information content of environmental risk characteristics, spatial accessibility, congestion level characteristics, environmental dynamic changes, and trend characteristics is systematically associated, so that the information of each feature complements each other and has no duplication or redundancy, forming a unified feature set, which is the risk quantification characteristics.

[0041] The beneficial effects are as follows: by separating the descriptive categories of the 3D dynamic raster map, the airspace type attribute and risk identification attribute are accurately separated, laying a clear foundation for subsequent feature extraction. Based on the airspace type attribute, the distribution ratio of different types of airspace is statistically analyzed and integrated with the risk identification attribute to form environmental risk features that can accurately reflect the risk level of the current flight environment. With the current position and heading of the UAV as the core, the passable space analysis is carried out to clarify the spatial accessibility and congestion characteristics, providing a direct basis for flight path planning. By comparing the raster state change information over time, the dynamic changes and trend characteristics of the environment are captured, enabling the prediction of environmental changes. Finally, through multimodal feature fusion and integration of various features, a comprehensive risk quantification feature that fits the actual flight needs is formed, providing accurate and comprehensive feature support for UAV flight risk assessment and avoidance decision-making, and improving the risk management capability of UAVs in dealing with complex environments.

[0042] S4. Based on the flight stage and mission type in the flight status data, the risk quantification features are weighted and the weighted data is fused and evaluated to obtain the comprehensive risk assessment value of the UAV. In this embodiment of the invention, the step of assigning weights to the risk quantification features based on the flight phase and mission type in the flight status data, and then fusing and evaluating the weighted data to obtain the comprehensive risk assessment value of the UAV, includes: Extract the current flight stage and current mission type of the UAV from the flight status data; Based on a preset flight phase-mission type-weight mapping table, corresponding weight coefficients are assigned to different dimensions of the risk quantification features according to the current flight phase and the current mission type. Based on the weighting coefficients, the risk quantification features are weighted and evaluated to obtain the risk dimension values ​​of the risk quantification features; Based on a preset risk level benchmark, the risk dimension values ​​are fused from multiple dimensions and normalized to obtain the comprehensive risk assessment value of the drone.

[0043] The formula for calculating the risk dimension value is as follows: ; In the formula, The value of the aforementioned risk dimension. For the first The weighting coefficients for each risk dimension, The total number of risk dimensions included in the aforementioned risk quantification features. For the first Normalized eigenvalues ​​for each risk dimension For the first Normalized eigenvalues ​​for each risk dimension For the first The risk dimension for the first The correlation coefficient of each risk dimension.

[0044] No. The weight coefficients for each risk dimension are derived from a pre-defined flight phase-mission type-weight mapping table. This mapping table is established based on the degree of influence of each risk dimension feature on UAV flight safety under different combinations of flight phases and mission types. The degree of influence is determined by analyzing a large amount of historical UAV flight test data. By combining the extracted current flight phase and current mission type of the UAV and matching the corresponding entries in the mapping table, the weight coefficient corresponding to that risk dimension can be obtained. The total number of risk dimensions included in the risk quantification features is the actual number of independent risk dimensions in the risk quantification features. The risk quantification features are a unified feature set formed by fusing environmental risk features, spatial accessibility, congestion degree features, and environmental dynamic changes and trends through multimodal feature fusion. The total number is directly determined by this fused feature set.

[0045] No. The normalized eigenvalues ​​of the first risk dimension and the second risk dimension The normalized feature values ​​of each risk dimension are derived from the original feature values ​​of the corresponding dimensions in the risk quantification features. These are obtained by mapping the original feature values ​​of each dimension to a preset unified evaluation interval. The mapping is based on a preset risk level benchmark, which is established based on historical test data of UAV safe flight and risk threshold classification standards, ensuring the comparability of feature values ​​across different dimensions. The risk dimension for the first The correlation coefficient of each risk dimension is derived from the statistical analysis of the drone flight history data. It is determined by sorting out the interaction patterns between each risk dimension under different flight scenarios. If one risk dimension has a positive impact on another risk dimension, a corresponding fixed value is taken; if there is no impact, a zero value is taken, which accurately reflects the correlation between dimensions.

