Control method, device and storage medium for automatic landing of fixed-wing unmanned aerial vehicle

Abstract

The present invention provides a control method, device and storage medium for automatic landing of fixed-wing unmanned aerial vehicles, and relates to the technical field of automatic landing of fixed-wing unmanned aerial vehicles. The method includes real-time collection of landing area images, flight status data and landing failure data; analyzing and processing image data through target detection and semantic segmentation models, and obtaining an optimal control strategy in combination with flight status data; constructing a covariance matrix adaptive evolutionary strategy model based on re-parameterization method optimization to generate and optimize the landing path; optimizing the model using landing failure data to improve the accuracy of the landing path, thereby effectively improving the safety and reliability of automatic landing of unmanned aerial vehicles, and enhancing its ability to cope with complex landing environments. The present invention adjusts the landing path in real time through the optimal control strategy to obtain the optimal landing path, greatly reducing mislanding and other potential safety hazards, and improving the overall reliability and safety of automatic landing of unmanned aerial vehicles.

Description

Technical Field

[0001] The present invention relates to the technical field of automatic landing of fixed-wing unmanned aerial vehicles, and in particular to a control method, device and storage medium for automatic landing of fixed-wing unmanned aerial vehicles. Background Art

[0002] The origin of fixed-wing drones can be traced back to the First and Second World Wars, when they were originally used for reconnaissance and training of military targets. Early fixed-wing drones usually used simple remote control technology, with low operating accuracy and mission requirements. Drones like the "Radioplane OQ-2" are early representatives, mainly used for target training. With the advancement of technology, fixed-wing drones have gradually been used in scientific research, agricultural monitoring, environmental protection and other fields. At this time, the flight control system, power system and sensor technology of drones have been significantly improved, enabling automated flight and data collection. This has enabled drones to play an increasingly important role in geographic mapping, forest monitoring and atmospheric detection. With the continuous maturity of GPS and drone control technology, fixed-wing drones have begun to be widely used in civilian fields such as mapping, agriculture, search and rescue. Modern fixed-wing drones not only have basic flight capabilities, but also introduce advanced control algorithms (such as PID, fuzzy control), machine learning algorithms (for image recognition, obstacle avoidance, etc.) and integrate efficient sensors (such as high-definition cameras, multispectral sensors, etc.).

[0003] During the landing process, fixed-wing drones may fail to identify and avoid obstacles in time due to inaccurate obstacle recognition or insufficient environmental perception, and may collide with obstacles on the ground (such as buildings, trees, vehicles, etc.). In strong winds or sudden changes in airflow, the drone may lose control during landing or even overturn. It is also possible that the pilot or automatic control system fails to correctly judge the landing speed, causing the drone to slow down when approaching the ground, resulting in a stall and crash.

[0004] Therefore, how to solve the defects of incomplete environmental perception and path planning methods and lack of sufficient adaptive characteristics during the landing of fixed-wing UAVs still requires further improvement. Summary of the invention

[0005] The present invention provides a control method, device and storage medium for automatic landing of a fixed-wing unmanned aerial vehicle, which are used to solve the defects of the fixed-wing unmanned aerial vehicle in the prior art that the perception of the environment is not comprehensive during the landing process and the path planning method lacks sufficient adaptive characteristics.

[0006] In one aspect, the present invention provides a control method for automatic landing of a fixed-wing unmanned aerial vehicle, comprising:

[0007] S1: Collect landing area image data and flight status data in real time, and obtain landing failure data.

[0008] S2: Analyze the landing area image data through the target detection model to obtain obstacle data, refine the obstacle data according to the semantic segmentation model to obtain the data of factors affecting landing, adjust the control parameters according to the flight status data, and obtain the optimal control strategy in combination with the data of factors affecting landing.

[0009] S3: Construct a covariance matrix adaptive evolution strategy model based on the reparameterization method optimization, input the data of factors affecting landing and flight status data, obtain the landing path, and adjust the landing path in real time according to the optimal control strategy to obtain the optimal landing path.

