Laser weeding method capable of adaptively adjusting laser dosage and laser weeding machine thereof

Through the combination of deep learning model and laser galvanometer, the adaptive adjustment of laser dose and precise positioning of laser weed killers is achieved, solving the problems of laser weeding accuracy and environmental protection in the existing technology, and improving the herbicidal efficiency and environmental safety .

CN118923656AInactive Publication Date: 2024-11-12JILIN UNIVERSITY

Patent Information

Application Number
CN202411025183.2
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2024-07-29
Publication Date
2024-11-12
Estimated Expiration
Not applicable · inactive patent

AI Technical Summary

Technical Problem

Existing laser herbicidal technology is difficult to achieve accurate positioning of the laser beam on weeds and adjust the laser dose according to different types and growth periods to achieve the best inactivation effect.

Method used

Deep learning model is used to identify weed species and growth period, and through precise control of laser galvanometer, combined with a monocular camera and industrial control system, the laser weed killer adaptively adjusts laser dose and precise positioning.

Benefits of technology

It improves the accuracy and efficiency of weeding, protects the soil microecological environment, reduces the risks of energy waste and pollution, and achieves the convenience of all-weather operations.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN118923656A_ABST
    Figure CN118923656A_ABST
Patent Text Reader

Abstract

According to the laser weeding method capable of adaptively adjusting the laser dosage and the weeding machine thereof, the types and growth periods of weeds are accurately recognized through a deep learning model, precise positioning of the weeds is achieved, the light path of the laser is accurately controlled by controlling the laser galvanometer, and the weeding machine can accurately control the weed dosage according to the types and growth periods of the weeds. And the power and the working speed are automatically adjusted, so that the laser can stay on weeds at different time lengths, and the efficient and environment-friendly weeding effect with the optimal laser dosage is achieved. The equipment integrates machine vision, deep learning and machine control technologies, realizes intelligentization and automation of the weeding process, avoids chemical pollution and soil disturbance, and protects the ecological environment. The whole machine is provided with high-performance parts such as a hub motor and an industrial personal computer, stable and efficient operation is ensured, the device is suitable for various farmland environments, and the efficiency and quality of agricultural weed prevention and control are improved.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The invention belongs to the technical field of agricultural machinery automation, and specifically relates to a laser weeding method for adaptively adjusting laser dosage and a laser weeding machine thereof. Background Art

[0002] Weeding is an important part of agricultural production. Although biochemical herbicides have the advantages of high efficiency and convenience, they will pollute rivers, water sources and soil with the erosion of rainwater, and will also threaten the health of workers who apply pesticides. More importantly, biochemical herbicides will cause pesticide residues on crops, threatening the health of consumers and reducing the comprehensive income of agricultural producers. Mechanical weeding will disturb the soil and destroy the microecological environment of the soil. Laser weeding has the advantages of not contacting the soil and not using biochemical herbicides. It can protect the soil and ecological environment while preventing and controlling weeds. At present, the main problems faced by laser weeding technology are concentrated in the following aspects: first, how to achieve accurate positioning and alignment of the laser beam on the weeds to ensure the accuracy of the weeding operation; second, how to adjust the laser dose according to the characteristics of different types of weeds and their different growth periods to achieve the best inactivation effect. In order to solve the above problems, we proposed a laser weeding method and a laser weeding machine with adaptive adjustment of laser dose. Summary of the invention

[0003] The object of the present invention is to provide a laser weeding method and a laser weeding machine for adaptively adjusting laser dosage, comprising the following steps:

[0004] S1. Weeds of different types and growth periods are selected in the farmland. Lasers ranging from 100W to 2000W are used to irradiate the weeds in situ in the farmland in increments of 100W. The activity of the weeds after irradiation with different laser doses is observed to obtain the optimal laser dose for inactivating weeds of different types and growth periods.