[0046] The significance of this calculation method is that by integrating the weight coefficients, normalized eigenvalues, and correlation coefficients between various risk dimensions, the risk dimension values ​​are calculated. This approach strengthens the influence weight of each dimension under different scenarios through weight coefficients and considers the interaction between dimensions through correlation coefficients, avoiding the one-sidedness of single-dimensional assessment. This ensures that the calculated risk dimension values ​​accurately reflect the risk level of each dimension and the correlation effects between dimensions. It provides a precise single-dimensional risk assessment basis for subsequent multi-dimensional fusion based on risk level benchmarks to obtain a comprehensive risk assessment value, ensuring that the risk assessment results are consistent with the actual flight scenarios of drones.

[0047] When extracting the current flight stage and current mission type of the UAV from the flight status data, a preset structured field is retrieved from the flight status data. This structured field contains a flight stage field and a mission type field that correspond one-to-one with a timestamp. A timestamp consistent with the current flight time of the UAV is selected, and the content of the flight stage field corresponding to this timestamp is extracted as the current flight stage, and the content of the mission type field corresponding to this timestamp is extracted as the current mission type. Based on a preset flight stage-mission type-weight mapping relationship table, when assigning corresponding weight coefficients to different dimensions of the risk quantification features according to the current flight stage and the current mission type, the preset flight stage-mission type-weight mapping relationship table is retrieved. This mapping relationship table is established based on the degree of influence of each risk dimension feature on UAV flight safety under different combinations of flight stages and mission types. Each flight stage-mission type combination in the table corresponds to a unique weight coefficient for each dimension of the risk quantification feature. The extracted current flight stage and current mission type are combined, and the corresponding entries are matched in the mapping relationship table. The weight coefficients corresponding to each risk dimension feature within the entry are extracted.

[0048] Based on the weighting coefficients, the risk quantification feature is weighted and evaluated to obtain the risk dimension value of the risk quantification feature. Then, all the dimension features included in the risk quantification feature are traversed, and the information content of each dimension feature is associated with the corresponding weighting coefficient. The degree of influence of the corresponding dimension feature in the risk assessment is strengthened or weakened according to the magnitude of the weighting coefficient. The information of each dimension feature after association processing is classified and integrated to form the risk assessment result corresponding to each dimension feature. This result is the risk dimension value of the risk quantification feature.

[0049] Based on a preset risk level benchmark, the risk dimension values ​​are fused in multiple dimensions and normalized to obtain the comprehensive risk assessment value of the UAV. The preset risk level benchmark is then retrieved. This benchmark is established based on historical test data of UAV safe flight and risk threshold classification standards, covering assessment intervals corresponding to different risk levels. The risk dimension values ​​are integrated in multiple dimensions according to the benchmark's assessment rules, making the risk assessment results of each dimension form an interconnected whole. The integrated risk assessment results are then mapped to a preset unified assessment interval, ensuring the assessment results fall within the range of values ​​in that interval. The result after mapping is the comprehensive risk assessment value of the UAV.

[0050] The beneficial effects include accurately extracting the current flight stage and mission type of the UAV, providing a basis for weight allocation that fits the actual flight scenario, assigning corresponding weight coefficients to each dimension of risk quantification features based on a preset mapping table, ensuring that the weights are adapted to the impact of each risk feature under different scenarios, weighting the risk quantification features through weighted evaluation using weight coefficients, clarifying the risk level of each dimension and obtaining risk dimension values, and completing multi-dimensional fusion and normalization processing based on a preset risk level benchmark, so that the evaluation results are in a unified range and can comprehensively reflect the overall risk, ultimately obtaining a comprehensive risk evaluation value that accurately meets actual needs, providing a reliable basis for UAV flight risk management and decision-making, and improving the pertinence and accuracy of risk assessment.

[0051] S5. The comprehensive risk assessment value is compared with a safety threshold to obtain the flight safety level of the UAV; In this embodiment of the invention, the step of comparing the comprehensive risk assessment value with a safety threshold to obtain the flight safety level of the UAV includes: Based on the current flight stage and current mission type in the flight status data, the comprehensive risk assessment value is mapped to a preset configuration database to obtain the dynamic safety threshold configuration of the UAV. The comprehensive risk assessment value is compared with the dynamic security threshold configuration to obtain a specific risk threshold range for the comprehensive risk assessment value. Discrete safety level calibration is performed on the specific risk threshold range to obtain the flight safety level of the UAV.

[0052] Based on the current flight stage and current mission type in the flight status data, the comprehensive risk assessment value is mapped to a preset configuration database to obtain the dynamic safety threshold configuration of the UAV. The preset configuration database stores safety threshold configurations corresponding to different combinations of flight stages and mission types. This configuration is determined based on the safe flight history data, mission priority, and risk tolerance of the UAV in different scenarios. Each combination corresponds to a clear risk threshold interval division. The extracted current flight stage and current mission type are used as search conditions to find matching entries in the configuration database. The risk threshold interval division content corresponding to the entry is extracted, which is the dynamic safety threshold configuration.

