Unmanned aerial vehicle flight security early warning method and system facing complex environment

Abstract

The invention relates to the technical field of environment perception and three-dimensional modeling, and discloses an unmanned aerial vehicle flight security early warning method and system for a complex environment, and the method comprises the steps: collecting the environment data of a complex flight scene, and enabling the environment data to comprise a dynamic obstacle, a static topographic feature and a meteorological condition; generating an environment three-dimensional semantic map of the flight complex scene; constructing an initial global path of the flight unmanned aerial vehicle corresponding to the flight complex scene; establishing a dynamic threat identification model of the flight scene of the unmanned aerial vehicle to identify a sudden threat factor of the flying unmanned aerial vehicle, and analyzing a factor potential movement track of the sudden threat factor; calculating a risk coefficient of the sudden threat factor to the flying unmanned aerial vehicle; and generating a smooth obstacle avoidance trajectory of the flying unmanned aerial vehicle to execute security early warning of the flying unmanned aerial vehicle. The safety of autonomous flight of the unmanned aerial vehicle in a complex environment can be improved.

Description

Technical Field

[0001] This invention relates to a method and system for flight safety early warning of unmanned aerial vehicles (UAVs) in complex environments, belonging to the field of environmental perception and 3D modeling technology. Background Technology

[0002] Unmanned aerial vehicle (UAV) flight security early warning for complex environments refers to a set of intelligent early warning and protection mechanisms that utilize advanced technologies to monitor the flight process of UAVs in real time, identify threats, predict risks, and proactively intervene in dynamic, changeable, uncertain, and potentially threatening real-world scenarios. This ensures the safety of the UAVs themselves, the safety of mission execution, and the safety of the surrounding environment. It ensures that UAVs can perform missions safely, reliably, and efficiently in increasingly congested and complex low-altitude airspace, and is a key step for UAV technology to move from "being able to fly" to "being easy to use and daring to use".

[0003] Traditional drone flight safety early warning methods for complex environments mainly rely on static perception and response mechanisms based on rules and preset thresholds. For example, they use one or a few sensors (such as visible light cameras, ultrasonic sensors, or simple radar) to detect physical obstacles within a fixed distance. Once an obstacle is detected entering a preset safety boundary, the system triggers simple evasive commands (such as emergency hovering or fixed-angle turning). While this approach is simple to implement and has low computational overhead, it has significant drawbacks in complex and ever-changing environments: it lacks semantic understanding of the environment and cannot distinguish the nature and intent of dynamic obstacles (such as birds and hostile drones), thus preventing drones from flying autonomously and safely in complex environments. Summary of the Invention

[0004] This invention provides a method and system for flight safety early warning of unmanned aerial vehicles (UAVs) in complex environments, the main purpose of which is to improve the safety of autonomous flight of UAVs in complex environments.

[0005] To achieve the above objectives, the present invention provides a method for unmanned aerial vehicle (UAV) flight security early warning in complex environments, comprising:

[0006] Collect environmental data of complex flight scenarios, including dynamic obstacles, static terrain features, and meteorological conditions, to generate a three-dimensional semantic map of the complex flight scenarios.

[0007] Based on the three-dimensional semantic map of the environment, an initial global path is constructed for the UAV corresponding to the complex flight scenario;

[0008] A dynamic threat identification model for the UAV flight scenario is established to identify sudden threat factors of the UAV and analyze the potential motion trajectories of the sudden threat factors.

[0009] Based on the potential motion trajectory of the factors, calculate the risk coefficient of the sudden threat factors for the flying drone;

[0010] Based on the risk coefficient, a smooth obstacle avoidance trajectory for the flying drone is generated to perform security early warning for the flying drone.

[0011] Optionally, generating the environmental three-dimensional semantic map of the complex flight scenario includes:

[0012] Semantic segmentation is performed on the static terrain features of the complex flight scenario to obtain the passable area, static obstacles, and no-fly zone of the complex flight scenario;

[0013] Based on the passable area, static obstacles, and no-fly zone, a static terrain mesh for the complex flight scenario is generated;

[0014] Extract the 3D bounding boxes of the dynamic obstacles to generate a dynamic threat corridor for the complex flight scenario;

[0015] Based on the meteorological conditions, the wind field vector layer, precipitation particle distribution map, and fog visibility of the complex flight scenario are analyzed to generate a risk probability heat map of the complex flight scenario.

[0016] A three-dimensional semantic map of the complex flight scenario is generated using the static terrain grid, the dynamic threat corridor, and the risk probability heatmap.

[0017] Optionally, generating the static terrain mesh for the complex flight scenario based on the passable area, static obstacles, and no-fly zone includes:

[0018] Construct a three-dimensional mesh of the passable area to obtain the passable area mesh;

[0019] Analyze the occlusion coefficient of the static obstacle;

[0020] The static obstacle is compensated according to the occlusion coefficient to obtain the compensated static obstacle;

[0021] The material properties of the compensated static obstacles are labeled and integrated into the passable area mesh to obtain the obstacle area mesh;

[0022] A logical mask layer is used to map the no-fly zone to the obstacle area mesh to generate a static terrain mesh for the complex flight scenario.