[0005] S2. Obtain images of weeds of different types and different growth periods in the farmland, mark the weeds in the images, generate a marking file and corresponding labels, import the marking file and labels into a deep learning model, and train a deep learning model that can identify weed types, growth periods, and locate the positions of weeds in the images;

[0006] S3, using a monocular camera to obtain images in the farmland in the form of video streams, and parsing the video streams into single-frame weed images;

[0007] S4, using the deep learning model obtained in S2 to process each frame of the video stream, if no weeds appear in the image, continue to detect the next frame of the image until weeds are detected in the image;

[0008] S5, obtaining a bounding box of the weeds detected in S4 and coordinate values ​​of the corner points of the bounding box;

[0009] S6, obtaining the coordinates of the center of the weed in the image coordinate system through the coordinate values ​​of the corner points of the bounding box obtained in S5;

[0010] S7, based on the calibration results of the machine coordinate system and the image coordinate system, output a transformation matrix from the image coordinate system to the machine coordinate system;

[0011] S8, obtaining the coordinates of the center of the weed in the machine coordinate system;

[0012] S9, using the coordinates of the center of the weed obtained in S8 in the machine coordinate system, the control program automatically controls the deflection of the laser galvanometer, thereby reflecting the optical path of the laser so that the optical path of the laser is always aligned with the weed;

[0013] S10, based on the obtained optimal laser dose for inactivating weeds of different types and growth periods, and the detected weed types and growth periods output by the deep learning model, adjusting the working speed of the laser weeder, making the laser stay on the detected weeds within a certain operation period, and adjusting the power of the laser at the same time;

[0014] S11, continue to emit laser until the weeds are no longer detected in the image as the laser weeder moves forward, and then continue to detect the next frame of image.

[0015] Preferably, a laser weeding method for adaptively adjusting laser dosage and a laser weeding machine thereof include 4 wheel hub motors, a chassis, an industrial computer, a wheel hub motor power supply, a water cooler, a machine cover, a main generator, a laser power supply, a fill light power supply, a laser tube, a galvanometer, a fill light controller, an LED fill light, a monocular camera and a lens. The wheel hub motor includes a wheel hub and a tire, which can provide power to the tire without relying on a drive shaft. The wheel hub motor adopts advanced DC brushless technology, which not only improves the operating efficiency, but also significantly reduces the noise level and enhances the user experience. The chassis suspension system has been carefully adjusted to ensure the smooth driving of the equipment on complex terrain while reducing the damage to the surface. The GPS positioning system integrated in the industrial computer is closely integrated with the equipment control system to achieve accurate navigation and automatic control of weeding operations. The water cooler is installed in the hood, and the heat exchange system between the water cooler and the carbon dioxide laser has been optimized to ensure the temperature stability of the laser under long-term high-load operation. The design of the hood further optimizes the waterproof, dustproof and heat dissipation performance to ensure that the internal components can still work stably in harsh environments. The generator is the power supply source for the entire equipment, preferably a fuel generator, with automatic start and stop functions. The overcurrent protection mechanism of the laser power supply can effectively prevent potential damage to the laser due to abnormal current fluctuations, improving the safety and reliability of the equipment. The communication protocol between the fill light and the controller has been optimized to achieve the control of the fill light switch and the precise adjustment of the brightness to meet the weeding needs under different lighting conditions.

[0016] Compared with the prior art, the present invention provides a laser weeding method and a laser weeding machine capable of adaptively adjusting laser dosage, which have the following significant beneficial effects:

[0017] 1. Environmentally friendly and pollution-free: The present invention adopts laser weeding technology and does not require the use of biochemical herbicides, thereby avoiding the pollution of rivers, water sources and soil caused by rain erosion. At the same time, it protects the health of pesticide application workers, eliminates pesticide residues on crops, protects the health of consumers, and is expected to increase the comprehensive income of agricultural producers.

[0018] 2. Protect soil microecology: Since laser weeding technology does not contact the soil and does not disturb the soil, it can protect the soil's microecological environment and is conducive to maintaining the ecological balance of farmland.

[0019] 3. Precise positioning and weeding: Through the deep learning model, the present invention realizes the precise identification of weed types and growth periods and the precise positioning of positions. Combined with the precise control of the laser galvanometer, the laser light path can be accurately aimed at the weeds, thereby improving the accuracy of weeding.