[0010] The step of obtaining the landing path comprises:

[0011] S301: Preprocessing the flight status data and the landing factor data, and defining landing time, distance from obstacles, and avoiding obstacles as optimization goals;

[0012] S302: By defining the key point positions and control inputs of the flight path, initializing the sample individuals, setting the covariance matrix and the mean vector, and constructing a covariance matrix adaptation evolution strategy model;

[0013] S303: Generate a new candidate individual according to the current mean vector and the covariance matrix using a reparameterization algorithm, the formula is:

[0014]

[0015] In the formula, is the newly generated individual path parameter, is the step length, is a random number sampled from a standard normal distribution, is the mean vector;

[0016] S304: Calculating the fitness value of each candidate individual according to the flight status data and the landing influencing factor data to obtain a fitness value;

[0017] S305: According to the fitness value, the mean vector, the covariance matrix and the step size are updated and iterated, and the formula is expressed as follows:

[0018]

[0019]

[0020]

[0021] In the formula, It is the degree to which the drone deviates from the planned path during landing. is the success rate of fitness, is the updated mean vector, is the updated covariance matrix, is the updated step size, , is a constant coefficient, is the number of flight data samples, It is the dimensional parameter related to the movement of the drone in three-dimensional space. yes Power operations of ;

[0022] S306: After the update iteration reaches a preset number of times, multiple mean paths are obtained, and the best mean path is selected as the landing path.

[0023] S4: The covariance matrix adaptation evolution strategy model is optimized through landing failure data to improve the accuracy of the landing path.

[0024] According to the control method for automatic landing of fixed-wing unmanned aerial vehicles provided by the present invention, in step S2, the step of obtaining data on factors affecting landing includes: collecting image sequences, environmental data of different places and image data sets annotated with detailed semantic information in the landing area during the landing of the unmanned aerial vehicle, preprocessing the image sequences, and obtaining image sequence data. The target detection model is trained according to the environmental data, and the weight of the target detection model is optimized by the back propagation algorithm, and the image sequence data is input into the target detection model to obtain obstacle data. The semantic segmentation model is trained using the image data set to learn the semantic features of different objects and areas. The image blocks corresponding to the obstacle data are input into the semantic segmentation model, and the pixels of the image blocks are subdivided to obtain obstacle property data, and the landing area image data is semantically segmented to obtain data on factors affecting landing.

[0025] According to the control method for automatic landing of fixed-wing UAV provided by the present invention, in step S2, the step of obtaining the optimal control strategy includes: extracting key parameters from the flight status data to obtain key flight status data, and performing height division to obtain the flight status area. Factor evaluation is performed on the data of factors affecting landing to obtain the obstacle shape, runway status and meteorological conditions. According to the key flight status data and the flight status area, the height, airspeed and attitude of the UAV are adjusted to obtain control parameters, and the control parameters are distributed according to preset weights based on the obstacle shape, runway status and meteorological conditions to obtain the optimal control strategy.

[0026] According to the control method for automatic landing of a fixed-wing UAV provided by the present invention, in step S3, the step of obtaining the optimal landing path includes:

[0027] S311: Obtain situational awareness during the landing process of the UAV, obtain UAV status data, and calculate an error vector based on the difference between the landing path and the current path.

[0028] S312: Use the controller to dynamically adjust the landing path to obtain an updated strategy, and calculate the control instruction according to the error vector and the updated strategy.

[0029] S313: Based on the error vector and the control instruction, the landing path is updated in real time to obtain the optimal landing path.

[0030] According to the control method for automatic landing of a fixed-wing UAV provided by the present invention, in step S312, the formula of the control instruction is expressed as:

[0031]

[0032] In the formula, is a control instruction. is the proportional gain, is the error vector, is the integral gain, yes The integral term of is the differential gain, is the rate of change of the error vector.

