Unmanned aerial vehicle flight risk assessment method

Abstract

An unmanned aerial vehicle flight risk assessment method belongs to the technical field of unmanned aerial vehicle control. The objective of the invention is to solve the problem of accurate assessment of potential risks in a flight process of an unmanned aerial vehicle in a dynamic and complex environment. The method comprises the steps of identifying spatial positions of obstacles around the unmanned aerial vehicle; calculating the speed and acceleration of the obstacle; constructing a weighting model to predict the track of the obstacle, and then constructing a dynamic consistency index; the curvature and disturbance energy of the obstacle trajectory are calculated; establishing a spatial cross integral model; constructing a critical approach time index, and then combining the critical approach time index with a spatial cross integral model to establish a risk function to realize dynamic risk scoring; carrying out weighted fusion on the prediction tracks among the plurality of obstacles and the risk function, and constructing a multi-target coupling risk function; and constructing a comprehensive risk score of the obstacle through a weighted summation mode, and then carrying out grade division on the comprehensive risk score of the obstacle to complete unmanned aerial vehicle flight risk assessment.

Description

Technical Field

[0001] This invention belongs to the field of unmanned aerial vehicle (UAV) control technology, and specifically relates to a method for assessing the flight risks of UAVs. Background Technology

[0002] The safety of drones flying autonomously in complex environments is becoming increasingly prominent. Urban low-altitude airspace often contains numerous dynamic and static obstacles, such as pedestrians, vehicles, utility poles, and buildings, and these obstacles may exhibit unpredictable behavior patterns. Traditional methods relying on manual remote control or simple obstacle avoidance algorithms are insufficient to meet the safety requirements of flight paths in highly dynamic and dense environments. Therefore, there is an urgent need for a technology capable of real-time environmental perception, dynamic threat analysis, and quantitative assessment of flight risks. By modeling the spatial position and movement trends of obstacles before or during flight, and predicting their future trajectories and interactions with the drone, potential collision risks can be effectively predicted, enabling proactive avoidance and path optimization, thereby significantly improving the autonomous flight capabilities and mission completion efficiency of drones in complex scenarios. Most existing drone risk assessment methods rely on simplified models or static threshold strategies, making them ill-suited to the diversity and uncertainty of obstacles in real-world scenarios. On one hand, traditional methods often lack dynamic modeling of obstacle motion states (such as speed and acceleration), failing to accurately predict future behavior and leading to delayed or misjudgments in risk assessment. On the other hand, existing methods typically use a single indicator (such as distance or line-of-sight obstruction) for risk scoring, neglecting the combined effect of multiple factors such as obstacle trajectory morphology, approach speed, and spatial orientation. Furthermore, most methods cannot effectively handle the coupling relationships between multiple obstacles, potentially creating hidden risk blind spots in high-density areas. Summary of the Invention

[0003] The problem this invention aims to solve is to achieve a multi-dimensional, real-time, and quantifiable assessment of potential risks during UAV flight in dynamic and complex environments, and proposes a method for UAV flight risk assessment.

[0004] To achieve the above objectives, the present invention provides the following technical solution:

[0005] A method for assessing the flight risk of unmanned aerial vehicles (UAVs) includes the following steps:

[0006] S1. Capture image data of obstacles using a binocular camera mounted on the drone, and identify the spatial location of obstacles around the drone;

[0007] S2. Calculate the velocity and acceleration of the obstacle by measuring the change in its position over time;

[0008] S3. Using the velocity and acceleration of the obstacle obtained in step S2, construct the weight coefficients of the obstacle's historical position, construct a weighted model to predict the obstacle trajectory, and then construct a dynamic consistency index. By comparing the error between the velocity derivative and the acceleration, evaluate the physical continuity of the obstacle trajectory.