[0020] 4. Adaptive adjustment of laser dosage: The present invention targets weeds of different types and growth periods. The optimal laser dosage is obtained through preliminary experiments. In the actual weeding process, the laser power and the working speed of the weeder are automatically adjusted according to the detection results of the deep learning model, so that the laser can stay on the detected weed plants for different lengths of time. While ensuring the best weed control effect, it saves laser energy consumption and avoids unnecessary energy waste.

[0021] 5. Intelligence and automation: The present invention integrates advanced machine vision, deep learning, machine control and other technologies to realize the intelligence and automation of the weeding process, reduce the need for manual intervention, and improve the convenience and efficiency of weeding operations.

[0022] 6. Superior overall performance: The overall design of the laser weeder of the present invention fully considers the user's convenience of use and maintenance costs. The connections and layouts between the various components are carefully planned to facilitate daily operation and maintenance. The present invention uses a monocular camera to locate weeds, which greatly reduces costs compared to using binocular cameras, depth cameras, and stereo cameras. The laser weeder of the present invention is reasonably designed and equipped with advanced equipment such as hub motors, industrial computers, water coolers, and laser power supplies to ensure the stable operation and efficient operation of the weeder. The overall performance is superior and is suitable for various farmland environments. Through the LED fill light, the laser weeder of the present invention can achieve all-weather operation. BRIEF DESCRIPTION OF THE DRAWINGS

[0023] Figure 1 is a flow chart of the present invention;

[0024] Figure 2 A three-dimensional schematic diagram of the laser weed remover of the present invention;

[0025] Figure 3 It is an exploded view of a three-dimensional schematic diagram of the laser weeder of the present invention.

[0026] In the figure: 1. Hub motor; 2. Chassis; 3. Industrial computer; 4. Hub motor power supply; 5. Water cooler; 6. Machine cover; 7. Main generator; 8. Laser power supply; 9. Fill light power supply; 10. Laser tube; 11. Galvanometer; 12. Fill light controller; 13. LED fill light; 14. Monocular camera and lens. DETAILED DESCRIPTION

[0027] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative work are within the scope of protection of the present invention.

[0028] Please refer to the attached Figure 1-3 Before the laser weeder of the present invention works, weeds of different types and different growth periods are selected in the farmland, and the weeds are irradiated in situ in the farmland with lasers ranging from 100W to 2000W in an increasing range of 100W, and the activity of the weeds after laser irradiation is observed, so as to obtain the optimal laser dose for inactivating weeds of different types and growth periods;

[0029] Next, we acquired images of weeds of different types and growth periods in the farmland, used software to mark the weeds in the images, generated a marking file and corresponding labels, imported the marking file and labels into the deep learning model, and trained the deep learning model to identify the weed types, growth periods, and locate the weeds in the images.

[0030] Use a monocular camera to obtain images of farmland in the form of video streams. Based on an industrial computer, an automated program will parse the video stream into single-frame weed images.

[0031] Each frame of the video stream is processed using the deep learning model obtained in the present invention; if no weeds appear in the image, the next frame of the image is detected until weeds are detected in the image; then, the bounding box of the detected weeds and the coordinate values ​​of the corner points of the bounding box are obtained; then, the program calculates the coordinates of the center of the weed in the image coordinate system according to the obtained coordinate values ​​of the corner points of the bounding box;

[0032] Based on the calibration results of the machine coordinate system and the image coordinate system; obtaining the conversion relationship from the image coordinate system to the machine coordinate system; for example, adding, subtracting, multiplying and dividing the horizontal and vertical coordinates of the center of the weed in the image coordinate system by specific coefficients, thereby obtaining the coordinates of the weed in the machine coordinate system;

[0033] The laser galvanometer is precisely deflected by the control program of the laser galvanometer, and then the laser light path is accurately reflected, so that the laser light path is always aimed at the weeds;

[0034] Based on the obtained optimal laser dose for inactivating weeds of different types and growth periods, and the deep learning model detection results, the working speed of the laser weeder is adjusted based on the detected weed types and growth periods, so that the laser stays on the detected weeds within a certain operation period, and the laser power is adjusted at the same time;

[0035] The laser is continuously emitted until the weeds are no longer detected in the image as the laser weeder moves forward, and then the next frame of image is detected.