[0023] Optionally, the step of analyzing the wind field vector layer, precipitation particle distribution map, and fog visibility of the complex flight scenario based on the meteorological conditions includes:

[0024] Extract wind field parameters from the meteorological conditions, wherein the wind field parameters include wind speed and wind direction;

[0025] Based on the wind field parameters, the wind field vector layer of the complex flight scenario is analyzed;

[0026] Extract radar reflectivity data from the meteorological conditions to identify the precipitation granularity of the complex flight scenario;

[0027] Based on the precipitation particle size, a precipitation particle distribution map of the complex flight scenario is constructed;

[0028] Based on the aforementioned meteorological conditions, the extinction coefficient of the complex flight scenario is analyzed;

[0029] Based on the extinction coefficient, the fog visibility in the complex flight scenario is calculated.

[0030] Optionally, generating the dynamic threat corridor for the complex flight scenario includes:

[0031] Based on the three-dimensional bounding box of the complex flight scenario, determine the spatial coordinates of the dynamic obstacles corresponding to the complex flight scenario.

[0032] Analyze the acceleration of the dynamic obstacle to generate dynamic filtering coefficients for the dynamic obstacle;

[0033] Based on the spatial coordinates and the dynamic filtering coefficients, the smoothed velocity of the dynamic obstacle corresponding to the complex flight scenario is calculated using the following formula:

[0034]

[0035] in, Indicates the dynamic obstacle in time Smoothed speed, Indicates the dynamic obstacle in time speed, Represents the dynamic filtering coefficients. Indicates the dynamic obstacle in time spatial coordinates, Indicates the dynamic obstacle in time spatial coordinates, Indicates a time interval;

[0036] Based on the smoothed velocity, a continuous motion path for the dynamic obstacle is generated;

[0037] Based on the continuous motion path, a dynamic threat corridor for the complex flight scenario is constructed.

[0038] Optionally, constructing the initial global path for the UAV corresponding to the complex flight scenario based on the environmental 3D semantic map includes:

[0039] Invalid voxels in the aforementioned three-dimensional semantic map of the environment are extracted to obtain a voxel three-dimensional semantic map.

[0040] Define the graph nodes of the voxel 3D semantic map;

[0041] Establish a navigation graph for the aforementioned graph nodes;

[0042] Define the multipath flight target of the navigation chart;

[0043] By combining the navigation map and the multi-path flight target, the initial path of the UAV corresponding to the complex flight scenario is constructed;

[0044] The initial global path of the flying UAV is determined by detecting the conflict segments between the initial path and the dynamic threat corridor corresponding to the three-dimensional semantic map of the environment.

[0045] Optionally, establishing the dynamic threat identification model for the UAV flight scenario includes:

[0046] Obtain historical threat data for the drone flight scenario;

[0047] Define semantic tags for the drone flight scenario;

[0048] Based on the historical threat data and the semantic tags, the threat type of the drone flight scenario is determined, wherein the threat type includes geometric threats and non-geometric threats;

[0049] Establish geometric threat identification models and non-geometric threat identification models for both geometric and non-geometric threats respectively;

[0050] Based on the geometric threat identification model and the non-geometric threat identification model, a dynamic threat identification model for the UAV flight scenario is established.

[0051] Optionally, the analysis of the potential movement trajectory of the sudden threat factor includes:

[0052] Capture the initial state vector of the sudden threat factor;

[0053] The aforementioned sudden threat factors are divided into geometric threat factors and non-geometric threat factors;

[0054] Generate a potential motion set of the geometric threat factors to analyze the potential motion trajectories of the geometric threat factors;

[0055] Generate a potential diffusion set of the non-geometric threat factors to analyze their potential diffusion trajectories;

[0056] By combining the potential movement trajectory and the potential diffusion trajectory, the potential movement trajectory of the sudden threat factor is determined.

[0057] Optionally, calculating the risk coefficient of the sudden threat factor for the flying UAV based on the potential motion trajectory of the factor includes:

[0058] Based on the initial global path of the flying drone, the spatiotemporal occupancy area of ​​the sudden threat factor is defined;

[0059] Calculate the potential motion trajectory of the factors and the coordinate probabilities of the UAV and threat factors in the spatiotemporal occupied area;

[0060] Based on the UAV coordinate probability and the threat factor coordinate probability, the spatiotemporal overlap of the sudden threat factor with the flying UAV is calculated using the following formula:

[0061]

[0062] in, This indicates the spatiotemporal overlap of sudden threat factors on flying drones. Indicates the current time. Indicates a time window. Indicates the probability of drone coordinates. Indicates the time of flight of the drone Located in world coordinates The probability of drone coordinates, Indicates the sudden threat factor in time Located in world coordinates Threat factor coordinate probability, This represents the function that takes the maximum value.

[0063] The threat kinetic energy and functional interference intensity of the sudden threat factor on the flying UAV are calculated based on the spatiotemporal overlap.

[0064] Based on the threat kinetic energy and functional interference intensity, the risk coefficient of the sudden threat factor to the flying drone is determined.

[0065] To address the aforementioned problems, the present invention also provides a drone flight safety early warning system for complex environments, the system comprising:

[0066] A semantic map construction module is used to collect environmental data of complex flight scenarios, including dynamic obstacles, static terrain features and meteorological conditions, in order to generate a three-dimensional semantic map of the complex flight scenario.

[0067] A global path construction module is used to construct the initial global path of the UAV corresponding to the complex flight scenario based on the three-dimensional semantic map of the environment.

[0068] The factor trajectory analysis module is used to establish a dynamic threat identification model for the UAV flight scenario, in order to identify the sudden threat factors of the flying UAV and analyze the potential motion trajectory of the sudden threat factors.