[0033] On the other hand, the present invention also provides a control device for automatic landing of a fixed-wing unmanned aerial vehicle, the control device comprising: a memory and a processor, the memory storing a computer program, and the processor executing the computer program to implement the above-mentioned control method for automatic landing of a fixed-wing unmanned aerial vehicle.

[0034] On the other hand, a readable storage medium stores one or more programs, which enable a machine to execute the above-mentioned control method for automatic landing of a fixed-wing unmanned aerial vehicle.

[0035] The control method for automatic landing of fixed-wing UAV provided by the present invention can not only identify the location of obstacles, but also classify and understand various elements in the environment in detail by combining target detection and semantic segmentation models. It can also find a safer and more efficient landing path by introducing a covariance matrix adaptive evolutionary strategy model optimized based on a reparameterization method and utilizing its global search capability and adaptive characteristics. The optimal landing path is obtained by adjusting the landing path in real time through the optimal control strategy, which greatly reduces the safety hazards of stalling and collision and improves the overall reliability and safety of automatic landing of UAV. BRIEF DESCRIPTION OF THE DRAWINGS

[0036] In order to more clearly illustrate the technical solutions in the present invention or the prior art, a brief introduction will be given below to the drawings required for use in the embodiments or the description of the prior art. Obviously, the drawings described below are some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative work.

[0037] Figure 1 This is one of the flow charts of a control method for automatic landing of a fixed-wing UAV provided in an embodiment of the present invention;

[0038] Figure 2 This is the second flow chart of the control method for automatic landing of a fixed-wing UAV provided in an embodiment of the present invention. DETAILED DESCRIPTION

[0039] In order to make the purpose, technical solution and advantages of the present invention clearer, the technical solution of the present invention will be clearly and completely described below in conjunction with the drawings of the present invention. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0040] Combine the following Figure 1-Figure 2 The present invention describes a control method, device and storage medium for automatic landing of a fixed-wing unmanned aerial vehicle.

[0041] Figure 1 This is one of the flow charts of the control method for automatic landing of a fixed-wing UAV provided in an embodiment of the present invention.

[0042] like Figure 1 As shown, the control method for automatic landing of a fixed-wing UAV provided by an embodiment of the present invention mainly includes the following steps:

[0043] S1: Real-time collection of landing area image data and flight status data, and acquisition of landing failure data through software simulation. Through the high-resolution optical camera, infrared thermal imager, laser radar and other advanced sensor equipment installed on the drone, the image data of the landing area is collected in real time in an all-round and high-precision manner. These image data cover the terrain and geomorphic features of the landing area, the distribution of obstacles, ground markings and other rich information. At the same time, the built-in high-precision gyroscope, accelerometer, global positioning system (GPS) and barometric altimeter and other flight status monitoring equipment of the drone are used to continuously and accurately obtain the flight status data of the drone, including the current position coordinates, flight speed, acceleration magnitude and direction, attitude angle (such as pitch angle, roll angle, yaw angle) and altitude information. In addition, the surrounding environmental parameters are monitored with the help of various environmental sensors on the drone (such as anemometer, thermometer, hygrometer, etc.), and these data are integrated and analyzed with the detailed flight log information recorded by the flight control system to accurately obtain the landing failure data. These data include specific situations of landing failures caused by various reasons such as unreasonable path planning, insufficient consideration of environmental factors, and deviation in the execution of control instructions during previous landing processes. For example, the drone's out-of-control trajectory during landing, the instantaneous state before colliding with obstacles, and the moment when the landing was interrupted due to severe weather conditions are key information. These provide a valuable reference basis for the subsequent optimization of the landing process, thereby effectively improving the safety and success rate of automatic landing of drones and enabling them to better adapt to complex and changeable actual landing scenarios.

[0044] S2: Analyze the landing area image data through the target detection model to obtain obstacle data, refine the obstacle data according to the semantic segmentation model to obtain the data of factors affecting landing, adjust the control parameters according to the flight status data, and obtain the optimal control strategy in combination with the data of factors affecting landing.

