Vehicle-mounted unmanned aerial vehicle cross-country road condition dynamic recovery simulation system

Abstract

The invention provides a vehicle-mounted unmanned aerial vehicle cross-country road condition dynamic recovery simulation system. In the system, simulation environments corresponding to various cross-country road conditions are created through an environment construction module; creating an unmanned aerial vehicle simulation model and a vehicle simulation model through a model construction module; controlling a pod mechanism in the unmanned aerial vehicle simulation model to drive a camera to rotate through a pod control module so as to track and shoot a specified two-dimensional code; determining pose information of the specified two-dimensional code in a camera coordinate system of the unmanned aerial vehicle simulation model through an image recognition model; the unmanned aerial vehicle simulation model is controlled to fly and land on the vehicle simulation model through the recovery control module; and through a data analysis module, an analysis result is obtained based on a related data set of the unmanned aerial vehicle simulation model in the flight and landing processes. Therefore, nonlinear motion characteristics such as bumping and slippage in a real scene can be effectively simulated, and the problem that'landing of the unmanned aerial vehicle to the vehicle running on the off-road condition cannot be accurately simulated 'can be solved.

Description

Technical Field

[0001] The present invention relates to the technical field of unmanned aerial vehicle simulation, and in particular to a vehicle-mounted unmanned aerial vehicle off-road dynamic recovery simulation system. Background Art

[0002] With the rapid development of drone technology, the demand for its application in many fields such as agricultural monitoring, environmental protection, logistics distribution, and infrastructure inspection is growing. However, achieving efficient and stable recovery of drones in complex terrain and off-road environments still faces many technical challenges. In order to meet these challenges, the application of simulation testing software in drone recovery systems is particularly important. At present, there are some simulation software for drone landing and recovery on the market. These software mainly focus on basic flight dynamics simulation, simple environment modeling, and limited dynamic recovery strategy support. Basic flight dynamics simulation modules usually use basic flight dynamics models and can simulate the basic flight behaviors of drones, such as hovering, takeoff, and landing. However, most of these models are based on idealized environmental conditions and lack in-depth simulation of complex terrain, which makes it difficult for simulation results to accurately reflect the situation in actual applications. Summary of the Invention

[0003] In view of this, the purpose of the embodiments of the present application is to provide a vehicle-mounted drone off-road dynamic recovery simulation system that can simulate off-road terrain, simulate complex terrain, and improve the problem of being unable to accurately simulate "drone landing on a vehicle traveling on off-road conditions."

[0004] To achieve the above technical objectives, the technical solutions adopted in this application are as follows:

[0005] The present invention provides a vehicle-mounted drone off-road dynamic recovery simulation system, comprising:

[0006] Environment construction module, used to create simulation environments corresponding to various off-road conditions;

[0007] A model building module, for creating a UAV simulation model and a vehicle simulation model, wherein the vehicle simulation model is provided with a recovery platform for landing the UAV simulation model, and the recovery platform is provided with a designated QR code for locating the UAV simulation model;

[0008] An image recognition module is used to recognize an image of the vehicle simulation model taken by a camera in the drone simulation model, and use the image with the specified QR code as a target image;

[0009] a pod control module, configured to control the pod mechanism in the UAV simulation model to rotate the camera based on the pixel position of the designated QR code in the target image, so as to track and photograph the designated QR code and update the target image;

[0010] The image recognition module is further configured to determine, based on the target image, the pose information of the designated QR code in the camera coordinate system of the UAV simulation model;

[0011] A vehicle control module, configured to control the vehicle simulation model to start and travel in the simulation environment after the drone simulation model takes off and when the distance from the drone simulation model exceeds a preset distance;

[0012] A recovery control module, configured to control the drone simulation model to fly and land in a designated QR code area of ​​the vehicle simulation model based on the posture information and the current position of the drone simulation model;

[0013] The data analysis module is used to obtain relevant data sets of the UAV simulation model during flight and landing, and obtain analysis results based on the relevant data sets, wherein the relevant data sets include the position, speed, attitude and expected trajectory of the UAV simulation model, and the errors corresponding to the speed, position and attitude of the analysis results.

[0014] In some optional embodiments, the recycling control module is further configured to:

[0015] Determining a position deviation based on the position of the UAV simulation model and the position of the vehicle simulation model obtained by the RTK positioning algorithm, and generating a first reference trajectory based on the position deviation and a proportional guidance strategy, so that the UAV simulation model flies toward the vehicle simulation model based on the first reference trajectory;

[0016] During the flight of the UAV simulation model, the position deviation is updated, and based on the proportional integral strategy, the lateral speed and the longitudinal speed of the UAV simulation model are adjusted so that the lateral distance and the longitudinal distance between the UAV simulation model and the vehicle simulation model are both less than a first preset distance, and the lateral speed is x Expressed as:

[0017] ιv x =k px Δx+k ix ∫Δxdt

[0018] The longitudinal velocity ιv y Expressed as:

[0019] ιv y =k py Δx+kiy ∫Δydt

[0020] Where k px 、k py Represents the horizontal and vertical proportional coefficients respectively; k ix 、k iy represent the lateral and longitudinal integral coefficients respectively; Δx and Δy represent the lateral distance and longitudinal distance between the UAV simulation model and the vehicle simulation model respectively;

[0021] When both the lateral distance and the longitudinal distance are less than the first preset distance, and the height difference between the UAV simulation model and the vehicle simulation model is less than a second preset distance, triggering the image recognition module to recognize the image of the vehicle simulation model taken by the camera in the UAV simulation model to obtain the posture information;

[0022] The recovery control module is also used to generate a second reference trajectory based on the posture information and the current position of the drone simulation model through a preset visual servo control strategy, so that the drone simulation model lands on the recovery platform based on the second reference trajectory.

[0023] In some optional embodiments, the recovery control module is also used to control the locking of the drone simulation model when the drone simulation model reaches a preset locking range of the vehicle simulation model, so that the drone simulation model lands on the recovery platform of the vehicle simulation model.

[0024] In some optional embodiments, the relevant data set further includes image data captured by a camera in the drone simulation model, and the data analysis module is specifically configured to:

[0025] determining a velocity error based on a velocity of the UAV simulation model and a reference velocity;

[0026] determining a position error based on the position of the drone simulation model and a corresponding expected position in the expected trajectory;

[0027] Determining an attitude error based on the attitude of the UAV simulation model and the corresponding attitude in the desired trajectory;

[0028] Determining a recognition rate and a recognition time of the designated QR code based on the image data;

[0029] The velocity error, the position error, and the attitude error are visually displayed.

[0030] In some optional embodiments, the system further includes a performance evaluation module for determining tracking accuracy based on the position of the drone simulation model and the position of the vehicle simulation model during the flight of the drone simulation model accompanying the vehicle simulation model. The tracking accuracy includes lateral and longitudinal errors, and height error.