[0069] The risk coefficient determination module is used to calculate the risk coefficient of the sudden threat factor to the flying drone based on the potential motion trajectory of the factor;

[0070] The drone security early warning module is used to generate a smooth obstacle avoidance trajectory for the flying drone based on the risk coefficient, so as to perform security early warning for the flying drone.

[0071] This solution constructs a high-precision 3D semantic map of the environment by comprehensively collecting dynamic obstacles, static terrain features, and meteorological conditions in complex flight scenarios. This enables all-round perception and modeling of the flight environment. Based on this map, the system can scientifically plan the initial global path of the UAV, ensuring efficient flight under ideal conditions. Simultaneously, by establishing a dynamic threat identification model, the system can identify sudden threat factors in real time and accurately predict their potential trajectories, providing proactive risk warnings for the UAV. Combining the trajectories of sudden threat factors, the system dynamically calculates their risk coefficients to the UAV and generates smooth obstacle avoidance trajectories accordingly, effectively avoiding potential dangers. This solution significantly improves the autonomous decision-making ability and safety of UAVs in complex and dynamic environments, ensuring rapid response and precise obstacle avoidance when performing security early warning tasks, greatly reducing the risk of flight accidents. Furthermore, the system's intelligent processing capabilities reduce the need for manual intervention and improve task execution efficiency, providing solid technical support for the application of UAVs in high-risk scenarios such as emergency rescue, border patrol, and urban security. It has broad practical value and promising prospects for promotion. Therefore, this invention can improve the safety of autonomous flight of UAVs in complex environments. Attached Figure Description

[0072] Figure 1 This is a flowchart illustrating a method for flight security early warning of unmanned aerial vehicles (UAVs) in complex environments, provided in an embodiment of the present invention.

[0073] Figure 2 This is a schematic diagram of a module for implementing the UAV flight security early warning method for complex environments, provided in an embodiment of the present invention.

[0074] The objectives, features, and advantages of this invention will be further explained in conjunction with the embodiments and with reference to the accompanying drawings. Detailed Implementation

[0075] It should be understood that the specific embodiments described herein are merely illustrative of the invention and are not intended to limit the invention.

[0076] This application provides a method for drone flight security early warning in complex environments. 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 drone flight security early warning in complex environments 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.

[0077] Reference Figure 1 The diagram shown is a flowchart illustrating a drone flight security early warning method for complex environments according to an embodiment of the present invention. In this embodiment, the drone flight security early warning method for complex environments includes:

[0078] S1. Collect environmental data of complex flight scenarios, including dynamic obstacles, static terrain features, and meteorological conditions, to generate a three-dimensional semantic map of the complex flight scenario.

[0079] It should be explained that the complex flight scenario refers to an aerial or near-ground space area with a complex environment (dynamic obstacles, static terrain undulation, sudden weather changes). The dynamic obstacles refer to obstacles such as other drones, birds, and manned aircraft. The static terrain features include geometric structures (building outlines, slender objects with a diameter of <5cm for bridge cables, spatial distribution of high-voltage power lines), material reflection characteristics (specular reflection coefficient of glass curtain walls, multispectral characteristics of vegetation), and regulatory restricted areas (geofence zone coordinates, height restriction threshold). The meteorological conditions include real-time meteorological parameters (turbulent areas with wind speed ≥8m / s, fog areas with horizontal visibility <1km, rain and snow range with precipitation intensity >5mm / h), meteorological threats (probability of lightning strikes within the next 5 minutes, sudden changes in vertical wind speed of downbursts), and microenvironmental effects (wind shear gradient of building clusters, local updrafts caused by the heat island effect).

[0080] The present invention generates a three-dimensional semantic map of the complex flight scenario, realizing comprehensive perception and accurate modeling of the flight environment.

[0081] Specifically, generating the environmental 3D semantic map of the complex flight scenario includes:

[0082] Semantic segmentation is performed on the static terrain features of the complex flight scenario to obtain the passable area, static obstacles, and no-fly zone of the complex flight scenario;

[0083] Based on the passable area, static obstacles, and no-fly zone, a static terrain mesh for the complex flight scenario is generated;

[0084] Extract the 3D bounding boxes of the dynamic obstacles to generate a dynamic threat corridor for the complex flight scenario;

[0085] Based on the meteorological conditions, the wind field vector layer, precipitation particle distribution map, and fog visibility of the complex flight scenario are analyzed to generate a risk probability heat map of the complex flight scenario.

[0086] A three-dimensional semantic map of the complex flight scenario is generated using the static terrain grid, the dynamic threat corridor, and the risk probability heatmap.

[0087] The accessible area refers to the area where the drone can fly safely in complex flight scenarios, such as areas with high visibility and no reflective surfaces with a reflectivity of ≥80%. Static obstacles refer to objects or structures that are fixed in position and do not move in the flight scenario, such as buildings, mountains, towers, and utility poles. No-fly zones refer to areas where drones are prohibited from entering due to regulations, safety, or mission requirements, such as military control zones, areas surrounding airports, and airspace above sensitive facilities. The static terrain mesh refers to a structure that discretizes and models the static terrain features (including accessible areas, static obstacles, and no-fly zones) in the flight scenario using a mesh format. The dynamic threat corridor refers to the "walkway" formed in space by potential threat areas composed of dynamic obstacles (such as moving vehicles, aircraft, and birds). The structure is described as follows: the wind field vector layer is a data layer describing the intensity, direction, and distribution of wind in the flight scenario; the precipitation particle distribution map is the spatial distribution density of precipitation particles (such as rain, snow, hail, etc.) in the scenario; the fog visibility refers to the degree of visual limitation caused by fog or other low visibility meteorological conditions; the risk probability heat map is a risk distribution map presented in the form of a heat map after comprehensive analysis of various environmental risk factors (such as dynamic threats, wind field interference, precipitation impact, visibility limitation, etc.); and the environmental three-dimensional semantic map is a three-dimensional spatial environment model with semantic annotation constructed by integrating multi-dimensional information such as static terrain grid, dynamic threat corridor, meteorological conditions (wind field vector layer, precipitation particle distribution, fog visibility), and risk probability heat map.