[0045] In step S2, the step of obtaining data on factors affecting landing includes:

[0046] During the landing process of the UAV, image sequences in the landing area, environmental data in different places, and image datasets annotated with detailed semantic information are collected, and the image sequences are preprocessed to obtain image sequence data.

[0047] The target detection model is trained according to the environmental data, and the weight of the target detection model is optimized through the back propagation algorithm. The image sequence data is input into the target detection model to obtain obstacle data, which includes buildings, trees, vehicles, pedestrians, animals, etc.

[0048] The semantic segmentation model is trained using image datasets to learn the semantic features of different objects and regions.

[0049] The image blocks corresponding to the obstacle data are input into the semantic segmentation model, and the pixels of the image blocks are subdivided to obtain the obstacle property data. At the same time, the landing area image data is semantically segmented to obtain the data of factors affecting landing, including runway, grass, wind speed, floating objects, etc.

[0050] The steps to obtain the optimal control strategy include:

[0051] Key parameters are extracted from the flight status data to obtain key flight status data, and flight status areas are obtained by performing altitude division.

[0052] The factors affecting landing are evaluated to obtain obstacle shape, runway status and meteorological conditions.

[0053] According to the key flight status data and flight status area, the altitude, airspeed and attitude of the UAV are adjusted to obtain the control parameters. The control parameters are allocated according to the preset weights based on the obstacle shape, runway status and weather conditions to obtain the optimal control strategy.

[0054] S3: Construct a covariance matrix adaptive evolution strategy model based on the reparameterization method optimization, input the data of factors affecting landing and flight status data, obtain the landing path, and adjust the landing path in real time according to the optimal control strategy to obtain the optimal landing path.

[0055] Figure 2 This is the second flow chart of the control method for automatic landing of a fixed-wing UAV provided in an embodiment of the present invention.

[0056] like Figure 2 As shown, in step S3, the step of obtaining the landing path includes:

[0057] S301: Preprocess the flight status data and the data of factors affecting landing, and define the landing time, distance from obstacles and avoiding obstacles as optimization goals. The preprocessing process involves data cleaning, removing noise interference and outliers; data normalization, unifying data of different dimensions into the same numerical range for efficient processing of subsequent models; and feature engineering construction, extracting key features in the data that have a significant impact on landing path planning. At the same time, the landing time (pursuing the completion of the landing operation within a safe and efficient time range, preventing the position and attitude of the fixed-wing UAV from being adjusted in time before landing, thereby reducing energy consumption and risks), the distance from obstacles (ensuring that the UAV always maintains a sufficient safe distance from various obstacles, preventing collision accidents, and ensuring the safety of the landing process) and avoiding obstacles (fundamentally reducing the probability of landing failure caused by obstacles, and cleverly guiding the UAV to bypass potential dangerous areas through reasonable path planning and intelligent decision-making) are clearly defined as optimization goals, thereby laying a solid foundation for the subsequent construction of an accurate and efficient landing path planning model, effectively improving the reliability and stability of the automatic landing of the UAV, and adapting to the complex and changeable actual landing scene requirements to the greatest extent.

[0058] S302: By defining the key point positions and control inputs of the flight path, initializing the sample individuals, setting the covariance matrix and mean vector, a covariance matrix adaptation evolution strategy model is constructed, and the formula is expressed as:

[0059]

[0060] In the formula, is the covariance matrix, is the mean vector, is a constant, is the identity matrix, It is the key point position of the flight status area.

[0061] S303: Use the reparameterization algorithm to generate new candidate individuals according to the current mean vector and covariance matrix. The formula is:

[0062]

[0063] In the formula, is the newly generated individual path parameter, is the step length, is a random number sampled from a standard normal distribution, is the mean vector.

[0064] S304: Calculate the fitness of each candidate individual according to the flight status data and the data of factors affecting landing to obtain a fitness value, which is expressed as:

[0065]

[0066] In the formula, is the fitness value, is the distance assessment to the obstacle, is the landing time assessment, It is to avoid encountering obstacles assessment.