[0009] S4. Calculate the curvature and disturbance energy of the obstacle trajectory to characterize the stability of the obstacle's motion trajectory from both geometric and dynamic perspectives;

[0010] S5. Establish a spatial cross-integral model to measure the degree of projection of obstacles relative to the flight direction of the UAV;

[0011] S6. Construct a critical approach time index to estimate the time required for the UAV to make contact with the obstacle under the current speed conditions. Then, combine the critical approach time index with the spatial cross-integral model obtained in step S5 to establish a risk function and realize dynamic risk scoring.

[0012] S7. For multiple obstacles, the predicted trajectories of the multiple obstacles are weighted and fused with the risk function to construct a multi-objective coupled risk function;

[0013] S8. Integrate the multi-objective coupled risk function of obstacles, trajectory curvature, critical approach time, disturbance energy, and dynamic consistency, and construct a comprehensive risk score for obstacles by weighted summation. Then, classify the comprehensive risk score of obstacles to complete the UAV flight risk assessment.

[0014] Furthermore, the specific implementation method of step S1 includes the following steps:

[0015] S1.1. Using a binocular camera mounted on an unmanned aerial vehicle (UAV), image data of obstacles is captured, pixel coordinates of obstacles are extracted, and then the depth value of obstacles is calculated by combining binocular vision principles and parallax information.

[0016] S1.2. Based on the depth value of the obstacle obtained in step S1.1, the image coordinates of the obstacle are converted into the obstacle position in the world coordinate system using the camera's intrinsic parameter matrix and pose information.

[0017] Furthermore, the specific implementation method of step S3 includes the following steps:

[0018] S3.1. Considering the dynamic characteristics of the obstacle's time distance, velocity, and acceleration, construct the weighting coefficients for the obstacle's historical position. The calculation formula is as follows:

[0019]

[0020] in, The weighting coefficients for the historical locations of obstacles represent the k-th historical time point. The degree of contribution to predicting unknown future events; The time kernel width is determined by expert experience; , The obstacles at the k-th historical time point are respectively Velocity and acceleration; , These are reference values ​​for velocity and acceleration, determined by expert experience combined with historical data; This is the weight normalization term, determined by expert experience. For a moment, and These are the velocity contribution coefficient and the acceleration contribution coefficient, respectively, and they also serve to adjust the dimensions.

[0021] S3.2. Construct a weighted model for obstacle trajectory prediction. The calculation formula is as follows:

[0022]

[0023] in, For obstacles in Predicted location at any given time; For the obstacle at the k-th historical time point The historical position is obtained from historical data; K is the total number of historical time points, determined by expert experience combined with historical data.

[0024] S3.3. Construct a dynamic consistency index, the calculation formula of which is:

[0025]

[0026] in, The total time is determined by the collected data; As a dynamic consistency indicator; The time interval is determined by the actual data acquisition frequency; Let t be the acceleration of the obstacle. Let be the velocity of the obstacle at time t-1.

[0027] Furthermore, the specific implementation method of step S4 includes the following steps:

[0028] S4.1. Calculate the curvature of the obstacle's trajectory using the following formula:

[0029]

[0030] in, Let t be the curvature value of the obstacle; To prevent the elimination of zero terms, and to set empirical values, the dimensions are... Maintain consistency; Let be the velocity of the obstacle at time t. It serves as a curvature adjustment coefficient and also functions as a dimension adjustment factor.

[0031] S4.2. Calculate the disturbance energy of the obstacle using the following formula:

[0032]

[0033] in, , These are the weighting factors corresponding to velocity and acceleration, respectively. For perturbation energy, , These are the velocity reference value and the acceleration reference value, respectively, determined by expert experience.

[0034] Furthermore, the expression for establishing the spatial cross-integral model in step S5 is as follows:

[0035]

[0036] in, Let t be the position of the drone, obtained from the drone monitoring system; Let t be the speed of the drone at time t, obtained from the drone monitoring system; The spatial cross-integral value at time t; Let t be the position of the obstacle. To and The corresponding zero-prevention term is an empirical setpoint, with dimensions equal to... Maintain consistency.

