High-precision low-cost unmanned aerial vehicle target interception method and system

Abstract

The invention relates to the technical field of unmanned aerial vehicle interception, in particular to a high-precision low-cost unmanned aerial vehicle target interception method and system. The method comprises the following steps: S1, target identification; images in the visual area range are collected, the geometric center of a target is calculated, and the target is tracked; s2, attitude closed-loop control; the airborne camera is used for detecting the pixel deviation between the center of the target and the optical center of the camera in real time, and the attitude of the unmanned aerial vehicle is controlled; s3, track closed-loop control; the flight control system combines deviation correction to realize trajectory closed-loop tracking, and ensures that the unmanned aerial vehicle always advances towards the target direction; s4, adjusting in real time; continuously monitoring target deviation and attitude adjustment, and updating the control quantity for real-time correction; s5, controlling the unmanned aerial vehicle to adjust the flight attitude and approach the target; and S6, repeating the steps S1 to S5 until the target is intercepted. According to the method, attitude adjustment of the unmanned aerial vehicle is directly driven through the pixel-level deviation, the problem of accumulative errors in traditional coordinate conversion is abandoned, and the target interception precision can be remarkably improved. According to the method, the real-time performance and stability of target tracking in the interception process are greatly improved, so that the probability of mistaken interception or missing interception is effectively reduced.

Description

Technical Field

[0001] The present invention relates to the technical field of unmanned aerial vehicle (UAV) interception, and particularly to a high-precision and low-cost UAV target interception method and system. Background Art

[0002] In recent years, with the rapid development of UAV technology, its applications have been widely penetrated into fields such as aerial photography, agricultural plant protection, disaster relief, logistics transportation, and entertainment performances, and the market scale has been continuously expanding. However, the frequent occurrence of "black flight" and "unauthorized flight" incidents caused by the abuse of UAVs seriously threatens airspace safety, public privacy, and the safety of critical infrastructure. For example, problems such as UAVs illegally entering the no-fly zone of an airport causing flight delays and maliciously carrying dangerous items to carry out terrorist attacks have made the research and development of anti-UAV technology the focus of global attention.

[0003] Currently, the mainstream anti-UAV technologies mainly include:

[0004] Signal interference technology: By using directional radio frequency or sound waves to interfere with the communication link or navigation system of UAVs, forcing them to land or return. However, this method is easily affected by environmental electromagnetic interference and may accidentally damage legitimate UAVs or surrounding electronic devices.

[0005] System control technology: Taking advantage of the vulnerabilities in the UAV communication protocol to hijack the control right, but it is necessary to customize attack strategies for different models of UAVs, with poor universality and legal risks.

[0006] Physical interception technology: Directly intercepting the target by means of net capture, laser destruction, or impact, etc., but relying on high-precision positioning devices (such as lidar, multi-axis gimbals) and complex mechanical systems, with high hardware costs and difficult to meet the interception requirements for large-scale low-cost targets.

[0007] In recent years, the autonomous interception technology based on visual recognition has gradually emerged. It real-time locates the target through image processing technology and combines with the flight control system to adjust the UAV attitude to complete the interception. However, the existing solutions still have significant defects:

[0008] Although depth point cloud technology can obtain the three-dimensional coordinates of the target, the algorithm complexity is high and it relies on expensive sensors, resulting in a sharp increase in system costs. Although the three-axis gimbal camera can align with the target in real time, its mechanical structure is complex and the response delay is high, making it difficult to deal with high-speed dynamic targets. Traditional control algorithms have cumulative errors in coordinate transformation and attitude adjustment, and lack a real-time compensation mechanism, resulting in insufficient interception accuracy. Traditional algorithms rely on on-board binocular or monocular cameras. After capturing the image, coordinate transformation and depth calculation are required, which not only significantly increases the complexity of the calculation, greatly prolongs the response time of the adjustment speed, but also causes coordinate accuracy deviation due to camera image distortion and built-in algorithms. If high-precision algorithms and cameras are selected, the cost of the drone will increase significantly. Traditional algorithms often use PID control. This algorithm calculates the corresponding speed adjustment amount by transmitting the deviation (i.e., error) between the target position and the drone position to the controller. However, due to the continuous accumulation of the target position error caused by coordinate transformation, overshoot and overshoot phenomena will occur. At the same time, the calculation of the drone's own position also increases the complexity of the PID calculation.

[0009] The above technical bottlenecks seriously restrict the actual deployment efficiency of the anti-drone system. Therefore, there is an urgent need for a low-cost, high-precision, and fully autonomous drone interception solution to balance technical performance and economy and meet the application requirements of multiple scenarios. Summary of the Invention

[0010] The object of the present invention is to propose a high-precision and low-cost drone target interception method and system for the problems existing in the background technology. The present invention is applied to scenarios such as civilian security, public safety protection, and military defense. Through innovative visual processing algorithms and real-time attitude adjustment strategies, it realizes the rapid identification, precise tracking, and autonomous interception of illegally invading or threatening drones, effectively solving the bottleneck problems in traditional countermeasure technologies such as high hardware costs, poor control accuracy, and low automation, and providing an efficient and economical solution for drone interception tasks in complex dynamic environments.

