Attitude control method and system for landing gear during descent of an autogyro

By identifying ground features to construct a three-dimensional ground model and dynamically adjusting the landing gear attitude, the problem of inaccurate attitude control during the descent of the autogyro was solved, enabling the autogyro to land safely in complex terrain.

CN123114013APending Publication Date: 2026-11-17YIKAILAI (YANTAI) AVIATION EQUIPMENT CO LTD
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
CN202611025613.X
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2026-07-10
Publication Date
2026-11-17

AI Technical Summary

Technical Problem

In the existing technology, the gyroplane fails to effectively consider the dynamic descent action of the landing gear and multiple attitude parameters during the descent process, resulting in insufficient accuracy of attitude control events.

Method used

By identifying ground features in the descent area of ​​the autogyro, a three-dimensional ground model is constructed to determine the descent position and mode of the autogyro. Based on real-time images and attitude parameters of the landing gear, the descent action of the landing gear is dynamically adjusted, and attitude compensation is performed based on the attitude delay, thereby achieving autonomous attitude control.

Benefits of technology

It improves the accuracy of landing gear attitude control events, ensuring safe landing of autogyros in different terrains.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN123114013A_ABST
    Figure CN123114013A_ABST
Patent Text Reader

Abstract

This invention discloses a method and system for attitude control of a landing gear during the descent of an autogyro. The invention relates to the technical field of attitude control methods. Based on real-time images of the landing gear and the descent mode of the autogyro, the dynamic descent action of the landing gear is determined. Attitude control events are determined based on the dynamic descent action, multiple attitude parameters of the landing gear, and the corresponding overall shape, thus improving the accuracy of attitude control events. Therefore, the attitude delay portion of the landing gear is determined based on multiple sub-attitude adjustment events, the position corresponding to the ground 3D model, and the real-time shape of the landing gear. Attitude compensation events are determined based on multiple damaged parts of the landing gear and the attitude delay portion. Multiple attitude compensation actions are determined based on the identification of attitude compensation events. Finally, autonomous attitude control events of the landing gear are determined based on the current attitude of the landing gear, multiple attitude compensation actions, and the action efficiency of the landing gear.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] This invention relates to the technical field of attitude control methods, and more particularly to an attitude control method and system for a landing gear during the descent of an autogyro. Background Technology

[0002] With the development of technology, landing gear has been applied to autogyros. As an integral part of the autogyro, the landing gear can withstand the enormous impact force during landing and adapt to the landing requirements of different terrains. In existing technologies, autogyros use landing gear during descent and collect the descent motion of the landing gear. This descent motion is not dynamic but a single form, ignoring the dynamic descent motion of the landing gear and the consideration of multiple attitude parameters of the landing gear. This makes it impossible to realize attitude control events and affects the accuracy of autonomous attitude control events of the landing gear. Summary of the Invention

[0003] The purpose of this invention is to overcome the shortcomings of the prior art. This invention provides a method and system for attitude control of the landing gear during the descent process of an autogyro.

[0004] This invention provides an attitude control method for the landing gear during the descent of an autogyro, comprising: when the autogyro is in a descent state, determining multiple ground features based on the identification of ground images of the descent area of ​​the autogyro; determining a ground 3D model based on the feature positions and corresponding feature shapes of the multiple ground features and the regional shape of the descent area, and marking the descent position of the autogyro on the ground 3D model; determining the descent mode of the autogyro based on the descent position of the autogyro and multiple descent data of the autogyro; and determining the dynamic descent motion of the landing gear based on real-time images of the landing gear and the descent mode of the autogyro. The system determines attitude control events based on the dynamic descent of the landing gear, multiple attitude parameters of the landing gear, and the corresponding overall shape. Based on the identification of these attitude control events, multiple sub-attitude adjustment events are determined. Based on these sub-attitude adjustment events, the corresponding positions of the ground 3D model, and the real-time shape of the landing gear, the attitude delay portion of the landing gear is determined. Based on the multiple damaged parts of the landing gear and the attitude delay portion of the landing gear, attitude compensation events are determined. Based on the identification of these attitude compensation events, multiple attitude compensation actions are determined. Finally, based on the current attitude of the landing gear, the multiple attitude compensation actions, and the action efficiency of the landing gear, the autonomous attitude control event of the landing gear is determined.

[0005] This invention provides an attitude control system for a landing gear during the descent of an autogyro. This attitude control system is applied to the aforementioned attitude control method for a landing gear during the descent of an autogyro. The attitude control system for the landing gear during the descent of the autogyro includes: The ground feature module is used to determine multiple ground features based on the recognition of ground images of the descent area of ​​the autogyro when it is in the descent state. The descent position module is used to determine the ground 3D model based on the feature positions and corresponding feature shapes of multiple ground features and the regional shape of the descent area, and to mark the descent position of the autogyro on the ground 3D model; The attitude control event module is used to determine the descent mode of the gyrocopter based on its descent position and multiple descent data, determine the dynamic descent action of the landing gear based on the real-time image of the landing gear and the descent mode of the gyrocopter, and determine the attitude control event based on the dynamic descent action of the landing gear, multiple attitude parameters of the landing gear and the corresponding overall shape. The attitude delay module is used to determine multiple sub-attitude adjustment events based on the identification of the attitude control event, and to determine the attitude delay part of the landing gear according to the multiple sub-attitude adjustment events, the position corresponding to the ground 3D model, and the real-time shape of the landing gear. The attitude autonomous control event module is used to determine attitude compensation events based on multiple damaged parts of the landing gear and the attitude delay of the landing gear, determine multiple attitude compensation actions based on the identification of attitude compensation events, and determine the attitude autonomous control event of the landing gear based on the current attitude of the landing gear, multiple attitude compensation actions and the action efficiency of the landing gear.

[0006] Compared with the prior art, the beneficial effects of the present invention are: In this embodiment of the invention, the descent mode of the gyrocopter is determined based on its descent position and multiple descent data. The dynamic descent action of the landing gear is determined based on the real-time image of the landing gear and the descent mode of the gyrocopter. The attitude control event is determined based on the dynamic descent action of the landing gear, multiple attitude parameters of the landing gear, and the corresponding overall shape. A ground 3D model is introduced, which is compatible with the overall consideration of the dynamic descent action of the landing gear, multiple attitude parameters of the landing gear, and the corresponding overall shape, thereby improving the accuracy of the attitude control event.

[0007] Therefore, attitude compensation events are determined based on multiple damaged parts of the landing gear and the attitude delay portion of the landing gear. Multiple attitude compensation actions are determined based on the identification of attitude compensation events. The autonomous attitude control events of the landing gear are determined based on the current attitude of the landing gear, multiple attitude compensation actions, and the action efficiency of the landing gear. The attitude delay portion is introduced to further control the attitude delay portion, realizing the overall consideration of the current attitude of the landing gear, multiple attitude compensation actions, and the action efficiency of the landing gear, thus improving the accuracy of the autonomous attitude control events of the landing gear. Attached Figure Description

[0008] Figure 1 This is a flowchart illustrating the attitude control method of the landing gear during the descent of an autogyro in an embodiment of the present invention. Figure 2 This is a flowchart illustrating step S11 of the attitude control method for the landing gear during the descent of an autogyro in an embodiment of the present invention. Figure 3 This is a flowchart illustrating step S12 of the attitude control method for the landing gear during the descent of an autogyro in an embodiment of the present invention. Figure 4 This is a flowchart illustrating step S13 of the attitude control method for the landing gear during the descent of an autogyro in an embodiment of the present invention. Figure 5 This is a flowchart illustrating step S14 of the attitude control method for the landing gear during the descent of an autogyro in an embodiment of the present invention. Figure 6 This is a flowchart illustrating step S15 of the attitude control method for the landing gear during the descent of an autogyro in an embodiment of the present invention. Figure 7 This is a schematic diagram of the structural composition of the attitude control system of the landing gear during the descent process of an autogyro in an embodiment of the present invention. Detailed Implementation

[0009] The technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings in the embodiments of the present invention.

[0010] Please see Figures 1 to 7 An attitude control method for a landing gear during the descent of an autogyro includes: Step S11: When the autogyro is in the descent state, multiple ground features are determined based on the identification of the ground image of the descent area of ​​the autogyro; Step S12: Determine the ground 3D model based on the feature positions and corresponding feature shapes of multiple ground features and the regional shape of the descent area, and mark the descent position of the autogyro on the ground 3D model; Step S13: Determine the descent mode of the gyrocopter based on its descent position and multiple descent data. Determine the dynamic descent action of the landing gear based on the real-time image of the landing gear and the descent mode of the gyrocopter. Determine the attitude control event based on the dynamic descent action of the landing gear, multiple attitude parameters of the landing gear, and the corresponding overall shape. Step S14: Based on the identification of the attitude control event, determine multiple sub-attitude adjustment events, and determine the attitude delay part of the landing gear according to the multiple sub-attitude adjustment events, the position corresponding to the ground three-dimensional model, and the real-time shape of the landing gear. Step S15: Determine attitude compensation events based on multiple damaged parts of the landing gear and attitude delays of the landing gear; determine multiple attitude compensation actions based on the identification of attitude compensation events; and determine autonomous attitude control events of the landing gear based on the current attitude of the landing gear, multiple attitude compensation actions, and the action efficiency of the landing gear.

[0011] refer to Figure 2 In step S11, the specific steps are as follows: S111: Real-time monitoring of the descent process of the autogyro, determining the descent state of the autogyro based on multiple descent data and corresponding descent patterns; simultaneously, collecting data on the descent area of ​​the autogyro. S112: Based on the camera of the autogyro capturing images of the descent area, a ground image of the descent area is determined. Multiple sub-ground regions are determined based on the recognition of the ground image of the descent area. In the multiple sub-ground regions, corresponding sub-ground features are determined based on the detection of each sub-ground region. Multiple ground features are determined based on the feature morphology of multiple sub-ground features, the regional morphology of the corresponding sub-ground regions, and the transition region between two adjacent sub-ground features.