[0031] The lateral and longitudinal errors E xy for:

[0032]

[0033] The height error E z for:

[0034]

[0035] Where x drone 、y drone 、z drone Respectively represent the coordinate values ​​of the x-axis, y-axis, and z-axis of the UAV simulation model in the world coordinate system of the simulation environment; car 、y car 、z car They respectively represent the coordinate values ​​of the x-axis, y-axis, and z-axis of the vehicle simulation model in the world coordinate system of the simulation environment.

[0036] In some optional embodiments, the performance evaluation module is further used to determine the recovery success rate of the UAV simulation model, expressed as:

[0037]

[0038] Where S represents the recovery success rate; n1 represents the number of times the UAV simulation model successfully lands on the recovery platform; and n2 represents the total number of landing tests of the UAV simulation model.

[0039] In some optional embodiments, the system further includes a calculation example generation module for generating the positioning accuracy of the RTK positioning algorithm, the positioning accuracy including high precision, medium precision and low precision, the error of the high precision is less than 5 cm; the error of the medium precision is greater than or equal to 5 cm and less than 10 cm; the error of the low precision is greater than or equal to 10 cm and less than 20 cm;

[0040] The performance evaluation module is also used to calculate the recovery success rate of the UAV simulation model under different positioning accuracies.

[0041] In some optional embodiments, the environment construction module is specifically used to create a simulation environment corresponding to the corresponding bumpiness level and the corresponding off-road road condition based on an operation instruction, and the operation instruction carries parameters corresponding to the bumpiness level and the off-road road condition.

[0042] In some optional embodiments, the multiple types of off-road road conditions include dirt off-road road conditions, grass off-road road conditions, rock off-road road conditions, and sand off-road road conditions, and each type of off-road road condition includes multiple simulation environments with different degrees of bumpiness.

[0043] In some optional embodiments, the environment construction module is also used to generate randomly distributed dynamic obstacles and environmental disturbances in the simulation environment, the dynamic obstacles include animal models and rolling stone models, and the environmental disturbances include ground vibrations and wind field changes.

[0044] The invention adopting the above technical solution has the following advantages:

[0045] In the technical solution provided by this application, the environment construction module overcomes the limitations of traditional simulations, which rely on a single terrain and cannot cover complex terrain, by meticulously modeling various off-road conditions (such as dirt, grass, and rocks). This module effectively simulates nonlinear motion characteristics such as bumps and slips found in real-world scenarios, addressing the issue of accurately simulating a drone landing on a vehicle traveling on an off-road surface. Using QR codes as visual markers, the image recognition module's real-time pose calculation ensures stable relative positioning even in bumpy vehicle conditions. The pod control module adjusts the camera angle in real time based on the QR code pixel position to continuously track the target, ensuring the continuity and accuracy of positioning information during landing. The vehicle control module and the recovery control module establish a dynamic "drone-vehicle" interactive link, ensuring that simulation results more closely align with the actual control logic of "dynamic tracking and precise landing." The data analysis module collects multi-dimensional data, including the drone's flight trajectory, speed, and attitude, to quantitatively evaluate the deviation between the simulation results and the expected trajectory, providing a basis for algorithm optimization. BRIEF DESCRIPTION OF THE DRAWINGS

[0046] The present application may be further illustrated by the non-limiting embodiments provided in the accompanying drawings. It should be understood that the following drawings illustrate only certain embodiments of the present application and are therefore not to be construed as limiting the scope of the present application. It is understood that a person skilled in the art can derive other relevant drawings from these drawings without inventive effort.

[0047] Figure 1 This is a functional block diagram of a vehicle-mounted drone off-road dynamic recovery simulation system provided in an embodiment of the present application.

[0048] FIG2( a ) is a schematic diagram of a soil surface off-road road condition simulation environment provided in an embodiment of the present application.

[0049] FIG2( b ) is a schematic diagram of a grass off-road road condition simulation environment provided in an embodiment of the present application.

[0050] FIG2( c ) is a schematic diagram of a rocky off-road road condition simulation environment provided in an embodiment of the present application.

[0051] FIG2( d ) is a schematic diagram of a sandy off-road road condition simulation environment provided in an embodiment of the present application.

[0052] FIG3( a ) is a schematic diagram of a target and a QR code located in the target provided in an embodiment of the present application.

[0053] FIG3( b ) is a schematic diagram of a vehicle simulation model provided in an embodiment of the present application.

[0054] FIG4( a ) is a schematic diagram of a simulation of the position of a vehicle tracked by a drone in the x-axis direction according to an embodiment of the present application.

[0055] FIG4( b ) is a schematic diagram of a simulation of the tracking error of a UAV tracking a vehicle in the x-axis direction provided in an embodiment of the present application.

[0056] FIG5( a ) is a schematic diagram of a simulation of the position of a vehicle tracked by a drone in the y-axis direction according to an embodiment of the present application.

[0057] FIG5( b ) is a schematic diagram of a simulation of the tracking error of a UAV tracking a vehicle in the y-axis direction provided in an embodiment of the present application.

[0058] FIG6( a ) is a schematic diagram of a simulation of the position of a vehicle tracked by a drone in the z-axis direction according to an embodiment of the present application.

[0059] FIG6( b ) is a schematic diagram of a simulation of the tracking error of a UAV tracking a vehicle in the z-axis direction provided in an embodiment of the present application.

[0060] Figure 7 A schematic diagram of the simulation of the relative posture and error curves of the drone and vehicle provided in the embodiment of the present application.

[0061] FIG8( a ) is a schematic diagram of a speed simulation of a drone tracking a vehicle in the x-axis direction provided in an embodiment of the present application.

[0062] FIG8( b ) is a schematic diagram of a simulation of the speed error in the x-axis direction of a vehicle tracked by a drone according to an embodiment of the present application.

[0063] FIG9( a ) is a schematic diagram of a speed simulation of a UAV tracking a vehicle in the y-axis direction provided in an embodiment of the present application.

[0064] FIG9( b ) is a schematic diagram of a simulation of the speed error in the y-axis direction of a vehicle tracked by a drone according to an embodiment of the present application.

[0065] Icons: 100-vehicle-mounted UAV off-road dynamic recovery simulation system; 110-environment construction module; 120-model construction module; 130-image recognition module; 140-pod control module; 150-vehicle control module; 160-recovery control module; 170-data analysis module. DETAILED DESCRIPTION

[0066] The present application will be described in detail below with reference to the accompanying drawings and specific embodiments. It should be noted that similar or identical parts in the drawings or descriptions are numbered the same. Implementations not shown or described in the drawings are known to those of ordinary skill in the art. In the description of this application, the terms "first," "second," etc. are used solely to distinguish descriptions and are not to be construed as indicating or implying relative importance.