[0037] Furthermore, the specific implementation method of step S6 includes the following steps:

[0038] S6.1. Construct the critical approach time index, and calculate it using the following formula:

[0039]

[0040] in, This refers to the critical approach time. For obstacles in Predicted location at time For the zero-prevention term corresponding to the critical approach time, an empirical setpoint is provided, with dimensions equal to... Maintain consistency;

[0041] S6.2. Construct the risk function, the calculation formula is as follows:

[0042]

[0043] in, This is the value of the risk function; This is the adjustment coefficient, set by expert experience.

[0044] The beneficial effects of this invention are:

[0045] This invention discloses a method for assessing the flight risks of unmanned aerial vehicles (UAVs). Based on image information acquired by a binocular camera, it accurately reconstructs the position and motion state of obstacles in three-dimensional space through image analysis and geometric reconstruction. Subsequently, by combining key indicators such as trajectory prediction, dynamic consistency analysis, spatial relativity measurement, and critical approach time calculation, a multi-dimensional and multi-scale risk assessment mechanism is constructed. Furthermore, through risk function mapping, multi-objective coupled modeling, and a hierarchical scoring system, a comprehensive quantification and classification output of the potential threats from different obstacles is achieved. The entire method exhibits good real-time performance, interpretability, and scalability, and is particularly suitable for high-dynamic operating scenarios such as urban low-altitude areas, complex terrain, or multi-object interference, providing an effective risk protection means for autonomous UAV flight.

[0046] This invention discloses a method for assessing the flight risk of unmanned aerial vehicles (UAVs). First, it employs binocular vision to acquire the three-dimensional position of obstacles, combining velocity, acceleration, and trajectory prediction to comprehensively construct a dynamic state model of the obstacles. Second, it introduces multiple indicators such as curvature, disturbance energy, and critical approach time to three-dimensionally characterize the behavioral features of the obstacles. Third, it proposes a spatial cross-integration and multi-objective coupling mechanism to effectively identify high-density risk areas. Finally, through weighted comprehensive scoring and risk level classification, it achieves a quantitative output of flight threats. Compared to traditional methods, this invention not only offers richer assessment dimensions and a more reasonable calculation model, but also possesses excellent real-time performance, scalability, and engineering deployment capabilities, making it widely applicable to flight safety tasks such as automatic obstacle avoidance and trajectory optimization.

[0047] The UAV flight risk assessment method described in this invention constructs a full-process mechanism from speed estimation and trajectory prediction to coupled risk aggregation, which enables a comprehensive judgment of the degree of obstacle threat, significantly improving the accuracy, timeliness and adaptability of the assessment, and is especially suitable for dynamic interaction scenarios such as urban low-altitude areas. Attached Figure Description

[0048] Figure 1 This is a flowchart of a method for assessing the flight risk of an unmanned aerial vehicle (UAV) according to the present invention;

[0049] Figure 2 This is a comparison chart of the comprehensive flight risk scores for the UAVs of this invention. Detailed Implementation

[0050] To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be further described in detail below with reference to the accompanying drawings and specific embodiments. It should be understood that the specific embodiments described herein are only for explaining the invention and are not intended to limit the invention; that is, the described specific embodiments are merely a part of the embodiments of the invention, and not all of them. The components of the specific embodiments of the invention described and shown in the accompanying drawings can generally be arranged and designed in various different configurations, and the invention may also have other embodiments.

[0051] Therefore, the following detailed description of specific embodiments of the invention provided in the accompanying drawings is not intended to limit the scope of the claimed invention, but merely to illustrate selected specific embodiments of the invention. All other specific embodiments obtained by those skilled in the art based on these specific embodiments without inventive effort are within the scope of protection of this invention.

