Precise plant protection unmanned aerial vehicle system based on machine vision

Abstract

The invention discloses a precise plant protection unmanned aerial vehicle system based on machine vision, and relates to the technical field of unmanned aerial vehicles. The unmanned aerial vehicle has the beneficial effects that the unmanned aerial vehicle body, the precise spraying holder carried below the unmanned aerial vehicle body, the flight control system and the visual recognition system are arranged, precise recognition and fixed-point spraying are achieved during use, machine vision and the two-degree-of-freedom spraying holder are combined, pests and diseases can be precisely attacked, and the spraying efficiency is improved. The traditional large-area and undifferentiated spraying mode is thoroughly changed; as only the identified pest and disease damage area is sprayed, the environmental pollution caused by pesticide abuse is effectively reduced, and the farmland ecology is protected; a distributed architecture is adopted, flight control and holder control are separated, mutual interference between system tasks is avoided, and the flight stability of the unmanned aerial vehicle during execution of complex recognition and spraying tasks is guaranteed.

Description

Technical Field

[0001] This invention relates to the field of drone technology, specifically to a precision agricultural drone system based on machine vision. Background Technology

[0002] Currently, pesticide spraying is widely used for pest and disease control in agricultural production. Traditional control methods mostly involve manual backpack spraying or indiscriminate, large-area spraying by large machinery. This method has two major drawbacks: First, it consumes huge amounts of pesticides, which not only increases agricultural costs but also leads to a large amount of pesticides being lost into the soil and water, causing serious environmental pollution problems such as soil acidification and eutrophication. Second, it is inefficient and labor-intensive.

[0003] To address the aforementioned issues, large agricultural drones, such as the DJI MG-1, have emerged on the market. While these drones achieve automated spraying and improve operational efficiency, they were originally designed for large-scale farmland. Therefore, they generally suffer from the following drawbacks:

[0004] 1. Expensive and bulky: The entire set of equipment is expensive, and the machine is large and heavy, making it unsuitable for operation in small-scale, scattered farmland or plots. It is not cost-effective for ordinary small farmers.

[0005] 2. Insufficient spraying precision: Most existing plant protection drones still use a wide-area spraying mode, which cannot identify the specific location of crop pests and diseases, resulting in pesticide waste and potential harm to healthy plants. This fails to fundamentally solve the problem of excessive pesticide and fertilizer application.

[0006] Therefore, this invention requires the design of a machine vision-based precision agricultural drone system to solve the aforementioned problems. Summary of the Invention

[0007] The purpose of this invention is to provide an agricultural drone that is suitable for small-scale farmland, can accurately identify and target pests and diseases, and is cost-effective, thus solving the problems mentioned in the background art.

[0008] To achieve the above objectives, the present invention provides the following technical solution: a precision agricultural drone system based on machine vision, comprising:

[0009] A drone body;

[0010] A precision spraying platform mounted under the aircraft;

[0011] A flight control system is used to achieve stable flight control and intelligent operation of the UAV;

[0012] A visual recognition system for real-time capture and analysis of crop images;

[0013] The drone body includes a drone frame, an installation top plate is installed inside the drone frame, four reinforcing rods are fixedly connected to the bottom of the installation top plate, a drive motor is fixedly connected to one end of each of the four reinforcing rods, a rotating rotor is fixedly connected to the output end of each of the drive motors, a bottom positioning block is fixedly connected to the bottom of each of the four reinforcing rods, and a bottom positioning plate is fixedly connected to the bottom of each of the bottom positioning blocks.

[0014] The precision spraying gimbal includes an industrial camera and a first brushless motor. A gimbal fixing frame is fixedly connected to the bottom of the bottom positioning plate. The first brushless motor is installed at the bottom of the gimbal fixing frame. A rotating plate is fixedly connected to the output end of the first brushless motor. A bottom connecting frame is fixedly connected to the outside of the rotating plate. Two limiting plates are fitted on the outside of the bottom connecting frame. Connecting side plates are fixedly connected to the bottom of the two limiting plates. A camera mounting bracket is fixedly connected to the top of the connecting side plates. An industrial camera is fixedly connected inside the camera mounting bracket. Storage boxes are fixedly connected to both sides of the camera mounting bracket. Spray nozzles are fixedly connected to the bottom of each storage box.