[0012] In the embodiments of this application, the descent process of the autogyro is monitored in real time, and multiple descent data of the autogyro are collected, such as: vertical descent speed, horizontal movement speed, fuselage attitude, rotor status, and environmental parameters.

[0013] Multiple descent data points are fused using Kalman filtering to eliminate measurement errors from single sensors. Based on historical descent data and the current state, the descent trajectory is predicted for the next 3-5 seconds. Descent states are categorized according to descent speed, attitude stability, and environmental conditions: stable descent (vertical speed stable at 2-3 m / s, fuselage attitude change less than 5°); adjustment descent (requiring adjustment of descent parameters to adapt to ground conditions); and emergency descent (excessive descent speed or unstable attitude requiring immediate intervention). Simultaneously, the current descent state is comprehensively assessed based on descent data and morphological analysis results. A risk index for the current descent state is calculated, including ground impact risk and attitude loss of control risk. The descent state is updated at a frequency of 50 Hz to ensure real-time performance.

[0014] Based on the current altitude, descent speed, and horizontal movement speed, the basic landing area is calculated; a safety boundary is extended around the basic area, typically 1.5 times the size of the basic area; the area's extent and shape are dynamically adjusted according to changes in the descent state; simultaneously, terrain types within the area are identified using image recognition and a terrain database; obstacle distribution within the area is detected using LiDAR and a vision system; elevation changes within the area are analyzed to assess ground slope; complete regional data is constructed by combining satellite imagery, UAV aerial imagery, and real-time captured images; and a three-dimensional model of the descent area is built based on the collected data.

[0015] Furthermore, the autogyro is equipped with at least three high-definition cameras (front-facing, downward-facing, and side-facing), with a resolution of no less than 1920×1080 and a frame rate of 30fps. Images captured from multiple angles are stitched together in real time to form a complete panoramic image of the descent area. For ground image recognition, a deep learning-based semantic segmentation algorithm (such as U-Net) is used to segment the ground image into different regions. Based on color, texture, and shape features, the segmented regions are classified as follows: Grassland: predominantly green with uniform texture; Soil: brown or yellow with coarse texture; Water: blue or black with strong surface reflectivity; Obstacles: irregularly shaped with a clear contrast to the surrounding environment. Region boundary extraction: The Canny edge detection algorithm is used to extract the boundary lines of each region, providing a basis for subsequent sub-ground region division.

[0016] The ground image was divided into 0.5m × 0.5m grid cells, each cell serving as a basic sub-ground region. Based on the similarity (color, texture, height, etc.) of adjacent grid cells, similar grid cells were merged into larger sub-ground regions. A unique identifier was assigned to each sub-ground region, and its location, size, and basic attributes were recorded. Simultaneously, geometric parameters such as area, perimeter, and shape factor of each sub-ground region were calculated. The texture features of each sub-ground region, including contrast, correlation, energy, and uniformity, were analyzed using the Gray-Level Co-occurrence Matrix (GLCM). The hue, saturation, and brightness characteristics of each sub-ground region were analyzed in the HSV color space. Finally, the height variation of each sub-ground region was analyzed by combining stereo vision or laser ranging data.

[0017] Additionally, the following features are extracted: Grass height: grass height is estimated through image analysis and stereo vision measurement; Grass density: grass density is assessed by analyzing the proportion and distribution of green pixels in the image; Grass type: grass type (e.g., short grass, long grass, withered grass) is identified based on texture and color features; Land feature extraction: Soil moisture: soil moisture is assessed through color analysis (wet soil is darker); Soil texture: different textures such as sandy soil and clay are distinguished based on texture analysis; Obstacle feature extraction: Obstacle type: different types of obstacles such as stones, tree stumps, and mounds are identified; Obstacle size: the height, width, and length of obstacles are measured; Obstacle shape: the outline features of obstacles are analyzed to determine their shape regularity.

[0018] For the quantification and parameterization of sub-surface features: the extracted sub-surface features are converted into numerical parameters, and different features are assigned weights based on their impact on landing safety, such as obstacles having the highest weight and grass height having the second highest. Simultaneously, transition recognition is performed between adjacent sub-surface features. This involves identifying the boundary lines between adjacent sub-surface regions; measuring the width of the transition region from one sub-surface feature to another; calculating the gradient of feature changes within the transition region, such as the rate of change of grass height and soil moisture; and classifying transition types: abrupt transitions: features change significantly over a short distance, such as a direct transition from grassland to bare ground; gradual transitions: features change gradually over a longer distance, such as a gradual decrease in grass height; transition risk assessment: abrupt transitions are generally higher risk, causing sudden changes in the landing gear's stress; gradual transitions are lower risk, giving the system more time to adjust its attitude; and establishing a mathematical model of the transition region to describe the pattern of feature changes within the transition region.

[0019] Sub-ground features and transition area features are fused to form a comprehensive ground feature description; the spatial relationships between different ground features are analyzed, such as the relative positions of obstacles and grass, and the distribution of grass at different heights; at the same time, each ground feature is represented as a multi-dimensional feature vector, containing information such as location, type, intensity, and range, to generate a ground feature distribution map, which intuitively displays the distribution of various features in the descent area.

[0020] refer to Figure 3 In step S12, the specific steps are as follows: S121: Collect multiple ground features, determine the feature location of each ground feature based on the recognition of each ground feature, and determine the corresponding feature shape based on the multi-directional recognition of the feature location of the ground feature; S122: Determine the first layer of stereo model based on the feature location of multiple ground features and the regional morphology of the descending area; determine the second layer of stereo model based on the feature morphology of multiple ground features and the regional morphology of the descending area. S123: Determine the ground stereo model based on the first stereo model and the second stereo model; determine the preset descent area based on the traversal of the ground stereo model; determine the descent position of the gyrocopter based on the preset descent area, the flight direction of the gyrocopter and the descent speed of the gyrocopter, so as to mark the descent position of the gyrocopter on the ground stereo model.

[0021] In the embodiments of this application, ground feature acquisition and location determination are key steps for the safe landing of the autogyro. The system acquires comprehensive ground information through multi-source data acquisition technology; high-definition cameras provide image data with a resolution of no less than 1920×1080, lidar scanning obtains accurate distance and altitude information, and infrared sensors detect ground temperature distribution to help identify different types of ground features; in the feature recognition stage, the system uses a convolutional neural network to recognize ground images. The training model contains more than 10,000 images of grassland scenes, and combines the Canny edge detection algorithm to identify feature boundaries, and applies the GLCM texture analysis algorithm to distinguish different types of grassland.

[0022] The feature selection process evaluates features based on metrics such as size, significance, and stability, eliminating temporary features like the shadow of wind-blown grass while retaining stable ground features. Each feature is quantified using a 0-100 importance scoring system. For feature location determination, the system establishes a three-dimensional coordinate system with the gyroplane's current position as the origin, defining the X-axis as the flight direction, the Y-axis as the lateral direction perpendicular to the flight direction, and the Z-axis as the altitude direction. The three-dimensional coordinates are then projected onto a two-dimensional ground plane. Position calculation employs a monocular vision ranging algorithm and a stereo vision matching algorithm, combined with GPS positioning information, to map the feature location to a global coordinate system.

[0023] To improve positioning accuracy, the system uses the RANSAC algorithm to remove outliers, applies Kalman filtering to smooth feature locations, and establishes a position confidence evaluation model, assigning a reliability score of 0-1 to each feature location. In the multi-directional recognition and feature morphology determination stage, the system collects data on the same ground feature from at least three different angles, adjusts the camera focal length and exposure parameters to ensure consistent image quality, and increases the acquisition angle for important features to improve the accuracy of morphology recognition.

[0024] Morphological parameter extraction includes geometric morphological parameters (height, width, area, perimeter, roundness, etc.), physical morphological parameters (density, hardness, elasticity, etc. estimated through image analysis), and texture morphological parameters (roughness, directionality, contrast, etc.). The system uses the K-means clustering algorithm to automatically classify feature morphologies, establishes a morphological description model, describes feature morphologies using natural language, and assigns a standardized morphological code to each feature morphology to facilitate subsequent processing and analysis. This complete ground feature acquisition and location determination process provides the autogyro with accurate ground environment information, supporting its safe landing decisions.

[0025] Specifically, when the autogyro was preparing to land over mixed grass, the system identified five grass areas of different densities, three obstacles, and two transition zones through data collection from high-definition cameras and lidar. After feature filtering, the system retained four important features: G2 high-density grass area (importance score 85), G4 medium-density grass area (importance score 90), O1 rocks (importance score 95), and T1 transition zone from tall grass to low grass (importance score 75).

[0026] In the feature location determination stage, the system establishes a coordinate system with the current position of the autogyro as the origin and calculates the specific positions of each feature: the center of region G2 is located at (25.3, 18.7), with a confidence level of 0.92; the center of region G4 is located at (30.2, 25.5), with a confidence level of 0.95; the stone in O1 is located at (28.5, 30.2), with a confidence level of 0.98; the transition region T1 extends from (22.1, 15.3) to (24.8, 17.9), with a confidence level of 0.85. The system uses Kalman filtering to smooth the feature positions, and the position of region G4 is optimized and adjusted to (30.15, 25.52), with the confidence level increased to 0.96.

[0027] In the multi-directional recognition and feature morphology determination stage, the system performs a detailed analysis of each feature from three perspectives. The front view image of region G2 shows grass height of about 20cm and density of about 80%. The bottom view image shows grass color uniform without obvious mottled spots. The side view image shows grass stems straight and elastic. The overall morphological description is "high-density, medium-height grassland area with straight grass stems and good elasticity", and the morphological code is GH-MD-SE. The front view image of region G4 shows grass height of about 15cm and density of about 65%. The bottom view image shows grass color with slight variations and minor mottled spots. The side view image shows grass stems slightly bent and with moderate elasticity. The overall morphological description is "medium-density, medium-height grassland area with slightly bent grass stems and moderate elasticity", and the morphological code is GH-MD-ME.