[0011] Existing drone interception technologies generally have problems such as high-cost hardware dependence, poor control algorithm accuracy, low automation, and serious resource waste, which affect the universality and application efficiency of interception technologies. Current vision-based interception methods often rely on expensive depth point cloud technology or complex three-axis gimbal systems, which not only greatly increase the hardware cost but also are difficult to effectively apply in many low-cost actual application scenarios. In addition, traditional control algorithms often have significant errors in target coordinate transformation, drone attitude adjustment, and real-time feedback processing, which directly affect the accuracy and efficiency of interception, especially the dynamic adjustment ability in complex environments is weak and it is difficult to meet the requirements of high-precision interception targets. Based on this, the following technical solutions are proposed:

[0012] The first aspect of the present invention provides a high-precision and low-cost method for intercepting unmanned aerial vehicle (UAV) targets, including the following specific steps:

[0013] S1. Target recognition: Collect images within the viewing area, preprocess the captured target images based on OpenCV, calculate the geometric center of the target, and track the target.

[0014] S2. Attitude closed-loop control: Use the on-board camera to detect the pixel deviation between the target center and the optical center of the camera in real time, and directly map it to the attitude adjustment control parameters of the UAV to control the attitude of the UAV.

[0015] S3. Trajectory closed-loop control: The flight control system calculates the speed at the next moment according to the current motion state, drives the UAV to continuously approach the target by accelerating to the forward speed, and combines deviation correction to achieve trajectory closed-loop tracking, ensuring that the UAV always moves forward in the direction of the target.

[0016] S4. Real-time adjustment: Continuously monitor the target deviation and attitude adjustment, and update the control amount for real-time correction.

[0017] S5. Control the UAV to adjust the flight attitude and approach the target.

[0018] S6. Repeat steps S1 - S5 until the target is intercepted.

[0019] Preferably, in step S1, the on-board camera of the UAV is used to collect images of the area where the target is located; image preprocessing operations are performed based on OpenCV; the preprocessing steps include filtering for noise reduction, image enhancement, and sharpening.

[0020] Preferably, key features of the target are extracted from the preprocessed images, and edge detection is performed on the target to calculate the geometric center of the target.

[0021] Preferably, the control parameters for UAV attitude adjustment in step S2 include pitch, roll, and yaw angular velocities.

[0022] Preferably, the specific steps of step S2 are as follows:

[0023] S21. Set parameters: According to the model of the UAV and the type of the camera carried, set the optical center coordinates (x 0 , y 0 ) and the camera rotation matrix in advance according to the actual device parameters of the camera.

[0024] S22. Preset the proportional control coefficients Kp_roll, Kp_pitch, Kp_yaw; the proportional control coefficients are used to adjust the flight speed of the UAV according to the target position deviation to achieve accurate tracking and interception of the target.

[0025] S23. Preset the forward speed v; Set the maximum forward speed in advance. After the UAV recognizes the target, it will directly accelerate to the maximum speed for impact;

[0026] S24. Feed back the geometric center (x, y) of the recognized target and the optical center of the camera (x 0 , y 0 ) to the flight control of the UAV;

[0027] S25. Calculate the pixel error between the geometric center (x, y) of the target and the optical center of the camera (x 0 , y 0 ), Δx = x - x 0 , Δy = y - y 0 ;

[0028] S26. The flight control calculates the flight control parameters of the UAV based on the pixel error (Δx, Δy); including roll, pitch, and yaw speeds; and adjusts the flight speed and attitude of the UAV.

[0029] Preferably, step S3 includes the following specific steps:

[0030] S31. The flight control combines the current horizontal yaw angular velocity yaw_speed and the vertical pitch angular velocity vert_speed to calculate the speed at the next moment;

[0031] S32. For the horizontal error Δx, control the lateral movement of the UAV in the horizontal plane through the yaw angular velocity;

[0032] S33. For the vertical error Δy, adjust the longitudinal movement of the UAV through the pitch angular velocity to approach the target;

[0033] S34. After the speed command is issued, the UAV flies according to the new speed and attitude.

[0034] The second aspect of the present invention provides a high-precision and low-cost UAV target interception system, which uses the above method to intercept the target, including an image preprocessing module, a target recognition module, a flight control module, a calculation module, and an interactive display module;

[0035] The image preprocessing module is used to preprocess the collected specific target images;

[0036] The target recognition module is used to recognize the geometric center of the target in the preprocessed image and feed back the recognized geometric center and the optical center of the camera to the calculation module;

[0037] The calculation module calculates the error between the geometric center and the optical center of the camera, calculates the speed and motion state at the next moment, and outputs them to the flight control module;

[0038] The flight control module controls the UAV to fly at a new speed and motion state, continuously approaching the target;

[0039] The interactive display module is communicatively connected to the UAV, and is used to input control programs to the UAV and communicate, and obtain the real-time image information, motion trajectory, and motion state information of the UAV.