[0088] Optionally, the extraction of the 3D bounding box of the dynamic obstacle can be achieved by detecting the bounding box of the dynamic target using the YOLOv7-3D model.

[0089] Furthermore, the step of generating a static terrain mesh for the complex flight scenario based on the passable area, static obstacles, and no-fly zone includes:

[0090] Construct a three-dimensional mesh of the passable area to obtain the passable area mesh;

[0091] Analyze the occlusion coefficient of the static obstacle;

[0092] The static obstacle is compensated according to the occlusion coefficient to obtain the compensated static obstacle;

[0093] The material properties of the compensated static obstacles are labeled and integrated into the passable area mesh to obtain the obstacle area mesh;

[0094] A logical mask layer is used to map the no-fly zone to the obstacle area mesh to generate a static terrain mesh for the complex flight scenario.

[0095] The accessible area mesh refers to the mesh structure generated after three-dimensional discretization modeling of the accessible area in the flight scene. The occlusion coefficient refers to the geometric or probabilistic compensation of slender static obstacles (such as utility poles, antenna towers, bridge cables, etc.) which may be partially missed or underestimated in the three-dimensional mesh due to resolution limitations or sensor perspective issues. The material properties refer to the physical or electromagnetic characteristics of the slender obstacles, such as metal, wood, concrete, glass, etc. The obstacle area mesh is a mesh structure generated by integrating the compensation of slender obstacles and their material properties on the basis of the accessible area mesh. The logical mask layer is a data layer that maps the no-fly zone to the obstacle area mesh in a logical "mask" manner. Each mesh cell in the mask layer is marked as "No Entry" or "Allowed Passage". The static terrain mesh is the final generated comprehensive three-dimensional mesh model that integrates the accessible area mesh, the obstacle area mesh (including compensation of slender obstacles and material properties), and the no-fly zone logical mask layer.

[0096] Optionally, the construction of the three-dimensional mesh of the passable area can be achieved by dividing the passable area into three-dimensional meshes using an octree structure.

[0097] Furthermore, the generation of the dynamic threat corridor for the complex flight scenario includes:

[0098] Based on the three-dimensional bounding box of the complex flight scenario, determine the spatial coordinates of the dynamic obstacles corresponding to the complex flight scenario.

[0099] Analyze the acceleration of the dynamic obstacle to generate dynamic filtering coefficients for the dynamic obstacle;

[0100] Based on the spatial coordinates and the dynamic filtering coefficients, the smoothed velocity of the dynamic obstacle corresponding to the complex flight scenario is calculated using the following formula;

[0101] Based on the smoothed velocity, a continuous motion path for the dynamic obstacle is generated;

[0102] Based on the continuous motion path, a dynamic threat corridor for the complex flight scenario is constructed.

[0103] Furthermore, as another embodiment of the present invention, the smoothed speed is calculated using the following formula:

[0104]

[0105] in, Indicates the dynamic obstacle in time Smoothed speed, Indicates the dynamic obstacle in time speed, Represents the dynamic filtering coefficients. Indicates the dynamic obstacle in time spatial coordinates, Indicates the dynamic obstacle in time spatial coordinates, Indicates a time interval.

[0106] Wherein, the spatial coordinates refer to the position of the dynamic obstacle in three-dimensional space, the dynamic filtering coefficients refer to the weighting factors used to smooth the velocity estimation, the smoothed velocity refers to the velocity estimate after filtering, the time interval refers to the time difference between two measurements, the continuous motion path refers to the motion trajectory of the obstacle over a future period of time generated based on the smoothed velocity, and the dynamic threat corridor refers to a dynamically changing three-dimensional spatial region used to represent the range of threats that the obstacle may pose to the UAV.

[0107] Furthermore, the analysis of the wind field vector layer, precipitation particle distribution map, and fog visibility in the complex flight scenario based on the meteorological conditions includes:

[0108] Extract wind field parameters from the meteorological conditions, wherein the wind field parameters include wind speed and wind direction;

[0109] Based on the wind field parameters, the wind field vector layer of the complex flight scenario is analyzed;

[0110] Extract radar reflectivity data from the meteorological conditions to identify the precipitation granularity of the complex flight scenario;

[0111] Based on the precipitation particle size, a precipitation particle distribution map of the complex flight scenario is constructed;

[0112] Based on the aforementioned meteorological conditions, the extinction coefficient of the complex flight scenario is analyzed;

[0113] Based on the extinction coefficient, the fog visibility in the complex flight scenario is calculated.

[0114] Wherein, wind speed refers to the speed of air flow, wind direction refers to the direction from which the wind blows, reflectivity data refers to the signal strength reflected back after the electromagnetic waves emitted by the radar encounter precipitation particles, precipitation particle size refers to the size or diameter distribution of precipitation particles (such as raindrops, snowflakes, hail), and extinction coefficient refers to the degree to which light or electromagnetic waves are weakened due to scattering and absorption when propagating in a medium.