[0036] like Figure 2As shown, when working, the laser weeder of the present invention is placed along the direction of the crop planting row, and the hub motor 1 is placed between the rows of the crop planting row, and along the forward direction, as shown in FIG. Figure 3 As shown, the four wheel hub motors 1 can provide power for the weeder to move forward, the monocular camera and the lens 14 vertically downward obtain the farmland image, the industrial computer 3 will automatically run the steps S3 to S11, and the laser beam will accurately irradiate the weeds within the wheel base of the wheel hub motor 1, thereby realizing intelligent laser weeding operations.

[0037] Although embodiments of the present invention have been shown and described, it will be appreciated by those skilled in the art that various changes, modifications, substitutions and variations may be made to the embodiments without departing from the principles and spirit of the present invention, and that the scope of the present invention is defined by the appended claims and their equivalents.

Claims

1. A laser weeding method and laser weeding machine capable of adaptively adjusting laser dosage, characterized in that: The following steps are involved: S1. Weeds of different types and growth periods are selected in the farmland. Lasers ranging from 100W to 2000W are used to irradiate the weeds in situ in the farmland in increments of 100W. The activity of the weeds after irradiation with different laser doses is observed to obtain the optimal laser dose for inactivating weeds of different types and growth periods. S2. Obtain images of weeds of different types and different growth periods in the farmland, mark the weeds in the images, generate a marking file and corresponding labels, import the marking file and labels into a deep learning model, and train a deep learning model that can identify weed types, growth periods, and locate the positions of weeds in the images; S3, using a monocular camera to obtain images in the farmland in the form of video streams, and parsing the video streams into single-frame weed images; S4, using the deep learning model obtained in S2 to process each frame of the video stream, if no weeds appear in the image, continue to detect the next frame of the image until weeds are detected in the image; S5, obtaining a bounding box of the weeds detected in S4 and coordinate values ​​of the corner points of the bounding box; S6, obtaining the coordinates of the center of the weed in the image coordinate system through the coordinate values ​​of the corner points of the bounding box obtained in S5; S7, based on the calibration results of the machine coordinate system and the image coordinate system, output a transformation matrix from the image coordinate system to the machine coordinate system; S8, obtaining the coordinates of the center of the weed in the machine coordinate system; S9, using the coordinates of the center of the weed obtained in S8 in the machine coordinate system, the control program automatically controls the deflection of the laser galvanometer, thereby reflecting the optical path of the laser so that the optical path of the laser is always aligned with the weed; S10, based on the obtained optimal laser dose for inactivating weeds of different types and growth periods, and the detected weed types and growth periods output by the deep learning model, adjusting the working speed of the laser weeder, making the laser stay on the detected weeds within a certain operation period, and adjusting the power of the laser at the same time; S11, continue to emit laser until the weeds are no longer detected in the image as the laser weeder moves forward, and then continue to detect the next frame of image.

2. A laser weeding method and laser weeding machine capable of adaptively adjusting laser dosage, characterized in that: The invention comprises four wheel hub motors (1), a chassis (2), an industrial computer (3), a wheel hub motor power supply (4), a water cooler (5), a hood (6), a main generator (7), a laser power supply (8), a fill light power supply (9), a laser tube (10), a galvanometer (11), a fill light controller (12), an LED fill light (13), a monocular camera and a lens (14); the wheel hub motor (1) comprises a wheel hub and a tire, and the wheel hub motor (1) adopts a DC brushless technology; the industrial computer (3) is integrated with a GPS positioning system; the main generator (7) has an automatic start and stop function; and the laser power supply (8) has an overcurrent protection function.

Citation Information

Patent Citations

  • Intelligent laser weeding device and method based on machine vision

    CN116250523A

  • Laser head pose adjuster suitable for laser weeding machine

    CN117898278A

  • Full-automatic weeding machine weed recognition method and system based on machine vision

    CN118015458A

  • Weeding method and device of laser weeding robot, medium and laser weeding robot

    CN118235753A

Cited By

  • Farmland weed laser weeding device and weeding method based on deep learning vision

    CN120959223A

  • A deep learning-based visual farmland weed laser weeding method

    CN120959223B

  • Light emitting device and weeding method by light supplementation

    CN122834812A