[0040] Preferably, the image preprocessing module preprocesses the image by filtering denoising, image enhancement, and sharpening.

[0041] Preferably, the target recognition module uses two-dimensional discrete Fourier transform, Gaussian filtering denoising, an image enhancement algorithm based on the multi-scale Retinex theory, and a LOG filter based on Gaussian filtering and Laplacian operator edge detection to extract the key features of the target, detect the target edge, and calculate the geometric center (x, y).

[0042] The third aspect of the present invention provides a high-precision and low-cost intercept UAV, which uses the above system and the above method to control the UAV for intercepting a target object in the airspace.

[0043] Compared with the prior art, the present invention has the following beneficial technical effects:

[0044] 1. The present invention directly drives the attitude adjustment of the UAV through pixel-level deviation, abandoning the cumulative error problem in traditional coordinate transformation, and can significantly improve the target interception accuracy. This method greatly improves the real-time performance and stability of target tracking during the interception process, thereby effectively reducing the probability of misinterception or missed interception.

[0045] 2. Low-cost deployment: Due to the adoption of lightweight vision processing algorithms and general hardware components (such as monocular cameras), the present invention significantly reduces the hardware cost, especially suitable for low-cost target interception scenarios. Even under limited resources, the system can still achieve efficient target recognition and precise interception control.

[0046] 3. Fully autonomous operation: The entire interception process requires no manual intervention, and operations such as target recognition, attitude adjustment, and trajectory tracking are all completed by the system independently. This technical solution fully improves the automation level of the system, enabling the UAV to adapt to more complex and dynamic interception environments, and further reducing the operation difficulty and error rate.

[0047] 4. Strong robustness and adaptability: Based on the closed-loop control strategy of proportional control, it can not only effectively compensate for the deviation of the target during flight, but also cope with environmental interference and the uncertainty of target movement. Even under relatively harsh weather conditions or complex environments, the system can still achieve precise target interception, with excellent robustness and adaptability.

[0048] 5. Through innovative visual processing technology and efficient control algorithm design, the present invention effectively optimizes the hardware cost and computing resource usage while ensuring the interception effectiveness, providing an efficient and economical solution for anti-drone technology. This technology can be widely deployed in multiple application scenarios, greatly promoting the development of drone interception technology, and has important application value especially in civil, security, and military fields. BRIEF DESCRIPTION OF THE DRAWINGS

[0049] Figure 1 It is a schematic diagram of a coordinate system and an imaging model;

[0050] Figure 2 It is a high-precision, low-cost automated interception / collision flow chart based on vision-based guidance method;

[0051] Figure 3 It is a simulation effect diagram of the target being struck under a circular trajectory.

[0052] Figure 4 It is the real effect of visual recognition in the simulation system

[0053] Figure 5 It is the process of the drone's camera view approaching the target continuously when chasing a moving object. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0054] Embodiment 1

[0055] As Figure 2 shown, the present invention provides a high-precision and low-cost drone target interception method, including the following specific steps:

[0056] S1. Target recognition; collect images within the viewing area, perform preprocessing on the collected target images based on OpenCV, calculate the geometric center of the target, and track the target; in step S1, use the drone's on-board camera to collect images of the target area; perform image preprocessing operations based on OpenCV; the preprocessing steps include filtering for noise reduction, image enhancement, and sharpening. Extract the key features of the target from the preprocessed images, and perform edge detection on the target to calculate the geometric center; calculate the geometric center and adjust the impact to achieve accurate target recognition and positioning. The process of identifying the geometric center of an object is as follows:

[0057] Step 1: Collect images: Run the code based on OpenCV through the drone's on-board camera to collect images in real time. Using the target recognition algorithm, once an object with preset features appears within the camera's field of view, it is quickly recognized and marked with a bounding box. The collected images serve as the basic data for subsequent processing (such as Figure 4 )

[0058] Step 2 Image Processing: Input the collected images into the OpenCV image preprocessing code module. First, perform Gaussian filtering for denoising, smooth the image through convolution operations, and reduce the impact of noise. Finally, sharpen the image using a LOG filter based on Gaussian filtering and Laplacian edge detection to highlight the edges of the target object.

[0059] Step 3 Calculate the Center: Use an edge detection algorithm (such as the Canny algorithm or the LOG filter) to extract the edge contour of the target object, separate it from the background, and calculate the geometric center coordinates (such as Figure 4 ), providing a key basis for UAV control

[0060] Step 4 Real-time Recognition: Continuously repeat the above steps during the process of the UAV hitting the target. The camera continuously captures new images, calculates the geometric center through processing, and obtains the pixel deviation by comparing it with the camera optical center coordinates. The flight control system calculates the flight control parameters based on this combined with the preset proportional control coefficient, and adjusts the attitude and speed of the UAV until it successfully hits the target. (For example, Figure 5 the UAV continuously approaches the target while continuously recognizing).