[0015] In a preferred embodiment of the present invention, a first rotating column and a second rotating column are rotatably connected inside the bottom connecting frame. The second rotating column is located on the side of the first rotating column away from the battery. A second brushless motor that works with the second rotating column is fixedly connected to the outside of the bottom connecting frame. A limiting bolt that works with the first rotating column is installed on the outside of the bottom connecting frame.

[0016] In a preferred embodiment of the present invention, a light is fixedly connected to the top of the connecting side plate and to one side of one of the storage boxes, and a feeding port is installed on the top of the storage box.

[0017] In a preferred embodiment of the present invention, a battery is fixedly connected to the top of the camera mounting bracket, an electrical control box is fixedly connected to the bottom of the gimbal fixing frame, and a motor reinforcement plate is fixedly connected to one side of the electrical control box and located outside the first brushless motor.

[0018] In a preferred embodiment of the present invention, four positioning rods are fixedly connected at equal intervals inside the drone frame and around the mounting top plate. Reinforcing connecting sleeves are fitted at the connection points between the four positioning rods and the drone frame. Limiting frames are installed at the bottom of the drive motors, and one end of each positioning rod extends into the limiting frame.

[0019] In a preferred embodiment of the present invention, the electrical control box is equipped with a main controller, and the electrical control box, the first brushless motor, the second brushless motor, the drive motor, the battery and the lighting are all electrically connected to the main controller.

[0020] In a preferred embodiment of the present invention, the flight control system includes a main control module, an attitude control module, a communication and coordination module, and a control mode coordination module. The output terminals of the attitude control module, the communication and coordination module, and the control mode coordination module are all communicatively connected to the input terminal of the main control module. The communication and coordination module is bidirectionally communicatively connected to the visual recognition system. The flight control system is bidirectionally communicatively connected to the main controller.

[0021] In a preferred embodiment of the present invention, the main control module is used to employ an STMF series high-performance microcontroller as the main controller, which is responsible for receiving sensor data, running flight control algorithms, and sending instructions to various execution components.

[0022] The attitude control module is used to perform closed-loop control of the UAV's flight attitude using a PID control algorithm, and to perform fusion processing of sensor data such as gyroscope and accelerometer using a Kalman filter algorithm.

[0023] The communication and coordination module is used to control the main controller to communicate with the electronic speed controllers of the four rotors via the CAN bus protocol, ensuring the real-time performance and accuracy of control commands.

[0024] The control mode coordination module is used to support a combination of manual control by remote control and autonomous control by host computer.

[0025] In a preferred embodiment of the present invention, the visual recognition system includes a hardware configuration control module, an image processing and recognition process control module, and a target positioning module. The output end of the image processing and recognition process control module is communicatively connected to the input end of the target positioning module. The hardware configuration control module and the image processing and recognition process control module are bidirectionally connected. The visual recognition system is bidirectionally connected to the main controller.

[0026] In a preferred embodiment of the present invention, the hardware configuration control module consists of a high-performance MINIPC and a high-frame-rate industrial camera.

[0027] The image processing and recognition process control module is used to acquire image data from an industrial camera and transmit it to MINIPC. MINIPC is based on the OpenCV framework and uses a combination of lightweight convolutional neural networks and traditional image processing methods to perform real-time image analysis.

[0028] After the target is identified, the visual recognition system calculates the pixel coordinates of the target in the image.

[0029] Compared with the prior art, the beneficial effects of the present invention are:

[0030] This invention comprises a drone body, a precision spraying gimbal mounted beneath the drone body, a flight control system, and a visual recognition system, which, when in use, achieve the following:

[0031] 1. Precise identification and targeted spraying: Combining machine vision and a dual-degree-of-freedom spraying platform, it can accurately target pests and diseases, completely changing the traditional large-area, indiscriminate spraying method;

[0032] 2. Significantly reduce pesticide use: Since spraying is only applied to identified pest and disease areas, the amount of pesticides used can be greatly reduced, effectively mitigating environmental pollution caused by pesticide abuse and protecting farmland ecology;

[0033] 3. Controllable cost and wide applicability: Compared with large and high-end agricultural drones, the overall design and selection of this solution focuses more on cost-effectiveness. It is small in size and light in weight, making it particularly suitable for use in small-scale, scattered, and complex terrain farmland, filling a market gap.