[0067] The inventors have found that for simulation scenarios where a drone lands on a moving vehicle, in terms of environmental modeling, existing simulation systems can construct basic terrain models, but they lack sufficient support for the diversity and complexity of off-road surfaces, such as irregular undulations, surfaces of different materials, and dynamic obstacles. This results in a large gap between the simulation results and the actual off-road environment, limiting the effectiveness evaluation of the recovery strategy. In addition, existing simulation software has limited functionality in the design and testing of dynamic recovery strategies. It can usually only support predefined recovery paths and strategies, lacks adaptive and optimization capabilities, and is difficult to cope with the changing environmental conditions in actual applications. More importantly, existing simulation systems have obvious deficiencies in simulating the complex multi-body dynamic coupling relationship between drones and ground vehicles. This dynamic coupling relationship includes the impact of the dynamic motion of the ground vehicle on the drone recovery process and the flight dynamics characteristics of the drone itself. Existing systems find it difficult to achieve accurate simulation, resulting in the inability to fully consider the interaction of multi-body dynamics in the recovery strategy design.

[0068] In addition, the simulation of environmental disturbances is also a major shortcoming of existing simulation software. Various environmental disturbance factors such as wind field changes, ground vibrations, and noise interference have a significant impact on the UAV recovery process in actual off-road environments. However, existing simulation systems can usually only simulate a single or a few disturbance factors, and lack the ability to comprehensively simulate multiple disturbance factors, which limits the robustness and adaptability of the recovery strategy. Although existing UAV recovery simulation software has certain capabilities in basic flight simulation and simple environmental modeling, it still has significant deficiencies in high simulation, multi-body dynamics coupling, and complex environmental disturbance simulation, and cannot meet the actual needs of UAV dynamic recovery in complex off-road road environments.

[0069] Furthermore, the drone recovery process generates a large amount of flight and environmental data. Existing simulation systems have limited capabilities for data recording, analysis, and feedback, making it impossible to fully monitor and optimize the recovery process in real time. This results in slow iterative optimization of recovery strategies and difficulty adapting quickly to environmental changes. Finally, existing simulation software is limited in its user interface and visualization of simulation results. This makes it difficult to intuitively demonstrate the dynamic changes and strategy effects of the drone recovery process in complex environments, hindering users' understanding and analysis of simulation results.

[0070] To address the aforementioned technical issues, this application proposes a simulation system for the dynamic recovery of UAVs on off-road surfaces. This system aims to improve the efficiency and reliability of UAV recovery in complex off-road environments, while reducing testing risks and costs, by constructing a highly realistic off-road environment, accurately simulating multi-body dynamics coupling and comprehensive environmental disturbances, and combining advanced recovery strategy optimization tools with real-time data analysis mechanisms.

[0071] Please refer to Figure 1 Referring to FIG9 , the present application provides a vehicle-mounted drone off-road dynamic recovery simulation system 100 (hereinafter referred to as the present system), which includes at least one software function module that can be stored in a memory in the form of software or firmware or fixed in the operating system (OS) of an electronic device. The processor in the electronic device is used to execute the executable modules stored in the memory, such as the software function modules and computer programs included in the system. The electronic device can be, but is not limited to, a personal computer, a server, and other devices.

[0072] In this embodiment, the electronic device's computer system can be equipped with the Ubuntu 20.04 operating system and the ROS1 framework, and is configured with a high-performance multi-core processor, ≥32GB of DDR4 memory, and a discrete graphics card that supports CUDA acceleration. The Gazebo 11 simulation platform is used to build a highly realistic off-road scene, simulating terrain physical properties and coupling multi-body dynamics calculations. Based on the PX4 SITL (Software-in-the-Loop) simulation framework and the Gazebo Classic physics engine, the electronic device provides high-precision simulation of drones, vehicles, terrain, and environmental models.

[0073] Please refer to Figure 1 , the functions of each unit in this system can be as follows:

[0074] An environment construction module 110 is used to create simulation environments corresponding to various off-road conditions;

[0075] A model building module 120 is used to create a drone simulation model and a vehicle simulation model, wherein the vehicle simulation model is provided with a recovery platform for landing the drone simulation model, and the recovery platform is provided with a designated QR code for locating the drone simulation model;

[0076] An image recognition module 130 is configured to recognize an image of the vehicle simulation model taken by a camera in the drone simulation model, and use the image having the designated QR code as a target image;

[0077] The pod control module 140 is configured to control the pod mechanism in the UAV simulation model to rotate the camera based on the pixel position of the designated QR code in the target image, so as to track and capture the designated QR code and update the target image;

[0078] The image recognition module 130 is further configured to determine, based on the target image, the pose information of the designated QR code in the camera coordinate system of the drone simulation model;

[0079] The vehicle control module 150 is configured to control the vehicle simulation model to start and travel in the simulation environment after the drone simulation model takes off and when the distance from the drone simulation model exceeds a preset distance;

[0080] A recovery control module 160 is configured to control the drone simulation model to fly and land in a designated QR code area of ​​the vehicle simulation model based on the posture information and the current position of the drone simulation model;

[0081] The data analysis module 170 is used to obtain relevant data sets of the drone simulation model during flight and landing, and obtain analysis results based on the relevant data sets, wherein the relevant data sets include the position, speed, attitude and expected trajectory of the drone simulation model, and the errors corresponding to the speed, position and attitude of the analysis results.

[0082] The functions of each functional module in this system will be further explained below.

[0083] To ensure the construction of a simulation scenario for testing and optimizing the landing performance of drones, first, create an off-road terrain with appropriate bump intensity, accurately simulate the ups and downs, slopes and potholes in a real off-road environment, and cover different ground materials such as mud and sand, adding complexity and authenticity to the system. Secondly, design a vehicle (simulation model) with a suitable landing platform. The size and flatness of the recovery platform in the vehicle are adapted to the drone (simulation model). The vehicle can simulate movement at different speeds and trajectories to examine the drone's ability to track and land dynamic targets. Furthermore, the drone is equipped with a three-axis controllable pod to achieve flexible adjustment at multiple angles and accurately capture target information (such as a designated QR code on the vehicle). Finally, select a drone with fast and stable flight speed, good attitude control and anti-interference capabilities to ensure efficient and safe completion of landing missions in complex environments.

[0084] Please refer to Figure 2(a) to Figure 2(d) In this embodiment, multiple types of off-road road conditions include, but are not limited to, dirt off-road road conditions, grass off-road road conditions, rock off-road road conditions, and sand off-road road conditions. Each type of off-road road condition includes multiple simulation environments with different degrees of bumpiness.

[0085] The environment construction module 110 is specifically used to create a simulation environment corresponding to a corresponding bumpiness degree and a corresponding off-road road condition based on an operation instruction, wherein the operation instruction carries parameters corresponding to the bumpiness degree and the off-road road condition.

[0086] The degree of bumpiness can be set to multiple levels according to the height of the road surface. As an example, the degree of bumpiness can be as follows:

[0087] Level 1 bump, slight undulation: [0, 5] cm;

[0088] Level 2 bumps, small fluctuations: (5, 10] cm;

[0089] Level 3 bumps, medium ups and downs: (10, 20] cm;

[0090] Level 4 turbulence, large fluctuations: (20, +∞]cm.

[0091] During simulation testing, the tester can input corresponding operating instructions, allowing the environment construction module 110 to create a simulation environment with a corresponding degree of bumpiness and off-road conditions based on the user's different needs. For example, the environment construction module 110 can use the Gazebo robot simulation platform to build a highly realistic simulation scene, such as creating a simulation environment with level one bumpiness and dirt off-road conditions.