[0052] To further understand the invention's content, features, and effects, the following specific embodiments are provided, along with accompanying drawings. Figure 1 - Appendix Figure 2 Detailed explanation is as follows:

[0053] Example 1:

[0054] A method for assessing the flight risk of unmanned aerial vehicles (UAVs) includes the following steps:

[0055] S1. Capture image data of obstacles using a binocular camera mounted on the drone, and identify the spatial location of obstacles around the drone;

[0056] Furthermore, the specific implementation method of step S1 includes the following steps:

[0057] S1.1. Using a binocular camera mounted on an unmanned aerial vehicle (UAV), image data of obstacles is captured, pixel coordinates of obstacles are extracted, and then the depth value of obstacles is calculated by combining binocular vision principles and parallax information.

[0058] S1.2. Based on the depth value of the obstacle obtained in step S1.1, the image coordinates of the obstacle are converted into the obstacle position in the world coordinate system using the camera's intrinsic parameter matrix and pose information.

[0059] Furthermore, identifying the spatial location of obstacles around the drone is fundamental to flight risk assessment. By capturing image data using a binocular camera mounted on the drone, and employing an intelligent obstacle recognition method embedded in the drone's monitoring system, the pixel coordinates of obstacles are identified and extracted. Combined with binocular vision principles and parallax information, the depth value of the obstacle can be calculated. Subsequently, using the camera's intrinsic parameter matrix and attitude information, the image coordinates can be converted into the actual position in the world coordinate system. This conversion process ensures that the spatial position of the obstacle has a unified reference frame, facilitating subsequent analyses such as velocity, acceleration, trajectory prediction, and collision estimation. This step provides the initial spatial geometric foundation for the entire assessment process.

[0060]

[0061]

[0062] in, This is the depth value; The focal length is the camera's focal length and can be obtained from the camera's parameters. The camera baseline length can be obtained from camera parameters; The disparity value of the target can be directly obtained by a binocular camera; , The rotation matrix and translation vector of the camera can be obtained from the camera parameters; This is the camera intrinsic parameter matrix, which can be obtained from the camera parameters; , These are the x and y coordinates of the image pixels, which can be read directly from the image; This refers to the position in the world coordinate system.

[0063] S2. Calculate the velocity and acceleration of the obstacle by measuring the change in its position over time;

[0064] Furthermore, based on the position changes at two consecutive moments, the velocity vector can be calculated; further differential calculations on the velocity yield the acceleration vector. These physical quantities reflect the movement trend of the obstacle, helping to determine whether it is approaching the drone or accelerating. The calculated velocity and acceleration results will play a crucial role in subsequent trajectory prediction, curvature estimation, collision detection, and risk scoring. This step uses time-synchronized position data for differential calculations to ensure that the dynamic information is consistent with the actual scene.

[0065]

[0066]

[0067] in, The time interval is determined by the actual data acquisition frequency; The speed of the obstacle can be calculated by a binocular camera; The acceleration of the obstacle can be calculated by the binocular camera;

[0068] S3. Using the velocity and acceleration of the obstacle obtained in step S2, construct the weight coefficients of the obstacle's historical position, construct a weighted model to predict the obstacle trajectory, and then construct a dynamic consistency index. By comparing the error between the velocity derivative and the acceleration, evaluate the physical continuity of the obstacle trajectory.

[0069] Furthermore, the specific implementation method of step S3 includes the following steps:

[0070] S3.1. Considering the dynamic characteristics of the obstacle's time distance, velocity, and acceleration, construct the weighting coefficients for the obstacle's historical position. The calculation formula is as follows:

[0071]

[0072] in, The weighting coefficients for the historical locations of obstacles represent the k-th historical time point. The degree of contribution to predicting unknown future events; The time kernel width is determined by expert experience; , The obstacles at the k-th historical time point are respectively Velocity and acceleration; , These are reference values ​​for velocity and acceleration, determined by expert experience combined with historical data; This is the weight normalization term, determined by expert experience. For a moment, and These are the velocity contribution coefficient and the acceleration contribution coefficient, respectively, and they also serve to adjust the dimensions.

