A small unmanned aerial vehicle automatic detection tracking and striking system based on computer vision

Abstract

The application discloses a kind of small unmanned aerial vehicle automatic detection tracking attack systems based on computer vision.The system integrates target identification, gimbals control, interceptor, data processing and transmission, energy supply module.Target identification module utilizes YOLOV5s model, through multidimensional data acquisition, training optimization and TensorRT quantization compression, realizes high-precision rapid identification.Gimbals control module accurately controls gimbals tracking target according to identification result.Interceptor module launches interceptor at appropriate time, and controls attack through unique attitude algorithm.Data processing and transmission module ensures data interaction of each module, and energy supply module provides power support.This system can be widely applied in areas with high requirements for low-altitude safety, such as airports and military bases, and can effectively detect, track and attack illegal intruding small unmanned aerial vehicles, ensuring regional safety.

Description

Technical Field

[0001] This invention relates to the field of security monitoring and drone countermeasures technology, specifically to a small drone automatic detection, tracking and strike system based on computer vision. Background Technology

[0002] In recent years, with the rapid development and popularization of drone technology, the number of small drones in use has increased dramatically. Drones have played an important role in aerial photography, logistics, agriculture, and other fields, but they have also brought a series of safety hazards. For example, unauthorized drones entering airport airspace may interfere with the normal take-off and landing of civil flights, seriously threatening aviation safety; in sensitive areas such as military bases and government agencies, drones may be used for reconnaissance, intelligence gathering, and other activities, posing a threat to national and information security; at large-scale events, disorderly drone flights may affect the normal conduct of the event and even cause harm to personnel on site.

[0003] Currently, countermeasures against small drones mainly include electromagnetic interference, laser strikes, and net capture. However, these technologies have certain limitations. Electromagnetic interference may affect surrounding electronic equipment, leading to malfunctions or misoperations; laser strike technology requires high precision and power from equipment, is expensive, and is greatly affected by environmental factors; net capture has a limited range and is difficult to deal with fast-moving drones. Furthermore, most existing countermeasure systems lack automatic detection and tracking capabilities for drones, requiring manual intervention, resulting in slow reaction times and an inability to respond promptly and effectively to emergencies.

[0004] Therefore, this invention develops an efficient and intelligent automatic detection, tracking and strike system for small unmanned aerial vehicles (UAVs). Summary of the Invention

[0005] The purpose of this section is to outline some aspects of embodiments of the present invention and to briefly describe some preferred embodiments. Simplifications or omissions may be made in this section, as well as in the abstract and title of this application, to avoid obscuring the purpose of these documents; however, such simplifications or omissions should not be construed as limiting the scope of the invention.

[0006] In view of the problems existing in the above and / or prior art, the present invention is proposed.

[0007] Therefore, the purpose of this invention is to overcome the shortcomings of the prior art and provide a small unmanned aerial vehicle (UAV) automatic detection, tracking and strike system based on computer vision, characterized in that it includes a target recognition module, a gimbal control module, an interceptor missile module, a data processing and transmission module and an energy supply module.

[0008] The target recognition module is used to identify drones using the YOLOv5s model. It includes a data acquisition and processing unit, a model training and optimization unit, and a model deployment and application unit. The data acquisition and processing unit collects and annotates multi-type drone image data and augments the dataset. The model training and optimization unit trains and debugs the YOLOv5s model. The model deployment and application unit uses the TensorRT framework to quantize and compress the model and apply it.

[0009] The gimbal control module controls the movement of the gimbal to track the drone based on the target recognition result. It includes a gimbal hardware design unit, a control algorithm implementation unit, and a system integration and testing unit. The gimbal hardware design unit models and simulates the gimbal to determine the selection and installation position of each component. The control algorithm implementation unit develops the control algorithm using the Adafruit_servokit library. The system integration and testing unit integrates and tests the hardware and algorithm.

[0010] The interceptor module is used to launch interceptor missiles to strike targets at appropriate times. It includes a hardware design unit, a software design (non-attitude algorithm) unit, and a software design (attitude algorithm) unit. The hardware design unit uses specific development boards, gyroscopes, cameras, and control boards. The software design (non-attitude algorithm) unit uses the Tracker algorithm to track targets. The software design (attitude algorithm) unit uses unique formulas and parameters to set the control servo motor offset angle.