[0028] The front view of stone O1 shows a height of approximately 30cm and a width of approximately 40cm. The bottom view shows an irregular shape with sharp edges, and the side view shows a rough, angular surface. The overall morphological description is "irregularly shaped stone, 30cm high, 40cm wide, with a rough surface and sharp edges," and the morphological code is ST-IR-SE. The front view of the T1 transition area shows a transition width of approximately 3m, with grass height gradually decreasing from 20cm to 10cm. The bottom view shows a density gradually decreasing from 80% to 50%, and the side view shows a gradual decrease in the straightness of the grass stems. The overall morphological description is "a gradual transition area from tall grass to short grass, with a transition width of 3m and uniform changes in density and height," and the morphological code is TR-GH-HD.

[0029] The system further extracted and quantified the morphological parameters, obtaining the following results: G2 region height 20cm±2cm, density 80%±5%, elastic coefficient 0.85±0.05; G4 region height 15cm±2cm, density 65%±5%, elastic coefficient 0.65±0.05; O1 stone height 30cm±1cm, width 40cm±1cm, hardness coefficient 0.95±0.02; T1 transition area transition width 3m±0.2m, height change gradient 3.3cm / m, density change gradient 10% / m. This precise ground feature information provides key data support for the autogyro to select the optimal landing point, ensuring its safe and stable landing on the mixed grass.

[0030] Furthermore, a first-level stereo model is determined based on the feature locations of multiple ground features and the regional morphology of the descending region. A second-level stereo model is determined based on the feature morphology of multiple ground features and the regional morphology of the descending region. This comprehensive consideration of the feature morphology of multiple ground features and the regional morphology of the descending region ensures the accuracy of the second-level stereo model.

[0031] At this point, the system constructs a dual-dimensional model based on the location and shape information of ground features, providing comprehensive environmental perception for the safe landing of the autogyro. The first-level dimensional model focuses on the geometric structure of the terrain. The system receives multiple ground features and their precise locations identified in S121, including feature points such as G2, G4, O1, and T1. Each point has three-dimensional coordinates (x, y, z) obtained by the fusion system of lidar and vision. In the regional morphology analysis stage, the system evaluates the geometry of the entire descent area, including parameters such as slope, curvature, and undulation changes, and uses a triangular mesh algorithm to connect feature points into spatial triangular patches to construct the basic terrain surface.

[0032] The first layer of the 3D model is constructed using the Delaunay triangulation algorithm, which connects feature points into a continuous triangular mesh and calculates the normal vector for each triangular facet to determine the tilt direction and angle of the ground. The system calculates the height of non-feature points through interpolation to form a continuous surface and stores the model in the form of a 3D mesh. Each vertex contains complete (x,y,z) coordinate information. This geometric model can be visualized through elevation maps or 3D topographic maps, providing basic terrain data for subsequent landing path planning.

[0033] The second layer of the 3D model focuses on the physical properties of the ground. The system integrates various feature morphological parameters extracted from S121, such as grass height, density, hardness, and elasticity, so that each feature point not only contains location information but also comes with a set of physical attribute vectors. During the regional morphological attribute mapping process, the system maps these physical attributes to each triangular facet or grid cell in the first layer of the model and uses inverse distance weighted or kriging interpolation methods to spatially interpolate the attributes, ensuring the continuity and accuracy of attribute distribution.

[0034] The second 3D model, as a physical property model, reflects the characteristics of the ground, such as hardness, friction, and elasticity. It is not an independent geometric structure, but rather an "attribute layer" attached to the geometric model. During the construction process, the system associates each grid cell with a set of physical properties, including parameters such as grass height (cm), density (%), hardness coefficient (0~1), and elasticity coefficient (0~1). The distribution of these properties can be visually displayed through color mapping or heatmaps. The output 3D model with attribute annotations can support the formulation of landing impact prediction and attitude compensation strategies, providing a comprehensive decision-making basis for the safe landing of autogyros in complex grassland environments.

[0035] Specifically, assuming an autogyro is descending over a grassy area, the system has identified the following characteristics, illustrated in Table 1: Table 1. Schematic diagram of each feature First layer of 3D model construction: Delaunay triangulation is performed using these points to form the following triangular mesh: Triangular facet 1: G2-G4-T1; Triangular facet 2: G2-O1-T1; Triangular facet 3: G4-O1-T1; Calculate the normal vector of each facet. For example, the normal vector of facet G2-G4-T1 is (0.12, -0.08, 0.99), indicating that the area is slightly tilted, generating a continuous elevation surface, showing that the grassland is generally gently undulating, but there is a local bulge at the O1 stone.

[0036] The second layer of 3D model construction: Mapping each feature attribute to a mesh cell: G2 area: grass height 20cm, density 80%, hardness 0.3 (soft); G4 area: grass height 15cm, density 65%, hardness 0.4 (medium); O1 stone: hardness 0.95 (extremely hard); T1 transition zone: attributes between G2 and G4; Using IDW interpolation to generate a continuous attribute distribution map: around O1, the hardness value increases sharply; in the T1 area, grass height and density show a gradual transition; The output is a 3D model with physical attributes, which can be used to determine the impact force distribution when the landing gear contacts the ground.

[0037] Therefore, a ground stereo model is determined based on the first and second stereo models; a preset descent area is determined by traversing the ground stereo model; and the descent position of the gyrocopter is determined based on the preset descent area, the gyrocopter's flight direction, and its descent speed. The descent position of the gyrocopter is then marked on the ground stereo model. This approach takes into account the preset descent area, the gyrocopter's flight direction, and its descent speed, ensuring the accuracy of the gyrocopter's descent position.

[0038] At this point, during the construction of the ground 3D model, the system merges the first geometric model with the second physical attribute model to form a unified ground 3D model. The two models share the same spatial grid structure, and each grid cell contains both geometric information (elevation, slope) and physical attributes (grass height, density, hardness). The system adopts a weighted fusion strategy, assigning weights according to the importance of features, with geometric features having a weight of 0.6 (because terrain undulations directly affect landing safety) and physical attributes having a weight of 0.4 (affecting buffering and stability). A comprehensive score is calculated for each grid cell. The fused model is stored in the form of a three-dimensional grid, with each vertex containing coordinates (x, y, z), slope, normal vector, and complete information such as grass height, density, and hardness.

[0039] To ensure model accuracy, the system performs cross-validation, using lidar point clouds and visual images to verify model accuracy, requiring the mean square error between model predictions and actual measurements to be less than 5 cm. During the descent of the autogyro, the system continuously acquires new ground data and uses a Kalman filter algorithm to update the model in real time, continuously improving model accuracy.

[0040] In the ground 3D model traversal and preset descent area determination stage, the system starts from the model boundary and uses a spiral traversal algorithm to gradually search towards the central area. A safety score is calculated for each grid cell, which comprehensively considers three factors: slope, hardness, and obstacle distance. The slope scoring criteria are: 100 points for a slope less than 5°, 60 points for 5°-10°, and 0 points for a slope greater than 10°. The hardness scoring criteria are: 100 points for a hardness of 0.2-0.6, 60 points for 0.6-0.8, and 20 points for a hardness greater than 0.8. The obstacle distance scoring criteria are: 100 points for a distance greater than 5m, 60 points for 2-5m, and 0 points for a distance less than 2m.

[0041] The system selects continuous grid cells with a safety score greater than 80 to form candidate regions, requiring the area of ​​the candidate regions to be no less than 3 times the area occupied by the landing gear; morphological processing is performed on the candidate regions to eliminate isolated points and small holes, and the centroid position of the region is calculated as the center point of the preset descent region. The system determines the final descent position based on flight parameters. It acquires the gyrocopter's current heading angle (e.g., 45° northeast), vertical descent speed (2.5 m / s), and horizontal speed (1.2 m / s), predicts the flight trajectory for the next 5 seconds, and calculates the horizontal displacement of 6 m and the vertical displacement of 12.5 m. Within the preset descent area, the system finds the grid cell closest to the predicted trajectory endpoint, prioritizing the safe area ahead of the heading, and finally determines the optimal descent position coordinates (x, y, z). This position is marked on the ground 3D model, and a safety boundary with a radius of 3 meters is set, providing precise guidance for the safe landing of the gyrocopter.

[0042] Specifically, during the construction of the ground 3D model, the system merges the first geometric model with the second physical attribute model to form a unified ground 3D model. The two models share the same spatial grid structure, with each grid cell containing both geometric information (elevation, slope) and physical attributes (grass height, density, hardness). The system employs a weighted fusion strategy, assigning weights based on feature importance. Geometric features have a weight of 0.6 (because terrain undulations directly affect landing safety), while physical attributes have a weight of 0.4 (affecting cushioning and stability). A comprehensive score is calculated for each grid cell. The fused model is stored in the form of a 3D mesh, with each vertex containing coordinates (x, y, z), slope, normal vector, and complete information such as grass height, density, and hardness.

[0043] To ensure model accuracy, the system performs cross-validation, using lidar point clouds and visual images to verify model accuracy, requiring the mean square error between model predictions and actual measurements to be less than 5 cm. During the descent of the autogyro, the system continuously acquires new ground data and uses a Kalman filter algorithm to update the model in real time, continuously improving model accuracy.

[0044] In the ground 3D model traversal and preset descent area determination stage, the system starts from the model boundary and uses a spiral traversal algorithm to gradually search towards the central area. A safety score is calculated for each grid cell, which comprehensively considers three factors: slope, hardness, and obstacle distance. The slope scoring criteria are: 100 points for a slope less than 5°, 60 points for 5°-10°, and 0 points for a slope greater than 10°. The hardness scoring criteria are: 100 points for a hardness of 0.2-0.6, 60 points for 0.6-0.8, and 20 points for a hardness greater than 0.8. The obstacle distance scoring criteria are: 100 points for a distance greater than 5m, 60 points for 2-5m, and 0 points for a distance less than 2m.

