Self-propelled electric laser weeding machine and adaptive weeding method

Abstract

The application discloses a self-propelled electric laser weeding machine and a self-adaptive weeding method, and relates to the technical field of intelligent agricultural equipment. The device comprises a machine body, an electric walking system, a path planning module, an image processing module and a laser weeding system. The path planning module is composed of a laser radar and a GPS positioning system and can realize autonomous avoidance function. The image processing module is composed of a binocular camera and a bottom acquisition camera and can identify weeds in real time, position the weeds and transmit data to the processor. The laser weeding system is mainly composed of a laser and can effectively kill weeds under the scheduling of the internal processor of the weeding machine. The application mainly provides a laser weeding machine which can autonomously plan a path and adjust laser power and scanning speed according to different weeds, and aims to solve the pain points of traditional weeding in the aspects of environmental protection, accuracy, work efficiency, adaptability and intelligence.

Description

Technical Field

[0001] This invention relates to the field of intelligent agricultural equipment technology, specifically to a self-propelled electric laser weeder and an adaptive weeding method. Background Technology

[0002] Weed control is a crucial aspect of agricultural production and forestry maintenance, directly impacting crop growth quality and yield, and has long been a common challenge for agriculture worldwide. Traditional weeding methods are mainly divided into three categories: manual weeding, chemical weeding, and mechanical weeding, all of which have significant limitations and cannot meet the demands of modern agriculture for precision, green development, and large-scale operations.

[0003] While manual weeding can accurately distinguish between weeds and crops and avoid accidental damage, it is extremely inefficient, labor-intensive, and has continuously rising labor costs, making it unsuitable for large-scale operations.

[0004] Chemical weeding, with its advantages of high efficiency, convenience, and low cost, was once widely used in large-scale agricultural production. However, long-term and large-scale use can cause serious environmental problems, including soil and water pollution, disruption of soil microbial balance, pesticide residues, and impact on the quality and safety of agricultural products.

[0005] Traditional mechanical weeding equipment, such as rotary and reciprocating weeders, can improve work efficiency, but they have poor adaptability to terrain, insufficient operating precision, are prone to damaging crop seedlings, are not effective in treating weed roots, require frequent operation, and have high maintenance costs.

[0006] To address the shortcomings of traditional weeding methods and realize the development trend of smart agriculture, laser weeding machines have emerged. Relying on cutting-edge technologies such as artificial intelligence recognition, high-precision positioning, and laser technology, they achieve precise, efficient, and green weeding without the need for chemical agents, resulting in no soil pollution or pesticide residues. They not only fill the gap in precise weeding during the seedling stage but also adapt to various complex operating scenarios, effectively balancing weeding efficiency and ecological protection, becoming an important direction for the green transformation of agriculture and the upgrading of intelligent agricultural machinery. Summary of the Invention

[0007] This invention primarily provides a laser weeding machine that can autonomously plan its path and adjust laser power and scanning speed according to different weed growth heights. It aims to address the pain points of traditional weeding methods in five aspects: environmental friendliness, precision, operational efficiency, adaptability, and intelligence.

[0008] It includes binocular cameras, lidar, satellite antenna, body, hydraulic telescopic mechanism, walking mechanism, machine vision system, and laser mechanism.

[0009] Furthermore, the hydraulic telescopic mechanism includes a bracket that is fixed to the bottom of the machine body by screws in a nested manner. The bracket contains uniform holes, and the other end is connected to the hydraulic telescopic device. The hydraulic telescopic rod is connected to the hydraulic telescopic device, and the extension and retraction of the hydraulic telescopic rod can be realized through the hydraulic telescopic device.

[0010] Furthermore, the walking mechanism includes wheels connected to a rotating shaft via bearings, an electric motor mounted on a wheel housing, and the wheel housing connected to a hydraulic telescopic rod via bolts.

[0011] Furthermore, the laser mechanism is installed below the machine body, and the laser is installed on the laser slide rail (which is installed on the laser guide rail and can realize the horizontal parallel movement of the laser).

[0012] Furthermore, the solar panel is mounted on top of the machine body.