[0034] 4. High stability and reliability: The distributed architecture separates flight control from gimbal control, avoiding mutual interference between system tasks and ensuring the flight stability of the UAV when performing complex identification and spraying tasks. Attached Figure Description

[0035] Figure 1 This is a schematic diagram of the overall structure of a machine vision-based precision agricultural drone system according to the present invention. Figure 1 ;

[0036] Figure 2 This is a schematic diagram of the overall structure of a machine vision-based precision agricultural drone system according to the present invention. Figure 2 ;

[0037] Figure 3 This is an enlarged schematic diagram of the drone body structure of a precision agricultural drone system based on machine vision according to the present invention;

[0038] Figure 4 This is a magnified schematic diagram of the precision spraying gimbal structure of a machine vision-based precision agricultural drone system according to the present invention.

[0039] Figure 5 This is a schematic diagram of the hardware connection of a precision agricultural drone system based on machine vision according to the present invention.

[0040] Figure 6This is a flowchart of the PID algorithm for a precision agricultural drone system based on machine vision, according to the present invention.

[0041] Figure 7 This is a schematic diagram of the system software architecture of a precision agricultural drone system based on machine vision according to the present invention;

[0042] Figure 8 This is a schematic diagram of the image processing flow of a precision agricultural drone system based on machine vision according to the present invention.

[0043] Figure 9 This is a schematic diagram of the gimbal control board of a machine vision-based precision agricultural drone system according to the present invention.

[0044] In the picture:

[0045] 1. Drone frame; 11. Positioning rod; 12. Reinforcing connecting sleeve;

[0046] 2. Gimbal mounting frame; 21. Electrical control box; 22. First brushless motor; 23. Rotating plate; 24. Bottom connecting frame; 25. Limiting plate; 26. Limiting bolt; 27. Connecting side plate; 28. Motor reinforcement plate; 29. ​​Second brushless motor;

[0047] 3. Install top plate; 31. Reinforcing rod; 32. Drive motor; 33. Limiting frame; 34. Rotating rotor; 35. Bottom positioning block; 36. Bottom positioning plate;

[0048] 4. Battery; 41. Camera mounting bracket; 42. Industrial camera; 43. First rotating column; 44. Lighting lamp; 45. Storage box; 46. Feed port; 47. Second rotating column; 48. Internal connecting plate; 49. Spray nozzle. Detailed Implementation

[0049] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0050] Please see Figures 1-9 This invention provides a technical solution: a precision agricultural drone system based on machine vision, comprising:

[0051] The drone airframe adopts a quadcopter structure, which has good flight stability and controllability. Some parts of the airframe frame adopt a hollow design, which effectively reduces the overall weight while ensuring structural strength.

[0052] A precision spraying gimbal mounted under the machine body is the core execution component of this system. The gimbal is driven by two mutually perpendicular brushless motors, realizing two degrees of freedom of movement: pitch and yaw. The gimbal integrates a spray nozzle 49 and an industrial camera 42. By precisely controlling the two brushless motors, the nozzle and camera can be pointed in any direction, thereby realizing the fixed-point and directional spraying of pesticides.

[0053] A flight control system is used to achieve stable flight control and intelligent operation of UAVs. This includes integrating sensor data and running control algorithms through the main control module, using the attitude control module to ensure the stability of flight attitude, interacting with the vision recognition system and execution components in real time through the communication and collaboration module, and supporting flexible switching between manual and autonomous control to ensure the accurate execution of plant protection tasks.

[0054] A visual recognition system is used to capture and analyze crop images in real time. It acquires high-quality image data through a hardware configuration control module, identifies pests or specific plant targets through an image processing and recognition process control module, and calculates the physical coordinates of the target through a target positioning module, providing precise spraying positioning information for drones.

[0055] The drone body includes a drone frame 1. A mounting top plate 3 is installed inside the drone frame 1. Four reinforcing rods 31 are fixedly connected to the bottom of the mounting top plate 3. A drive motor 32 is fixedly connected to one end of each of the four reinforcing rods 31. A rotating rotor 34 is fixedly connected to the output end of each of the drive motors 32. A bottom positioning block 35 is fixedly connected to the bottom of each of the four reinforcing rods 31. A bottom positioning plate 36 is fixedly connected to the bottom of each bottom positioning block 35.