[0115] Optionally, the wind field vector layer based on the wind field parameters analyzes the complex flight scenario by synthesizing the wind speed and wind direction of each grid point into a three-dimensional vector.

[0116] S2. Based on the three-dimensional semantic map of the environment, construct the initial global path of the UAV corresponding to the complex flight scenario.

[0117] Based on the three-dimensional semantic map of the environment, this invention constructs the initial global path of the UAV corresponding to the complex flight scenario, which can determine the safe path of the UAV.

[0118] Specifically, the step of constructing the initial global path for the UAV corresponding to the complex flight scenario based on the 3D semantic map of the environment includes:

[0119] Invalid voxels in the aforementioned three-dimensional semantic map of the environment are extracted to obtain a voxel three-dimensional semantic map.

[0120] Define the graph nodes of the voxel 3D semantic map;

[0121] Establish a navigation graph for the aforementioned graph nodes;

[0122] Define the multipath flight target of the navigation chart;

[0123] By combining the navigation map and the multi-path flight target, the initial path of the UAV corresponding to the complex flight scenario is constructed;

[0124] The initial global path of the flying UAV is determined by detecting the conflict segments between the initial path and the dynamic threat corridor corresponding to the three-dimensional semantic map of the environment.

[0125] Wherein, the voxel 3D semantic map refers to the discretized flyable space representation obtained by removing invalid voxels from the environmental 3D semantic map; the graph node refers to the abstract representation of the valid voxels in the voxel 3D semantic map that have not been removed in the navigation graph; the navigation graph refers to the directed weighted graph constructed based on the graph nodes and their connection relationships; the multi-path flight target refers to the set of optimization targets for path planning, such as minimizing distance, minimizing risk, minimizing energy consumption, etc.; the initial path refers to the preliminary feasible path obtained by searching the navigation graph using the improved A* algorithm; the conflict segment refers to the set of continuous path points in the initial path that spatiotemporally overlap with the dynamic threat corridor; and the initial global path refers to the final safe path output after conflict detection and local optimization.

[0126] Optionally, the initial path of the UAV corresponding to the complex flight scenario is constructed by combining the navigation map and the multi-path flight target and retrieved from the navigation map using an improved A* algorithm.

[0127] S3. Establish a dynamic threat identification model for the UAV flight scenario to identify sudden threat factors of the UAV and analyze the potential motion trajectories of the sudden threat factors.

[0128] This invention establishes a dynamic threat identification model for the aforementioned UAV flight scenario, providing a basis for subsequent analysis of sudden threat factors.

[0129] Specifically, establishing the dynamic threat identification model for the drone flight scenario includes:

[0130] Obtain historical threat data for the drone flight scenario;

[0131] Define semantic tags for the drone flight scenario;

[0132] Based on the historical threat data and the semantic tags, the threat type of the drone flight scenario is determined, wherein the threat type includes geometric threats and non-geometric threats;

[0133] Establish geometric threat identification models and non-geometric threat identification models for both geometric and non-geometric threats respectively;

[0134] Based on the geometric threat identification model and the non-geometric threat identification model, a dynamic threat identification model for the UAV flight scenario is established.

[0135] The historical threat data refers to a collection of multi-source threat-related data recorded by UAVs in past flight missions. The semantic tags refer to the manual or automatic annotation of each threat instance in the historical threat data. The geometric threat refers to a dynamic obstacle with a physical entity that can be directly detected by sensors, such as a swarm of UAVs that suddenly appears. The non-geometric threat refers to intangible environmental factors that affect UAV navigation or communication, such as spectral distortion or sudden changes in air pressure. The geometric threat identification model refers to a multi-target detection and tracking system based on neural networks. The non-geometric threat identification model refers to a signal and time series analysis system for intangible threats. The dynamic threat identification model refers to a multimodal decision-making system that integrates the results of geometric and non-geometric threat identification.

[0136] Optionally, in the separate establishment of geometric threat identification models and non-geometric threat identification models for geometric threats and non-geometric threats, the geometric threat identification model is constructed using the YOLOv7-3D algorithm and LSTM, while the non-geometric threat identification model is constructed using a signal propagation model combined with a temporal Transformer model to predict turbulence.

[0137] It should be explained that the aforementioned sudden threat factors refer to threats that suddenly appear during the flight of the drone and are not planned in advance, such as sudden drone swarms, sudden birds, sudden airflow changes, signal interference sources, etc.

[0138] This invention analyzes the potential motion trajectory of the sudden threat factors to accurately analyze their impact on UAV flight, thereby improving the reliability of flight path optimization for UAVs in later stages.

[0139] In detail, the analysis of the potential movement trajectory of the sudden threat factors includes:

[0140] Capture the initial state vector of the sudden threat factor;

[0141] The aforementioned sudden threat factors are divided into geometric threat factors and non-geometric threat factors;

[0142] Generate a potential motion set of the geometric threat factors to analyze the potential motion trajectories of the geometric threat factors;

[0143] Generate a potential diffusion set of the non-geometric threat factors to analyze their potential diffusion trajectories;

[0144] By combining the potential movement trajectory and the potential diffusion trajectory, the potential movement trajectory of the sudden threat factor is determined.