[0061] S2. Attitude Closed-loop Control; Use the on-board camera to continuously detect the pixel deviation between the target center and the camera optical center, and directly map it to the attitude adjustment control parameters of the UAV to control the attitude of the UAV; the control parameters for UAV attitude adjustment in step S2 include pitch, roll, and yaw angular velocities;

[0062] The specific steps of step S2 are as follows:

[0063] S21. Set Parameters; According to the model of the UAV and the type of the camera it carries, set the optical center coordinates (x 0 , y 0 ) and the camera rotation matrix in advance according to the actual device parameters of the camera;

[0064] The determination of the camera internal parameters and the optical center uses the Zhang Zhengyou calibration method, combined with the built-in algorithm of opencv

[0065] Step 1 Select a Suitable Chessboard: Select a chessboard pattern with black and white alternating colors, whose inner corner points are easy to identify and locate, and the grid size needs to be known and have high precision.

[0066] Step 2 Shoot from Multiple Angles: Use the camera to be calibrated to shoot the chessboard from different positions and angles, ensure that the chessboard occupies a certain proportion in the image and is clear and complete, avoid tilting, occlusion, and blurring, and the number of shots is generally 10 to 20.

[0067] Step 3 Obtain Camera Internal Parameters: Store the pictures in a specified folder, run the python code in opencv to read the images, detect the corner points, and perform camera calibration to finally obtain the internal parameter matrix where $f_x$ and $f_y$ are the focal lengths of the camera in the $x$-axis and $y$-axis directions respectively, and $C_x$ and $C_y$ represent the coordinates of the optical center in the image coordinate system, that is, the position of the optical center of the image. The last row $(0, 0, 1)$ is for the internal parameter matrix to perform correct operations in homogeneous coordinate transformation, which is a mathematical standard representation.

[0068] S22. Preset proportional control coefficients $K_{p\_roll}$, $K_{p\_pitch}$, and $K_{p\_yaw}$; the proportional control coefficients are used to adjust the flight speed of the UAV according to the target position deviation to achieve accurate tracking and interception of the target.

[0069] S23. Preset forward speed $v$; set the maximum forward speed in advance. After the UAV recognizes the target, it will directly accelerate to the maximum speed for impact.

[0070] S24. Feed back the recognized geometric center $(x, y)$ of the target and the camera optical center $(x$ 0 , $y$ 0 ) to the flight control of the UAV.

[0071] S25. Calculate the pixel error between the geometric center $(x, y)$ of the target and the camera optical center $(x$ 0 , $y$ 0 ), $\Delta x = x - x$ 0 , $\Delta y = y - y$ 0 .

[0072] S26. The flight control calculates the flight control parameters of the UAV based on the pixel error $(\Delta x, \Delta y)$; including roll, pitch, and yaw speeds; and adjusts the flight speed and attitude of the UAV.

[0073] S3. Trajectory closed-loop control; the flight control system calculates the speed at the next moment according to the current motion state, drives the UAV to continuously approach the target by accelerating to the forward speed, and combines deviation correction to achieve trajectory closed-loop tracking, ensuring that the UAV always moves towards the target direction.

[0074] Step S3 includes the following specific steps:

[0075] S31. The flight control combines the current horizontal yaw angular velocity $yaw\_speed$ and the vertical pitch angular velocity $vert\_speed$ to calculate the speed at the next moment.

[0076] S32. For the horizontal error $\Delta x$, control the lateral movement of the UAV in the horizontal plane through the yaw angular velocity.

[0077] S33. For the vertical error $\Delta y$, adjust the longitudinal movement of the UAV through the pitch angular velocity to approach the target.

[0078] S34. After the speed command is issued, the UAV flies according to the new speed and attitude.

[0079] S4. Real-time adjustment: Continuously monitor the target deviation and attitude adjustment, update the control quantity for real-time correction;

[0080] S5. Control the UAV to adjust its flight attitude and approach the target;

[0081] S6. Repeat steps S1 - S5 until the target is intercepted.

[0082] In view of the drawback that the coordinate conversion process in traditional algorithms is overly complex, the algorithm of this patent innovatively skips this link. Only by relying on the optical center coordinates of the camera parameters and the coordinates of the UAV pixel points, can the gap be quickly calculated and real-time differential proportional adjustment be implemented. This improvement omits the complicated calculation processes such as coordinate conversion, effectively improves the real-time response speed of UAV attitude control, reduces the requirements for the computing power of the on-board flight control and equipment, and reduces the cost expenditure.