[0092] The environment construction module 110 is further configured to generate randomly distributed dynamic obstacles and environmental disturbances in the simulation environment. The dynamic obstacles include animal models and falling stone models, and the environmental disturbances include ground vibrations and wind field changes.

[0093] In this embodiment, when creating a simulation environment that simulates off-road conditions, high-precision terrain model data can be imported to accurately reproduce the irregular undulating features of the off-road road surface, such as the steep slopes of the mountains and the low-lying terrain of the valleys. At the same time, according to the physical properties of different ground materials in actual off-road scenes, the ground in the simulation environment is given corresponding friction coefficients, elastic moduli and other parameters to simulate a variety of ground material effects such as grass, sand, and rocky ground. In addition, dynamic obstacles are randomly distributed in the simulation scene, such as moving wild animal models, rolling stones, etc., and environmental disturbances (ground vibrations and wind field changes) are added to increase the complexity and authenticity of the environment. Among them, ground vibrations include the vibration amplitude and vibration duration of the ground. Wind field changes include wind direction, wind speed and duration.

[0094] Model building module 120 utilizes the Gazebo simulation platform, selecting the Iris drone model as the drone simulation model (or simply, drone) to meet flight requirements in complex environments. To achieve precise visual perception and data acquisition, the Iris drone model is equipped with a controllable three-axis gimbal pod mechanism, the CGo3. This pod mechanism is capable of high-definition image acquisition, and its controllable three-axis gimbal allows for flexible adjustment of shooting angles, providing the hardware foundation for subsequent target (specified QR code) recognition and tracking.

[0095] Model building module 120 uses the Gazebo simulation platform to create a vehicle model suitable for off-road driving, serving as the vehicle simulation model (or simply, the vehicle), as shown in Figure 3(b). The vehicle simulation model features large off-road tires and a high-ground-clearance chassis design, giving the vehicle excellent driving stability and maneuverability, capable of meeting the requirements of driving in complex off-road conditions. The vehicle's suspension system has been tuned to effectively buffer road bumps, ensuring smooth driving on rough terrain, laying the hardware foundation for the vehicle's reliable off-road driving.

[0096] In the vehicle simulation model, the recovery platform serves as the landing platform for the drone simulation model. A designated QR code is set on the recovery platform, which can be an Aruco QR code. As a visual identifier, the Aruco QR code has the characteristics of rapid recognition and high positioning accuracy. By parametrically designing the vehicle simulation model, its speed can be flexibly adjusted within a certain range to simulate mobile vehicles with different driving speeds and motion trajectories. In order to facilitate the camera to capture the Aruco QR code at a distance (such as 5 to 10 meters), a larger target can be set on the recovery platform, and the Aruco QR code is located at the center of the target. In other words, the target can be used to initially locate the vehicle process, and the Aruco QR code is used for precise positioning. The target and QR code can be shown in Figure 3(a).

[0097] In terms of the recognition of the specified QR code, the image recognition module 130 pre-processes the image collected by the camera, including grayscale, filtering and other operations to enhance the clarity and contrast of the image. Then, the detection function in the Aruco library is used to quickly detect whether the specified QR code exists in the image. If the specified QR code (such as the Aruco QR code) exists, the image is used as the target image. Next, the QR code in the target image is located, and the position and posture information of the specified QR code in the image coordinate system are solved, and finally the coordinates are transformed to the world coordinate system through the rotation matrix for UAV flight control. By comparing with the pre-set QR code parameters, accurate recognition of the specified QR code can be achieved. In order to improve the stability and accuracy of recognition, the image recognition module 130 can introduce a multi-frame image fusion and tracking algorithm so that during the movement of the vehicle, even if the specified QR code is partially blocked or the lighting conditions change, the position of the specified QR code can still be continuously and stably tracked.

[0098] In this embodiment, the pod control module 140 controls the pod mechanism in the drone simulation model to rotate along three axes, thereby ensuring that the Aruco QR code is always centered in the camera's field of view. Specifically, the center of the captured Aruco QR code is at or near the center of the entire image. Due to perspective distortion in camera imaging, when the target is at the edge of the field of view, the imaging deviation is large, resulting in increased errors in position and attitude calculations. However, at the center of the field of view, the camera's optical axis is perpendicular to the target plane, minimizing the effects of distortion and providing high-precision positioning information, which helps the drone adjust its trajectory and attitude. Furthermore, during actual drone landing, both the target and the drone are in motion, making it easy for the target to deviate from the field of view. The pod control module 140 controls the pod mechanism's rotation, compensating for relative motion in real time to ensure continuous and stable tracking, even in complex environments such as off-roading. Because the target information at the center of the camera's field of view is accurate and reliable, the drone can calculate the optimal landing path based on its own state, thus facilitating risk prediction and flight strategy adjustment.

[0099] The implementation process of the pod control module 140 tracking and photographing the specified QR code may be:

[0100] Determining a pixel error between the designated two-dimensional code (or target) and a center point of the target image based on a pixel position of the designated two-dimensional code (or target) in the target image;

[0101] Determine, based on the pixel error, a horizontal azimuth error and a vertical elevation error of the designated QR code (or target) in the camera viewing angle;

[0102] Determining a horizontal yaw angular velocity and a vertical angular velocity of the camera driven by the pod mechanism based on the horizontal azimuth error and the vertical elevation error;

[0103] Based on the horizontal yaw angular velocity and the vertical angular velocity, the pod mechanism is controlled to drive the camera to turn, so as to track and shoot the designated QR code (or target).

[0104] Wherein, determining the horizontal yaw angular velocity and the vertical angular velocity of the camera driven by the pod mechanism based on the horizontal azimuth error and the vertical elevation error includes:

[0105] Based on the horizontal azimuth error and the vertical elevation error, the horizontal yaw angular velocity and the vertical angular velocity are determined by a preset control law, wherein the preset control law is:

[0106] ω x =k r1 e ax +V Tx (1)

[0107] ω y =k r2 e ay +V Ty (2)

[0108] e ax =ψ a +k1∫ψ a dt (3)

[0109] e ay =θ a +k2∫θ a dt (4)

[0110] Where, ω x represents the horizontal yaw angular velocity; ω y represents the vertical angular velocity; k r1 represents the first gain constant; k r2 represents the second gain constant; e axRepresents the horizontal auxiliary error signal; e ay Represents the auxiliary error signal in the vertical direction; V Tx V represents the first compensation term used to compensate for interference and uncertainty; Ty represents the second compensation term used to compensate for interference and uncertainty; ψ a Indicates the horizontal azimuth error; θ a represents the vertical elevation angle error; k1 represents the third gain constant, which is used to adjust the integral term weight; k2 represents the fourth gain constant, which is used to adjust the integral term weight; t represents time.