[0067] S305: According to the fitness value, update the iterative mean vector, covariance matrix and step size, the formula is expressed as:

[0068]

[0069]

[0070]

[0071] In the formula, It is the degree to which the drone deviates from the planned path during landing. is the success rate of fitness, is the updated mean vector, is the updated covariance matrix, is the updated step size, , is a constant coefficient, is the number of flight data samples, It is the dimensional parameter related to the movement of the drone in three-dimensional space. yes Power operation.

[0072] S306: After the update iteration reaches a preset number of times, multiple mean paths are obtained, and the best mean path is selected as the landing path.

[0073] The steps to obtain the optimal landing path include:

[0074] S311: Obtain situational awareness during the landing process of the UAV, obtain UAV status data, and calculate an error vector based on the difference between the landing path and the current path.

[0075] S312: Use the controller to dynamically adjust the landing path to obtain an updated strategy, and calculate the control instruction according to the error vector and the updated strategy. The formula of the control instruction is expressed as:

[0076]

[0077] In the formula, is a control instruction. is the proportional gain, is the error vector, is the integral gain, yes The integral term of is the differential gain, is the rate of change of the error vector. In this embodiment, a fixed-wing UAV with three-dimensional underactuated characteristics is selected as the controlled object, and a track angle controller, a heading angle controller and an airspeed thrust controller are designed to control the landing of the fixed-wing UAV. The mathematical model formula of the fixed-wing UAV is expressed as:

[0078]

[0079] In the formula, is the airspeed of the drone (scalar speed), is the heading angle (the angle between the nose direction and the north direction in the inertial coordinate system), is the flight path angle (the angle between the velocity vector and the horizontal angle, Indicates climbing indicates a decrease), , is the angular velocity of the track angle, is the angular velocity of the heading angle, is the gravity coefficient, are the linear speeds of the drone in the x-axis, y-axis, and z-axis directions, is the control quantity of the track angle controller used for the UAV attitude, It is the control quantity of the heading angle controller used for the UAV attitude.

[0080] The controller design for the track angle, heading angle and airspeed of the fixed-wing UAV is expressed as follows:

[0081]

[0082]

[0083]

[0084] In the formula, is the control quantity of the thrust controller that controls the airspeed, is the proportional gain of the track angle controller, is the heading angle controller proportional gain, is the speed controller proportional gain, is the integral gain of the track angle controller, is the heading angle controller integral gain, is the speed controller integral gain, is the differential gain of the track angle controller, is the heading angle controller differential gain, is the speed controller differential gain, is the track angle error, is the heading angle error, is the speed error, , , are the rates of change of track angle error, heading angle error and velocity error respectively.

[0085] S313: Based on the error vector and the control command, the landing path is updated in real time to obtain the optimal landing path. The optimal landing path includes the expected position, heading angle, track angle and airspeed information. At the same time, the track angle error is obtained based on the expected track angle in the optimal landing path. The formula is expressed as:

[0086]

[0087] In the formula, is the desired track angle, is the track angle error.

[0088] The heading angle error is obtained based on the expected heading angle of the optimal landing path. The formula is expressed as:

[0089]

[0090] In the formula, is the desired heading angle, is the heading angle error.

[0091] The airspeed error is obtained based on the expected airspeed in the optimal landing path, and the formula is expressed as:

[0092]

[0093] In the formula, is the desired airspeed, is the airspeed error.

[0094] S4: The covariance matrix adaptive evolutionary strategy model is optimized through landing failure data to improve the accuracy of the landing path. In the process of automatic landing of fixed-wing UAVs, landing failure data contains key improvement information. Through in-depth mining and analysis of these landing failure data, the characteristic information about flight trajectory deviation, the degree of influence of environmental factors, and poor control parameter effects is accurately extracted. These processed landing failure data are fed back to the covariance matrix adaptive evolutionary strategy model, so that it can accurately identify the shortcomings and defects of the current model in landing path planning, and then the key parameters, search strategies, fitness functions and other core elements of the model are optimized and adjusted in a targeted manner, so as to continuously iterate and upgrade the model, significantly improve the accuracy of the landing path, enhance the safety and reliability of automatic landing of UAVs, and effectively ensure that UAVs can stably and accurately complete landing operations in complex and changeable landing environments, and effectively reduce various risks and potential losses caused by inaccurate landing paths.