[0013] Furthermore, the machine vision system includes a binocular camera mounted on the top of the machine body and a data acquisition camera mounted on the bottom of the machine body. The binocular camera and the data acquisition camera acquire weed image information, and determine the weed variety, age, and growth point through OpenCV image processing and YOLOv8 system recognition.

[0014] Furthermore, it also includes a path planning system, which includes a binocular camera mounted on the mobile platform and a lidar and satellite antenna mounted on the top of the machine. The Beidou differential positioning system communicates with satellite signals to perform real-time positioning of the weeding robot and transmits coordinate information to the control system. The control system compares and analyzes the collected coordinate information with the initial reference map, collects image information of the direction of travel through the binocular camera, and identifies and detects it to ensure that the weeding robot does not deviate from the row. The lidar scans the surrounding obstacles to ensure that the weeding robot can avoid obstacles and ensure that the weeding robot does not deviate from the row. The control system makes a final decision through data fusion and issues corresponding turning commands to the mobile platform.

[0015] An adaptive weeding method for a self-propelled electric laser weeder includes the following steps:

[0016] S1: The operator moves the weeder to the work area, inputs the boundary coordinates of the work area through the external terminal, and sets the weeding parameters; checks the working status of each module to ensure that the GPS positioning unit, image acquisition unit, laser emission unit, etc. are working properly; fully charges the power supply system, starts the weeder, and switches to automatic control mode.

[0017] S2: The GPS positioning unit of the path planning module obtains the current position coordinates of the weeder and the boundary coordinates of the work area in real time. The terrain detection unit detects the terrain and obstacle information of the work area in real time. Based on the real-time terrain and obstacle information, the global path is dynamically corrected, real-time driving instructions are generated and transmitted to the electric walking system to control the weeder to drive smoothly along the planned path.

[0018] S3: As the weeder travels along the planned path, the binocular camera and bottom acquisition camera of the image acquisition unit acquire images of the work area in real time and transmit them to the image preprocessing unit. The image preprocessing unit performs noise reduction, enhancement, and segmentation processing on the acquired raw images. The super green algorithm and grayscale processing are used to enhance image contrast and highlight the difference between crops and weeds. The OTSU algorithm is used, combined with the differences in plant spectral fingerprints, to segment the image into crop areas, weed areas, and background areas, eliminating background interference.

[0019] S4: Input the preprocessed image into the weed recognition unit, and identify the type, growth stage and growth density of weeds through the improved YOLOv8 model. The position positioning unit combines GPS positioning information and image shooting parameters, and calculates the three-dimensional coordinates of the weeds and the optimal scorching point position through the principle of binocular visual parallax, and transmits the weed information to the control system.

[0020] S5: After receiving weed information, the control system controls the laser adjustment unit to adaptively adjust the laser power, irradiation time and spot size according to the type, growth stage and density of the weeds through a power adjustment algorithm, and establishes a matching relationship between the laser parameters and the weed parameters.

[0021] S6: During the operation of the lawnmower, the path planning module detects terrain and obstacle information in real time. If an obstacle is detected, the path is immediately corrected to avoid it. At the same time, the inertial navigation unit supplements GPS positioning information in real time to avoid path deviation due to signal interruption. After the lawnmower completes the weeding operation of a section of the path, it automatically switches to the next section of the path until the weeding task of the entire work area is completed.

[0022] S7: After the weeding in the work area is completed, the weeder automatically shuts down the laser emission unit and the electric walking system, enters standby mode, and the operator checks the weeding effect, cleans the dust and debris on the surface of the weeder, charges the power supply system, and completes the operation.