[0011] The data processing and transmission module is responsible for data interaction and processing between modules. It includes a data acquisition and preprocessing unit, a data transmission and interaction unit, and a data storage and management unit. The data acquisition and preprocessing unit acquires and preprocesses data from each module. The data transmission and interaction unit transmits data through a high-speed communication link. The data storage and management unit stores and manages important system data.

[0012] The energy supply module provides power support for the entire system and includes a power selection and configuration unit, an energy management and distribution unit, and a charging and endurance guarantee unit. The power selection and configuration unit selects and configures suitable power equipment. The energy management and distribution unit manages and distributes the power energy. The charging and endurance guarantee unit designs a charging scheme for the power supply.

[0013] Preferably, in the data acquisition and processing unit of the target recognition module, a multi-dimensional data acquisition strategy is used to collect UAV image data covering 7 mainstream aircraft types, including rotorcraft and fixed-wing aircraft. The samples cover flight attitudes with elevation angles from 30° to 90° and include 12 typical environmental conditions such as fog, strong light, and nighttime infrared. The data is labeled using the VOC2012 standard, with the labeling error controlled within ±2 pixels. The dataset is amplified by 3 times using methods such as random flipping, color perturbation, and noise injection.

[0014] Preferably, in the model training and optimization unit of the target recognition module, the YOLOv5s model is selected to build the target detection system. Based on the RTX4060 graphics card, the parameters such as the convolution kernel size, stride, and number of layers of the model are repeatedly adjusted. At the same time, different combinations of activation functions and loss functions are tried. After training, the mean GIoU loss function of the model is less than 0.04, the mean target detection loss is less than 0.016, the precision reaches 0.8, the recall reaches 0.6, and the F1 value is about 0.84.

[0015] Preferably, in the model deployment and application unit of the target recognition module, the TensorRT framework is used to quantize and compress the model, which improves the inference speed by 40% while maintaining 98% of the original model accuracy. The recognition accuracy reaches 97.3% within an effective detection range of 500 meters, and the false alarm rate is less than 0.5 times / hour. When integrated into the low-altitude security platform, the system response time from image acquisition to coordinate output is controlled within 80ms.

[0016] Preferably, in the gimbal hardware design unit of the gimbal control module, AutodeskCAD software is used to perform precise modeling and simulation analysis of the gimbal. The gimbal is mainly composed of modules such as the projectile launch tube, servo motor 1, servo motor 2, vision camera and bearings. Each component is selected and its installation position and angle are determined according to the corresponding requirements and standards.

[0017] Preferably, in the control algorithm implementation unit of the gimbal control module, the Adafruit_servokit library is used to develop the gimbal control algorithm. A "Follow_the_face" class is defined, and servo control parameters self.Servo1 and self.Servo2 are set to correspond to the servos on the x-axis and y-axis, respectively. The initial angle is set to 90 degrees. The x_offset and y_offset parameters are introduced to represent the center position coordinates of the target detection box. The servo angle is adjusted according to these parameters and assigned a weight of 0.02 to achieve precise control.

[0018] Preferably, in the hardware design unit of the interceptor missile module, the RV1106 development board developed by Xinghu Technology is selected as the core hardware, the MPU6050 gyroscope is used as the attitude control module, a high-resolution CSI camera is selected, and a 16-channel PCA9685 control board is used to control the servo motor based on pulse width modulation (PWM) technology.

[0019] The automatic detection, tracking and strike system for small unmanned aerial vehicles based on computer vision according to claim 1 is characterized in that the data acquisition and preprocessing unit of the data processing and transmission module performs normalization, noise reduction and other operations on the image data acquired by the target recognition module; the data transmission and interaction unit realizes data transmission between modules through a USB-CAN adapter; and the data storage and management unit uses a database or file system to store and manage data such as unmanned aerial vehicle recognition datasets, model training parameters, and system operation logs.

[0020] This invention discloses an automatic detection, tracking, and strike system for small unmanned aerial vehicles (UAVs) based on computer vision, which has the following beneficial effects:

[0021] 1. This invention employs the YOLOv5s model for target recognition and improves the model's training performance through multi-dimensional data collection, sample labeling, and dataset augmentation. Simultaneously, the TensorRT framework is used for model quantization and compression, significantly enhancing inference speed while maintaining high recognition accuracy. Within an effective detection range of 500 meters, the recognition accuracy reaches 97.3%, with a false alarm rate of less than 0.5 times per hour, enabling rapid and accurate identification of various small unmanned aerial vehicle (UAV) targets.