[0111] V Tx 、V Ty is the solution of the following differential equation:

[0112]

[0113]

[0114] Where k r3 Represents the preset gain constant; β represents the switching gain constant of the sliding mode control, which is used to adjust the system's compensation strength for interference and uncertainty; the function sign() is a sign function, which is mathematically defined as: when the input is greater than zero, the output is +1, when it is less than zero, the output is -1, and when it is equal to zero, the output is 0. The drone can control ω x and ω y The pod can track the target.

[0115] A pixel coordinate system can be established within the target image. In this pixel coordinate system, the upper left corner of the target image is used as the origin, with the x-axis pointing rightward and the y-axis pointing downward. When the target enters the camera's field of view, the pod's visual feedback is obtained. The following algorithm calculates the pixel error between the target and the center of the field of view, and thus the horizontal and vertical errors of the target relative to the pod.

[0116] Assume that the normalized pixel coordinates of the center point of the target on the target image are (c x , c y ), ranging from 0 to 1. The width of the target image is w, and the height is h, in pixels. The pixel error between the center point of the target and the center of the field of view (that is, the center point of the target image) is expressed as:

[0117] e px =(0.5-c x )·w (7)

[0118] e py =(0.5-c y )·h (8)

[0119] Where, e px represents the x-axis pixel error; e py Represents the pixel error on the y-axis. From this, the horizontal azimuth error of the target relative to the camera's viewing angle can be calculated. a and vertical elevation error θ a ,for:

[0120]

[0121]

[0122] Among them, H FOV is the horizontal field of view, F FOV is the vertical field of view angle. a and θ a As the input of the UAV controller, the angular velocity of the pod is controlled to control the pod's rotation to make c x →0.5, c y →0.5, the target can be placed in the center of the field of view for tracking and photographing the target, which is conducive to providing low-latency and high-definition target images, facilitating subsequent pose calculation.

[0123] In this embodiment, tracking and photographing a designated QR code improves system adaptability and robustness. Under varying lighting conditions, the center of the field of view provides uniform illumination and good contrast, facilitating target identification. Flexible tracking of targets at varying speeds and trajectories ensures accurate and safe landing.

[0124] The image recognition module 130 may determine the pose information of the specified QR code in the camera coordinate system of the drone simulation model in the following manner:

[0125] Determining the distance corresponding to the designated QR code (or target) in the most recently updated target image according to a pre-established mapping relationship between the distance between the designated QR code (or target) and the camera and the size of the designated QR code (or target) in the image, and obtaining the posture of the designated QR code (or target) based on the target image, wherein the distance corresponding to the designated QR code (or target) in the target image represents the distance between the designated QR code (or target) and the camera;

[0126] The distance corresponding to the specified QR code (or target) in the target image and the posture of the specified QR code (or target) are converted into the position and posture in the camera coordinate system to obtain the posture information.

[0127] In this embodiment, the recovery control module 160 can complete the drone takeoff, accompanying flight, landing recovery functions through a script, and print and record data. During the drone flight control process, speed control is achieved with the help of the PX4 open source flight control system and Mavros middleware. Mavros serves as a communication bridge between ROS (Robot Operating System) and PX4, and can efficiently transmit control instructions and sensor data. The control program is written in C++ code.

[0128] When conducting a drone landing and recovery simulation test, initialization is required. The initialization process includes: starting the simulation environment; initializing the program files, setting the maximum lateral and longitudinal speeds of the drone, and setting the average vehicle speed; clearing all flags; positioning the drone and vehicle; setting controller parameters; and initializing the time.

[0129] After successful initialization, the drone unlocks the takeoff phase. As an example, if the drone takes off to an altitude of 10 meters above the ground, while continuously determining whether the distance between the drone and the vehicle is greater than a preset distance (for example, 10 meters), if it is greater than the preset distance, the vehicle control module 150 controls the vehicle to start and drive in an off-road simulation environment. In addition, the drone enters the first flight mode of accompanying the vehicle.

[0130] In the first flight mode, the drone can accompany the vehicle based on GPS data. For example, after remotely acquiring GPS information from a target vehicle (i.e., a vehicle with a designated QR code for the drone to land on), the system plans the drone's initial flight path based on this information, directing the drone toward the approximate location of the target vehicle. During flight, the drone uses its GPS module to obtain its own real-time position information, compares it with the target vehicle's position, and calculates the position deviation. Based on this deviation, MAVROS sends speed control commands to the PX4 to adjust the drone's flight speed and direction, gradually closing the distance between the drone and the target vehicle. The drone continuously monitors the vehicle's position, speed, and attitude information, sets a maximum tracking speed limit and a distance error, and controls the drone's accompanying flight. The drone's mode is controlled based on an error condition. If the error condition is not met, the drone continues accompanying the vehicle; if the error condition is met, the drone initiates the landing function. This error condition can be flexibly configured based on actual conditions. For example, if the horizontal distance difference between the drone and the vehicle is less than 1 meter and the vertical distance difference is less than 10 meters, the error condition is considered met, at which point the drone automatically enters the second flight mode, indicating landing.

[0131] In this embodiment, when the drone's onboard camera recognizes the Aruco QR code on the vehicle, flight control enters the precision tracking phase. At this point, the drone's offset relative to the target vehicle is calculated based on the Aruco QR code position information fed back by the camera. MAVROS adjusts the drone's lateral, longitudinal, and vertical speeds, gradually bringing the drone closer to the target vehicle while maintaining an appropriate altitude and position to ensure the QR code remains within the camera's field of view. As the drone approaches the target vehicle, its speed control instructions are continuously updated based on data from the camera and other sensors. For example, if the relative speed between the drone and the target vehicle is detected to be excessive, the flight speed is appropriately reduced to improve landing safety and accuracy. Furthermore, the drone's attitude and speed are adjusted in real time to ensure flight stability, taking into account potential airflow disturbances and terrain undulations in off-road environments. Once the set error range is reached, the drone begins landing, maintaining its lateral and lateral speeds, adjusting its altitude to enter the landing platform range, and continuously calculating the position error. If the lock conditions are met, the drone locks and lands. When the drone reaches the lock range, it locks directly in mid-air and lands on the landing platform, while the vehicle brakes. Terminal data recording stops, and the data log file is output.