[0053] When comparing the comprehensive risk assessment value with the dynamic security threshold configuration to obtain a specific risk threshold range for the comprehensive risk assessment value, the dynamic security threshold configuration contains multiple continuous and non-overlapping risk threshold ranges. Each range has a clear start and end value. Starting from the lowest value range, it is determined whether the comprehensive risk assessment value is within the start and end value range of the range. If it is, the comparison stops, and the range is the specific risk threshold range for the comprehensive risk assessment value. If it is not, the comparison continues to the next range until a matching range is found.

[0054] When the specific risk threshold range is discretely calibrated to obtain the flight safety level of the UAV, the preset correspondence between the risk threshold range and the discrete safety level is retrieved. This correspondence assigns different threshold ranges to fixed safety levels. The safety levels are divided into four levels, corresponding to safe, basic safe, risky, and high risk, respectively. The specific risk threshold range is matched with this correspondence, and the successfully matched safety level is extracted, which is the flight safety level of the UAV.

[0055] The beneficial effects include: combining the current flight stage and mission type of the UAV, retrieving the corresponding dynamic safety threshold configuration from the preset configuration database, ensuring that the safety thresholds are accurately adapted to the actual flight scenario and mission requirements, and avoiding assessment deviations caused by general thresholds. By comparing the comprehensive risk assessment value with the dynamic safety threshold configuration one by one, the corresponding specific risk threshold range is accurately identified, ensuring the accuracy and uniqueness of the range matching. Based on the preset correspondence, the specific risk threshold range is discretely calibrated for safety level, clarifying the UAV flight safety status, providing a clear and intuitive basis for subsequent flight safety management decisions, improving the pertinence and reliability of UAV flight safety assessments, and helping to avoid flight risks in different scenarios.

[0056] S6. Map the flight safety level to a preset safe flight command library to obtain the flight adjustment control command for the UAV.

[0057] In this embodiment of the invention, mapping the flight safety level to a preset safe flight command library to obtain the flight adjustment control commands for the UAV includes: Read the flight safety level of the UAV, the current standard flight status data of the UAV, and the current mission objective; Based on the flight safety level, a preset safe flight instruction library is searched and matched to obtain a candidate instruction set for the UAV. The safe flight instruction library stores flight operation instructions that can be executed under different safety levels and their execution conditions. Based on the current standard flight status data and the current mission objective, the candidate instruction set is verified for feasibility and prioritized to obtain the optimized instruction sequence for the UAV. The optimized instruction sequence is encapsulated into flight adjustment control instructions for the UAV.

[0058] When reading the flight safety level of the UAV, the current standard flight status data of the UAV, and the current mission objective, the calibration result storage module of the UAV flight safety level is retrieved to extract the calibrated flight safety level. At the same time, the storage unit of the standard flight status data is retrieved to select the standard flight status data consistent with the current time as the current standard flight status data. Then, the UAV mission parameter storage module is retrieved to extract the pre-set mission objective of this flight mission as the current mission objective.

[0059] Based on the flight safety level, a preset safe flight instruction library is searched and matched to obtain a candidate instruction set for the UAV. When the safe flight instruction library stores executable flight operation instructions and their execution conditions under different safety levels, the preset safe flight instruction library is classified and stored according to the flight safety level. Each safety level category contains multiple sets of executable flight operation instructions, and each set of instructions is bound to specific execution conditions. These execution conditions are set based on the operational requirements of the UAV under different safety states. Using the extracted flight safety level as the search keyword, the corresponding category entry is searched in the safe flight instruction library, and all flight operation instructions and their execution conditions under that entry are extracted and integrated to form the candidate instruction set.

[0060] Based on the current standard flight status data and the current mission objective, the candidate instruction set is subjected to feasibility verification and priority ranking to obtain the optimized instruction sequence for the UAV. First, feasibility verification is performed by comparing the execution conditions of each instruction in the candidate instruction set with the current standard flight status data. If the current standard flight status data meets the execution conditions of the instruction and executing the instruction will not conflict with the current mission objective, then the instruction is determined to be a feasible instruction, and instructions that do not meet the conditions are eliminated. Then, priority ranking is performed by retrieving a preset instruction priority determination rule. This rule is established based on the importance of the instruction to ensuring flight safety and its contribution to completing the mission objective. All feasible instructions are ranked according to this rule to form an ordered instruction arrangement result, which is the optimized instruction sequence.