[0056] The precision spraying gimbal includes an industrial camera 42 and a first brushless motor 22. A gimbal fixing frame 2 is fixedly connected to the bottom of the bottom positioning plate 36. The first brushless motor 22 is installed at the bottom of the gimbal fixing frame 2. A rotating plate 23 is fixedly connected to the output end of the first brushless motor 22. A bottom connecting frame 24 is fixedly connected to the outside of the rotating plate 23. Two limiting plates 25 are fitted on the outside of the bottom connecting frame 24. A connecting side plate 27 is fixedly connected to the bottom of the two limiting plates 25. A camera mounting bracket 41 is fixedly connected to the top of the connecting side plate 27. An industrial camera 42 is fixedly connected inside the camera mounting bracket 41. Storage boxes 45 are fixedly connected to both sides of the camera mounting bracket 41. Spray nozzles 49 are fixedly connected to the bottom of each storage box 45. Spraying liquid is stored in the storage box 45, and the spray nozzles 49 are used for spraying operations.

[0057] Please see Figures 1-4In this design, the bottom connecting frame 24 is internally connected to a first rotating column 43 and a second rotating column 47. The second rotating column 47 is located on the side of the first rotating column 43 away from the battery 4. The bottom connecting frame 24 is fixedly connected to a second brushless motor 29 that works with the second rotating column 47. The bottom connecting frame 24 is also equipped with a limiting bolt 26 that works with the first rotating column 43. The second brushless motor 29 rotates, causing the second rotating column 47 inside the bottom connecting frame 24 to rotate. Under the connection of the internal connecting plate 48, the first rotating column 43 rotates. Limited by the limiting plate 25, the angle of the connecting side plate 27 is slightly adjusted, thereby improving the operation monitoring, spraying range, and the flexibility of the equipment.

[0058] A light 44 is fixedly connected to the top of the connecting side plate 27 and to one side of one of the storage boxes 45. A feeding port 46 is installed on the top of the storage box 45, through which the required spray liquid is replenished into the storage box 45 in a timely manner.

[0059] Please see Figures 1-4 In this solution, a battery 4 is fixedly connected to the top of the camera mounting bracket 41, and an electrical control box 21 is fixedly connected to the bottom of the gimbal mounting frame 2. A motor reinforcement plate 28 is fixedly connected to one side of the electrical control box 21 and outside the first brushless motor 22. The motor reinforcement plate 28 is used to reinforce the connection between the first brushless motor 22 and the electrical control box 21, thereby improving the installation stability of the equipment.

[0060] Four positioning rods 11 are fixedly connected at equal intervals inside the drone frame 1 and around the mounting top plate 3. Each of the four positioning rods 11 is fitted with a reinforcing connecting sleeve 12 at the connection point with the drone frame 1. Each of the drive motors 32 has a limit frame 33 installed at the bottom. One end of each positioning rod 11 extends into the limit frame 33 to reinforce the connection between the positioning rod 11 and the drive motor 32.

[0061] The electrical control box 21 is equipped with a main controller. The electrical control box 21, the first brushless motor 22, the second brushless motor 29, the drive motor 32, the battery 4, and the lighting 44 are all electrically connected to the main controller. The main controller is used to control the operation of the electrical control box 21, the first brushless motor 22, the second brushless motor 29, the drive motor 32, the battery 4, and the lighting 44, thereby realizing unified management of electrical equipment.

[0062] Please see Figures 1-9 In this scheme, the flight control system includes a main control module, an attitude control module, a communication and coordination module, and a control mode coordination module. The output terminals of the attitude control module, the communication and coordination module, and the control mode coordination module are all connected to the input terminal of the main control module. The communication and coordination module is bidirectionally connected to the vision recognition system. The flight control system is bidirectionally connected to the main controller.

[0063] The main control module uses an STM32F407 series high-performance microcontroller as the main controller, which is responsible for receiving sensor data, running flight control algorithms, and sending instructions to various execution components.

[0064] The attitude control module is used to perform closed-loop control of the UAV's pitch, roll, and yaw using a PID control algorithm, and to fuse data from sensors such as gyroscopes and accelerometers using a Kalman filter algorithm to obtain more accurate attitude information and ensure the stability of the UAV during hovering and flight.