[0022] 2. The gimbal control module uses advanced control algorithms to achieve real-time tracking of the UAV based on target recognition results. By precisely calculating the servo adjustment angles, it ensures that the visual camera and interceptor missile launcher are always aligned with the target. The interceptor missile module launches interceptor missiles at the appropriate time; its unique attitude algorithm effectively controls the servo offset angle, improving strike accuracy and enabling timely and effective engagement of illegally intruding UAVs, thus reducing security risks.

[0023] 3. This invention organically integrates multiple modules, including target identification, gimbal control, interceptor missile launch, data processing and transmission, and energy supply. These modules work collaboratively to achieve automated system operation. Furthermore, by optimizing hardware selection and software algorithms, the system's stability and reliability are improved, while costs and energy consumption are reduced.

[0024] 4. The data processing and transmission module can process and transmit data collected by each module in real time, ensuring smooth information flow between all parts of the system. Through high-speed communication links and efficient data processing algorithms, the system can quickly respond to the appearance and movement of UAVs, adjust tracking and strike strategies in a timely manner, and improve the system's reaction speed and combat effectiveness. Attached Figure Description

[0025] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on the structures shown in these drawings without creative effort.

[0026] Figure 1 This is a framework diagram of the computer vision-based automatic detection, tracking, and strike system for small unmanned aerial vehicles (UAVs) of the present invention.

[0027] The objectives, features, and advantages of this invention will be further explained in conjunction with the embodiments and with reference to the accompanying drawings. Detailed Implementation

[0028] The technical solutions of the present invention will now be clearly and completely described with reference to the accompanying drawings of the embodiments of the present invention. It should be understood that the specific embodiments described herein are for illustration and explanation only and are not intended to limit the present invention.

[0029] To make the above-mentioned objects, features and advantages of the present invention more apparent and understandable, the specific embodiments of the present invention will be described in detail below with reference to the examples in the specification.

[0030] Many specific details are set forth in the following description in order to provide a full understanding of the invention. However, the invention may also be practiced in other ways different from those described herein, and those skilled in the art can make similar extensions without departing from the spirit of the invention. Therefore, the invention is not limited to the specific embodiments disclosed below.

[0031] Secondly, the term "one embodiment" or "embodiment" as used herein refers to a specific feature, structure, or characteristic that may be included in at least one implementation of the present invention. The phrase "in one embodiment" appearing in different places in this specification does not necessarily refer to the same embodiment, nor is it a single or selective embodiment that is mutually exclusive with other embodiments.

[0032] This invention provides a computer vision-based automatic detection, tracking, and strike system for small unmanned aerial vehicles (UAVs). The system mainly consists of a target recognition module, a gimbal control module, an interceptor module, a data processing and transmission module, and a power supply module. These modules work collaboratively to achieve automatic detection, tracking, and strike of UAVs. The target recognition module uses a YOLOv5s model to identify the UAV; the gimbal control module controls the gimbal movement based on the target recognition results to track the UAV; the interceptor module launches interceptor missiles to strike the target at appropriate times; the data processing and transmission module is responsible for data interaction and processing between the modules; and the power supply module provides power support for the entire system.

[0033] In a preferred embodiment of the present invention, the target recognition module includes a data acquisition and processing unit, a model training and optimization unit, and a model deployment and application unit.

[0034] The data acquisition and processing unit collects UAV image data covering seven mainstream aircraft types, including rotorcraft and fixed-wing aircraft, using a multi-dimensional data acquisition strategy. The samples cover flight attitudes with elevation angles ranging from 30° to 90° and include 12 typical environmental conditions such as fog, strong light, and nighttime infrared. The VOC2012 standard is adopted, and semi-automated tools are used to complete bounding box annotation and category labeling, ensuring that the annotation error is controlled within ±2 pixels. The dataset is tripled using techniques such as random flipping, color perturbation, and noise injection to ultimately form a high-quality training set.

[0035] In the model training and optimization unit, the YOLOv5s model was selected to construct the object detection system, and the model underwent 15 rounds of training. During the training process, based on an RTX 4060 graphics card, parameters such as the model's convolutional kernel size, stride, and number of layers were repeatedly adjusted, while different combinations of activation and loss functions were tried. After training, the model's mean GIoU loss function was less than 0.04, the mean object detection loss was less than 0.016, the precision reached 0.8, the recall reached 0.6, and the F1 score was approximately 0.84, demonstrating accurate and reliable identification and localization of drones in complex environments.