[0083] Embodiment 2

[0084] The present invention provides a high-precision and low-cost UAV target interception system, which uses the method in Embodiment 1 to intercept the target, including an image preprocessing module, a target recognition module, a flight control module, a calculation module, and an interactive display module;

[0085] The image preprocessing module is used to preprocess the collected specific target images;

[0086] The target recognition module is used to recognize the geometric center of the target from the preprocessed image, and feedback the recognized geometric center and the camera optical center to the calculation module;

[0087] The calculation module calculates the error between the geometric center and the camera optical center, calculates the speed and motion state of the next moment, and outputs them to the flight control module;

[0088] The flight control module controls the UAV to fly according to the new speed and motion state, and continuously approaches the target;

[0089] The interactive display module is communicatively connected to the UAV, and is used to input control programs to the UAV and communicate, and obtain the real-time image information, motion trajectory, and motion state information of the UAV.

[0090] Preferably, the image preprocessing module preprocesses the image by using filtering denoising, image enhancement, and sharpening. The target recognition module uses two-dimensional discrete Fourier transform, Gaussian filtering denoising, an image enhancement algorithm based on the multi-scale Retinex theory, and a LOG filter based on Gaussian filtering and Laplacian operator edge detection to extract the key features of the target, detect the target edge, and calculate the geometric center (x, y).

[0091] The following uses a specific case to introduce the solution of this embodiment in detail.

[0092] As Figure 1 shown, there is currently 1 unmanned aerial vehicle equipped with a monocular camera and 1 moving target in an unknown environment. It is necessary to complete target recognition and autonomous interception / collision control. The implementation steps are as follows:

[0093] Step 1. Impact test design: The target is 30m away and moves in a circular trajectory with a radius of 15m. The maximum forward speed is 5m / s, and the initial parameters are as follows:

[0094] Camera rotation matrix:

[0095] The matrix plays a key role in the subsequent image acquisition and target positioning processes

[0096] Optical center coordinates: (320.5, 240.5)

[0097] Proportional control coefficients: Kp_roll = 0.3, Kp_pitch = 0.3

[0098] Step 2. Preprocess the collected specific target images, such as filtering and denoising, image enhancement, and sharpening. Use two-dimensional discrete Fourier transform, Gaussian filtering for denoising, image enhancement algorithm based on multi-scale Retinex theory, and LOG filter based on Gaussian filtering and Laplacian operator edge detection to extract key features of the target, detect the target edge, and calculate the geometric center (x, y).

[0099] Step 3. Data feedback to the calculation module: The target recognition module feeds back the recognized geometric center (x, y) information and the camera optical center to the calculation module. As the core control unit of the unmanned aerial vehicle, the calculation module will calculate the flight control parameters of the unmanned aerial vehicle based on this information to achieve target tracking and interception

[0100] Step 4. Error calculation: Calculate the error between the optical center and the target geometric center (Δx = x - 320.5, Δy = y - 240.5)

[0101] Step 5. Speed calculation: The flight control module combines the current horizontal yaw angular velocity yaw_speed and the vertical pitch angular velocity vert_speed to calculate the new speed (yaw_speed = -Kp_roll × Δx, vert_speed = -Kp_pitch × Δy) at the next moment (interval of about 0.02s) and issues commands to control the rotor. At the same time, the forward speed remains 5m / s. In this way, the flight speed of the unmanned aerial vehicle is dynamically adjusted according to the target position deviation, enabling the unmanned aerial vehicle to accurately approach the target.

[0102] Step 6. Repeat instructions: After the speed instruction is issued, the UAV flies according to the new speed and attitude. During this process, steps 2, 3, 4, and 5 are continuously repeated, that is, image preprocessing and recognition, data feedback, and speed control are continuously performed to form a closed-loop control process. As the UAV continuously adjusts its flight attitude and speed, the distance from the target is gradually reduced until the target is successfully struck. The interception trajectory is shown in Figure 3 , demonstrating the flexibility and accuracy of this strike system.

[0103] The following is a comparison test between this solution and traditional physical interception methods:

[0104] Experimental process

[0105] UAV target interception experiment of traditional method and pixel deviation method

[0106] Simulation experiment equipment and environment

[0107] 1. UAV: Equipped with a monocular camera and corresponding flight control system to ensure attitude adjustment and speed control can be achieved.

[0108] 2. Target object: Different motion trajectories and speeds can be set to simulate targets in different scenarios.

[0109] 3. Environment: The experiment is carried out in an open field to minimize the influence of environmental interference on the experimental results.

[0110] II. Experiment preparation

[0111] 1. Parameter setting: According to the experimental requirements, parameters such as the camera rotation matrix, optical center coordinates, and proportional control coefficient are calibrated. For example, the optical center coordinates are calibrated as (320.5, 240.5), and the proportional control coefficients are Kp_roll = 0.3 and Kp_pitch = 0.3.