[0045] The system selects continuous grid cells with a safety score greater than 80 to form candidate regions, requiring the area of ​​the candidate regions to be no less than three times the area occupied by the landing gear; the candidate regions are morphologically processed to eliminate isolated points and small holes, and the centroid of the region is calculated as the center point of the preset descent region.

[0046] The system determines the final descent position based on flight parameters. It acquires the gyrocopter's current heading angle (e.g., 45° northeast), vertical descent speed (2.5 m / s), and horizontal speed (1.2 m / s), predicts the flight trajectory for the next 5 seconds, and calculates the horizontal displacement of 6 m and the vertical displacement of 12.5 m. Within the preset descent area, the system finds the grid cell closest to the predicted trajectory endpoint, prioritizing the safe area ahead of the heading, and finally determines the optimal descent position coordinates (x, y, z). This position is marked on the ground 3D model, and a safety boundary with a radius of 3 meters is set, providing precise guidance for the safe landing of the gyrocopter.

[0047] refer to Figure 4 In step S13, the specific steps are as follows: S131: Real-time monitoring of the descent process of the autogyro and collection of multiple descent data of the autogyro. Based on the multiple descent data and the descent position of the autogyro, multiple descent combinations are determined. Based on the identification of multiple descent combinations, the corresponding descent action behavior is determined. Based on the multiple descent action behaviors, the altitude position of the autogyro and the ground three-dimensional model, the descent mode of the autogyro is determined. S132: Acquire real-time images of the landing gear, determine the real-time shape and trajectory of the landing gear based on the recognition of the real-time images, and determine the dynamic descent action of the landing gear based on the real-time shape and trajectory of the landing gear and the descent mode of the autogyro. S133: Collect multiple attitude parameters of the landing gear, determine the first attitude control coefficient based on the multiple attitude parameters of the landing gear and the dynamic descent action of the landing gear, determine the second attitude control coefficient based on the multiple attitude parameters of the landing gear and the overall shape of the landing gear, and determine the attitude control event based on the mapping relationship between the first attitude control coefficient, the second attitude control coefficient and the attitude control event.

[0048] In the embodiments of this application, the descent process of the autogyro is monitored in real time, and multiple descent data of the autogyro are collected. The descent data collected by the system includes basic motion data (absolute altitude, relative altitude, vertical speed, horizontal speed), attitude data (pitch angle, roll angle, yaw angle) and environmental data (wind speed, wind direction).

[0049] Based on the collected descent data and descent position information, the system generates multiple descent combinations; the position data is combined with GPS and ground stereo model to generate accurate three-dimensional position information, and position features such as slope, roughness and obstacle distance are extracted; the system generates a complete data combination at a time point every 0.5 seconds, each combination containing motion data, attitude data, position features and environmental data. A typical descent process (from altitude 30m to 0m) generates about 60 combinations, each combination containing complete flight status information.

[0050] The system identifies descent behaviors based on multiple descent combinations, employing a temporal pattern recognition algorithm and a pre-trained LSTM neural network model trained on over 1000 real landing records. Behavior types include stable descent, lateral drift, pitch oscillation, roll instability, and wind disturbance. During identification, the system extracts variation features from five consecutive combinations, calculates the rate of change and variance of each parameter, and determines the behavior based on preset logic. For example, a stable descent is identified when the vertical velocity variance is less than 0.1 and the roll angle variance is less than 0.5; lateral drift is identified when the horizontal velocity exceeds 1.0 m / s; pitch oscillation is identified when the rate of change of pitch angle is greater than 0.5° / s; roll instability is identified when the roll angle variance is greater than 1.0; and wind disturbance is identified when the wind speed is greater than 3.0 m / s and the trajectory deviation exceeds 2 m.

[0051] The system determines the descent mode based on the descent behavior, altitude, and ground 3D model. The descent modes are divided into standard descent mode, compensated descent mode, emergency descent mode, and precision descent mode. The mode determination logic comprehensively considers the main behavior type, the current altitude range, and the characteristics of the ground 3D model. The standard descent mode is used when the behavior is a stable descent with an altitude greater than 10m and a flat ground surface. The compensated descent mode is used when the behavior is unstable with roll or pitch oscillations. The emergency descent mode is used when the behavior is affected by wind disturbance and the altitude is less than 5m. The precision descent mode is used when the ground complexity is greater than 0.7 and the altitude is less than 3m. Each mode corresponds to different control parameter configurations: the standard descent mode uses standard control gain and conservative adjustments; the compensated descent mode adds attitude feedback gain and active compensation; the emergency descent mode uses maximum control authority and rapid response; and the precision descent mode uses fine control and small-step adjustments to ensure the autogyro can land safely and smoothly under various conditions.

[0052] Specifically, during the autogyro's landing on the grass, the system executed the complete S131 procedure to ensure a safe landing. During the data acquisition phase from 14:35:00 to 14:35:30, the system recorded key flight parameters, including altitude of 8.5m, descent speed of 2.3m / s, horizontal speeds Vx=0.3m / s and Vy=0.1m / s, attitude angles θ=1.5°, φ=2.1°, ψ=190° at 14:35:15, as well as positional characteristics such as slope of 3°, roughness of 0.4, obstacle distance of 4.8m, and environmental data including wind speed of 2.8m / s and wind direction of 60°.

[0053] Based on this data, the system generated five consecutive descent combinations, T123 to T127. Each combination contained complete motion data, attitude data, position features, and environmental data. In the behavior recognition stage, the system analyzed the characteristics of the T123-T127 combinations and found that the vertical velocity was stable in the range of 2.2-2.4 m / s, the roll angle φ varied from 1.8° to 2.5° with a variance of 0.9, and the wind speed exceeded 2.5 m / s, causing the trajectory to deviate to the right by 1.5 m. Through a temporal pattern recognition algorithm, the system determined that the primary behavior was roll instability (RU) and the secondary behavior was wind disturbance (WD), with confidence levels of 82% and 65%, respectively.

[0054] During the descent mode determination phase, the system comprehensively considers the current altitude of 8.5m (medium altitude range), the main roll instability behaviors, and the moderate complexity displayed by the ground stereo model (grass + slight slope). Based on the decision logic "IF(behavior = roll instability AND altitude = 8.5m) THENCDM", it determines to adopt the compensated descent mode (CDM). This mode is configured with an attitude feedback gain of 1.5 times the standard value, an adjustment frequency of 15Hz (10Hz higher than the standard), and a maximum adjustment angle of ±5°.

[0055] The system outputs a compensated descent mode (CDM), triggered by roll instability (φ variance = 0.9). The control strategy is to enhance attitude feedback and actively compensate for the roll angle, with the expected action being to adjust the roll angle to within the range of 0° ± 1°. Through the complete execution of step S131, the system can accurately identify the flight status of the autogyro during its landing on grass and select the most suitable descent mode, providing an accurate mode basis for subsequent attitude control and greatly improving the landing safety and reliability of the autogyro in complex grass environments.

[0056] Furthermore, during the autogyro's landing, the system monitors the landing gear's status and formulates dynamic descent maneuvers by acquiring real-time images of the landing gear. The image acquisition system deploys a high-definition industrial camera on each side of the fuselage bottom, with a resolution of 1920×1080, a frame rate of 30fps, and a wide angle of 120°. It is also equipped with built-in infrared LEDs to adapt to twilight or nighttime landing environments. The system automatically adjusts the exposure time (1 / 1000-1 / 60s) and white balance to adapt to the green grass environment, while using H.264 encoding to balance image clarity and transmission bandwidth. The acquired images undergo preprocessing, including using adaptive median filtering to remove noise, histogram equalization to enhance contrast, Sobel operators to strengthen the landing gear outline, and HSV color space to separate the grass background from the landing gear.

[0057] Based on the preprocessed images, the system performs real-time morphological recognition, identifies landing gear joints using the SIFT algorithm, obtains the landing gear shape using Canny edge detection, and calculates morphological parameters such as main strut angle, tire orientation, and shock absorber travel. Simultaneously, the system employs Lucas-Kanade optical flow and inter-frame difference techniques for behavioral trajectory analysis, calculating trajectory parameters such as descent speed, swing amplitude, and oscillation frequency. These morphological and trajectory data, along with descent patterns (such as CDM) from S131, are input into a multi-source data fusion system for comprehensive analysis using weighted decision trees and fuzzy logic algorithms.

[0058] Based on the fused data, the system makes dynamic action decisions, defining three action types: pre-compression, attitude adjustment, and damping control. The decision logic triggers corresponding actions according to specific conditions: when the swing amplitude exceeds a threshold and the current descent mode is compensated, damping control is executed; when the main strut angle exceeds a threshold, attitude adjustment is executed. The system also accurately calculates various action parameters: the pre-compression amount is calculated based on the descent speed and ground hardness, the attitude adjustment angle is determined based on the main strut angle deviation, and the damping force is set based on the swing amplitude and oscillation frequency. This dynamic descent action control system based on real-time image monitoring and multi-source data fusion can effectively cope with various landing challenges in complex grassland environments, ensuring that the autogyro completes the landing process safely and smoothly.

[0059] Specifically, in the actual application of the grass landing scenario, the autogyro faced environmental conditions of medium-height grass (grass height 15-20cm), slight slope (5°) and light wind (3m / s). It was currently in compensated descent mode (CDM) at an altitude of 5m and a descent speed of 1.2m / s. The system collected real-time images of the landing gear through high-definition cameras on the left and right sides. The left side showed the main strut deflected 3° to the right, the shock absorber compressed 25%, and the tire tilted 2° to the left. The right side showed the main strut deflected 2° to the left, the shock absorber compressed 20%, and the tire tilted 1° to the right. The image processing results showed a 95% success rate in key point recognition, clear and unobstructed contour extraction, and slight swaying in motion detection.