[0036] In the model deployment and application unit, the TensorRT framework is used to quantize and compress the model, improving inference speed by 40% while maintaining 98% of the original model accuracy. Actual testing shows that the system achieves a 97.3% recognition accuracy within a 500-meter effective detection range, with a false alarm rate of less than 0.5 times per hour. When integrated into a low-altitude security platform, the system response time from image acquisition to coordinate output is controlled within 80ms, meeting the real-time requirements of civil aviation organizations for airport airspace protection.

[0037] In a preferred embodiment of the present invention, the gimbal control module includes a gimbal hardware design unit, a control algorithm implementation unit, and a system integration and testing unit.

[0038] In the gimbal hardware design unit, AutodeskCAD software is used for precise modeling and simulation analysis of the gimbal. The gimbal mainly consists of modules such as the projectile launch tube, servo motor 1, servo motor 2, vision camera, and bearings. The projectile launch tube design fully considers the projectile size, launch power requirements, and launch stability and accuracy; servo motor 1 and servo motor 2 control the angles in the X and Y axes respectively, and their selection and installation positions are determined through extensive mechanical calculations and motion simulations; the vision camera uses a USB camera module, which is comprehensively evaluated and optimized in terms of parameters such as image sensor pixel size, resolution, frame rate, lens focal length, aperture, and field of view through cooperation with multiple manufacturers, and the installation position and angle are optimized; the bearing module plays a crucial supporting and lubricating role in the movement of the gimbal.

[0039] In the control algorithm implementation unit, the Adafruit_servokit library is used to develop the gimbal control algorithm. A "Follow_the_face" class (used for testing) is defined, setting servo control parameters self.Servo1 and self.Servo2, corresponding to the servos on the x and y axes respectively, with an initial angle of 90 degrees. The x_offset and y_offset parameters are introduced, representing the coordinates of the target detection box center. The servo adjustment angle is calculated based on these coordinates and assigned a certain weight (0.02 for this gimbal) to achieve precise control. Testing shows that this algorithm has high accuracy in face localization and can effectively achieve target tracking and aiming.

[0040] The system integration and testing unit integrates the gimbal hardware and control algorithms, using an i5-13490F + NVIDIA RTX 3060 Ti industrial control host as the core computing unit, paired with a Raspberry Pi 5 and a PCA9685 control board. Testing showed that the YOLOv5s model maintained a stable inference frame rate of over 240 FPS under Windows 11, supporting simultaneous processing of four 4K video streams. The system was tested in various environments to verify its drone tracking performance in complex scenarios, ensuring stable gimbal operation under diverse conditions.

[0041] In a preferred embodiment of the present invention, the interceptor missile module includes a hardware design unit, a software design (non-attitude algorithm) unit, and a software design (attitude algorithm) unit.

[0042] The hardware design unit utilizes the RV1106 development board developed by Xinghu Technology as its core hardware. This board integrates a high-performance processor chip, possessing powerful computing capabilities and abundant hardware resources. It supports various storage devices and communication interfaces, allowing for flexible configuration and secondary development according to project requirements. An MPU6050 gyroscope is used as the attitude control module to accurately measure the angular velocity and acceleration information of the interceptor missile in real time, acquiring attitude parameters. A high-resolution CSI camera is used to acquire high-quality target image information under different lighting and weather conditions. A 16-channel PCA9685 control board, based on advanced pulse width modulation (PWM) technology, enables precise control of the servos, meeting the needs of different flight phases and mission scenarios.

[0043] In the software design (non-attitude algorithm) unit, the Tracker algorithm is used on the RV1106 development board for target tracking. By extracting features such as target color, texture, and shape, the target position is tracked in a continuous image sequence. After the data is accurately aimed and identified by the control gimbal, it is transmitted to the RV1106 development board. The board activates its processing capabilities to select and track the target, calculate the difference between the target frame and the center of the image, and transmit the information to the STM32 development board. The STM32 development board controls the missile's aiming and simultaneously uses the MPU6050 sensor for attitude control to ensure that the interceptor missile flies stably and accurately toward the target.