[0073] S3.2. Construct a weighted model for obstacle trajectory prediction. The calculation formula is as follows:

[0074]

[0075] in, For obstacles in Predicted location at any given time; For the obstacle at the k-th historical time point The historical position is obtained from historical data; K is the total number of historical time points, determined by expert experience combined with historical data.

[0076] S3.3. Construct a dynamic consistency index, the calculation formula of which is:

[0077]

[0078] in, The total time is determined by the collected data; As a dynamic consistency indicator; The time interval is determined by the actual data acquisition frequency; Let t be the acceleration of the obstacle. Let be the velocity of the obstacle at time t-1.

[0079] Furthermore, the future movement trend of obstacles is a crucial basis for risk assessment. Based on historical position sequences, a weighted model can be constructed to predict their future positions. The weighting function considers both time distance and the dynamic characteristics of velocity and acceleration, making the prediction closer to the actual motion state. To verify the rationality of the prediction, a dynamic consistency index is introduced. By comparing the errors of the velocity derivative and acceleration, the physical continuity of the trajectory is evaluated. This index can be used to determine whether there are abnormal jumps in the trajectory. Using the above formula, the dynamic consistency index value of the i-th obstacle can be calculated and denoted as: .

[0080] S4. Calculate the curvature and disturbance energy of the obstacle trajectory to characterize the stability of the obstacle's motion trajectory from both geometric and dynamic perspectives;

[0081] Furthermore, the specific implementation method of step S4 includes the following steps:

[0082] S4.1. Calculate the curvature of the obstacle's trajectory using the following formula:

[0083]

[0084] in, Let t be the curvature value of the obstacle; To prevent the elimination of zero terms, and to set empirical values, the dimensions are... Maintain consistency; Let be the velocity of the obstacle at time t. It serves as a curvature adjustment coefficient and also functions as a dimension adjustment factor.

[0085] S4.2. Calculate the disturbance energy of the obstacle using the following formula:

[0086]

[0087] in, , These are the weighting factors corresponding to velocity and acceleration, respectively. For perturbation energy, , These are the velocity reference value and the acceleration reference value, respectively, determined by expert experience.

[0088] Furthermore, the shape characteristics of an obstacle's trajectory have a significant impact on flight safety. Trajectory curvature can be used to determine whether sharp turns or evasive maneuvers are occurring; a higher curvature indicates a more pronounced path change. Simultaneously, a disturbance energy index is introduced to measure the severity of changes in velocity and acceleration. These two indices, from both geometric and dynamic perspectives, characterize the stability of the trajectory and are crucial for assessing potential threats. High curvature or high disturbance energy both indicate an unstable target trajectory, requiring attention. Using the above formula, the curvature value of the i-th obstacle can be calculated and denoted as... The calculated disturbance energy value of the i-th obstacle can be denoted as: .

[0089] S5. Establish a spatial cross-integral model to measure the degree of projection of obstacles relative to the flight direction of the UAV;

[0090] Furthermore, the expression for establishing the spatial cross-integral model in step S5 is as follows:

[0091]

[0092] in, Let t be the position of the drone, obtained from the drone monitoring system; Let t be the speed of the drone at time t, obtained from the drone monitoring system; The spatial cross-integral value at time t; Let t be the position of the obstacle. To and The corresponding zero-prevention term is an empirical setpoint, with dimensions equal to... Maintain consistency.

[0093] Furthermore, whether an obstacle is located in the drone's direction of travel is a key factor in determining its threat level. To address this, a spatial cross-integral index is introduced to measure the degree to which an obstacle is projected relative to the flight direction. This index calculates the cosine of the angle between the obstacle's relative position and the drone's velocity direction; a larger value indicates that the target is more directly ahead of the drone's flight path. This index can serve as a directional adjustment factor for the risk function, allowing the assessment method to focus more on potential obstacles located along the flight path.