[0061] When encapsulating the optimized instruction sequence into flight adjustment control instructions for the UAV, a preset UAV flight control system instruction receiving format is retrieved. This format includes fixed entries such as instruction header, instruction execution sequence identifier, instruction content, and instruction verification information. Each instruction in the optimized instruction sequence is sequentially filled into the corresponding entry according to this format. At the same time, instruction verification information is added to ensure the integrity of the instruction transmission process. All entries are integrated into an instruction data packet that meets the reception requirements of the flight control system. This data packet is the flight adjustment control instruction for the UAV.

[0062] The beneficial effects are as follows: by accurately reading the current standard flight status data of the UAV's flight safety level and the current mission objective, a comprehensive and realistic basis for subsequent command matching and selection is provided. A candidate command set is obtained by searching a preset safe flight command library based on the flight safety level, ensuring that the candidate commands are accurately adapted to the UAV's current safety status. Based on the current standard flight status data and the current mission objective, the candidate command set is subjected to feasibility verification and priority ranking. Feasible commands that meet the flight status requirements and fit the mission objective are selected, and their execution order is determined to form an optimized command sequence. Finally, the optimized command sequence is encapsulated into flight adjustment control commands that conform to the flight control system's receiving format, ensuring that the commands can be executed accurately. This provides a clear and reliable operational basis for the UAV to adjust its flight status in real time, effectively improving the UAV's adaptability to different safety level scenarios and helping to ensure flight safety and successful mission completion.

[0063] like Figure 2 The diagram shown is a functional block diagram of a low-altitude flight safety assessment system for unmanned aerial vehicles (UAVs) provided in an embodiment of the present invention.

[0064] The UAV low-altitude flight safety assessment system 100 described in this invention can be installed in an electronic device. Depending on the functions implemented, the UAV low-altitude flight safety assessment system 100 may include a data preprocessing module 101, an adaptive rasterization processing module 102, a multi-dimensional feature extraction module 103, a weight allocation and fusion evaluation module 104, a safety threshold comparison module 105, and an instruction mapping module 106. The modules described in this invention can also be referred to as units, which are a series of computer program segments that can be executed by the processor of an electronic device and can perform a fixed function, and are stored in the memory of the electronic device.

[0065] In this embodiment, the functions of each module / unit are as follows: The data preprocessing module 101 is used to preprocess the flight status data and three-dimensional environment data of the UAV to obtain the standard flight status data and standard three-dimensional environment data of the UAV. The adaptive rasterization processing module 102 is used to perform adaptive rasterization processing on the standard three-dimensional environment data based on the standard flight status data to obtain a three-dimensional dynamic rasterized map of the UAV. The multi-dimensional feature extraction module 103 is used to extract multi-dimensional features from the three-dimensional dynamic rasterized map to obtain the risk quantification features of the UAV. The weight allocation and fusion evaluation module 104 is used to allocate weights to the risk quantification features based on the flight stage and mission type in the flight status data, and to fuse and evaluate the weighted data to obtain the comprehensive risk evaluation value of the UAV. The safety threshold comparison module 105 is used to compare the comprehensive risk assessment value with a safety threshold to obtain the flight safety level of the UAV. The instruction mapping module 106 is used to map the flight safety level to a preset safe flight instruction library to obtain the flight adjustment control instructions of the UAV.

[0066] In the several embodiments provided by this invention, it should be understood that the disclosed methods and systems can be implemented in other ways. For example, the system embodiments described above are merely illustrative; for instance, the division of modules is only a logical functional division, and other division methods may be used in actual implementation.

[0067] The modules described as separate components may or may not be physically separate. The components shown as modules may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the modules can be selected to achieve the purpose of this embodiment according to actual needs.

[0068] Furthermore, the functional modules in the various embodiments of the present invention can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit. The integrated unit can be implemented in hardware or in the form of hardware plus software functional modules.

[0069] It will be apparent to those skilled in the art that the present invention is not limited to the details of the exemplary embodiments described above, and that the present invention can be implemented in other specific forms without departing from the spirit or essential characteristics of the present invention.

[0070] This application embodiment can acquire and process relevant data based on artificial intelligence technology. Artificial intelligence is the theory, method, technology, and application system that uses digital computers or machines controlled by digital computers to simulate, extend, and expand human intelligence, perceive the environment, acquire knowledge, and use that knowledge to obtain optimal results.