[0060] Based on image analysis, the system calculates real-time morphological parameters: average main support angle deviation of 2.5°, average shock absorber compression of 22.5%, and average tire tilt of 1.5°, as well as behavioral trajectory parameters: falling speed of 1.2 m / s, swing amplitude of 0.3 m, and oscillation frequency of 1.5 Hz. In the data fusion analysis stage, the system uses the support angle of 2.5° (exceeding the threshold of 2°), the swing amplitude of 0.3 m (approaching the threshold of 0.4 m), and the CDM mode as input parameters for decision-making. Since the support angle exceeds the threshold, the system determines that attitude adjustment is required, while the swing amplitude does not reach the threshold, so damping control is not required.

[0061] The system accurately calculates the motion parameters: the attitude adjustment angle is 0.8 × 2.5 = 2°, and the pre-compression is 0.5 × 1.2 + 0.3 × 0.6 = 0.78m; it generates motion commands, using a combination of attitude adjustment and pre-compression, with the left strut adjusted by -2° and the right strut adjusted by +2°, and the shock absorber pre-compressed by 78%, scheduled to be executed 2 seconds before landing; the expected execution effect is that the landing gear will be close to vertical at touchdown (deviation < 0.5°), the shock absorber will effectively absorb the impact energy, the impact of the grass slope will be compensated, and in terms of safety indicators, the load factor will be controlled within 2.0g, the attitude stabilization time will be less than 3 seconds, and there will be no sideslip or bouncing, ensuring the safe and stable landing of the autogyro in complex grass environments.

[0062] Therefore, multiple attitude parameters of the landing gear are collected. Based on these parameters and the landing gear's dynamic descent, a first attitude control coefficient is determined. A second attitude control coefficient is determined based on these parameters and the overall shape of the landing gear. The attitude control event is then determined based on the mapping relationship between the first and second attitude control coefficients and the attitude control event. This approach takes into account the overall consideration of the mapping relationship between the first and second attitude control coefficients and the attitude control event, ensuring the accuracy of the attitude control event. Furthermore, a ground 3D model is introduced, taking into account the landing gear's dynamic descent, multiple attitude parameters, and corresponding overall shape, further improving the accuracy of the attitude control event.

[0063] During the landing of the autogyro, the system achieves precise control by collecting multiple attitude parameters of the landing gear. The attitude parameter acquisition system is equipped with various sensors such as tilt sensors, displacement sensors, pressure sensors, accelerometers, and gyroscopes, which sample at a high frequency of 100Hz to ensure the capture of rapid changes. The data accuracy reaches 0.1 for angle, 0.1mm for displacement, and 0.5kPa for pressure, with a synchronization error of less than 1ms for all sensor timestamps. The attitude parameters collected by the system include geometric parameters (main strut angle, crossbeam tilt angle, wheel track change), dynamic parameters (shock absorber compression, compression speed, ground reaction force), and vibration parameters (vertical and horizontal vibration acceleration, vibration frequency).

[0064] Based on these parameters, the system determines the first attitude control coefficient K1, which reflects the degree of matching between the current attitude and the ideal dynamic action, with a value range of [0,2], where 1 represents a perfect match. K1 is calculated using a weighted allocation method, with the main support angle weighted at 0.3, the shock absorber compression weighted at 0.25, the ground reaction force weighted at 0.25, and the vibration parameter weighted at 0.2, calculated using a weighted formula. The system also determines the second attitude control coefficient K2 to evaluate the stability and safety of the current overall form, with a value range of [0,1], where 1 represents the most stable state. K2 is calculated based on three form evaluation indicators: symmetry, balance, and stability, taking into account the angle difference between the left and right supports, the ratio of the left and right ground reaction forces, and the vibration amplitude and frequency, respectively.

[0065] The system determines attitude control events based on the values ​​of K1 and K2 and a preset mapping relationship, defining a total of 9 control events, from E1 over-control adjustment to E9 emergency correction. For example, when K1>1.2 and K2>0.9, the E1 over-control adjustment event is triggered, reducing the shock absorber pre-compression by 10-20% and lowering the control gain by 15%. When K1<0.8 and K2<0.7, the E9 emergency correction event is triggered, adjusting the strut angle by ±5-10 degrees, setting the maximum shock absorber pre-compression to 90-100%, and controlling the adjustment time within 0.5 seconds. This dual-coefficient control method can dynamically adjust the control strategy according to real-time attitude parameters, ensuring that the autogyro can safely and stably complete the landing process under various complex conditions.

[0066] Specifically, in the autogyro landing scenario on grass, the system faced grass with a height of 15-20cm, medium density, a ground slope of 3 degrees to the left and right, a descent speed of 1.2m / s, and a crosswind of 5m / s. It performed dynamic descent maneuvers, including left strut adjustment -2, right strut adjustment +2, and shock absorber pre-compression of 70%. The attitude parameters collected by the system showed that the left strut angle was -1.8, the right strut angle was +2.3, the crossbeam tilt angle was 2.1, the wheel track change was +5mm, the left shock absorber compression was 68mm, the right shock absorber compression was 72mm, the left ground reaction force was 2.8kN, the right ground reaction force was 3.2kN, the vertical vibration acceleration was 0.3g, and the dominant vibration frequency was 12Hz.

[0067] Based on these parameters, the system calculated the first attitude control coefficient K1. By comparing it with the ideal parameters (θ_idealL=-2, θ_idealR=+2, C_idealL=C_idealR=70mm, F_idealL=F_idealR=3.0kN), the matching degree of each parameter was obtained: support angle matching degree 0.95, shock absorber compression matching degree 0.8, ground reaction force matching degree 0.8, vibration parameter matching degree 0.7, and K1 value 0.825. At the same time, the system calculated the second attitude control coefficient K2, and evaluated the symmetry (Δθ=4.1), balance (R=0.875), and stability (A=0.3g, f=12Hz) of the morphology, obtaining a symmetry score of 0.59, a balance score of 0.734, a stability score of 0.562, and K2 value 0.625.

[0068] Based on K1=0.825 (within the range of 0.8-1.2) and K2=0.625 (within the range of 0.5-0.7), the system determines the control event as E6: attitude correction through the mapping table. The execution parameters include left strut adjustment +1.5, right strut adjustment -1.5, and shock absorber pre-compression increase of 5% to 75%, with an adjustment time of 1.5 seconds. The expected execution effect will adjust the left strut angle from -1.8 to -0.3, the right strut angle from +2.3 to +0.8, the crossbeam tilt angle from 2.1 to 0.5, and the ground reaction force balance close to 1.0. In terms of safety indicators, K1 is expected to increase to 0.95, K2 is expected to increase to 0.8, and the vibration amplitude is expected to decrease to below 0.2g, thereby effectively improving the attitude stability and safety of the autogyro during the landing process on grass.

[0069] refer to Figure 5 In step S14, the specific steps are as follows: S141: Based on the identification of the attitude control event, multiple attitude control items are determined, and multiple sub-attitude adjustment events are determined according to the multiple attitude control items and the real-time form of the landing gear. In the sub-attitude adjustment event, the sub-attitude adjustment event presents the adjustment content of the attitude of the landing gear at the corresponding time node. S142: Acquire the ground stereo model and determine the corresponding attitude influencing factors based on multiple sub-attitude adjustment events and the ground stereo model; S143: Determine the corresponding attitude delay factors based on multiple sub-attitude adjustment events and the real-time shape of the landing gear, and determine the attitude delay portion of the landing gear based on the attitude influencing factors, attitude delay factors, and the overall shape of the landing gear.

[0070] In the embodiments of this application, in the autogyro landing control system, the identification and execution of attitude control events are key to ensuring a safe landing. The system receives attitude control events (such as E6: "attitude correction") output from step S133, classifies them, extracts targets, and identifies constraints through an event parsing process, and clarifies the control targets (such as adjusting the main strut angle to 0°±0.5°) and physical and temporal constraints. Based on this information, the system determines multiple attitude control items, including P1 strut angle adjustment, P2 shock absorber pre-compression, P3 crossbeam leveling correction, and P4 wheel track dynamic adjustment, and prioritizes them according to their impact on landing stability, with P1 being the highest priority and P4 being the lowest priority.

[0071] The system matches real-time morphological data with control targets and calculates various deviations: strut angle deviation 2.3°, shock absorber compression deviation 10%, and crossbeam tilt deviation 1.5°, determining that three control items, P1, P2, and P3, need to be executed. Subsequently, the system divides the total control time window of 3 seconds into 6 time nodes, with each node lasting 0.5 seconds, of which 1 second before landing contact (T4) is the critical adjustment point. Based on the time allocation principle and parameter calculation formula, the system generates a specific sub-attitude adjustment event sequence for each control item.

[0072] For the P1 strut angle adjustment, the system distributes the total adjustment amount of -2.3° to the T0-T4 stages according to the time ratio, generating a sub-event sequence that gradually adjusts from 2.3° to 0.0°. For the P2 damper pre-compression, the system distributes the total adjustment amount of +10% evenly to the T0-T4 stages, generating a sub-event sequence that gradually increases from 5% to 15%. For the P3 crossbeam leveling, the system mainly distributes the total adjustment amount of -1.5° to the T1-T5 stages, generating a sub-event sequence that gradually adjusts from 1.5° to 0.0°. This sub-attitude adjustment event generation method based on time node division ensures the continuity and smoothness of attitude adjustment during the autogyro's landing process, providing precise control assurance for safe landing.

[0073] Furthermore, the stereo model construction process involves raw data acquisition, obtaining LiDAR point clouds containing ground elevation, slope, and roughness information, providing stereo image pairs with texture, color, and vegetation cover information, as well as real-time flight altitude change data. Subsequent data preprocessing includes point cloud filtering to remove noise points and outliers, image correction to eliminate lens distortion and lighting effects, and time synchronization processing to ensure consistent timestamps across multiple data sources. In the model construction phase, the point cloud data is converted into a 1m x 1m grid, and characteristic parameters such as average height, height standard deviation, slope, roughness, and vegetation cover are calculated for each grid. Finally, cross-validation, accuracy evaluation, and real-time update mechanisms ensure the model's accuracy and timeliness.