[0094] S6. Construct a critical approach time index to estimate the time required for the UAV to make contact with the obstacle under the current speed conditions. Then, combine the critical approach time index with the spatial cross-integral model obtained in step S5 to establish a risk function and realize dynamic risk scoring.

[0095] Furthermore, the specific implementation method of step S6 includes the following steps:

[0096] S6.1. Construct the critical approach time index, and calculate it using the following formula:

[0097]

[0098] in, This refers to the critical approach time. For obstacles in Predicted location at time For the zero-prevention term corresponding to the critical approach time, an empirical setpoint is provided, with dimensions equal to... Maintain consistency;

[0099] S6.2. Construct the risk function, the calculation formula is as follows:

[0100]

[0101] in, This is the value of the risk function; The adjustment coefficient is set based on expert experience.

[0102] Furthermore, determining whether an obstacle poses a threat to flight requires considering not only its distance but also its relative motion trend. This step proposes a critical approach time index to estimate the time required for contact with the obstacle under current speed conditions. A smaller critical approach time value indicates a higher approach speed and a greater risk. Further, a risk function is introduced that combines the rate of change of the critical approach time with spatial cross-integral to establish a dynamic risk score. This function nonlinearly amplifies the negative derivative of the critical approach time, allowing for a higher risk value when the obstacle approaches rapidly, which is used in subsequent aggregation and comprehensive scoring stages. Using the above formula, the critical approach time of the i-th obstacle can be calculated and denoted as... The calculated risk function value for the i-th obstacle can be denoted as: ;

[0103] S7. For multiple obstacles, the predicted trajectories of the multiple obstacles are weighted and fused with the risk function to construct a multi-objective coupled risk function;

[0104] Furthermore, in complex environments, multiple obstacles may appear simultaneously, creating a combined impact on flight paths. Relying solely on the risk value of a single target is insufficient to comprehensively reflect the overall threat. Therefore, a coupled risk function is introduced, which weights and fuses the spatial distance and risk values ​​between multiple targets. If multiple high-risk targets exist around a certain obstacle, the coupling effect will significantly enhance its threat level. This function adjusts for distance through exponential decay and can also incorporate a directional consistency factor to improve the response capability to clustered, high-density risk areas.

[0105]

[0106] Where i is the number of the i-th obstacle currently being evaluated, and j is the number of other obstacles that are different from the i-th obstacle. , These are the risk function values ​​for the i-th and j-th obstacles, respectively. The coupling adjustment coefficient between the i-th obstacle and the j-th obstacle is determined by expert experience; , The predicted positions of the i-th and j-th obstacles are respectively calculated by step S3; This represents the coupling risk value. The distance attenuation control parameter is determined by experts based on the distance to obstacles, and... Dimensions are consistent;

[0107] S8. Integrate the multi-objective coupled risk function of obstacles, trajectory curvature, critical approach time, disturbance energy, and dynamic consistency, and construct a comprehensive risk score for obstacles by weighted summation. Then, classify the comprehensive risk score of obstacles to complete the UAV flight risk assessment.

[0108] Furthermore, after calculating risk indicators across multiple dimensions, they need to be integrated into a unified risk score for ranking and classification. The comprehensive scoring function integrates five indicators: coupled risk value, trajectory curvature, critical approach time, disturbance energy, and dynamic consistency, and outputs the final score through a weighted summation. A higher score indicates a greater threat to flight safety posed by the obstacle. The final score will be divided into five levels (1-5) to facilitate rapid response from the planner or manual operating system. This classification mechanism offers good interpretability and real-time performance, making it suitable for assessing UAV flight safety in highly dynamic environments.