[0095] The control method for automatic landing of fixed-wing UAV provided by the present invention can not only identify the location of obstacles, but also classify and understand various elements in the environment in detail by combining target detection and semantic segmentation models. It can also find a safer and more efficient landing path by introducing a covariance matrix adaptive evolutionary strategy model optimized based on a reparameterization method and utilizing its global search capability and adaptive characteristics. The optimal landing path is obtained by adjusting the landing path in real time through the optimal control strategy, which greatly reduces mislanding and other potential safety hazards and improves the overall reliability and safety of automatic landing of UAV.

[0096] On the other hand, the present invention also provides a control device for automatic landing of a fixed-wing unmanned aerial vehicle, including: a data collection module: collects landing area image data and flight status data in real time, and obtains landing failure data through software simulation. Data analysis and strategy generation module: analyzes the landing area image data through a target detection model to obtain obstacle data, refines the obstacle data according to a semantic segmentation model to obtain data of factors affecting landing, adjusts control parameters according to flight status data, and obtains an optimal control strategy in combination with the data of factors affecting landing. Path planning module: constructs a covariance matrix adaptive evolutionary strategy model optimized based on a reparameterization method, inputs data of factors affecting landing and flight status data, obtains a landing path, and adjusts the landing path in real time according to the optimal control strategy to obtain an optimal landing path. Model optimization module: optimizes the covariance matrix adaptive evolutionary strategy model through landing failure data to improve the accuracy of the landing path.

[0097] On the other hand, the present invention also provides a control device for automatic landing of a fixed-wing unmanned aerial vehicle, the control device comprising: a memory and a processor, the memory storing a computer program, and the processor executing the computer program to implement the above-mentioned control method for automatic landing of a fixed-wing unmanned aerial vehicle.

[0098] On the other hand, a readable storage medium stores one or more programs, which enable a machine to execute the above-mentioned control method for automatic landing of a fixed-wing unmanned aerial vehicle.

[0099] The device embodiments described above are merely illustrative, wherein the units described as separate components may or may not be physically separated, and the components displayed as units may or may not be physical units, that is, they may be located in one place, or they may be distributed on multiple network units. Some or all of the modules may be selected according to actual needs to achieve the purpose of the scheme of this embodiment. Ordinary technicians in this field can understand and implement it without paying creative labor.

[0100] Through the description of the above implementation methods, those skilled in the art can clearly understand that each implementation method can be implemented by means of software plus a necessary general hardware platform, and of course, can also be implemented by hardware. Based on this understanding, the above technical solution is essentially or the part that contributes to the prior art can be embodied in the form of a software product, and the computer software product can be stored in a computer-readable storage medium, such as ROM / RAM, a disk, an optical disk, etc., including a number of instructions for a computer device (which can be a personal computer, a server, or a network device, etc.) to execute the methods described in each embodiment or some parts of the embodiments.