[0065] The specific operation process of the PID control algorithm is as follows:

[0066] Data Acquisition and Filtering: The flight control system measures the UAV's angular velocity and linear acceleration in real time using onboard gyroscopes and accelerometers. Due to sensor noise and drift, the raw data cannot be used directly. The Kalman filter algorithm is used as an optimal estimator here, and data fusion is performed through the following steps:

[0067] Prediction: Based on the optimal attitude estimate of the previous moment and the angular velocity measured by the gyroscope, predict the attitude at the current moment;

[0068] Update: The predicted attitude is weighted and fused with the attitude information measured by the accelerometer, and the Kalman gain is dynamically adjusted for confidence based on the sensor noise characteristics;

[0069] Output: The final output is a smoother and more accurate estimate of the current attitude of the UAV, including pitch, roll and yaw angles;

[0070] PID control decision: The flight control system sets the desired attitude angles, such as the desired pitch and roll angles of 0 degrees when hovering;

[0071] The PID controller receives two inputs: the desired attitude angle; and the actual attitude angle after Kalman filtering.

[0072] The controller calculates the error between the two and outputs a control quantity according to the following formula:

[0073] Control quantity = Kp * error + Ki * integral (error) + Kd * derivative of error

[0074] Proportional term: Instantaneous response error, which determines the reaction speed;

[0075] Integral term: Accumulates historical errors and eliminates steady-state errors;

[0076] Differential term: predicts the trend of error change, suppresses overshoot, and increases stability;

[0077] The coefficients Kp, Ki, and Kd have been pre-tuned and optimized for the dynamic model of this UAV;

[0078] Command execution: The control quantity output by the PID controller is converted into control signals from the four electronic speed controllers. For example, when it is necessary to correct a positive pitch error, the system will increase the speed of the rear rotor and decrease the speed of the front rotor to generate a pitching torque until the actual pitch angle matches the desired value.

[0079] The communication and coordination module is used to control the main controller to communicate with the electronic speed controllers of the four rotors via the CAN bus protocol, ensuring the real-time performance and accuracy of control commands. At the same time, the main controller communicates with the vision recognition system via the UART serial port to receive target coordinate information.

[0080] The control mode coordination module supports a combination of manual control by remote control and autonomous control by the host PC, balancing flexibility and intelligence.

[0081] Please see Figures 1-9 In this solution, the visual recognition system includes a hardware configuration control module, an image processing and recognition process control module, and a target positioning module. The output end of the image processing and recognition process control module is connected to the input end of the target positioning module. The hardware configuration control module and the image processing and recognition process control module are connected in two directions. The visual recognition system is connected in two directions to the main controller.

[0082] The hardware configuration control module consists of a high-performance MINIPC and a high-frame-rate industrial camera 42. The industrial camera 42 is fixed on the precision spraying gimbal and moves with the gimbal to capture real-time images of crops below the drone.

[0083] The image processing and recognition process control module is used to acquire image data from an industrial camera and transmit it to MINIPC. MINIPC is based on the OpenCV framework and uses a combination of lightweight convolutional neural networks and traditional image processing methods such as color channel subtraction, binarization, and morphological operations to perform real-time image analysis and accurately identify the location of pests, diseases, or specific plants.

[0084] After the target is identified, the visual recognition system calculates the pixel coordinates of the target in the image and obtains its precise physical coordinates relative to the drone body through coordinate transformation.

[0085] Following the classic pinhole camera model and coordinate transformation principles in computer vision, the specific steps are as follows:

[0086] Obtaining pixel coordinates: After the image processing algorithm recognizes the target, it determines its position in the image frame. It is usually represented by the pixel coordinates (u,v) of the center point of the target area, where u is the column coordinate and v is the row coordinate. The origin is located at the upper left corner of the image.

[0087] Camera coordinate system transformation: Using the camera intrinsic parameter matrix K obtained from camera calibration, the pixel coordinates (u,v) are transformed to coordinates (Xc,Yc,Zc) in the camera coordinate system;

[0088] The conversion formula is based on the pinhole imaging model:

[0089] [u,v,1]^T=K*[Xc,Yc,Zc]^T

[0090] The intrinsic parameter matrix K contains inherent parameters such as the camera's focal length (fx, fy) and principal point coordinates (cx, cy). Since a monocular camera cannot directly obtain depth information Zc, Zc is set to a known fixed value, namely the preset average height H from the industrial camera lens to the crop canopy. This height H is determined by the drone's hovering height and gimbal angle. Therefore, by solving the above equations inversely, the three-dimensional coordinates (Xc, Yc, H) of the target in the camera coordinate system can be obtained.