[0074] Based on sub-attitude adjustment events and ground stereo model, a comprehensive framework for analyzing attitude influencing factors was established. The 15 sub-attitude adjustment events from S141 were divided into three categories: angle adjustment, compression control, and horizontal correction. The influencing factors include topographic geometric factors (slope, roughness, height difference), surface physical factors (soil hardness, vegetation characteristics, humidity conditions), and environmental dynamic factors (wind speed and direction, temperature changes, and light conditions).

[0075] The system uses specially designed algorithm functions to calculate the specific effects of various influencing factors: the slope influence analysis function calculates different influence coefficients according to the adjustment type, with angle adjustment being the most sensitive to slope, followed by horizontal correction, and compression control being relatively less sensitive; the roughness influence analysis function particularly emphasizes the impact on compression control, with its influence coefficient being 1.5 times that of other adjustment types; the soil hardness influence function dynamically calculates the hardness coefficient based on soil type and moisture conditions, with clay, loam, and sand having different basic hardness values ​​and moisture sensitivities; and the vegetation influence function comprehensively considers both vegetation cover and grass height.

[0076] The system integrates various influencing factors through a comprehensive impact calculation function, with different weighting schemes for different types of adjustment events. Angle adjustment events are most concerned with the impact of slope (weight 0.4), compression control events are most concerned with the impact of roughness (weight 0.4), and level correction events are also most concerned with the impact of slope (weight 0.5). The system can output the comprehensive impact coefficient of each sub-attitude adjustment event and identify the most important influencing factors, providing a scientific basis for the subsequent attitude control strategy formulation. This method based on multi-source data fusion and quantitative analysis of influencing factors significantly improves the attitude control accuracy and landing safety of autogyros in complex grassland environments.

[0077] Therefore, the corresponding attitude delay factors are determined based on multiple sub-attitude adjustment events and the real-time shape of the landing gear. The attitude delay portion of the landing gear is determined based on the attitude influencing factors, attitude delay factors, and the overall shape of the landing gear. This approach takes into account the overall consideration of attitude influencing factors, attitude delay factors, and the overall shape of the landing gear, ensuring the accuracy of the attitude delay portion of the landing gear.

[0078] At this point, during the autogyro landing process, the system needs to comprehensively analyze and determine the corresponding attitude delay factors based on multiple sub-attitude adjustment events and the real-time shape of the landing gear. The system reviews 15 sub-attitude adjustment events from S141, including main strut angle adjustment, crossbeam leveling, etc. Each event includes a time node, target attitude, and execution parameters. At the same time, it collects morphological data such as the current landing gear angle, tire compression, crossbeam tilt, and shock absorber stroke in real time, providing a basis for identifying delay factors.

[0079] The method for identifying attitude delay factors mainly focuses on three aspects: mechanical response delay, environmental disturbance delay, and control system delay. Mechanical response delay includes the response time of the hydraulic system (80-120ms for main support adjustment, 60-100ms for crossbeam leveling, and 40-80ms for shock absorber adjustment) and the clearance of the transmission mechanism (0.5-1° for universal joint clearance and 0.3-0.8mm for connecting rod clearance). Environmental disturbance delay considers the influence of grass resistance (an additional resistance increases by 10-15% when the grass height is 15-25cm, and the viscosity effect of wet grass increases the delay by 5-10%) and the influence of wind disturbance (a 20-30% compensation time is required when the crosswind is 5m / s, and the response delay of gusts is 50-100ms). Control system delay includes sensor sampling delay (20ms for angle sensors, 15ms for pressure sensors, and 10ms for height sensors) and control calculation cycle (50ms for main control cycle, 30ms for attitude calculation, and 20ms for command generation).

[0080] In the quantification stage of attitude delay factors, the system calculates the total delay using the formula: Total Delay = Mechanical Delay + Environmental Delay + Control Delay. For example, the total delay of main column adjustment reaches 200ms (100ms mechanical + 30ms environmental + 70ms control). Delay distribution characteristics include fixed delay (such as control calculation cycle, sensor sampling), variable delay (such as hydraulic response, wind disturbance impact), and cumulative delay (the superposition of delays from multiple consecutive adjustment events). The system also establishes a delay severity assessment system, classifying delays into three levels: slight (<10%), moderate (10-30%), and severe (>30%), and further categorizing them into three types based on their impact range: local, diffuse, and systemic.

[0081] Based on attitude influencing factors, attitude delay factors, and overall landing gear morphology, the system further determines the attitude delay component. It integrates the five types of attitude influencing factors identified in S142 (terrain slope 30%, ground roughness 25%, soil conditions 20%, vegetation characteristics 15%, and micro-topography 10%) and identifies delays through three dimensions: time delay (command execution time exceeds expectations), angle delay (deviation of attitude adjustment from the target value), and position delay (deviation of spatial position from the expected position). The system adopts delay compensation calculation methods, including time compensation (delay time = expected time × (1 + delay coefficient)) and angle compensation (compensation angle = target angle × (1 + comprehensive influence coefficient)).

[0082] refer to Figure 6 In step S15, the specific steps are as follows: S151: Collect past damage events of the landing gear, identify multiple damaged parts of the landing gear based on the identification of past damage events, and mark the damage data of each damaged part; determine the first attitude compensation amount based on the relative positions of the multiple damaged parts of the landing gear and the attitude delay of the landing gear. S152: Determine the second attitude compensation amount based on the damage data of multiple damaged parts of the landing gear and the attitude delay of the landing gear; determine the attitude compensation event based on the first attitude compensation amount, the second attitude compensation amount and the attitude compensation mapping relationship; S153: Based on the identification of attitude compensation events, determine multiple compensation action features; based on the multiple compensation action features, the corresponding compensation time nodes, and the overall shape of the landing gear, determine multiple attitude compensation actions; based on the current attitude of the landing gear and the multiple attitude compensation actions, determine the first level of autonomous control content; based on the multiple attitude compensation actions and the landing gear's action efficiency, determine the second level of autonomous control content; based on the first level of autonomous control content and the second level of autonomous control content, determine the attitude autonomous control event of the landing gear.

[0083] In the embodiments of this application, in the landing control system of the autogyro, the system collects past damage events of the landing gear to obtain comprehensive damage data; the system accesses the aircraft maintenance history database to extract all maintenance records, fault reports and abnormal event logs related to the landing gear in the past three years, and at the same time retrieves the structural health monitoring system records in each flight, paying particular attention to abnormal data that exceed the threshold, such as vertical impact force greater than 2.5 times the design load, lateral offset greater than 15 degrees, etc.

[0084] Based on the collected data, the system classifies and identifies damage events, categorizing them by damage type into structural deformation (such as bending of main supports and twisting of beams), loose connections (such as loose bolts and increased hinge gaps), functional degradation (such as decreased damping of shock absorbers and uneven tire wear), and external impacts (such as impacts from grass and stones stuck in the ground). The system uses pattern recognition technology to automatically identify recurring damage patterns from a large number of records. For example, if multiple records show that "the left main support experienced a slight shift after landing on the grass," it is identified as a high-frequency damage location.

[0085] In the stage of identifying multiple damaged parts of the landing gear and marking damage data, the system divides the landing gear into 8 key monitoring parts, each with a unique identifier and spatial coordinates: point A (connection point on the main strut), point B (hydraulic cylinder in the middle of the main strut), point C (lower wheel axle connection on the main strut), point D (left side connection of the crossbeam), point E (central hinge of the crossbeam), point F (right side connection of the crossbeam), point G (front shock absorber), and point H (rear shock absorber). By statistically analyzing the number of times each part was damaged in historical records, a damage frequency heatmap is generated. For example, point C (wheel axle connection) is damaged... The system records 3 instances of damage, with point E (central hinge) damaged twice, and the remaining parts damaged once or zero times. Each damaged area is marked with detailed data, including damage level (Level 1: Slight, Level 2: Moderate, Level 3: Severe), damage type (deformation, loosening, wear, impact), repair status (repaired, partially repaired, unrepaired), and performance impact (percentage of influence on attitude control). For example, point C (wheel-axle connection) is marked as damage level 2, damage type: deformation, repair status: partially repaired, and performance impact: 15%, indicating a 15% decrease in control accuracy at this location.

[0086] Based on the relative position of the damaged parts and the attitude delay, the system further determines the first attitude compensation amount; establishes a three-dimensional spatial model of the landing gear, and clarifies the relative positional relationship of each damaged part. For example, point C (wheel axle connection) is located at the bottom of the main support, and point D (left side connection of the crossbeam) is located at the left end of the crossbeam, with a horizontal distance of 1.2 meters between them; the system analyzes the damage propagation effect, such as the deformation of point C causing the entire main support to shift, which in turn affects the stress state of points D and F; at the same time, the system integrates the attitude delay data calculated in S143, including a mechanical response delay of 200ms, an environmental disturbance delay of 150ms, and a control system delay of 55ms, for a total delay of 405ms, and analyzes whether the attitude delay will aggravate the impact of the damaged parts.