[0071] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit it. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can be made to the technical solutions of the present invention without departing from the spirit and scope of the technical solutions of the present invention.

Claims

1. A method for assessing the safety of low-altitude flight of unmanned aerial vehicles (UAVs), characterized in that, The method includes: S1. Perform data preprocessing on the UAV's flight status data and three-dimensional environment data to obtain the UAV's standard flight status data and standard three-dimensional environment data. S2. Based on the standard flight status data, the standard three-dimensional environment data is subjected to adaptive rasterization processing to obtain a three-dimensional dynamic rasterized map of the UAV. S3. Perform multi-dimensional feature extraction on the three-dimensional dynamic rasterized map to obtain the risk quantification features of the UAV; S4. Based on the flight stage and mission type in the flight status data, the risk quantification features are weighted and the weighted data is fused and evaluated to obtain the comprehensive risk assessment value of the UAV. S5. The comprehensive risk assessment value is compared with a safety threshold to obtain the flight safety level of the UAV; S6. Map the flight safety level to a preset safe flight command library to obtain the flight adjustment control command for the UAV.

2. The method for determining the safety of low-altitude flight of a UAV as described in claim 1, characterized in that, The process of preprocessing the UAV's flight status data and 3D environment data to obtain the UAV's standard flight status data and standard 3D environment data includes: The data formats of the UAV's flight status data and 3D environment data are parsed and converted to obtain the UAV's regular flight status data and regular 3D environment data; Anomaly detection is performed on the regularized flight state data to obtain the abnormal data points and missing data segments of the regularized flight state data. Remove the abnormal data points and fill the missing data segments with linear interpolation to obtain the clean flight status data of the UAV; Based on the preset geographic coordinate system transformation rules, the coordinate system of the regular three-dimensional environment data is transformed to the navigation coordinate reference of the UAV, so as to obtain the corrected three-dimensional environment data of the UAV. The clean flight status data and the three-dimensional environment data are time-aligned, and the sampling rate of the aligned data is standardized and resampled to obtain the standard flight status data and the standard three-dimensional environment data.

3. The method for determining the safety of low-altitude flight of a UAV as described in claim 1, characterized in that, The process of adaptively rasterizing the standard 3D environment data based on the standard flight status data to obtain a 3D dynamic rasterized map of the UAV includes: Based on the real-time flight speed and altitude data in the standard flight status data, the grid granularity parameters and spatial range parameters for spatial division of the standard three-dimensional environment data are determined. Based on the determined grid granularity parameters and spatial range parameters, a three-dimensional cubic grid array with uniform size is established within the three-dimensional space of the standard three-dimensional environmental data. The terrain elevation information, static obstacle location information, and dynamic obstacle prediction location information in the standard three-dimensional environment data are mapped and associated with the three-dimensional cube grid array to obtain the comprehensive attribute information of the three-dimensional cube grid array. The three-dimensional cubic grid array and the comprehensive attribute information are structurally integrated to obtain a three-dimensional dynamic rasterized map of the UAV.

4. The method for determining the safety of low-altitude flight of a UAV as described in claim 3, characterized in that, The determination of raster granularity parameters and spatial range parameters for spatial division of the standard three-dimensional environment data based on real-time flight speed and altitude data from the standard flight status data includes: Extract the real-time flight speed and real-time flight altitude from the standard flight status data; The real-time flight speed is verified and matched with a preset speed threshold range, and the initial raster granularity parameters of the three-dimensional dynamic rasterized map are determined based on the matched speed threshold range. The real-time flight altitude is compared and matched with a preset altitude threshold range, and the initial spatial range parameters of the three-dimensional dynamic raster map are determined based on the matched altitude threshold range. Based on the mission requirements of the UAV in its current flight phase, the initial grid granularity parameters and the initial spatial range parameters are constrained and adjusted to obtain the grid granularity parameters and the spatial range parameters of the standard three-dimensional environment data.