[0023] Beneficial technical effects of the present invention:

[0024] Compared with existing technologies, this invention has comprehensive and superior improvements. Using a self-propelled chassis driven by pure electric low carbon as the carrier, combined with a machine vision adaptive system, it precisely targets weed tissues through laser, achieving zero pesticide residue and water pollution, effectively protecting soil fertility, and achieving millimeter-level precise weeding effect. It significantly reduces the rate of seedling damage and has the ability to operate continuously in all weather conditions. Its operating efficiency far exceeds that of manual labor and is not limited by weather or herbicide resistance in traditional chemical weeding. It also avoids the damage to soil structure caused by traditional mechanical weeding, achieving simultaneous improvement in environmental protection and safety, precision and efficiency, soil protection, adaptive adaptability, and long-term economic benefits. Attached Figure Description

[0025] The accompanying drawings, which are included to provide a further understanding of the invention and form part of this application, illustrate exemplary embodiments of the invention and, together with their description, serve to explain the invention and do not constitute an undue limitation thereof. In the drawings:

[0026] Figure 1 This is a schematic diagram of the structure of the self-propelled electric laser weeding machine of the present invention;

[0027] Figure 2 This is a bottom schematic diagram of the self-propelled electric laser weeding machine of the present invention;

[0028] Figure 3 This is a top view of the self-propelled electric laser weeding machine of the present invention;

[0029] Figure 4 This is a schematic diagram of the hydraulic telescopic device of the self-propelled electric laser weeding machine of the present invention;

[0030] Figure 5 This is a schematic diagram of the walking mechanism of the self-propelled electric laser weeding machine of the present invention;

[0031] Figure 6 This is a schematic diagram of the path planning for the weeding machine of the present invention;

[0032] Figure 7 This is a schematic diagram of the image processing flow of the present invention;

[0033] Figure 8 This is a schematic diagram illustrating the actual location of weeds and the division of image acquisition blocks according to the present invention;

[0034] Figure 9 This is a schematic diagram showing the coordinates of weeds to the laser emitter according to the present invention;

[0035] Figure 10 This is a schematic diagram of the working process of the weeding machine of the present invention;

[0036] Numbered in the diagram: 1. Binocular camera; 2. LiDAR; 3. Satellite antenna; 4. Body; 5. Hydraulic telescopic mechanism; 6. Walking mechanism; 7. Data acquisition camera; 8. Laser mechanism;

[0037] 41. Solar panels;

[0038] 51. Hydraulic telescopic device; 52. Hydraulic telescopic rod; 53. Support frame;

[0039] 61. Wheel; 62. Electric motor; 63. Axle; 64. Wheel shell;

[0040] 81. Laser; 82. Laser guide rail; 83. Laser slide rail. Detailed Implementation

[0041] The technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings of the embodiments of the present invention. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments.

[0042] Example 1: To achieve automatic planning of weeding paths and adjustment of the weeder height according to different crop heights for better laser weed removal, this example provides a self-propelled electric laser weeder. (See attached image) Figure 1 Specifically, it includes a binocular camera 1, a lidar 2, a satellite antenna 3, a body 4, a hydraulic telescopic mechanism 5, a walking mechanism 6, a machine vision system, and a laser mechanism 8.

[0043] The body 4 is made of lightweight, high-strength alloy material. The four limbs adopt a retractable hydraulic telescopic mechanism 5, which can automatically adjust the chassis height according to different conditions to adapt to different terrains such as fields and greenhouses. A binocular camera 1 is installed at the front of the body, and a solar panel 41 is set on the top of the body to improve the weeder's battery life. A laser guide rail 82 is set at the bottom of the body for mounting a laser slide rail 83. A storage compartment is set at the rear of the body for storing the power supply module and control module. The overall structure is compact and easy to move flexibly.

[0044] The walking mechanism 6 consists of wheels 61 plus servo motors, including four motors for walking and four motors for steering. Each pair of motors has an independent motor driver, and all motors can be controlled independently. This enables the weeder to move forward, backward, and turn. The steering mechanism uses a differential steering design, working in conjunction with instructions from the path planning module to achieve precise steering, ensuring the weeder travels smoothly along the planned path. The walking speed can be adaptively adjusted within a range according to weeding needs, avoiding missed weeds or inaccurate identification due to excessive speed.

[0045] The path planning section mainly consists of a GPS positioning unit and an obstacle avoidance unit, used to enable the weeder to autonomously plan and correct its path in real time. The specific structure is as follows:

[0046] GPS positioning unit: Includes satellite antenna 3 to obtain the current position coordinates of the weeder in real time and transmit the position information to the path calculation unit. At the same time, it can receive the boundary coordinates of the work area input by the external terminal through the satellite antenna 3 module to determine the weeding operation range. Obstacle avoidance unit: Includes lidar 2 to detect the terrain undulations and obstacles in the work area and transmit the terrain and obstacle information to the path calculation unit.