[0087] In the first attitude compensation calculation stage, the system calculates the required attitude compensation for each damaged part based on its damage severity, relative position, and attitude delay. This includes angle adjustment, force adjustment, and timing advance. Angle compensation uses the formula: Compensation Angle = Damage Influence Coefficient × Delay Amplification Coefficient. For example, if the damage influence coefficient for point C is 0.15 and the delay amplification coefficient is 1.2 (a 405ms delay amplifies the impact by 20%), then the compensation angle = 0.15 × 1.2 = 0.18 degrees. Force compensation uses the formula: Compensation Force = Base Force × (1 + Damage Influence Coefficient + Delay Influence Coefficient). For example, if the base force is 100N, the damage influence coefficient is 0.15, and the delay influence coefficient is 0.2, then the compensation force = 100 × (1 + 0.1). 5 + 0.2) = 135N; The time advance is calculated using the formula: advance time = total delay × safety factor. For example, if the total delay is 405ms and the safety factor is 1.1, then the advance time = 405 × 1.1 = 445.5ms; Finally, the system performs a weighted average of the compensation amounts for all damaged parts to obtain the first attitude compensation amount. For example, the angle compensation at point C is 0.18 degrees and the force compensation is 135N, the angle compensation at point E is 0.12 degrees and the force compensation is 120N, and the comprehensive compensation amount is an angle compensation of 0.15 degrees, a force compensation of 127.5N, and a time advance of 445.5ms. This comprehensive compensation mechanism based on historical damage data and real-time attitude delay can effectively improve the attitude control accuracy and landing safety of the autogyro in complex grassland environments.

[0088] Furthermore, in the landing control system of the autogyro, the second attitude compensation amount is determined based on the damaged data and attitude delay. The damaged data needs to be quantified and normalized. The system extracts data on damaged parts from S151, such as the bending amount of the main strut (point C) of 2.5mm and the performance impact of 15%, the loosening amount of the crossbeam (point E) of 0.8mm and the performance impact of 10%, and the wear amount of the left front wheel (point F) of 3mm and the performance impact of 8%. The system converts different types of damaged data into standard values ​​in the range of 0-1 through normalization formulas. For example, the normalized value of bending at point C is 2.5mm / 5mm=0.5, the loosening at point E is 0.8mm / 2mm=0.4, and the wear at point F is 3mm / 10mm=0.3.

[0089] Simultaneously, the system extracts attitude delay data from S143, including a mechanical response delay of 300ms, an environmental disturbance delay of 80ms, and a control system delay of 25ms, for a total delay of 405ms. It then assesses the impact of each delay on attitude control, determining the influence coefficients for mechanical delay (0.6), environmental delay (0.3), and control delay (0.1). Based on this data, the system employs a second compensation calculation model, using the formula Σ(normalized damage value × influence coefficient × delay coefficient) to calculate angle compensation and force compensation respectively. For angle compensation, point C is 0.5×0.8×0.6=0.24, point E is 0.4×0.6×0.3=0.072, and point F is 0.3×0.4×0.1=0.012, for a total angle compensation of 0.324 degrees. For force compensation, point C is 0.5×0.7×0.6=0.21, point E is 0.4×0.5×0.3=0.06, and point F is 0.3×0.3×0.1=0.009, for a total force compensation of 0.279 (normalized to 27.9%).

[0090] In the stage of determining the attitude compensation event based on the first and second compensation amounts and the mapping relationship, the system reviews the first compensation amount obtained from S151, including an angle compensation of 0.15 degrees, a force compensation of 132.5N, and a time advance of 445.5ms. It establishes the mapping relationship between the compensation amounts and the control actions, defining angle compensation as corresponding to the attitude adjustment angle (1 degree maps to 5 degrees of adjustment), force compensation as corresponding to the actuator output force (1N maps to 10N output), and time compensation as corresponding to the action execution time (1ms maps to 1ms advance). Through comprehensive compensation amount calculation, the system obtains a total angle compensation of 0.15 + 0.324 = 0.474 degrees, with a mapped output of 2.37 degrees; a total force compensation of 132.5 + (100 × 0.279) = 160.4N, with a mapped output of 1604N; and a total time compensation of 445.5 + (405 × 0.324) = 576.72ms, with a mapped output of 576.72ms advance.

[0091] The system generates an attitude compensation event (Event ID: EC-2024-001). The event type is attitude compensation, and the triggering condition is the detection of a damaged part and the existence of attitude delay. The execution parameters include target angle adjustment of 2.37 degrees, actuator output force of 1604N, and execution time advance of 576.72ms. This event is assigned a high safety level and is executed in priority over conventional attitude control to ensure that the autogyro can effectively cope with the challenges brought by damaged parts and attitude delay in complex grass environments and achieve a safe and stable landing.

[0092] Therefore, the attitude autonomous control events of the landing gear are determined based on the first and second levels of autonomous control content. This approach is compatible with the overall consideration of both levels of autonomous control content, ensuring the accuracy of the attitude autonomous control events of the landing gear. At the same time, an attitude delay component is introduced to further control the attitude delay component. This approach achieves an overall consideration of the current attitude of the landing gear, multiple attitude compensation actions, and the action efficiency of the landing gear, thereby improving the accuracy of the attitude autonomous control events of the landing gear.

[0093] At this time, during the autogyro's grass landing control process, the system determines multiple compensation action characteristics based on attitude compensation events. The attitude compensation event (such as EC-2024-001) includes a total compensation angle of 2.37 degrees, a total compensation force of 1604N, and a total compensation time of 576.72ms advance. From this, the system extracts angle compensation characteristics (main strut angle adjustment of 1.5 degrees, crossbeam horizontal adjustment of 0.87 degrees, and wheel axle alignment adjustment of 0.3 degrees), force compensation characteristics (main strut actuator 800N, crossbeam actuator 500N, and wheel axle actuator 304N), and time compensation characteristics (main strut advance of 300ms, crossbeam advance of 200ms, and wheel axle advance of 76.72ms). These action characteristics are divided into preparatory actions, main actions, and closing actions according to the execution order, and into core actions, auxiliary actions, and optimization actions according to their importance, laying the foundation for the subsequent control strategy formulation.

[0094] Based on the characteristics of the compensation actions, time points, and overall shape, the system further determines multiple attitude compensation actions; the time axis is divided into T0 (current time point), T1 (main support adjustment begins), T2 (crossbeam adjustment begins), T3 (wheel axle adjustment begins), and T4 (landing time point), and the start time, duration, and end time of each action are planned in detail; the overall shape analysis shows that the current angle of the main support is 85 degrees (target 87.5 degrees), the current tilt of the crossbeam is 1.2 degrees (target 0.33 degrees), and the current offset of the wheel axle is 0.5 degrees (target 0.2 degrees), while also considering the maximum adjustment speed. The system is subject to constraints such as a maximum force change of 1000 N / s and a structural safety limit not exceeding 90% of the design limit. It generates three main attitude compensation actions: A1 (main support adjustment) adjusts the angle from 85 degrees to 87.5 degrees within T1 to T1+250 ms, with a force output of 800 N; A2 (crossbeam adjustment) adjusts the tilt from 1.2 degrees to 0.33 degrees within T2 to T2+150 ms, with a force output of 500 N; and A3 (wheel and axle adjustment) adjusts the offset from 0.5 degrees to 0.2 degrees within T3 to T3+100 ms, with a force output of 304 N.

[0095] Based on the current attitude and compensation actions, the system determines the first level of autonomous control content. The current attitude assessment shows a pitch angle of 2.1 degrees (target 0 degrees), a roll angle of 1.5 degrees (target 0 degrees), and a yaw angle of 0.3 degrees (acceptable range). Based on this, the system generates a control strategy with the primary objective of eliminating pitch and roll deviations and the secondary objective of maintaining yaw stability. The control content is ordered by priority: pitch control (highest priority) executes action A1, roll control (medium priority) executes action A2, and yaw control (low priority) executes action A3. Each control parameter and corresponding compensation action are executed synchronously.

[0096] Based on the compensation actions and action efficiency, the system further determines the second level of autonomous control content. The efficiency evaluation indicators include response speed, execution accuracy, energy consumption, and structural load. The evaluation results show that A1 has a response speed of 95ms (excellent) and execution accuracy of 98% (excellent), A2 has a response speed of 120ms (good) and execution accuracy of 95% (good), and A3 has a response speed of 80ms (excellent) and execution accuracy of 92% (good). Based on this, the system generates the second level of control content with the goals of minimizing the total response time, maximizing execution accuracy, optimizing energy consumption, and controlling structural load. It adopts action sequence optimization (prioritizing the execution of high-efficiency action A3, then the execution of medium-efficiency action A2, and finally the execution of critical action A1) and parameter optimization (increasing the force of A2 by 10% to 550N and advancing the time of A3 by 20ms to 80ms). At the same time, it adds intermediate state checkpoints for A1, optimizes the force curve of A2, and simplifies the control logic of A3.

[0097] The system determines attitude autonomous control events based on dual autonomous control content. The first level of control is safety-oriented, ensuring the accuracy and safety of attitude correction. The second level of control is efficiency-oriented, optimizing the efficiency and resource utilization of the control process. The integration principle follows the principles of safety priority, performance balance, and real-time adjustment. The generated attitude autonomous control event (event number: AC-2024-001) is of the attitude autonomous control type, triggered by a grass landing scenario, and has the highest execution priority. The control parameters include a total control time of 576.72ms, involved components such as main support / crossbeam / wheel axle, and the control target of eliminating attitude deviation.

[0098] Please see Figure 7 The attitude control system of the landing gear during the descent of the autogyro includes: Ground feature module 21 is used to determine multiple ground features based on the recognition of ground images of the descent area of ​​the autogyro when it is in the descent state. The descent position module 22 is used to determine the ground stereo model based on the feature positions and corresponding feature shapes of multiple ground features and the regional shape of the descent area, and to mark the descent position of the autogyro on the ground stereo model; The attitude control event module 23 is used to determine the descent mode of the gyrocopter based on the descent position of the gyrocopter and multiple descent data of the gyrocopter, determine the dynamic descent action of the landing gear based on the real-time image of the landing gear and the descent mode of the gyrocopter, and determine the attitude control event based on the dynamic descent action of the landing gear, multiple attitude parameters of the landing gear and the corresponding overall shape. The attitude delay module 24 is used to determine multiple sub-attitude adjustment events based on the identification of the attitude control event, and to determine the attitude delay part of the landing gear according to the multiple sub-attitude adjustment events, the position corresponding to the ground three-dimensional model and the real-time shape of the landing gear. The attitude autonomous control event module 25 is used to determine attitude compensation events based on multiple damaged parts of the landing gear and attitude delays of the landing gear, determine multiple attitude compensation actions based on the identification of attitude compensation events, and determine the attitude autonomous control event of the landing gear based on the current attitude of the landing gear, multiple attitude compensation actions and the action efficiency of the landing gear.