[0091] Gimbal coordinate system transformation: The camera coordinate system is fixed on the camera and needs to be transformed to the center of the gimbal. This is described by a rotation matrix R_cp and a translation vector T_cp. These parameters are known after the mechanical design is completed.

[0092] The target's coordinates (Xp, Yp, Zp) in the gimbal coordinate system are calculated as follows:

[0093] [Xp,Yp,Zp,1]^T=[R_cp|T_cp]*[Xc,Yc,Zc,1]^T

[0094] UAV body coordinate system transformation: The gimbal coordinates are transformed into the body coordinate system with the UAV's center of gravity as the origin. This transformation is dynamically determined by the real-time rotation angles (pitch angle θ and yaw angle ψ) of the two brushless motors of the gimbal. The transformation matrix is ​​a rotation matrix R_pb composed of θ and ψ.

[0095] Therefore, the final coordinates (Xb, Yb, Zb) of the target in the body coordinate system are:

[0096] [Xb,Yb,Zb,1]^T=[R_pb|0]*[Xp,Yp,Zp,1]^T

[0097] The vision system obtains the precise physical coordinates (Xb, Yb, Zb) of the target relative to the center of the UAV body. These coordinates are sent to the flight control main controller, which can then calculate the following:

[0098] The displacement that the drone needs to move;

[0099] The pan-tilt unit needs to be rotated at an angle so that the spray nozzles can be precisely aimed at the target.

[0100] Please see Figures 1-9 The working principle of this invention is as follows:

[0101] This invention comprises a drone fuselage, a precision spraying gimbal mounted beneath the fuselage, a flight control system, and a vision recognition system. In operation, the drone fuselage, employing a quadcopter structure, offers excellent flight stability and maneuverability. The frame components feature a hollow design, effectively reducing overall weight while maintaining structural strength. The precision spraying gimbal, mounted beneath the fuselage, is the core actuator of the system. Driven by two perpendicular brushless motors, the gimbal achieves pitch and yaw motion. The gimbal integrates a spray nozzle 49 and an industrial camera 42. Precise control of the two brushless motors allows the nozzle and camera to be pointed in any direction. This enables targeted and directional spraying of pesticides; a flight control system is used to achieve stable flight control and intelligent operation of the drone, including integrating sensor data and running control algorithms through the main control module, using the attitude control module to ensure flight attitude stability, interacting in real time with the vision recognition system and execution components through the communication and collaboration module, and supporting flexible switching between manual and autonomous control to ensure the precise execution of plant protection tasks; a vision recognition system is used to capture and analyze crop images in real time, acquire high-quality image data through the hardware configuration control module, identify pests or specific plant targets through the image processing and recognition process control module, and calculate the physical coordinates of the target through the target positioning module to provide the drone with accurate spraying positioning information;

[0102] Spraying liquid is stored in the storage box 45, and the spray nozzle 49 is used for spraying. The second brushless motor 29 drives the second rotating column 47 inside the bottom connecting frame 24 to rotate. Under the connection of the internal connecting plate 48, the first rotating column 43 rotates. Limited by the limiting plate 25, the angle of the connecting side plate 27 is slightly adjusted, thereby improving the operation monitoring, spraying range, and the flexibility of the equipment. The required spraying liquid is replenished into the storage box 45 in a timely manner through the feeding port 46. The motor reinforcement plate 28 is used to reinforce the connection between the first brushless motor 22 and the electrical control box 21, thereby improving the installation stability of the equipment. One end of the positioning rod 11 extends into the limiting frame 33 to reinforce the connection between the positioning rod 11 and the drive motor 32. The main controller is used to control the operation of the electrical control box 21, the first brushless motor 22, the second brushless motor 29, the drive motor 32, the battery 4, and the lighting 44, realizing unified management of electrical equipment.

[0103] Overall Workflow

[0104] a. The drone cruises or hovers at a predetermined altitude above the farmland;

[0105] b. The industrial camera on the precision spraying platform captures high-definition images of the crops below in real time and immediately calculates their precise coordinates;

[0106] c. The visual algorithm on MINIPC analyzes the image and immediately calculates its precise coordinates once the target of pests and diseases is identified.