[0112] 2. Define the motion parameters of the target object. For example, in a dynamic experiment, set the starting coordinates and different speeds of the target object.

[0113] III. Experimental process of traditional method

[0114] 1. Image acquisition and processing: Use the UAV-borne camera to collect images of the area where the target is located, and perform preprocessing operations such as filtering and denoising, and image enhancement through the opencv processing algorithm.

[0115] 2. Target coordinate calculation: Based on the collected images, use traditional coordinate transformation methods and combine camera parameters to calculate the coordinates of the target in the world coordinate system. In this process, complex coordinate transformation and depth calculation are required to obtain the three-dimensional coordinate information of the target.

[0116] 3. PID Control Calculation: Input the deviation between the target coordinates and the current position coordinates of the UAV into the flight controller to calculate the speed adjustment amount of the UAV. The PID algorithm controller outputs corresponding control signals according to the magnitude and change trend of the deviation to adjust the flight speed and attitude of the UAV.

[0117] 4. UAV Control and Interception: The flight control system controls the UAV to adjust its flight attitude and speed according to the PID calculation results, approach the target object and conduct interception. During the flight, continuously collect images and repeat the above steps to continuously update the target coordinates and adjust the flight state of the UAV.

[0118] 5. Data Recording: In each experiment, record data such as the actual coordinates of the target, the coordinates obtained from the camera coordinate transformation calculation, the relative error, and whether the interception is successful, and fill them into the corresponding data table.

[0119] V. Experimental Process of Pixel Deviation Method

[0120] 1. Image Acquisition and Preprocessing: Also use the UAV-borne camera to collect images of the area where the target is located, and perform filtering denoising, image enhancement, and sharpening based on OpenCV to prepare for subsequent target recognition.

[0121] 2. Calculation of Target Geometric Center: For the preprocessed image, use a script based on opencv to extract the key features of the target, detect the target edge, and thus calculate the geometric center of the target.

[0122] 3. Pixel Error Calculation: Feed the recognized target geometric center coordinates (x, y) and the pre-set camera optical center coordinates (x0, y0) back to the flight control system of the UAV, and calculate the pixel errors Δx = x - x0, Δy = y - y0.

[0123] 4. Calculation of Flight Control Parameters: The flight control system takes the pixel errors (Δx, Δy) as the basis and combines the preset proportional control coefficients Kp_roll, Kp_pitch, Kp_yaw to calculate the flight control parameters of the UAV, including roll, pitch, and yaw speeds. Calculate the horizontal yaw angular velocity yaw_speed = -Kp_roll × Δx, and the vertical pitch angular velocity vert_speed = -Kp_pitch × Δy.

[0124] 5. UAV Control and Real-time Adjustment: The flight control system controls the UAV to adjust its flight attitude and speed according to the calculated flight control parameters, approach the target object. During the flight, continuously monitor the target deviation and attitude adjustment situation, and update the control amount for real-time correction.

[0125] 6. Data recording: In each experiment, record data such as the actual pixel coordinates of the target, the coordinates calculated by the pixel deviation algorithm, the relative error, and whether the interception is successful. Record the speed adjustment frequency and fill it into the corresponding data table.

[0126] Table 1 Data table of recognition and impact tests using the pixel deviation method under static conditions for ten times

[0127]

[0128]

[0129] Table 2 Data table of recognition and impact tests using the coordinate transformation method under static conditions for ten times

[0130]

[0131] Table 3 Data table of impact comparison tests of two methods under dynamic conditions for ten times

[0132]

[0133]

[0134] Table 4 Comparison table of simulation impact effects of different methods

[0135]

[0136] Through the analysis of the data in Tables 1 - 4, it can be seen that in the UAV target interception experiment, the pixel deviation algorithm shows significant advantages in terms of accuracy, speed adjustment frequency, and success rate, providing a more efficient and economical solution for UAV target interception technology.

[0137] Analysis of the results of recognition and impact tests under static conditions

[0138] Pixel deviation method: From the data in Table 1, in the ten static condition experiments, the relative error between the coordinates calculated by the pixel deviation algorithm and the actual pixel coordinates of the object is relatively low, all within 1%. For example, when the static target coordinates are (4, 3, 4), the relative error is only (0.05%, 0.02%). This indicates that the algorithm can accurately identify the target coordinates, effectively reduce error accumulation, and thus greatly improve the accuracy of target interception. In the ten experiments, all interceptions were successful, and the success rate reached 100%, strongly proving the reliability of the algorithm in the static target interception scenario.

[0139] (2) Coordinate transformation method: As can be seen from Table 2, the relative error between the coordinates calculated by the coordinate transformation method and the actual coordinates is relatively large, above 1% - 10%. Taking the target coordinates (4, 3, 4) as an example, the relative errors are (1.05%, 1.02%, 0.92%), and with the change of the target coordinates, the error fluctuations are relatively obvious. In ten trials, the last two interceptions failed, and the success rate was 80%. This shows that the coordinate transformation method has certain limitations in dealing with the calculation of relatively distant coordinates, which affects the success rate of interception.