[0109]

[0110] in, The overall risk score for the i-th obstacle; Let i be the coupling risk value of the i-th obstacle; Let be the curvature value of the i-th obstacle; Let be the critical approach time of the i-th obstacle; Let be the disturbance energy of the i-th obstacle; Let be the dynamic consistency index of the i-th obstacle; , , , , These are the weighting coefficients corresponding to the coupling risk value, curvature value, critical approach time, disturbance energy, and dynamic consistency index of the i-th obstacle, respectively. They are determined by expert experience and also have the function of adjusting the dimensions.

[0111]

[0112] in, , , , The thresholds for classifying levels 1 through 4 can be determined by experts based on historical data or experience. Let be the risk level of the i-th obstacle.

[0113] This embodiment is illustrated with an example as follows:

[0114] The drone is equipped with a binocular camera with an image resolution of 1280×720, a baseline length of 0.12 m, a focal length of 700 pixels, and the image center point coordinates as (640, 360). The pixel coordinates of the obstacle identified in the current frame are (720, 380), and the parallax value is 24 pixels.

[0115] The current position of obstacle ZA1 is (3.0, 1.2, 2.0), and the position of the obstacle 0.1s ago was (2.6, 0.9, 2.0). The reference velocity and acceleration are set to (3.5, 2.5) and (1, 3) respectively.

[0116] Using the method in this embodiment, the overall risk score for the current obstacle to the drone's flight is calculated to be 2.61 points; threshold , , , The values ​​are 1, 2, 3, and 4 respectively. Therefore, the current obstacle corresponds to a drone flight risk level of 3. Similarly, the comprehensive drone flight risk scores for obstacles ZA2 (25.5, 31.4, 49.3), ZA3 (59.1, 70.4, 20.7), ZA4 (31.6, 56.3, 55.5), and ZA5 (21.6, 48.9, 69.7) are 3.55, 1.68, 4.51, and 2.69 respectively, corresponding to drone flight risk levels of 3, 2, 5, and 3.

[0117] It should be noted that relational terms such as "first" and "second" are used merely to distinguish one entity or operation from another, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Furthermore, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or apparatus. Without further limitations, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or apparatus that includes said element.

[0118] Although this application has been described above with reference to specific embodiments, various modifications can be made and components can be replaced with equivalents without departing from the scope of this application. In particular, as long as there is no structural conflict, the features in the specific embodiments disclosed in this application can be combined with each other in any way. The lack of an exhaustive description of these combinations in this specification is merely for the sake of brevity and resource conservation. Therefore, this application is not limited to the specific embodiments disclosed herein, but includes all technical solutions falling within the scope of the claims.

Claims

1. A method for assessing the flight risk of unmanned aerial vehicles (UAVs), characterized in that, Includes the following steps: S1. Capture image data of obstacles using a binocular camera mounted on the drone, and identify the spatial location of obstacles around the drone; S2. Calculate the velocity and acceleration of the obstacle by measuring the change in its position over time; S3. Using the velocity and acceleration of the obstacle obtained in step S2, construct the weight coefficients of the obstacle's historical position, construct a weighted model to predict the obstacle trajectory, and then construct a dynamic consistency index. By comparing the error between the velocity derivative and the acceleration, evaluate the physical continuity of the obstacle trajectory. S4. Calculate the curvature and disturbance energy of the obstacle trajectory to characterize the stability of the obstacle's motion trajectory from both geometric and dynamic perspectives; S5. Establish a spatial cross-integral model to measure the degree of projection of obstacles relative to the flight direction of the UAV; S6. Construct a critical approach time index to estimate the time required for the UAV to make contact with the obstacle under the current speed conditions. Then, combine the critical approach time index with the spatial cross-integral model obtained in step S5 to establish a risk function and realize dynamic risk scoring. S7. For multiple obstacles, the predicted trajectories of the multiple obstacles are weighted and fused with the risk function to construct a multi-objective coupled risk function; S8. Integrate the multi-objective coupled risk function of obstacles, trajectory curvature, critical approach time, disturbance energy, and dynamic consistency, and construct a comprehensive risk score for obstacles by weighted summation. Then, classify the comprehensive risk score of obstacles to complete the UAV flight risk assessment.