[0107] d.MINIPC sends the processed target coordinate data to the UAV's main controller (STM32F407) via UART serial port;

[0108] e. After receiving the coordinates, the main controller performs two actions:

[0109] e.1 Fine-tune the drone's flight attitude to keep it stably hovering directly above the target;

[0110] e.2 Control the two brushless motors of the precision spraying gimbal to make the nozzle and camera accurately point to the target;

[0111] f. After confirming aiming, the main controller starts the spraying device (such as a micro water pump) to spray pesticides onto the target in a short time and in a fixed quantity;

[0112] g. After completing one spraying cycle, the system continues to cruise, searching for the next target.

[0113] Hardware design innovation

[0114] To ensure flight control stability and real-time response of the spraying gimbal, this system adopts a distributed system architecture, with a dedicated control circuit board designed for the precision spraying gimbal. Figure 9 It is specifically responsible for receiving main control commands and driving the gimbal motor and nozzle, separating the gimbal control task from the main controller, avoiding interference from complex calculation tasks on flight attitude control, and significantly improving the stability and reliability of the system.

[0115] Advantages of the present invention

[0116] 1. Precise identification and targeted spraying: Combining machine vision and a dual-degree-of-freedom spraying platform, it can accurately target pests and diseases, completely changing the traditional large-area, indiscriminate spraying method;

[0117] 2. Significantly reduce pesticide use: Since spraying is only applied to identified pest and disease areas, the amount of pesticides used can be greatly reduced, effectively mitigating environmental pollution caused by pesticide abuse and protecting farmland ecology;

[0118] 3. Controllable cost and wide applicability: Compared with large and high-end agricultural drones, the overall design and selection of this solution focuses more on cost-effectiveness. It is small in size and light in weight, making it particularly suitable for use in small-scale, scattered, and complex terrain farmland, filling a market gap.

[0119] 4. High stability and reliability: The distributed architecture separates flight control from gimbal control, avoiding mutual interference between system tasks and ensuring the flight stability of the UAV when performing complex identification and spraying tasks.

[0120] Alternative solutions

[0121] 1. UAV airframe structure: In addition to quadcopters, other multi-rotor configurations such as hexacopters or octocopters can be selected according to the requirements of payload and stability.

[0122] 2. Spraying gimbal drive method: In addition to using a brushless motor, the drive motor of the gimbal can also be replaced with a servo motor according to the accuracy requirements of cost.

[0123] 3. Vision Processing Unit: The MINIPC can be replaced with other embedded AI computing platforms, such as NVIDIA Jetson series development boards, which may have advantages in AI computing. Alternatively, to further reduce the drone's payload, image data can be wirelessly transmitted to a ground workstation for processing, and then control commands can be transmitted back to the drone.

[0124] 4. Main control chip: The STM32F407 main control chip can also be replaced with other platforms or other series of high-performance microprocessors, such as GigaDevice's GD32F407 series chips.

[0125] Although embodiments of the invention have been shown and described, it will be understood by those skilled in the art that various changes, modifications, substitutions and alterations can be made to these embodiments without departing from the principles and spirit of the invention, the scope of which is defined by the appended claims and their equivalents.

Claims

1. A precision agricultural drone system based on machine vision, characterized in that, Including: A drone fuselage for normal flight; A precision spraying platform mounted under the machine body is used to achieve targeted and directional spraying of pesticides; A flight control system is used to achieve stable flight control and intelligent operation of the UAV; A visual recognition system for real-time capture and analysis of crop images; The drone body includes a drone frame (1), and a mounting top plate (3) is installed inside the drone frame (1). Four reinforcing rods (31) are fixedly connected to the bottom of the mounting top plate (3). A drive motor (32) is fixedly connected to one end of each of the four reinforcing rods (31). A rotating rotor (34) is fixedly connected to the output end of each drive motor (32). The precision spraying gimbal includes a gimbal fixing frame (2), the bottom of the drive motor (32) is equipped with the gimbal fixing frame (2), the bottom of the gimbal fixing frame (2) is rotatably connected to a bottom connecting frame (24), the outer side of the bottom connecting frame (24) is fitted with two limiting plates (25), the bottom of the two limiting plates (25) is fixedly connected to a connecting side plate (27), the top of the connecting side plate (27) is equipped with an industrial camera (42), both sides of the industrial camera (42) are fixedly connected to a storage box (45), and the bottom of the storage box (45) is fixedly connected to a spray nozzle (49).