[0140] Analysis of impact test results under dynamic conditions: From the data in Table 3, it can be seen that under dynamic conditions, the interception success rates of the traditional method and the pixel deviation algorithm are greatly affected by the target speed. When the target speed is relatively low, such as 3 m / s and 8 m / s, both methods can successfully intercept some targets; but as the target speed increases to 12 m / s and above, the interception success rate of the traditional method drops significantly, and there are many cases of interception failure. In contrast, the pixel deviation algorithm still has a relatively high success rate at 12 m / s. For example, when the starting coordinates of the target are (4, 3, 4) and (11, 12, 8), the pixel deviation algorithm can successfully intercept at a speed of 12 m / s, while the traditional method fails. This fully shows that the pixel deviation algorithm has better real-time adjustment ability and adaptability when dealing with high-speed dynamic targets, and can complete the interception task more effectively.

[0141] Comparative analysis of simulation impact effects of different methods

[0142] Speed adjustment frequency: As can be seen from Table 4, the speed adjustment frequency of the image coordinate transformation + pid method is about 0.2 s, while the speed adjustment frequency of the optical center recognition (pixel deviation algorithm) is about 0.02 s. The speed adjustment frequency of optical center recognition is faster, which means it can more quickly adjust the flight attitude and speed of the UAV according to the target deviation, respond more promptly to the changes of the target, and thus significantly improve the interception efficiency and accuracy.

[0143] Static target interception success rate: The static target interception success rate of the image coordinate transformation + pid method is 80%, while the success rate of optical center recognition is 100%. Optical center recognition performs better in intercepting static targets and can more stably achieve precise strikes on targets.

[0144] Dynamic target interception success rate: The dynamic target interception success rate of the image coordinate transformation + pid method is 60%, and the success rate of optical center recognition is 90%. Optical center recognition also has obvious advantages in the scenario of intercepting dynamic targets, can better adapt to the dynamic changes of the target, and effectively improves the interception success rate.

[0145] In the simulation system, the traditional PID control method and the coordinate transformation of the camera are used and compared with the pixel deviation algorithm. It can be seen that the recognition effect is better than the traditional method in both static and dynamic situations.

[0146] The experimental data of this experiment are only the experimental results under simulation conditions, and the parameters in the actual situation may vary slightly (only one recognition and direct impact are carried out in the static state, and real-time recognition is carried out in the dynamic state).

[0147] Compared with previous physical interception devices, for the present invention, a monocular camera is added to the unmanned aerial vehicle (UAV) and can be used in combination with the flight control. The cost of using a small UAV is only two thousand to three thousand yuan, but the cost of using physical interception devices such as lidar and multi-axis gimbals will reach more than five thousand yuan. If a high-power laser strike device is used, it will start from 100,000 to 300,000 yuan, or even more than one million yuan. When using the net capture method, the mechanical structure is complex, which interferes with the flexibility of the UAV and the interception success rate is low.

[0148] Low-coupling hardware architecture: The system architecture of the present invention makes full use of the data fusion technology of the flight control system and the monocular camera, avoids high-cost depth sensors or complex three-axis gimbal systems, and significantly reduces the overall cost of the hardware. By simplifying the hardware design and computing resource configuration, this solution can be effectively applied in large-scale low-cost target interception scenarios, especially showing strong adaptability and flexibility in UAV interception tasks.

[0149] Full-autonomous control: This solution not only realizes full automation from target recognition to attitude adjustment, but also can complete the interception and tracking of the target in a dynamic environment without manual intervention. With the highly integrated automated control system, the overall operation efficiency is greatly improved. Especially in a complex environment, the automated system can make real-time adjustments and compensations, greatly improving the response speed and interception ability of the UAV.

[0150] Strong robustness and real-time compensation: The proportional control closed-loop strategy adopted by the present invention has strong anti-interference ability and can compensate the deviation in the target movement in real time to ensure that the UAV always maintains accurate tracking of the target. Whether the target is moving rapidly or the environment changes greatly, the system can quickly adjust the flight attitude of the UAV to ensure the smooth completion of the interception task.

[0151] Embodiment 3

[0152] The present invention provides a high-precision and low-cost UAV interceptor, which uses the system in Embodiment 2 and the method in Embodiment 1 to control the UAV for intercepting the target object in the airspace.

[0153] Through innovative visual processing technology and efficient control algorithm design, the present invention effectively optimizes the hardware cost and computing resource usage while ensuring the interception effectiveness, providing an efficient and economical solution for anti-drone technology. This technology can be widely deployed in multiple application scenarios, greatly promoting the development of drone interception technology, and has important application value especially in civil, security, and military fields.