[0047] The image processing module includes an image acquisition unit, an image preprocessing unit, a weed recognition unit, and a location positioning unit. It is used to achieve accurate weed recognition and location positioning, providing precise target information for laser weed control. The specific structure is as follows:

[0048] Image acquisition unit: The acquisition camera 7 acquires real-time images of crops and weeds in the work area and transmits them to the image preprocessing unit;

[0049] Image preprocessing unit: performs denoising, enhancement, and segmentation on the acquired raw image; uses the super green algorithm and grayscale processing to enhance image contrast and highlight the difference between crops and weeds; uses the OTSU algorithm, combined with plant spectral fingerprint differences, to segment the image into crop region, weed region, and background region to eliminate background interference;

[0050] Weed Recognition Unit: Built-in improved YOLOv8 lightweight model, pre-trained with a large number of image samples of different crops, different weeds, different growth stages, and different light conditions to establish a weed recognition model; input the pre-processed image into the weed recognition model, the model can quickly identify the type of weed, growth stage (seedling stage, mature stage) and growth density, to meet the needs of real-time weeding.

[0051] Positioning Unit: Combining GPS positioning information from the path planning module and shooting parameters (focal length, shooting angle) from the image acquisition unit, the three-dimensional coordinates (x, y, z) of the weeds are calculated using the principle of binocular visual parallax. This accurately locates the specific position of the weeds and determines the optimal scorching point. When shooting from above, the stem is located; when shooting from below, the root is located. The weed position coordinates and optimal scorching point information are transmitted to the laser weeding system and control system.

[0052] The laser mechanism 8 includes a laser emitting unit and a laser adjustment unit, used to achieve precise laser irradiation of weeds, and to adaptively adjust the laser power according to the weed information. The specific structure is as follows:

[0053] Laser 81: Mounted on the laser slide rail 83 at the bottom of the machine body, the laser emission direction can be adjusted by the electric gimbal. Combined with the weed location information, it can achieve precise aiming at the weeds. The laser emission unit is equipped with a galvanometer scanning system and the focusing lens can achieve adjustable spot diameter to ensure that the laser energy is concentrated on the weeds.

[0054] Laser adjustment unit: Connected to the image processing module and control system, it is used to receive weed identification information, adaptively adjust laser power, irradiation time and spot size, and has a built-in power adjustment algorithm to establish the correspondence between laser power and weed parameters based on the thermodynamic characteristics and spectral absorption rate of weeds, so as to realize power graded adjustment.

[0055] Example 2:

[0056] like Figures 1-9 As shown, based on Embodiment 1, this embodiment discloses an adaptive weeding method for a self-propelled electric laser weeding machine, including the following steps:

[0057] S1: In scenarios where no crops are growing or have not yet emerged from the soil, a satellite map of the area to be weeded can be imported first, and the work area can be divided into grids. The weeding robot traverses the grids along an S-shaped path to carry out weeding operations. Grid division not only accurately grasps the distribution of weeds, enabling systematic and orderly weeding, but also intuitively reflects the equipment's working efficiency. If there are no weeds in a certain grid, the robot can directly skip that area and move to the next grid to continue working, ensuring no weeds are missed through the full grid coverage mode.

[0058] When crops are present in the field, the robot still employs a grid-based S-shaped path strategy, but the crop planting areas are separated. The weeding robot uses LiDAR to perceive and avoid crop-growing areas in real time. Simultaneously, the onboard BeiDou differential positioning system performs real-time positioning of the robot, transmitting its coordinates to the control system for comparison and analysis with a preset initial map, completing real-time position calibration. Furthermore, the inter-row navigation camera continuously collects images of the path ahead, identifying crop rows and detecting its trajectory, further ensuring the robot's stable movement along the crop rows and preventing deviation from the working path.