[0145] Wherein, the initial state vector refers to the complete motion state of the sudden threat factor when it is first detected, including position, velocity, acceleration, type, etc.; the geometric threat factor refers to a dynamic obstacle with a measurable physical contour and motion state; the non-geometric threat factor refers to an invisible environmental threat that affects the navigation or communication of the UAV; the potential motion set refers to the set of possible motion assumptions of the geometric threat factor; the potential motion trajectory refers to the spatiotemporal path probability distribution generated based on the potential motion set; the potential diffusion set refers to the set of assumptions about the influence range of the non-geometric threat factor; the potential diffusion trajectory refers to the time-varying influence range boundary of the non-geometric threat factor; and the factor potential motion trajectory refers to the comprehensive spatiotemporal threat domain that integrates the motion of the geometric threat and the diffusion of the non-geometric threat.

[0146] Optionally, the potential motion set for generating the geometric threat factor can be dynamically weighted using an interactive multi-model (IMM) to output a cluster of probabilistic motion trajectories.

[0147] S4. Based on the potential motion trajectory of the factors, calculate the risk coefficient of the sudden threat factors for the flying drone.

[0148] Based on the potential motion trajectory of the aforementioned factors, this invention calculates the risk coefficient of the sudden threat factors for the flying UAV, providing a basis for subsequent obstacle avoidance route analysis.

[0149] Specifically, calculating the risk coefficient of the sudden threat factor for the flying drone based on the potential motion trajectory of the factor includes:

[0150] Based on the initial global path of the flying drone, the spatiotemporal occupancy area of ​​the sudden threat factor is defined;

[0151] Calculate the potential motion trajectory of the factors and the coordinate probabilities of the UAV and threat factors in the spatiotemporal occupied area;

[0152] Based on the UAV coordinate probability and the threat factor coordinate probability, the spatiotemporal overlap of the sudden threat factor with the flying UAV is calculated using the following formula;

[0153] The threat kinetic energy and functional interference intensity of the sudden threat factor on the flying UAV are calculated based on the spatiotemporal overlap.

[0154] Based on the threat kinetic energy and functional interference intensity, the risk coefficient of the sudden threat factor to the flying drone is determined.

[0155] Furthermore, as another embodiment of the present invention, the spatiotemporal overlap is calculated using the following formula:

[0156]

[0157] in, This indicates the spatiotemporal overlap of sudden threat factors on flying drones. Indicates the current time. Indicates a time window. Indicates the probability of drone coordinates. Indicates the time of flight of the drone Located in world coordinates The probability of drone coordinates, Indicates the sudden threat factor in time Located in world coordinates Threat factor coordinate probability, This represents the function that takes the maximum value.

[0158] The spatiotemporal occupied area refers to the three-dimensional spatial region that the UAV may occupy within a certain period of time during the execution of its initial global path, and the UAV coordinate probability refers to the probability of the flying UAV in time... Located in world coordinates The probability of a sudden threat factor's coordinates refers to the probability of a sudden threat factor occurring in time. Located in world coordinates The probability, spatiotemporal overlap refers to the likelihood of a conflict between a sudden threat factor and a drone within a spatiotemporally occupied area; the time window refers to the time period used to predict the future state of the drone and the threat factor; the world coordinates refer to a globally unified coordinate system used to describe the absolute position of the drone and the threat factor; the threat kinetic energy refers to the potential destructive capability of the threat factor due to its movement; the functional interference intensity refers to the ability of the threat factor to interfere with the drone's communication, navigation, or perception systems; and the risk coefficient refers to the overall threat level of a sudden threat factor to the drone by combining threat kinetic energy and functional interference intensity.

[0159] Optionally, the calculation of the potential motion trajectory of the factors and the probability of UAV coordinates and threat factor coordinates of the spatiotemporal occupied area can be performed by particle filtering.

[0160] Optionally, determining the risk coefficient of the sudden threat factor to the flying UAV based on the threat kinetic energy and functional interference intensity can be achieved by weighted summation.

[0161] S5. Based on the risk coefficient, generate a smooth obstacle avoidance trajectory for the flying drone to perform security early warning for the flying drone.

[0162] This invention generates a smooth obstacle avoidance trajectory for the flying drone based on the aforementioned risk coefficient, effectively mitigating potential hazards. The smooth obstacle avoidance trajectory refers to a safe and smooth flight path planned based on the current environmental conditions (including risk coefficient, obstacle distribution, aircraft dynamic constraints, etc.). Specifically, the smooth obstacle avoidance trajectory can be optimized using the CHOMP algorithm based on gradient descent.

[0163] Finally, based on the risk coefficient, the present invention generates a smooth obstacle avoidance trajectory for the flying drone to perform security early warning, ensuring that the flying drone can respond quickly and avoid obstacles accurately when performing security early warning tasks, thereby significantly reducing the risk of flight accidents.

[0164] This solution constructs a high-precision 3D semantic map of the environment by comprehensively collecting dynamic obstacles, static terrain features, and meteorological conditions in complex flight scenarios. This enables all-round perception and modeling of the flight environment. Based on this map, the system can scientifically plan the initial global path of the UAV, ensuring efficient flight under ideal conditions. Simultaneously, by establishing a dynamic threat identification model, the system can identify sudden threat factors in real time and accurately predict their potential trajectories, providing proactive risk warnings for the UAV. Combining the trajectories of sudden threat factors, the system dynamically calculates their risk coefficients to the UAV and generates smooth obstacle avoidance trajectories accordingly, effectively avoiding potential dangers. This solution significantly improves the autonomous decision-making ability and safety of UAVs in complex and dynamic environments, ensuring rapid response and precise obstacle avoidance when performing security early warning tasks, greatly reducing the risk of flight accidents. Furthermore, the system's intelligent processing capabilities reduce the need for manual intervention and improve task execution efficiency, providing solid technical support for the application of UAVs in high-risk scenarios such as emergency rescue, border patrol, and urban security. It has broad practical value and promising prospects for promotion. Therefore, this invention can improve the safety of autonomous flight of UAVs in complex environments.