[0154] The embodiments of the present invention have been described in detail above in conjunction with the accompanying drawings. However, the present invention is not limited thereto. Various changes can be made without departing from the spirit of the present invention within the scope of knowledge possessed by those skilled in the art to which the present invention pertains.

Claims

1. A high-precision and low-cost UAV target interception method, characterized in that: The specific steps include: S1. Target recognition: collect images within the field of view, pre-process the collected target images based on opencv, calculate the geometric center of the target, and track the target; S2, attitude closed-loop control: Use the onboard camera to detect the pixel deviation between the target center and the camera optical center in real time, and directly map it to the attitude adjustment control parameters of the drone to control the attitude of the drone; S3, trajectory closed-loop control; The flight control system calculates the speed at the next moment based on the current motion state, drives the drone to continuously approach the target by accelerating to the forward speed, and realizes closed-loop trajectory tracking in combination with deviation correction to ensure that the drone always moves in the direction of the target; S4, real-time adjustment: continuously monitor target deviation and posture adjustment, update control amount and make real-time correction; S5, control the drone to adjust the flight attitude and approach the target; S6. Repeat steps S1-S5 until the target is intercepted.

2. The high-precision and low-cost UAV target interception method according to claim 1 is characterized in that: In step S1, the drone's onboard camera is used to collect images of the target area; opencv is used as the basis for performing image preprocessing operations; the preprocessing steps include filtering denoising, image enhancement and sharpening.

3. The high-precision and low-cost UAV target interception method according to claim 2 is characterized in that: The key features of the target are extracted from the preprocessed image, and the edge detection is performed on the target to calculate the geometric center of the target.

4. The high-precision and low-cost UAV target interception method according to claim 1 is characterized in that: The control parameters for the attitude adjustment of the drone in step S2 include pitch, roll and yaw angular velocity.

5. The high-precision and low-cost UAV target interception method according to claim 4 is characterized in that: The specific steps of step S2 are as follows: S21, setting parameters; according to the model of the drone and the type of the camera carried, set the optical center coordinates (x0, y0) and the camera rotation matrix in advance according to the actual device parameters of the camera; S22, preset proportional control coefficients Kp_roll, Kp_pitch, Kp_yaw; The proportional control coefficient is used to adjust the flight speed of the UAV according to the target position deviation to achieve accurate tracking and interception of the target; S23, preset forward speed v; Set the maximum forward speed in advance. After the drone identifies the target, it will directly accelerate to the maximum speed to hit it. S24, feeding back the identified target geometric center (x, y) and the camera optical center (x0, y0) to the flight control of the drone; S25, calculating the pixel error between the target geometric center (x, y) and the camera optical center (x0, y0), Δx = x-x0, Δy = y-y0; S26. The flight control calculates the flight control parameters of the UAV based on the pixel error (Δx, Δy), including roll, pitch, and yaw speeds, and adjusts the flight speed and attitude of the UAV.

6. The high-precision and low-cost UAV target interception method according to claim 1 is characterized in that: Step S3 includes the following specific steps: S31, the flight control calculates the speed at the next moment by combining the current horizontal rolling angular velocity yaw_speed and the vertical pitch angular velocity vert_speed; S32, horizontal error Δx, controls the lateral motion of the drone in the horizontal plane through the roll angular velocity; S33, vertical error Δy, adjusts the longitudinal motion of the drone through the pitch angular velocity to approach the target; S34. After the speed command is issued, the UAV flies at the new speed and attitude.

7. A high-precision and low-cost UAV target interception system, which uses the method described in any one of claims 1 to 6 to intercept the target, characterized in that: It includes image preprocessing module, target recognition module, flight control module, calculation module and interactive display module; The image preprocessing module is used to preprocess the acquired specific target image; The target recognition module is used to recognize the geometric center of the target in the preprocessed image, and feed the recognized geometric center and camera optical center back to the calculation module; The calculation module calculates the error between the geometric center and the optical center of the camera, and calculates the speed and motion state at the next moment, and outputs it to the flight control module; The flight control module controls the drone to fly at a new speed and motion state, continuously approaching the target; The interactive display module is connected to the UAV for inputting control programs and communicating with the UAV, and obtaining the real-time image information, motion trajectory, and motion status information of the UAV.

8. The high-precision and low-cost UAV target interception system according to claim 7 is characterized in that: The image preprocessing module preprocesses the image by filtering denoising, image enhancement and sharpening.

9. The high-precision and low-cost UAV target interception system according to claim 7 is characterized in that: The target recognition module uses two-dimensional discrete Fourier transform, Gaussian filter denoising, image enhancement algorithm based on multi-scale Retinex theory, and LOG filter based on Gaussian filter and Laplace operator edge detection to extract key target features, detect target edges, and calculate the geometric center (x, y).

10. A high-precision and low-cost drone interceptor, characterized in that: The system described in any one of claims 7 to 9 and the method described in any one of claims 1 to 6 are used to control the drone to intercept targets in the airspace.

Images