[0132] Specifically, when the UAV approaches the vehicle and prepares to land, the vehicle is used as the target point. The recovery control module 160 can determine the position deviation of the position of the UAV simulation model and the position of the vehicle simulation model based on the RTK (Real-Time Kinematic, real-time dynamic carrier phase difference technology) positioning algorithm, and generate a first reference trajectory based on the position deviation and the proportional guidance strategy, so that the UAV simulation model flies toward the vehicle simulation model based on the first reference trajectory; wherein, based on the position deviation, the UAV simulation model can generate speed instructions in the three-axis direction based on the proportional guidance strategy, and the desired speed of the three axes is:

[0133] v x =k px ·Δx0 (11)

[0134] v y =k py ·Δy0 (12)

[0135] v h =k ph ·Δz0 (13)

[0136] Where, v x 、v y 、v h They represent the expected speeds of the UAV simulation model in the x, y, and z axes in the world coordinate system of the simulation environment; k px 、k py、k ph Represent the proportional coefficients corresponding to the x, y, and z axes respectively; Δx0, Δy0, and Δz0 represent the initial position deviations between the UAV simulation model and the vehicle simulation model on the x, y, and z axes respectively;

[0137] At the initial landing node, the UAV simulation model flies toward the vehicle simulation model at the desired speed;

[0138] During the flight of the UAV simulation model, the position deviation can be continuously updated in the mid-distance adjustment stage, and the lateral speed and longitudinal speed of the UAV simulation model can be adjusted based on the proportional integral strategy, so that the lateral distance and longitudinal distance between the UAV simulation model and the vehicle simulation model are both less than the first preset distance, and the lateral speed i x Expressed as:

[0139] ιv x =k px Δx+k ix ∫Δxdt (14)

[0140] The longitudinal speed lv y Expressed as:

[0141] ιv y =k py Δx+k iy ∫Δydt (15)

[0142] Where k px 、k py Represents the horizontal and vertical proportional coefficients respectively; k ix 、k iy represent the lateral and longitudinal integral coefficients respectively; Δx and Δy represent the lateral distance and longitudinal distance between the UAV simulation model and the vehicle simulation model respectively;

[0143] The UAV simulation model is in the mid-range adjustment stage, according to the lateral speed ιv x and longitudinal velocity ιv y Approach the vehicle simulation model so that the UAV simulation model looks down at the vehicle simulation model. Ideally, the UAV simulation model is directly above the vehicle simulation model and flies along with the movement of the vehicle simulation model.

[0144] When both the lateral distance and the longitudinal distance are less than the first preset distance (which can be flexibly set according to actual conditions, for example, 1 meter), and the height difference between the drone simulation model and the vehicle simulation model is less than the second preset distance (which can be flexibly set according to actual conditions, for example, 10 meters), the image recognition module 130 is triggered to recognize the image of the vehicle simulation model taken by the camera in the drone simulation model to obtain the posture information;

[0145] The recovery control module 160 is further configured to generate a second reference trajectory based on the posture information and the current position of the UAV simulation model through a preset visual servo control strategy, so that the UAV simulation model lands on the recovery platform based on the second reference trajectory.

[0146] As an example, the implementation method of the drone simulation model landing based on the second reference trajectory can be:

[0147] When the UAV simulation model approaches the top of the vehicle simulation model and the vertical distance is less than 10m, the image recognition module 130 is triggered to recognize the image of the vehicle simulation model taken by the camera in the UAV simulation model. The image recognition module 130 recognizes the Aruco QR code on the vehicle recovery platform and obtains the lateral and longitudinal relative position deviation Δx between the UAV and the vehicle. v , Δy v , and switches to visual servo control. Under the visual servo control strategy, the image recognition module 130 calculates the control amount based on the image Jacobian matrix and generates a precise adjustment trajectory (i.e., the second reference trajectory) so that the drone can accurately align with the recovery platform on the landing vehicle, such as the horizontal displacement adjustment amount. (J x1 、J x2 is the image Jacobian matrix element), guiding the UAV to complete the landing.

[0148] The recovery control module 160 is further configured to lock the drone simulation model when it reaches a preset locking range of the vehicle simulation model, allowing it to land on the recovery platform of the vehicle simulation model. The locking range can be flexibly determined based on actual conditions. For example, when the height difference between the drone simulation model's landing gear and the recovery platform is less than 5 centimeters, the preset locking range is determined to be reached, and the drone simulation model is then locked. After locking, the drone falls onto the recovery platform, and the vehicle brakes.

[0149] In this embodiment, the relevant data set also includes image data captured by a camera in the drone simulation model. The data analysis module 170 is specifically configured to:

[0150] determining a velocity error based on a velocity of the UAV simulation model and a reference velocity;

[0151] determining a position error based on the position of the drone simulation model and a corresponding expected position in the expected trajectory;

[0152] Determining an attitude error based on the attitude of the UAV simulation model and the corresponding attitude in the desired trajectory;

[0153] Determining a recognition rate and a recognition time of the designated QR code based on the image data;

[0154] The speed error, the position error and the attitude error are visualized as shown in Figures 4 to Figure 7 .

[0155] In this embodiment, throughout the drone's flight and landing process, the system collects various data in real time, including the drone's position, speed, attitude, camera image data, and vehicle motion information, to form relevant data sets. These relevant data sets are transmitted to the ROS node via Mavros and stored in the electronic device's local database for subsequent analysis and processing. The data analysis module 170 can use Python to write data analysis scripts, combined with data processing libraries such as Pandas and NumPy to clean and preprocess the collected data. Outliers and noise interference are removed to ensure data accuracy and reliability. The data is then further mined and analyzed based on different research needs and analysis indicators. For example, by analyzing the drone's flight trajectory data, its flight stability and accuracy are evaluated. Parameters such as flight trajectory deviation, average speed, and maximum speed are calculated to determine whether the drone is flying along the predetermined path and the tracking accuracy when approaching the target vehicle. By processing the camera image data, the recognition rate and recognition time of the QR code can be analyzed to evaluate the performance of the visual navigation system. To intuitively display the data analysis results, the Matplotlib drawing library is used to draw various data curve graphs. For example, plotting the drone's position versus time, velocity versus time, and attitude angle versus time clearly illustrates the drone's changing state during flight. Furthermore, plotting the change in QR code recognition rate over time helps researchers understand the performance of the visual navigation system at different stages. By analyzing these data graphs, researchers can gain a deeper understanding of the drone's performance characteristics and challenges during off-road landings.

[0156] Please refer to Figures 4(a) to 9(b), as a data analysis example, shows the analysis results of the relative position error, relative attitude error, and relative velocity error of the UAV and vehicle during landing. Based on the performance analysis results, the UAV's flight control strategy, control parameters, visual navigation algorithm, and hardware parameters were optimized and adjusted to improve its landing performance in complex off-road environments, providing strong support for scientific research and experiments.

[0157] In this embodiment, the system may further include a performance evaluation module for determining tracking accuracy based on the position of the drone simulation model and the position of the vehicle simulation model during the flight of the drone simulation model and the vehicle simulation model. The tracking accuracy includes lateral and longitudinal errors, as well as height error.

[0158] The lateral and longitudinal errors E xy for:

[0159]

[0160] The height error E z for:

[0161]

[0162] Where x drone 、y drone 、z drone Respectively represent the coordinate values ​​of the x-axis, y-axis, and z-axis of the UAV simulation model in the world coordinate system of the simulation environment; car 、y car 、z car They respectively represent the coordinate values ​​of the x-axis, y-axis, and z-axis of the vehicle simulation model in the world coordinate system of the simulation environment.

[0163] In this embodiment, in the scenario where a UAV accompanies a vehicle, positioning and tracking accuracy are key indicators for measuring system performance. xy The square root of the sum of the squares of the horizontal coordinate differences between the drone and the target vehicle is calculated. It directly reflects the real-time relative position deviation between the two in the horizontal direction. For example, when the vehicle is traveling in a straight line at a speed of 8m / s and the drone is flying alongside it, continuously measuring this error can effectively evaluate the horizontal positioning and tracking capabilities. z The absolute value of the vertical height difference between the drone and the vehicle focuses on accuracy before touchdown and can be used to determine vertical positioning and tracking effectiveness. These two error metrics provide a quantitative basis for evaluating the positioning and tracking accuracy of the drone relative to the vehicle, both horizontally and vertically, helping to optimize system performance and ensure mission reliability.