[0165] like Figure 2 The diagram shown is a functional block diagram of a UAV flight security early warning system for complex environments according to the present invention.

[0166] The UAV flight security early warning system 200 for complex environments described in this invention can be installed in an electronic device. Depending on the functions implemented, the UAV flight security early warning system for complex environments may include a semantic map construction module 201, a global path construction module 202, a factor trajectory analysis module 203, a risk coefficient determination module 204, and a UAV security early warning module 205. The module described in this invention can also be referred to as a unit, which refers to 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.

[0167] In this embodiment of the invention, the functions of each module / unit are as follows:

[0168] The semantic map construction module 201 is used to collect environmental data of complex flight scenarios, wherein the environmental data includes dynamic obstacles, static terrain features and meteorological conditions, in order to generate a three-dimensional semantic map of the complex flight scenario.

[0169] The global path construction module 202 is used to construct the initial global path of the UAV corresponding to the complex flight scenario based on the three-dimensional semantic map of the environment.

[0170] The factor trajectory analysis module 203 is used to establish a dynamic threat identification model for the UAV flight scenario, in order to identify the sudden threat factors of the flying UAV and analyze the potential motion trajectory of the sudden threat factors.

[0171] The risk coefficient determination module 204 is used to calculate the risk coefficient of the sudden threat factor to the flying drone based on the potential motion trajectory of the factor;

[0172] The drone security early warning module 205 is used to generate a smooth obstacle avoidance trajectory for the flying drone based on the risk coefficient, so as to perform security early warning for the flying drone.

[0173] In detail, the modules in the UAV flight security early warning system 200 for complex environments described in this embodiment of the invention employ the same methods as described above during use. Figure 1 The method used is the same as the drone flight security early warning method for complex environments described in the article, and can produce the same technical effect, so it will not be repeated here.

[0174] 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.

[0175] Finally, it should be noted that in the above embodiments, each embodiment can be combined with each other or independent. Deleting any one of them will not affect the technical implementation of other embodiments. The above embodiments are only used to illustrate the technical solutions of the present invention and not 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 flight security early warning of unmanned aerial vehicles (UAVs) in complex environments, characterized in that, The method includes: Collect environmental data of complex flight scenarios, including dynamic obstacles, static terrain features, and meteorological conditions, to generate a three-dimensional semantic map of the complex flight scenarios. Based on the three-dimensional semantic map of the environment, an initial global path is constructed for the UAV corresponding to the complex flight scenario; A dynamic threat identification model for the UAV flight scenario is established to identify sudden threat factors of the UAV and analyze the potential motion trajectories of the sudden threat factors. Based on the potential motion trajectory of the factors, calculate the risk coefficient of the sudden threat factors for the flying drone; Based on the risk coefficient, a smooth obstacle avoidance trajectory for the flying drone is generated to perform security early warning for the flying drone.

2. The method for unmanned aerial vehicle (UAV) flight security early warning in complex environments as described in claim 1, characterized in that, The generation of the environmental 3D semantic map for the complex flight scenario includes: Semantic segmentation is performed on the static terrain features of the complex flight scenario to obtain the passable area, static obstacles, and no-fly zone of the complex flight scenario; Based on the passable area, static obstacles, and no-fly zone, a static terrain mesh for the complex flight scenario is generated; Extract the 3D bounding boxes of the dynamic obstacles to generate a dynamic threat corridor for the complex flight scenario; Based on the meteorological conditions, the wind field vector layer, precipitation particle distribution map, and fog visibility of the complex flight scenario are analyzed to generate a risk probability heat map of the complex flight scenario. A three-dimensional semantic map of the complex flight scenario is generated using the static terrain grid, the dynamic threat corridor, and the risk probability heatmap.

3. The method for unmanned aerial vehicle (UAV) flight security early warning in complex environments as described in claim 2, characterized in that, The step of generating a static terrain mesh for the complex flight scenario based on the passable area, static obstacles, and no-fly zone includes: Construct a three-dimensional mesh of the passable area to obtain the passable area mesh; Analyze the occlusion coefficient of the static obstacle; The static obstacle is compensated according to the occlusion coefficient to obtain the compensated static obstacle; The material properties of the compensated static obstacles are labeled and integrated into the passable area mesh to obtain the obstacle area mesh; A logical mask layer is used to map the no-fly zone to the obstacle area mesh to generate a static terrain mesh for the complex flight scenario.

4. The method for unmanned aerial vehicle (UAV) flight security early warning in complex environments as described in claim 2, characterized in that, The analysis of the wind field vector layer, precipitation particle distribution map, and fog visibility based on the meteorological conditions includes: Extract wind field parameters from the meteorological conditions, wherein the wind field parameters include wind speed and wind direction; Based on the wind field parameters, the wind field vector layer of the complex flight scenario is analyzed; Extract radar reflectivity data from the meteorological conditions to identify the precipitation granularity of the complex flight scenario; Based on the precipitation particle size, a precipitation particle distribution map of the complex flight scenario is constructed; Based on the aforementioned meteorological conditions, the extinction coefficient of the complex flight scenario is analyzed; Based on the extinction coefficient, the fog visibility in the complex flight scenario is calculated.