[0059] S2: The machine vision system consists of two parts: OpenCV and YOLOv8. OpenCV is mainly responsible for image processing, while YOLOv8 is responsible for the identification of weeds and crops.

[0060] The raw photos taken by the binocular cameras are processed by YOLOv8. If no weeds are detected, the system automatically skips this step and proceeds to the next block. If weeds are detected, the system will mark them and perform a series of operations.

[0061] Since most crops and weeds are green, while soil is brown, the purpose of field image background segmentation is to distinguish plants (including crops and weeds) from the soil or water layer background, so that the segmented image only retains crops and weeds. The normalized supergreen algorithm (2g-rb) is selected. This method is an extension of the weighted method, which can increase the weight of the green channel G component in the RGB image and suppress the background parts of non-green R and B components. Utilizing this characteristic in combination with threshold segmentation, plants can be effectively distinguished from the background.

[0062] The segmentation threshold T is obtained using the OTU algorithm, and the grayscale image is binarized. This method, compared to the traditional Super Green algorithm, can better eliminate noise during image acquisition, facilitating the extraction of object information later. Simultaneously, to reduce the interference of illumination intensity on image analysis, the color components of the red, blue, and green (RGB) color channels in the Super Green algorithm are normalized. The specific calculation formula for the Super Green algorithm is as follows:

[0063]

[0064]

[0065] In the formula: R, G, B are the pixel channel values ​​in the RGB color space. This is an ultra-green image.

[0066] Since grayscale images of weeds still contain small holes and noise after OTSU thresholding, this invention uses a combination of opening and closing operations in image morphology to filter image noise and fill holes, based on morphological processing methods. The morphological processing followed by opening and then closing operations makes the weed portion clearer and more complete, facilitating more accurate weed identification.

[0067] After image processing, other irrelevant targets in the photo can be largely removed, making it easier to determine the location of weeds. Furthermore, the original high-definition camera captured images with a large file size, but after a series of processing steps, the file size is greatly reduced. This ensures that the speed at which images are transmitted to the weed cutter's processor and the speed at which the processor processes and recognizes the images are accelerated.

[0068] S3: To convert the image coordinates of the weeds into ground coordinates to obtain the actual distance, the weeding machine uses a vertical overhead shooting method to acquire images. The camera has no forward tilt angle. Based on the pinhole imaging model, the mathematical relationship between the image coordinates and the ground coordinates can be derived:

[0069]

[0070] In the formula, Z is the distance from the center of the camera lens to the intersection of the principal optical axis and the ground, in mm; , These are the width and height of the image, in mm; , pixels represent the x and y coordinates of a pixel in the image coordinate system; , These represent the horizontal and vertical coordinates in the corresponding geo coordinate system after the image coordinates have been converted, in mm; θ is the camera's focal length, in mm; θ is the camera's tilt angle, in degrees (º).

[0071] Since laser weeding is based on laser surface scanning of the center point of the weed, cutting off the weed bud and main petiole can stop the weed from growing and eventually kill it, so it has a large tolerance for error.

[0072] The distance from the laser emitter to the center of the weed top can be approximated as a triangle. The Z-axis coordinate of the weed top center can be directly captured by the binocular camera. Therefore, the distance from the laser emitter to the weed top center can be determined using trigonometric functions.

[0073]

[0074] S4: The core principle of laser weed control is the photothermal effect: the laser transmits energy to plant cells (mainly meristematic tissues), instantly heating the water inside the cells to boiling point, causing the cell walls to rupture, thereby "cutting off" the growth ability of weeds.

[0075] The effective energy input by the laser must be greater than the energy required for the plant tissue to heat up and for water to vaporize.

[0076]

[0077] In the formula:

[0078] Energy output by the laser

[0079] The absorption rate of biological tissues, usually Take a value of 0.8 to 0.9.

[0080] The amount of heat required to heat a tissue from ambient temperature to its boiling point (100°C).

[0081] The heat required for water vaporization.

[0082] Heat conduction is the energy lost to the surrounding stems, leaves, or soil.

[0083] mm / s mm

[0084]

[0085] In the formula:

[0086] F: Actual output energy of the laser

[0087] T: Laser

[0088] P: Laser power, W.