[0164] In this embodiment, the performance evaluation module is further used to determine the recovery success rate of the UAV simulation model, which is expressed as:

[0165]

[0166] Where S represents the recovery success rate; n1 represents the number of times the UAV simulation model successfully lands on the recovery platform; and n2 represents the total number of landing tests of the UAV simulation model.

[0167] When evaluating drone performance, the recovery success rate (S) is a key metric. It's measured by calculating the percentage of successful landings out of the total number of tests. This metric comprehensively reflects the stability and reliability of the drone during landing, as well as its adaptability to the landing environment and control commands. For missions such as logistics and mapping that rely on precise drone landings, the success rate provides an important quantitative basis for evaluating and improving drone system performance. The success criterion for this system is a smooth landing of the drone on the vehicle. Simulation experiments have demonstrated that using high-precision RTK positioning (error less than 5 cm), the system's landing success rate exceeds 95%.

[0168] The system may further include a calculation example generation module for generating the positioning accuracy of the RTK positioning algorithm, wherein the positioning accuracy includes high precision, medium precision and low precision, wherein the error of the high precision is less than 5 cm; the error of the medium precision is greater than or equal to 5 cm and less than 10 cm; the error of the low precision is greater than or equal to 10 cm and less than 20 cm;

[0169] The performance evaluation module is also used to calculate the recovery success rate of the UAV simulation model under different positioning accuracies.

[0170] As an example, the following are the test results of the inventors for three positioning accuracy calculation examples:

[0171]

[0172] Under high-precision positioning, the drone's success rate is as high as 96%, indicating that accurate positioning can provide the drone with accurate location information, allowing it to effectively adjust its flight trajectory and complete landing. As positioning accuracy decreases, the drone adjusts its trajectory based on inaccurate location information, easily deviating from the correct landing path, resulting in an increased risk of landing failure. xy As accuracy decreases, the error increases. This clearly demonstrates the close correlation between positioning accuracy and landing deviation. The higher the positioning accuracy, the more precisely the drone can control its position, and the smaller the landing deviation. Overall, positioning accuracy has a critical impact on drone landing performance. Improving positioning accuracy can help improve the success rate and accuracy of drone landings.

[0173] In terms of the simulation environment's construction principles, the Gazebo platform imports high-precision terrain model data and assigns parameters to different ground materials based on their actual physical properties, making the simulated off-road conditions highly realistic. In this environment, the movements of the drone and ground vehicle follow the laws of physics, providing a reliable foundation for testing. In multi-body dynamics coupling simulation, by establishing a precise dynamic model that considers the interplay between the ground vehicle's motion and the drone's own flight dynamics, the recovery process is simulated more accurately. For example, the impact of vehicle acceleration and cornering on the drone's recovery trajectory can be precisely calculated, allowing researchers to optimize the recovery strategy. In environmental disturbance simulation, various factors, such as wind and noise, are comprehensively considered, and simulations based on their respective physical models are performed to more realistically reflect actual conditions, helping to improve the robustness of the recovery strategy.

[0174] This application implements several new features. In the simulation system, advanced optimization algorithms are integrated to achieve automated design and optimization of recovery strategies, which is not available in traditional systems. At the same time, a powerful real-time data analysis and feedback mechanism can comprehensively monitor and deeply analyze large amounts of data during the recovery process. Potential patterns are discovered from the data to provide support for rapid iterative optimization of recovery strategies and promote continuous improvement of drone technology. In addition, the optimized user interaction and visualization interface provides an intuitive and easy-to-use operating experience and rich simulation result display functions, making it easier for users to deeply understand and analyze the recovery process.

[0175] After multiple experimental verifications, the system of the present application has demonstrated excellent performance. In comparative experiments, the average landing success rate of drones tested using a traditional simulation system on off-road surfaces was 60%, while the landing success rate of drones optimized using the simulation system of the present application was increased to 90%. From the perspective of flight trajectory stability, the standard deviation of the flight trajectory deviation of the drone of the present application system was reduced to 0.25 meters, effectively improving the accuracy of the flight. In the environmental disturbance simulation experiment, when wind field changes, bumpy road conditions and noise interference were introduced at the same time, a large number of errors occurred in the recovery strategy of the traditional system. However, due to the comprehensive environmental disturbance simulation capability of the present application system, its optimized recovery strategy can still maintain a high degree of reliability, which fully demonstrates the significant effect of the present application in improving the landing performance of drones on off-road surfaces.

[0176] Based on the above design, this system can realistically reproduce complex off-road terrain features and achieve precise dynamic coupling simulation between the UAV and ground vehicles. Using Gazebo, a comprehensive and highly realistic off-road simulation environment is constructed, encompassing multiple factors such as complex terrain, dynamic obstacles, and varying ground materials, providing a near-realistic scenario for UAV testing. This system can effectively test the landing performance of UAVs on off-road surfaces, providing a highly realistic and repeatable testing platform for scientific research and experimentation. By constructing complex off-road conditions and dynamic targets, the system simulates various real-world challenges and uncertainties, making the test results more valuable. Combining Aruco QR code visual navigation with PX4 speed control enables the UAV to accurately track and land moving targets, improving the UAV's autonomous recovery capabilities in complex environments. A comprehensive data acquisition and analysis system has been established, capable of recording and analyzing UAV flight data in real time, providing strong support for performance evaluation and algorithm optimization.

[0177] In addition, this system supports the comprehensive simulation of multiple environmental disturbance factors to ensure the robustness and adaptability of the recovery strategy. By integrating advanced optimization algorithms, this system can automatically design and optimize the recovery strategy, improving recovery efficiency and success rate. In addition, this system also has a powerful real-time data analysis and feedback mechanism, which can comprehensively monitor and analyze flight data and environmental data during the recovery process, supporting rapid iteration and optimization of the strategy. Finally, this system has also been optimized in terms of user interaction and visualization, providing an intuitive and easy-to-use interface and rich simulation results display functions to help users better understand and analyze the dynamic changes and strategy effects during the recovery process.

[0178] Through the description of the above implementation methods, those skilled in the art can clearly understand that the present application can be implemented through hardware or by means of software plus a necessary general hardware platform. Based on this understanding, the technical solution of the present application can be embodied in the form of a software product, which can be stored in a non-volatile storage medium (which can be a CD-ROM, a USB flash drive, a mobile hard disk, etc.), including a number of instructions for enabling a computer device (which can be a personal computer, a server, or a network device, etc.) to execute the methods described in each implementation scenario of the present application.