2. The machine vision-based precision agricultural drone system according to claim 1, characterized in that: A first brushless motor (22) is installed at the bottom of the gimbal fixing frame (2). A rotating plate (23) is fixedly connected to the output end of the first brushless motor (22). The outer side of the rotating plate (23) is fixedly connected to the bottom connecting frame (24). A camera mounting bracket (41) is installed on the outer side of the industrial camera (42). A first rotating column (43) and a second rotating column (47) are rotatably connected inside the bottom connecting frame (24). A second brushless motor (29) used in conjunction with the second rotating column (47) is fixedly connected to the outer side of the bottom connecting frame (24). A limiting bolt (26) used in conjunction with the first rotating column (43) is installed on the outer side of the bottom connecting frame (24).

3. The machine vision-based precision agricultural drone system according to claim 1, characterized in that: A light (44) is fixedly connected to the top of the connecting side plate (27) and to one side of one of the storage boxes (45), and a feeding port (46) is installed on the top of the storage box (45).

4. The machine vision-based precision agricultural drone system according to claim 2, characterized in that: A battery (4) is fixedly connected to the top of the camera mounting bracket (41), an electrical control box (21) is fixedly connected to the bottom of the gimbal fixing frame (2), and a motor reinforcement plate (28) is fixedly connected to one side of the electrical control box (21) and outside the first brushless motor (22).

5. The machine vision-based precision agricultural drone system according to claim 1, characterized in that: Bottom positioning blocks (35) are fixedly connected to the bottom of each of the four reinforcing rods (31), and bottom positioning plates (36) are fixedly connected to the bottom of each of the bottom positioning blocks (35). The bottom positioning plates (36) are fixedly connected to the gimbal fixing frame (2). Four positioning rods (11) are fixedly connected at equal intervals inside the drone frame (1) and around the mounting top plate (3). Reinforcing connecting sleeves (12) are fitted at the connection points between the four positioning rods (11) and the drone frame (1). Limiting frames (33) are installed at the bottom of each of the drive motors (32), and one end of each positioning rod (11) extends into the limiting frame (33).

6. The machine vision-based precision agricultural drone system according to claim 4, characterized in that: The electrical control box (21) is equipped with a main controller. The electrical control box (21), the first brushless motor (22), the second brushless motor (29), the drive motor (32), the storage battery (4) and the lighting lamp (44) are all electrically connected to the main controller.

7. The machine vision-based precision agricultural drone system according to claim 1, characterized in that: The flight control system includes a main control module, an attitude control module, a communication and coordination module, and a control mode coordination module. The output terminals of the attitude control module, the communication and coordination module, and the control mode coordination module are all communicatively connected to the input terminal of the main control module. The communication and coordination module is bidirectionally communicatively connected to the visual recognition system. The flight control system is bidirectionally communicatively connected to the main controller.

8. The machine vision-based precision agricultural drone system according to claim 7, characterized in that: The main control module uses an STM32F407 series high-performance microcontroller as the main controller, and is responsible for receiving sensor data, running flight control algorithms, and sending instructions to various execution components. The attitude control module is used to perform closed-loop control of the UAV's flight attitude using a PID control algorithm, and to perform fusion processing of sensor data such as gyroscope and accelerometer using a Kalman filter algorithm. The communication and coordination module is used to control the main controller to communicate with the electronic speed controllers of the four rotors via the CAN bus protocol. The control mode coordination module is used to support a combination of manual control by remote control and autonomous control by host computer (PC).

9. The machine vision-based precision agricultural drone system according to claim 7, characterized in that: The visual recognition system includes a hardware configuration control module, an image processing and recognition process control module, and a target positioning module. The output end of the image processing and recognition process control module is communicatively connected to the input end of the target positioning module. The hardware configuration control module and the image processing and recognition process control module are bidirectionally connected. The visual recognition system is bidirectionally connected to the main controller.

10. The machine vision-based precision agricultural drone system according to claim 9, characterized in that: The hardware configuration control module consists of a high-performance MINIPC and a high-frame-rate industrial camera (42). The industrial camera (42) is fixed on the precision spraying gimbal and moves with the gimbal to capture crop images below the drone in real time. The image processing and recognition process control module is used to acquire image data from an industrial camera and transmit it to MINIPC. MINIPC is based on the OpenCV framework and uses a combination of lightweight convolutional neural networks and traditional image processing methods to perform real-time image analysis. After the target is identified, the visual recognition system calculates the pixel coordinates of the target in the image and obtains its precise physical coordinates relative to the drone body through coordinate transformation.

Images