[0101] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit it. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or make equivalent replacements for some of the technical features therein. However, these modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A control method for automatic landing of a fixed-wing UAV, characterized in that: include: S1: Collect landing area image data and flight status data in real time, and obtain landing failure data; S2: Analyze the landing area image data through the target detection model to obtain obstacle data, refine the obstacle data according to the semantic segmentation model to obtain landing factor data, adjust the control parameters according to the flight status data, and obtain the optimal control strategy in combination with the landing factor data; S3: constructing a covariance matrix adaptive evolution strategy model based on the optimization of the reparameterization method, inputting the data of the factors affecting landing and the flight status data, obtaining a landing path, and adjusting the landing path in real time according to the optimal control strategy to obtain an optimal landing path; The step of obtaining the landing path comprises: S301: Preprocessing the flight status data and the landing factor data, and defining landing time, distance from obstacles, and avoiding obstacles as optimization goals; S302: By defining the key point positions and control inputs of the flight path, initializing the sample individuals, setting the covariance matrix and the mean vector, and constructing a covariance matrix adaptation evolution strategy model; S303: Generate a new candidate individual according to the current mean vector and the covariance matrix using a reparameterization algorithm, the formula is: In the formula, is the newly generated individual path parameter, is the step length, is a random number sampled from a standard normal distribution, is the mean vector; S304: Calculating the fitness value of each candidate individual according to the flight status data and the landing influencing factor data to obtain a fitness value; S305: According to the fitness value, the mean vector, the covariance matrix and the step size are updated and iterated, and the formula is expressed as follows: In the formula, It is the degree to which the drone deviates from the planned path during landing. is the success rate of fitness, is the updated mean vector, is the updated covariance matrix, is the updated step size, , is a constant coefficient, is the number of flight data samples, It is the dimensional parameter related to the movement of the drone in three-dimensional space. yes Power operations of ; S306: After the update iteration reaches a preset number of times, multiple mean paths are obtained, and the best mean path is selected as the landing path; S4: Optimizing the covariance matrix adaptive evolution strategy model through the landing failure data to improve the accuracy of the landing path.

2. The control method for automatic landing of a fixed-wing UAV according to claim 1, characterized in that: In step S2, the step of obtaining the data of factors affecting landing includes: Collecting image sequences, environmental data of different places and image data sets annotated with detailed semantic information in the landing area during the landing process of the UAV, and preprocessing the image sequences to obtain image sequence data; The target detection model is trained according to the environmental data, and the weight of the target detection model is optimized by a back propagation algorithm, and the image sequence data is input into the target detection model to obtain obstacle data; Using the image dataset to train the semantic segmentation model to learn semantic features of different objects and regions; The image block corresponding to the obstacle data is input into the semantic segmentation model, the pixels of the image block are subdivided to obtain obstacle property data, and the landing area image data is semantically segmented to obtain the landing factor data.

3. The control method for automatic landing of a fixed-wing UAV according to claim 1, characterized in that: In step S2, the step of obtaining the optimal control strategy includes: Extracting key parameters from the flight status data to obtain key flight status data, and performing altitude division to obtain flight status areas; Performing factor evaluation on the landing influencing factor data to obtain obstacle shape, runway status and meteorological conditions; The control parameters are obtained by adjusting the altitude, airspeed and attitude of the UAV according to the key flight status data and the flight status area, and the control parameters are allocated according to preset weights based on the obstacle shape, the runway status and the meteorological conditions to obtain the optimal control strategy.

4. The control method for automatic landing of a fixed-wing UAV according to claim 1, characterized in that: In step S3, the step of obtaining the optimal landing path includes: S311: Acquire situational awareness during the landing process of the UAV, obtain UAV status data, and calculate an error vector based on the difference between the landing path and the current path; S312: using a controller to dynamically adjust the landing path to obtain an update strategy, and calculating a control instruction according to the error vector and the update strategy; S313: Based on the error vector and the control instruction, the landing path is updated in real time until the drone reaches the end point of the optimal landing path.

5. The control method for automatic landing of a fixed-wing UAV according to claim 4, characterized in that: In step S312, the control instruction is expressed as: In the formula, is a control instruction. is the proportional gain, is the error vector, is the integral gain, yes The integral term of is the differential gain, is the rate of change of the error vector.

6. A control device for automatic landing of a fixed-wing UAV, characterized in that: The control device includes: a memory and a processor, the memory stores a computer program, and the processor executes the computer program to implement the control method for automatic landing of a fixed-wing unmanned aerial vehicle according to any one of claims 1-5.

7. A readable storage medium, characterized in that: The readable storage medium stores one or more programs, which can be executed by one or more processors, and the programs enable the machine to execute the control method for automatic landing of a fixed-wing unmanned aerial vehicle according to any one of claims 1-5.

Images