[0179] In the embodiments provided in the present application, it should be understood that the disclosed system can also be implemented in other ways. The system embodiments described above are merely schematic. For example, the flowcharts and block diagrams in the accompanying drawings show the possible architecture, functions and operations of the systems and computer program products according to the multiple embodiments of the present application. In this regard, each box in the flowchart or block diagram can represent a module, a program segment or a part of code, and a part of the module, program segment or code includes one or more executable instructions for implementing the specified logical function. It should also be noted that each box in the block diagram and / or the flowchart, and the combination of the boxes in the block diagram and / or the flowchart, can be implemented by a dedicated hardware-based system that performs the specified function or action, or can be implemented by a combination of dedicated hardware and computer instructions. In addition, the functional modules in the various embodiments of the present application can be integrated together to form an independent part, or each module can exist alone, or two or more modules can be integrated to form an independent part.

[0180] The above description is merely an embodiment of the present application and is not intended to limit the scope of protection of the present application. For those skilled in the art, various modifications and variations of the present application are possible. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principles of the present application shall be included in the scope of protection of the present application.

Claims

1. A vehicle-mounted UAV off-road dynamic recovery simulation system, characterized by: The system comprises: Environment construction module, used to create simulation environments corresponding to various off-road conditions; A model building module, for creating a UAV simulation model and a vehicle simulation model, wherein the vehicle simulation model is provided with a recovery platform for landing the UAV simulation model, and the recovery platform is provided with a designated QR code for locating the UAV simulation model; An image recognition module is used to recognize an image of the vehicle simulation model taken by a camera in the drone simulation model, and use the image with the specified QR code as a target image; a pod control module, configured to control the pod mechanism in the UAV simulation model to rotate the camera based on the pixel position of the designated QR code in the target image, so as to track and photograph the designated QR code and update the target image; The image recognition module is further configured to determine, based on the target image, the pose information of the designated QR code in the camera coordinate system of the UAV simulation model; A vehicle control module, configured to control the vehicle simulation model to start and travel in the simulation environment after the drone simulation model takes off and when the distance from the drone simulation model exceeds a preset distance; A recovery control module, configured to control the drone simulation model to fly and land in a designated QR code area of ​​the vehicle simulation model based on the posture information and the current position of the drone simulation model; The data analysis module is used to obtain relevant data sets of the UAV simulation model during flight and landing, and obtain analysis results based on the relevant data sets, wherein the relevant data sets include the position, speed, attitude and expected trajectory of the UAV simulation model, and the errors corresponding to the speed, position and attitude of the analysis results.

2. The system according to claim 1, wherein: The recycling control module is further configured to: Determining a position deviation based on the position of the UAV simulation model and the position of the vehicle simulation model obtained by the RTK positioning algorithm, and generating a first reference trajectory based on the position deviation and a proportional guidance strategy, so that the UAV simulation model flies toward the vehicle simulation model based on the first reference trajectory; During the flight of the UAV simulation model, the position deviation is updated, and based on the proportional integral strategy, the lateral speed and the longitudinal speed of the UAV simulation model are adjusted so that the lateral distance and the longitudinal distance between the UAV simulation model and the vehicle simulation model are both less than a first preset distance, and the lateral speed is x Expressed as: ιv x =k px ·Δx+k ix ∫Δxdt The longitudinal velocity ιv y Expressed as: ιv y =k py ·Δx+k iy ∫Δydt Where k px 、k py Represents the horizontal and vertical proportional coefficients respectively; k ix 、k iy represent the lateral and longitudinal integral coefficients respectively; Δx and Δy represent the lateral distance and longitudinal distance between the UAV simulation model and the vehicle simulation model respectively; When both the lateral distance and the longitudinal distance are less than the first preset distance, and the height difference between the UAV simulation model and the vehicle simulation model is less than a second preset distance, triggering the image recognition module to recognize the image of the vehicle simulation model taken by the camera in the UAV simulation model to obtain the posture information; The recovery control module is also used to generate a second reference trajectory based on the posture information and the current position of the drone simulation model through a preset visual servo control strategy, so that the drone simulation model lands on the recovery platform based on the second reference trajectory.

3. The system according to claim 1, wherein: The recovery control module is also used to control the UAV simulation model to lock when the UAV simulation model reaches a preset locking range of the vehicle simulation model, so that the UAV simulation model lands on the recovery platform of the vehicle simulation model.

4. The system according to claim 1, wherein: The relevant data set also includes image data captured by a camera in the UAV simulation model, and the data analysis module is specifically used to: determining a velocity error based on a velocity of the UAV simulation model and a reference velocity; determining a position error based on the position of the drone simulation model and a corresponding expected position in the expected trajectory; Determining an attitude error based on the attitude of the UAV simulation model and the corresponding attitude in the desired trajectory; Determining a recognition rate and a recognition time of the designated QR code based on the image data; The velocity error, the position error, and the attitude error are visually displayed.

5. The system according to claim 1, wherein: The system further includes a performance evaluation module for determining tracking accuracy based on the position of the UAV simulation model and the position of the vehicle simulation model during the flight of the UAV simulation model accompanying the vehicle simulation model. The tracking accuracy includes lateral and longitudinal errors, and height error. The lateral and longitudinal errors E xy for: The height error E z for: Where x drone 、y drone 、z drone Respectively represent the coordinate values ​​of the x-axis, y-axis, and z-axis of the UAV simulation model in the world coordinate system of the simulation environment; car 、y car 、z car They respectively represent the coordinate values ​​of the x-axis, y-axis, and z-axis of the vehicle simulation model in the world coordinate system of the simulation environment.

6. The system according to claim 5, characterized in that The performance evaluation module is also used to determine the recovery success rate of the UAV simulation model, which is expressed as: Where S represents the recovery success rate; n1 represents the number of times the UAV simulation model successfully lands on the recovery platform; and n2 represents the total number of landing tests of the UAV simulation model.

7. The system according to claim 6, characterized in that The system also includes a calculation example generation module for generating the positioning accuracy of the RTK positioning algorithm, wherein the positioning accuracy includes high precision, medium precision and low precision, wherein the error of the high precision is less than 5 centimeters; the error of the medium precision is greater than or equal to 5 centimeters and less than 10 centimeters; and the error of the low precision is greater than or equal to 10 centimeters and less than 20 centimeters. The performance evaluation module is also used to calculate the recovery success rate of the UAV simulation model under different positioning accuracies.

8. The system according to any one of claims 1 to 7, characterized in that The environment construction module is specifically used to create a simulation environment corresponding to a corresponding degree of bumpiness and a corresponding off-road road condition based on an operation instruction, wherein the operation instruction carries parameters corresponding to the degree of bumpiness and the off-road road condition.

9. The system according to any one of claims 1 to 7, characterized in that The multiple types of off-road road conditions include dirt off-road road conditions, grass off-road road conditions, rock off-road road conditions and sand off-road road conditions, and each type of off-road road condition includes multiple simulation environments with different degrees of bumpiness.

10. The system according to any one of claims 1 to 7, characterized in that The environment construction module is further used to generate randomly distributed dynamic obstacles and environmental disturbances in the simulation environment. The dynamic obstacles include animal models and rolling stone models, and the environmental disturbances include ground vibrations and wind field changes.

Images