2. The method for assessing the flight risk of an unmanned aerial vehicle (UAV) according to claim 1, characterized in that, The specific implementation method of step S1 includes the following steps: S1.

1. Using a binocular camera mounted on an unmanned aerial vehicle (UAV), image data of obstacles is captured, pixel coordinates of obstacles are extracted, and then the depth value of obstacles is calculated by combining binocular vision principles and parallax information. S1.

2. Based on the depth value of the obstacle obtained in step S1.1, the image coordinates of the obstacle are converted into the obstacle position in the world coordinate system using the camera's intrinsic parameter matrix and pose information.

3. The method for assessing the flight risk of an unmanned aerial vehicle (UAV) according to claim 2, characterized in that, The specific implementation method of step S3 includes the following steps: S3.

1. Considering the dynamic characteristics of the obstacle's time distance, velocity, and acceleration, construct the weighting coefficients for the obstacle's historical position. The calculation formula is as follows: ； in, The weighting coefficients for the historical locations of obstacles represent the k-th historical time point. The degree of contribution to predicting unknown future events; The time kernel width is determined by expert experience; , The obstacles at the k-th historical time point are respectively Velocity and acceleration; , These are reference values ​​for velocity and acceleration, determined by expert experience combined with historical data; The weight normalization term is determined by expert experience. For a moment, and These are the velocity contribution coefficient and the acceleration contribution coefficient, respectively, and they also serve to adjust the dimensions. S3.

2. Construct a weighted model for obstacle trajectory prediction. The calculation formula is as follows: ； in, For obstacles in Predicted location at any given time; For the obstacle at the k-th historical time point The historical position is obtained from historical data; K is the total number of historical time points, determined by expert experience combined with historical data. S3.

3. Construct a dynamic consistency index, the calculation formula of which is: ； in, The total time is determined by the collected data; As a dynamic consistency indicator; The time interval is determined by the actual data acquisition frequency; Let t be the acceleration of the obstacle. Let be the velocity of the obstacle at time t-1.

4. The method for assessing the flight risk of a drone according to claim 3, characterized in that, The specific implementation method of step S4 includes the following steps: S4.

1. Calculate the curvature of the obstacle's trajectory using the following formula: ； in, Let t be the curvature value of the obstacle; To prevent the elimination of zero terms, and to set empirical values, the dimensions are... Maintain consistency; Let be the velocity of the obstacle at time t. It serves as a curvature adjustment coefficient and also functions as a dimension adjustment factor. S4.

2. Calculate the disturbance energy of the obstacle using the following formula: ； in, , These are the weighting factors corresponding to velocity and acceleration, respectively. For perturbation energy, , These are the velocity reference value and the acceleration reference value, respectively, determined by expert experience.

5. The method for assessing the flight risk of an unmanned aerial vehicle (UAV) according to claim 4, characterized in that, Step S5 establishes the expression for the spatial cross-integral model as follows: ； in, Let t be the position of the drone, obtained from the drone monitoring system; Let t be the speed of the drone at time t, obtained from the drone monitoring system; The spatial cross-integral value at time t; Let t be the position of the obstacle. To and The corresponding zero-prevention term is an empirical setpoint, with dimensions equal to... Maintain consistency.

6. The method for assessing the flight risk of an unmanned aerial vehicle (UAV) according to claim 5, characterized in that, The specific implementation method of step S6 includes the following steps: S6.

1. Construct the critical approach time index, and calculate it using the following formula: ； in, The critical approach time; For obstacles in Predicted location at time For the zero-prevention term corresponding to the critical approach time, an empirical setpoint is provided, with dimensions equal to... Maintain consistency; S6.

2. Construct the risk function, the calculation formula is as follows: ； in, This is the value of the risk function; This is the adjustment coefficient, set by expert experience.

Images