5. The method for unmanned aerial vehicle (UAV) flight security early warning in complex environments as described in claim 2, characterized in that, The dynamic threat corridor that generates the complex flight scenario includes: Based on the three-dimensional bounding box of the complex flight scenario, determine the spatial coordinates of the dynamic obstacles corresponding to the complex flight scenario. Analyze the acceleration of the dynamic obstacle to generate dynamic filtering coefficients for the dynamic obstacle; Based on the spatial coordinates and the dynamic filtering coefficients, the smoothed velocity of the dynamic obstacle corresponding to the complex flight scenario is calculated using the following formula: in, Indicates the dynamic obstacle in time Smoothed speed, Indicates the dynamic obstacle in time speed, Represents the dynamic filtering coefficients. Indicates the dynamic obstacle in time spatial coordinates, Indicates the dynamic obstacle in time spatial coordinates, Indicates a time interval; Based on the smoothed velocity, a continuous motion path for the dynamic obstacle is generated; Based on the continuous motion path, a dynamic threat corridor for the complex flight scenario is constructed.

6. The method for unmanned aerial vehicle (UAV) flight security early warning in complex environments as described in claim 1, characterized in that, The process of constructing the initial global path for the UAV corresponding to the complex flight scenario based on the 3D semantic map of the environment includes: Invalid voxels in the aforementioned three-dimensional semantic map of the environment are extracted to obtain a voxel three-dimensional semantic map. Define the graph nodes of the voxel 3D semantic map; Establish a navigation graph for the aforementioned graph nodes; Define the multipath flight target of the navigation chart; By combining the navigation map and the multi-path flight target, the initial path of the UAV corresponding to the complex flight scenario is constructed; The initial global path of the flying UAV is determined by detecting the conflict segments between the initial path and the dynamic threat corridor corresponding to the three-dimensional semantic map of the environment.

7. The method for unmanned aerial vehicle (UAV) flight security early warning in complex environments as described in claim 1, characterized in that, The establishment of the dynamic threat identification model for the UAV flight scenario includes: Obtain historical threat data for the drone flight scenario; Define semantic tags for the drone flight scenario; Based on the historical threat data and the semantic tags, the threat type of the drone flight scenario is determined, wherein the threat type includes geometric threats and non-geometric threats; Establish geometric threat identification models and non-geometric threat identification models for both geometric and non-geometric threats respectively; Based on the geometric threat identification model and the non-geometric threat identification model, a dynamic threat identification model for the UAV flight scenario is established.

8. The method for unmanned aerial vehicle (UAV) flight security early warning in complex environments as described in claim 1, characterized in that, The analysis of the potential movement trajectory of the sudden threat factors includes: Capture the initial state vector of the sudden threat factor; The aforementioned sudden threat factors are divided into geometric threat factors and non-geometric threat factors; Generate a potential motion set of the geometric threat factors to analyze the potential motion trajectories of the geometric threat factors; Generate a potential diffusion set of the non-geometric threat factors to analyze their potential diffusion trajectories; By combining the potential movement trajectory and the potential diffusion trajectory, the potential movement trajectory of the sudden threat factor is determined.

9. The method for unmanned aerial vehicle (UAV) flight security early warning in complex environments as described in claim 1, characterized in that, The calculation of the risk coefficient of the sudden threat factor for the flying drone based on the potential motion trajectory of the factor includes: Based on the initial global path of the flying drone, the spatiotemporal occupancy area of ​​the sudden threat factor is defined; Calculate the potential motion trajectory of the factors and the coordinate probabilities of the UAV and threat factors in the spatiotemporal occupied area; Based on the UAV coordinate probability and the threat factor coordinate probability, the spatiotemporal overlap of the sudden threat factor with the flying UAV is calculated using the following formula: in, This indicates the spatiotemporal overlap of sudden threat factors on flying drones. Indicates the current time. Indicates a time window. Indicates the probability of drone coordinates. Indicates the time of flight of the drone Located in world coordinates The probability of drone coordinates, Indicates the sudden threat factor in time Located in world coordinates Threat factor coordinate probability, This represents the function that takes the maximum value. The threat kinetic energy and functional interference intensity of the sudden threat factor on the flying UAV are calculated based on the spatiotemporal overlap. Based on the threat kinetic energy and functional interference intensity, the risk coefficient of the sudden threat factor to the flying drone is determined.

10. A flight security early warning system for unmanned aerial vehicles (UAVs) in complex environments, characterized in that: The system includes: A semantic map construction module is used to collect environmental data of complex flight scenarios, including dynamic obstacles, static terrain features and meteorological conditions, in order to generate a three-dimensional semantic map of the complex flight scenario. A global path construction module is used to construct the initial global path of the UAV corresponding to the complex flight scenario based on the three-dimensional semantic map of the environment. The factor trajectory analysis module is used to establish a dynamic threat identification model for the UAV flight scenario, in order to identify the sudden threat factors of the flying UAV and analyze the potential motion trajectory of the sudden threat factors. The risk coefficient determination module is used to calculate the risk coefficient of the sudden threat factor to the flying drone based on the potential motion trajectory of the factor; The drone security early warning module is used to generate a smooth obstacle avoidance trajectory for the flying drone based on the risk coefficient, so as to perform security early warning for the flying drone.

Images