[0044] In the software design (attitude algorithm) unit, the attitude algorithm employs unique formulas and parameter settings, involving general formulas, constraints, nonlinear response terms, linear stability terms, and Y-axis stability terms. By processing the angle value α and angular acceleration value β collected by the MPU6050, and combining multiple k-value weighting functions, precise control of the actual offset angle of the servo motor is achieved. A method of simulating the host computer's working process is used, employing an ANOTC host computer to simulate the detection state of the MPU6050, optimizing and adjusting the system to ensure the effectiveness and stability of attitude control.

[0045] In a preferred embodiment of the present invention, the data processing and transmission module includes a data acquisition and preprocessing unit, a data transmission and interaction unit, and a data storage and management unit.

[0046] In the data acquisition and preprocessing unit, image data acquired by the target recognition module, sensor data obtained by the gimbal control module, and attitude data from the interceptor missile module are all collected in this module. The collected data undergoes preprocessing, such as normalization and noise reduction of image data, to improve data quality and provide reliable data support for subsequent processing.

[0047] In the data transmission and interaction unit, high-speed communication links, such as a USB-CAN adapter, enable data transmission and interaction between modules. A stable communication connection is established between the host computer and the slave computer to ensure the timing consistency of target tracking, gimbal adjustment, and interceptor missile launch, making the entire system work more efficiently.

[0048] The data storage and management unit stores important data generated during system operation, such as UAV identification datasets, model training parameters, and system operation logs. Appropriate data storage technologies, such as databases or file systems, are employed to effectively manage the data, facilitating subsequent querying and analysis, and providing data support for system optimization and improvement.

[0049] In a preferred embodiment of the present invention, the energy supply module includes a power supply selection and configuration unit. 、 Energy Management and Distribution Unit 、 Charging and battery life protection unit.

[0050] In the power supply selection and configuration unit, appropriate power supply equipment is selected based on the power consumption requirements of each module in the system. For the gimbal control module and the interceptor missile module, a stable and reliable power supply, such as a lithium battery pack, is used to ensure stable operation under different environments. The power supply is rationally configured to meet the system's long-term operational needs.

[0051] The energy management and distribution unit includes an energy management system designed to effectively manage and distribute power. Based on the operating status and needs of each module, the system intelligently adjusts energy distribution to avoid energy waste, improve energy efficiency, and ensure the system can still operate normally even with limited energy resources.

[0052] The charging and endurance protection unit includes a charging solution designed for the power supply equipment, such as using solar charging panels or external power sources, to ensure the system's endurance. In outdoor or long-term working scenarios, it ensures the system can continuously obtain energy and maintain normal operation.

[0053] The computer vision-based automatic detection, tracking, and strike system for small unmanned aerial vehicles (UAVs) of this invention achieves efficient detection, tracking, and strike capabilities for UAVs through the coordinated operation of its various modules. It has broad application prospects in the fields of civilian security and military applications, effectively addressing the security threats posed by UAVs and ensuring public safety and the smooth conduct of military operations.

[0054] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit it. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can be made to the technical solutions of the present invention without departing from the spirit and scope of the technical solutions of the present invention, and all such modifications or substitutions should be covered within the scope of the claims of the present invention.

Claims

1. A computer vision-based automatic detection, tracking, and strike system for small unmanned aerial vehicles (UAVs), characterized in that, It includes a target identification module, a gimbal control module, an interceptor missile module, a data processing and transmission module, and an energy supply module; The target recognition module is used to identify drones using the YOLOv5s model. It includes a data acquisition and processing unit, a model training and optimization unit, and a model deployment and application unit. The data acquisition and processing unit collects and annotates multi-type drone image data and augments the dataset. The model training and optimization unit trains and debugs the YOLOv5s model. The model deployment and application unit uses the TensorRT framework to quantize and compress the model and apply it. The gimbal control module controls the movement of the gimbal to track the drone based on the target recognition result. It includes a gimbal hardware design unit, a control algorithm implementation unit, and a system integration and testing unit. The gimbal hardware design unit models and simulates the gimbal to determine the selection and installation location of each component; the control algorithm implementation unit develops the control algorithm using the Adafruit_servokit library; and the system integration and testing unit integrates and tests the hardware and algorithm. The interceptor module is used to launch interceptor missiles to strike targets at appropriate times. It includes a hardware design unit, a software design (non-attitude algorithm) unit, and a software design (attitude algorithm) unit. The hardware design unit uses specific development boards, gyroscopes, cameras, and control boards. The software design (non-attitude algorithm) unit uses the Tracker algorithm to track targets. The software design (attitude algorithm) unit uses unique formulas and parameters to set the control servo motor offset angle. The data processing and transmission module is responsible for data interaction and processing between modules. It includes a data acquisition and preprocessing unit, a data transmission and interaction unit, and a data storage and management unit. The data acquisition and preprocessing unit acquires and preprocesses data from each module. The data transmission and interaction unit transmits data through a high-speed communication link. The data storage and management unit stores and manages important system data. The energy supply module provides power support for the entire system and includes a power selection and configuration unit, an energy management and distribution unit, and a charging and endurance guarantee unit. The power selection and configuration unit selects and configures suitable power equipment. The energy management and distribution unit manages and distributes the power energy. The charging and endurance guarantee unit designs a charging scheme for the power supply.