5. The method for determining the safety of low-altitude flight of a UAV as described in claim 1, characterized in that, The step of extracting multi-dimensional features from the three-dimensional dynamic rasterized map to obtain the risk quantification features of the UAV includes: The three-dimensional dynamic rasterized map is subjected to description category separation to obtain the spatial type attribute and risk identification attribute of the three-dimensional dynamic rasterized map; Based on the airspace type attribute, the distribution ratio of the UAV in different types of airspace within the current flight airspace is statistically analyzed, and the statistical results are integrated with the risk identification attribute to form the environmental risk characteristics of the UAV. In the three-dimensional dynamic rasterized map, with the current position and heading of the UAV as the center direction, a passable space analysis is performed to obtain the spatial accessibility and congestion characteristics of the UAV. A time-series comparative analysis of the raster state change information in the three-dimensional dynamic rasterized map is performed to obtain the environmental dynamic changes and trend characteristics of the three-dimensional dynamic rasterized map; The risk quantification characteristics of the UAV are obtained by fusing the environmental risk characteristics, spatial accessibility, congestion level characteristics, environmental dynamic changes, and trend characteristics through multimodal feature fusion.

6. The method for determining the safety of low-altitude flight of a UAV as described in claim 1, characterized in that, The risk quantification features are weighted based on the flight phase and mission type in the flight status data, and the weighted data is then fused and evaluated to obtain the comprehensive risk assessment value of the UAV, including: Extract the current flight stage and current mission type of the UAV from the flight status data; Based on a preset flight phase-mission type-weight mapping table, corresponding weight coefficients are assigned to different dimensions of the risk quantification features according to the current flight phase and the current mission type. Based on the weighting coefficients, the risk quantification features are weighted and evaluated to obtain the risk dimension values ​​of the risk quantification features; Based on a preset risk level benchmark, the risk dimension values ​​are fused from multiple dimensions and normalized to obtain the comprehensive risk assessment value of the drone.

7. The method for determining the safety of low-altitude flight of a UAV as described in claim 6, characterized in that, The formula for calculating the risk dimension value is as follows: ； In the formula, The value of the aforementioned risk dimension. For the first The weighting coefficients for each risk dimension, The total number of risk dimensions included in the aforementioned risk quantification features. For the first Normalized eigenvalues ​​for each risk dimension For the first Normalized eigenvalues ​​for each risk dimension For the first The risk dimension for the first The correlation coefficient of each risk dimension.

8. The method for determining the safety of low-altitude flight of a UAV as described in claim 1, characterized in that, The step of comparing the comprehensive risk assessment value with a safety threshold to obtain the flight safety level of the UAV includes: Based on the current flight stage and current mission type in the flight status data, the comprehensive risk assessment value is mapped to a preset configuration database to obtain the dynamic safety threshold configuration of the UAV. The comprehensive risk assessment value is compared with the dynamic security threshold configuration to obtain a specific risk threshold range for the comprehensive risk assessment value. Discrete safety level calibration is performed on the specific risk threshold range to obtain the flight safety level of the UAV.

9. The method for determining the safety of low-altitude flight of a UAV as described in claim 1, characterized in that, The step of mapping the flight safety level to a preset safe flight command library to obtain the flight adjustment control commands for the UAV includes: Read the flight safety level of the UAV, the current standard flight status data of the UAV, and the current mission objective; Based on the flight safety level, a preset safe flight instruction library is searched and matched to obtain a candidate instruction set for the UAV. The safe flight instruction library stores flight operation instructions that can be executed under different safety levels and their execution conditions. Based on the current standard flight status data and the current mission objective, the candidate instruction set is verified for feasibility and prioritized to obtain the optimized instruction sequence for the UAV. The optimized instruction sequence is encapsulated into flight adjustment control instructions for the UAV.

10. A safety assessment system for low-altitude flight of unmanned aerial vehicles (UAVs), characterized in that, The system for implementing the method for determining the safety of low-altitude flight of a UAV as described in claim 1 includes: The data preprocessing module is used to preprocess the flight status data and three-dimensional environment data of the UAV to obtain the standard flight status data and standard three-dimensional environment data of the UAV. An adaptive rasterization processing module is used to perform adaptive rasterization processing on the standard three-dimensional environment data based on the standard flight status data to obtain a three-dimensional dynamic rasterized map of the UAV. A multi-dimensional feature extraction module is used to extract multi-dimensional features from the three-dimensional dynamic rasterized map to obtain the risk quantification features of the UAV. The weighting and fusion evaluation module is used to assign weights to the risk quantification features based on the flight phase and mission type in the flight status data, and to fuse and evaluate the weighted data to obtain the comprehensive risk assessment value of the UAV. The safety threshold comparison module is used to compare the comprehensive risk assessment value with a safety threshold to obtain the flight safety level of the UAV. The command mapping module is used to map the flight safety level to a preset safe flight command library to obtain the flight adjustment control commands of the UAV.

Images