[0089] t: Irradiation time, ms.

[0090] A: Laser

[0091] Substituting the values, we get:

[0092]

[0093] To control weeds, the energy density must exceed the lethal energy of the weeds. Then we can obtain:

[0094]

[0095] Laser spot diameter:

[0096] The laser is focused by a lens with a focal length of f (i.e., the designed maximum striking range). When the actual distance Z deviates from the focal length f, the spot diameter d(z) becomes:

[0097]

[0098] In the formula:

[0099] Z: Actual distance from the laser aperture to the top of the weed, mm.

[0100] f: The focal length of the lens, in mm, which is the system's preset optimal striking distance.

[0101] The diameter of the light spot waist at the finest point of the focal point is mm.

[0102] Rayleigh length. ( (Where the laser wavelength is used).

[0103] Then it can be known that when At that time, the light spot is the smallest. The energy is strongest at this point. As Z increases or decreases, the light spot size d(z) increases, and the energy density decreases rapidly. The previous energy density equation was... Now, substituting d(z) into d, we obtain the energy model that includes the distance variable:

[0104]

[0105] After simplification:

[0106]

[0107] Right now:

[0108]

[0109] in It is the maximum energy density at the focal point.

[0110] Therefore, the maximum speed of the lawnmower can be calculated as follows:

[0111]

[0112] The minimum required power is:

[0113]

[0114] Therefore, optimal weed control can only be achieved when the focal length f = S (the distance from the laser to the top of the weeds). The sensor measures the distance S, and the system needs to adjust other parameters based on this distance. Thus:

[0115] Control Strategy 1: Automatic Zoom

[0116] The lens position is adjusted by a mechanical structure, so that .

[0117] Advantages: Highest energy efficiency.

[0118] Disadvantages: Slow mechanical response, may not be able to keep up with high-speed movement.

[0119] Control Strategy 2: Power / Speed ​​Compensation

[0120] The lens remains stationary (f is fixed).

[0121] When detected When the light spot becomes larger, the moving speed v is reduced or the power P is increased to compensate for the loss of energy density.

[0122]

[0123] In the formula:

[0124] P1: Actual power required

[0125] P2: Laser base power

[0126] (If the distance is off-center, greater power must be output to achieve the same destructive effect.)

[0127] The above description is only a preferred embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any equivalent substitutions or modifications made by those skilled in the art within the scope of the technology disclosed in the present invention, based on the technical solution and inventive concept of the present invention, should be covered within the scope of protection of the present invention.

Claims

1. A self-propelled electric laser weeding machine, characterized in that, It includes a binocular camera (1), a lidar (2), a satellite antenna (3), a body (4), a hydraulic telescopic mechanism (5), a walking mechanism (6), a machine vision system, and a laser mechanism (8).

2. The self-propelled electric laser weed cutter according to claim 1, characterized in that, The hydraulic telescopic mechanism (5) includes a bracket (53) which is fixed to the bottom of the body (4) by screws in a nested manner. The bracket (53) contains uniform holes, and the other end is connected to the hydraulic telescopic device (51). The hydraulic telescopic rod (52) is connected to the hydraulic telescopic device (51), and the extension and retraction of the hydraulic telescopic rod (52) can be realized through the hydraulic telescopic device (51).

3. The self-propelled electric laser weeding machine according to claim 1, characterized in that, The walking mechanism (6) includes a wheel (61) connected to a rotating shaft (63) via a bearing, an electric motor (62) mounted on a wheel housing (64), and the wheel housing (64) connected to a hydraulic telescopic rod (52) via bolts.

4. The self-propelled electric laser weed cutter according to claim 1, characterized in that, The laser mechanism (8) is installed below the body (4), the laser (81) is installed on the laser slide rail (83), and the laser slide rail (83) is installed on the laser guide rail (82), which can realize the horizontal parallel movement of the laser.

5. A self-propelled electric laser weed cutter according to claim 1, characterized in that, The solar panel (41) is mounted on top of the body (4).