2. The automatic detection, tracking, and strike system for small unmanned aerial vehicles based on computer vision according to claim 1, characterized in that, In the data acquisition and processing unit of the target recognition module, a multi-dimensional data acquisition strategy collects UAV image data covering seven mainstream aircraft types, including rotorcraft and fixed-wing aircraft. The samples cover flight attitudes with elevation angles from 30° to 90° and include 12 typical environmental conditions such as fog, strong light, and nighttime infrared. The data is labeled using the VOC2012 standard, with the labeling error controlled within ±2 pixels. The dataset is augmented by 3 times using methods such as random flipping, color perturbation, and noise injection.

3. The automatic detection, tracking, and strike system for small unmanned aerial vehicles based on computer vision according to claim 1, characterized in that, In the model training and optimization unit of the target recognition module, the YOLOv5s model is selected to build the target detection system. Based on the RTX4060 graphics card, the parameters of the model such as convolution kernel size, stride, and number of layers are repeatedly adjusted. At the same time, different combinations of activation functions and loss functions are tried. After training, the mean GIoU loss function of the model is less than 0.04, the mean target detection loss is less than 0.016, the precision reaches 0.8, the recall reaches 0.6, and the F1 value is about 0.

84.

4. The automatic detection, tracking, and strike system for small unmanned aerial vehicles based on computer vision according to claim 1, characterized in that, In the model deployment and application unit of the target recognition module, the TensorRT framework is used to quantize and compress the model, which improves the inference speed by 40% while maintaining 98% of the original model accuracy. The recognition accuracy reaches 97.3% within an effective detection range of 500 meters, and the false alarm rate is less than 0.5 times / hour. When integrated into the low-altitude security platform, the system response time from image acquisition to coordinate output is controlled within 80ms.

5. The automatic detection, tracking, and strike system for small unmanned aerial vehicles based on computer vision according to claim 1, characterized in that, In the gimbal hardware design unit of the gimbal control module, AutodeskCAD software is used to perform precise modeling and simulation analysis of the gimbal. The gimbal is mainly composed of modules such as the projectile launch tube, servo motor 1, servo motor 2, vision camera and bearings. Each component is selected and its installation position and angle are determined according to the corresponding requirements and standards.

6. The automatic detection, tracking, and strike system for small unmanned aerial vehicles based on computer vision according to claim 1, characterized in that, In the control algorithm implementation unit of the gimbal control module, the Adafruit_servokit library is used to develop the gimbal control algorithm. The "Follow_the_face" class is defined, and the servo control parameters self.Servo1 and self.Servo2 are set to correspond to the servos on the x-axis and y-axis, respectively. The initial angle is set to 90 degrees. The x_offset and y_offset parameters are introduced to represent the center position coordinates of the target detection box. The servo angle is adjusted according to these parameters and a weight of 0.02 is assigned to achieve precise control.

7. The automatic detection, tracking, and strike system for small unmanned aerial vehicles based on computer vision according to claim 1, characterized in that, In the hardware design unit of the interceptor missile module, the RV1106 development board developed by Xinghu Technology is selected as the core hardware, the MPU6050 gyroscope is used as the attitude control module, a high-resolution CSI camera is selected, and a 16-channel PCA9685 control board is used to control the servo motor based on pulse width modulation (PWM) technology.

8. The automatic detection, tracking, and strike system for small unmanned aerial vehicles based on computer vision according to claim 1, characterized in that, The data acquisition and preprocessing unit of the data processing and transmission module performs normalization, noise reduction, and other operations on the image data acquired by the target recognition module. The data transmission and interaction unit realizes data transmission between modules through a USB-CAN adapter. The data storage and management unit uses a database or file system to store and manage data such as UAV recognition datasets, model training parameters, and system operation logs.

Images