[0099] The technical features of the above embodiments can be combined arbitrarily. For the sake of brevity, not all combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

Claims

1. A method for attitude control of a landing gear during the descent of an autogyro, characterized in that, include: When the autogyro is descending, multiple ground features are determined based on the identification of ground images of the autogyro's descent area; The ground 3D model is determined based on the feature location and corresponding feature shape of multiple ground features and the regional shape of the descent area, and the descent position of the autogyro is marked on the ground 3D model; The descent mode of the gyrocopter is determined based on its descent position and multiple descent data. The dynamic descent action of the landing gear is determined based on the real-time image of the landing gear and the descent mode of the gyrocopter. The attitude control event is determined based on the dynamic descent action of the landing gear, multiple attitude parameters of the landing gear, and the corresponding overall shape. Based on the identification of the attitude control event, multiple sub-attitude adjustment events are determined, and the attitude delay of the landing gear is determined according to the multiple sub-attitude adjustment events, the corresponding position of the ground stereo model, and the real-time shape of the landing gear. Attitude compensation events are determined based on multiple damaged parts of the landing gear and the attitude delay of the landing gear. Multiple attitude compensation actions are determined based on the identification of attitude compensation events. Attitude autonomous control events of the landing gear are determined based on the current attitude of the landing gear, multiple attitude compensation actions and the action efficiency of the landing gear.

2. The attitude control method for the landing gear during the descent of an autogyro, as described in claim 1, is characterized in that... When the autogyro is descending, multiple ground features are determined based on the identification of ground images of the autogyro's descent area, including: The descent process of the autogyro is monitored in real time, and the descent state of the autogyro is determined based on multiple descent data and corresponding descent patterns; at the same time, the descent area of ​​the autogyro is collected. The ground image of the descent region is determined by the camera of the autogyro capturing the descent region. Multiple sub-ground regions are determined by the recognition of the ground image of the descent region. In the multiple sub-ground regions, the corresponding sub-ground features are determined by the detection of each sub-ground region. Multiple ground features are determined by the feature morphology of multiple sub-ground features, the regional morphology of the corresponding sub-ground regions, and the transition region between two adjacent sub-ground features.

3. The attitude control method for the landing gear during the descent of an autogyro, as described in claim 1, is characterized in that... The process of determining a ground 3D model based on the feature locations and corresponding feature shapes of multiple ground features and the regional shape of the descent area, and marking the descent position of the autogyro on the ground 3D model, includes: Multiple ground features are collected, and the feature location of each ground feature is determined based on the recognition of each ground feature. The corresponding feature shape is determined based on the multi-directional recognition of the feature location of the ground feature. The first layer of stereo model is determined based on the feature location of multiple ground features and the regional morphology of the descending area; the second layer of stereo model is determined based on the feature morphology of multiple ground features and the regional morphology of the descending area. The ground 3D model is determined based on the first and second 3D models; the preset descent area is determined by traversing the ground 3D model; the descent position of the gyrocopter is determined based on the preset descent area, the flight direction of the gyrocopter, and the descent speed of the gyrocopter, so as to mark the descent position of the gyrocopter on the ground 3D model.

4. The attitude control method for the landing gear during the descent of an autogyro, as described in claim 1, is characterized in that... The process involves determining the gyrocopter's descent mode based on its descent position and multiple descent data points, determining the landing gear's dynamic descent action based on real-time images of the landing gear and the gyrocopter's descent mode, and determining attitude control events based on the landing gear's dynamic descent action, multiple attitude parameters of the landing gear, and its corresponding overall shape. This includes: The system monitors the descent process of the autogyro in real time and collects multiple descent data points. Based on these data points and the autogyro's descent position, it determines multiple descent combinations. Based on the identification of these combinations, it identifies the corresponding descent maneuvers and determines the descent mode of the autogyro based on these maneuvers, the autogyro's altitude, and a ground 3D model.

5. The attitude control method for the landing gear during the descent of an autogyro, as described in claim 4, is characterized in that... The process of determining the descent mode of the autogyro based on its descent position and multiple descent data points, determining the dynamic descent action of the landing gear based on real-time images of the landing gear and the descent mode, and determining attitude control events based on the dynamic descent action of the landing gear, multiple attitude parameters of the landing gear, and the corresponding overall shape, further includes: Real-time images of the landing gear are acquired, and the real-time shape and trajectory of the landing gear are determined based on the recognition of the real-time images. The dynamic descent action of the landing gear is determined based on the real-time shape and trajectory of the landing gear and the descent mode of the autogyro. Multiple attitude parameters of the landing gear are collected. A first attitude control coefficient is determined based on the multiple attitude parameters of the landing gear and the dynamic descent action of the landing gear. A second attitude control coefficient is determined based on the multiple attitude parameters of the landing gear and the overall shape of the landing gear. An attitude control event is determined based on the mapping relationship between the first attitude control coefficient, the second attitude control coefficient and the attitude control event.

6. The attitude control method for the landing gear during the descent of an autogyro, as described in claim 1, is characterized in that... The process of identifying multiple sub-attitude adjustment events based on the attitude control event, and determining the attitude delay portion of the landing gear based on the multiple sub-attitude adjustment events, the corresponding position of the ground 3D model, and the real-time shape of the landing gear, includes: Based on the identification of the attitude control event, multiple attitude control items are determined. Based on the multiple attitude control items and the real-time shape of the landing gear, multiple sub-attitude adjustment events are determined. In the sub-attitude adjustment events, the attitude adjustment content of the landing gear at the corresponding time node is presented.

7. The attitude control method for the landing gear during the descent of an autogyro, as described in claim 6, is characterized in that... The process of identifying multiple sub-attitude adjustment events based on the recognition of the attitude control event, and determining the attitude delay portion of the landing gear according to the multiple sub-attitude adjustment events, the corresponding position of the ground 3D model, and the real-time shape of the landing gear, further includes: Collect a ground stereo model and determine the corresponding attitude influencing factors based on multiple sub-attitude adjustment events and the ground stereo model; The corresponding attitude delay factors are determined based on multiple sub-attitude adjustment events and the real-time shape of the landing gear. The attitude delay portion of the landing gear is determined based on the attitude influencing factors, attitude delay factors, and the overall shape of the landing gear.

8. The attitude control method for the landing gear during the descent of an autogyro, as described in claim 1, is characterized in that... The process involves determining attitude compensation events based on multiple damaged parts of the landing gear and the attitude delay portion of the landing gear, determining multiple attitude compensation actions based on the identification of these events, and determining autonomous attitude control events for the landing gear based on the current attitude of the landing gear, the multiple attitude compensation actions, and the landing gear's action efficiency. This includes: Collect past damage events of the landing gear, identify multiple damaged parts of the landing gear based on the identification of past damage events, and mark the damage data of each damaged part; determine the first attitude compensation amount based on the relative position of the multiple damaged parts of the landing gear and the attitude delay of the landing gear. The second attitude compensation amount is determined based on the damage data of multiple damaged parts of the landing gear and the attitude delay of the landing gear; the attitude compensation event is determined based on the first attitude compensation amount, the second attitude compensation amount and the attitude compensation mapping relationship.

9. The attitude control method for the landing gear during the descent of an autogyro, as described in claim 8, is characterized in that... The process of determining attitude compensation events based on multiple damaged parts of the landing gear and the attitude delay of the landing gear, determining multiple attitude compensation actions based on the identification of attitude compensation events, and determining autonomous attitude control events of the landing gear based on the current attitude of the landing gear, multiple attitude compensation actions, and the action efficiency of the landing gear, further includes: Multiple compensation action features are determined based on the identification of attitude compensation events. Multiple attitude compensation actions are determined based on the multiple compensation action features, the corresponding compensation time nodes, and the overall shape of the landing gear. The first level of autonomous control content is determined based on the current attitude of the landing gear and the multiple attitude compensation actions. The second level of autonomous control content is determined based on the multiple attitude compensation actions and the action efficiency of the landing gear. The attitude autonomous control event of the landing gear is determined based on the first level of autonomous control content and the second level of autonomous control content.

10. An attitude control system for a landing gear during the descent of an autogyro, characterized in that, The attitude control system of the landing gear during the descent of the autogyro is applied to the attitude control method of the landing gear during the descent of the autogyro as described in any one of claims 1-9, wherein the attitude control system of the landing gear during the descent of the autogyro includes: The ground feature module is used to determine multiple ground features based on the recognition of ground images of the descent area of ​​the autogyro when it is in the descent state. The descent position module is used to determine the ground 3D model based on the feature positions and corresponding feature shapes of multiple ground features and the regional shape of the descent area, and to mark the descent position of the autogyro on the ground 3D model; The attitude control event module is used to determine the descent mode of the gyrocopter based on its descent position and multiple descent data, determine the dynamic descent action of the landing gear based on the real-time image of the landing gear and the descent mode of the gyrocopter, and determine the attitude control event based on the dynamic descent action of the landing gear, multiple attitude parameters of the landing gear and the corresponding overall shape. The attitude delay module is used to identify multiple sub-attitude adjustment events based on the recognition of the attitude control event, and to determine the attitude delay part of the landing gear according to the multiple sub-attitude adjustment events, the position corresponding to the ground 3D model, and the real-time shape of the landing gear. The attitude autonomous control event module is used to determine attitude compensation events based on multiple damaged parts of the landing gear and the attitude delay of the landing gear, determine multiple attitude compensation actions based on the identification of attitude compensation events, and determine the attitude autonomous control event of the landing gear based on the current attitude of the landing gear, multiple attitude compensation actions and the action efficiency of the landing gear.