6. A self-propelled electric laser weed cutter according to claim 1, characterized in that, The machine vision system includes a binocular camera (1) mounted above the machine body (4) and a data acquisition camera (7) mounted below the machine body (4). The binocular camera (1) and the data acquisition camera (7) acquire weed image information and determine the weed variety, age and growth point through OpenCV image processing and YOLOv8 system recognition.

7. A self-propelled electric laser weed cutter according to claim 5, characterized in that, It also includes a path planning system, which includes a binocular camera (1) set on the mobile end of the mobile platform and a laser radar (2) and a satellite antenna (3) set on the top of the body (4). The Beidou differential positioning system communicates with the satellite signal to locate the weeder in real time and transmits coordinate information to the control system. The control system compares and analyzes the collected coordinate information with the initial reference map. The binocular camera (1) collects image information in front of the direction of travel and identifies and detects it to ensure that the weeding robot does not deviate from the row. The laser radar (2) scans the surrounding obstacles to ensure that the weeder can avoid obstacles and ensure that the weeding robot does not deviate from the row. The control system makes a final decision through data fusion and issues corresponding turning commands to the mobile platform.

8. An adaptive weeding method for a self-propelled electric laser weeding machine, characterized in that, A self-propelled electric laser weed cutter according to any one of claims 1-7 includes the following steps: S1: The operator moves the weeder to the work area, inputs the boundary coordinates of the work area through the external terminal, and sets the weeding parameters; checks the working status of each module to ensure that the GPS positioning unit, image acquisition unit, laser emission unit, etc. are working properly; fully charges the power supply system, starts the weeder, and switches to automatic control mode. S2: The GPS positioning unit of the path planning module obtains the current position coordinates of the weeder and the boundary coordinates of the work area in real time. The terrain detection unit detects the terrain and obstacle information of the work area in real time. Based on the real-time terrain and obstacle information, the global path is dynamically corrected, real-time driving instructions are generated and transmitted to the electric walking system to control the weeder to drive smoothly along the planned path. S3: As the weeder travels along the planned path, the binocular camera and bottom acquisition camera of the image acquisition unit acquire images of the work area in real time and transmit them to the image preprocessing unit. The image preprocessing unit performs noise reduction, enhancement, and segmentation processing on the acquired raw images. The super green algorithm and grayscale processing are used to enhance image contrast and highlight the difference between crops and weeds. The OTSU algorithm is used, combined with the differences in plant spectral fingerprints, to segment the image into crop areas, weed areas, and background areas, eliminating background interference. S4: Input the preprocessed image into the weed recognition unit, and identify the type, growth stage and growth density of weeds through the improved YOLOv8 model. The position positioning unit combines GPS positioning information and image shooting parameters, and calculates the three-dimensional coordinates of the weeds and the optimal scorching point position through the principle of binocular visual parallax, and transmits the weed information to the control system. S5: After receiving weed information, the control system controls the laser adjustment unit to adaptively adjust the laser power, irradiation time and spot size according to the type, growth stage and density of the weeds through a power adjustment algorithm, and establishes a matching relationship between the laser parameters and the weed parameters. S6: During the operation of the lawnmower, the path planning module detects terrain and obstacle information in real time. If an obstacle is detected, the path is immediately corrected to avoid it. At the same time, the inertial navigation unit supplements GPS positioning information in real time to avoid path deviation due to signal interruption. After the lawnmower completes the weeding operation of a section of the path, it automatically switches to the next section of the path until the weeding task of the entire work area is completed. S7: After weeding in the work area is completed, the weeder automatically shuts down the laser emission unit and the electric walking system, enters standby mode, and the operator checks the weeding effect, cleans the dust and debris on the surface of the weeder, charges the power supply system, and completes the operation.

9. The adaptive weeding method for a self-propelled electric laser weeder according to claim 8, characterized in that, In step S4, the coordinates of the weeds are located and the distance between the weeds and the laser is determined.

10. The adaptive weeding method for a self-propelled electric laser weeder according to claim 8, characterized in that, In step S5, the laser power, irradiation time, and spot size are adaptively adjusted through a power adjustment algorithm to establish a matching relationship between the laser parameters and the weed parameters.

Images