A laser-chemical synergistic weed control method and system based on target detection

Abstract

This invention addresses the shortcomings of existing related technologies and equipment by disclosing a laser-chemical synergistic weeding method and system based on target detection. The system includes a control system and weeding equipment. The method utilizes an embedded computer composed of an ARM architecture processor and a system-on-a-chip (SoC) as the control system. A high-definition camera collects real-time field image data, and the SoC's target detection algorithm detects and locates the weeds. The laser emitter first strikes the main body of the weed based on the location coordinates, causing obvious damage. Then, a chemical spraying device sprays herbicides to remove the weeds. This invention uses a composite embedded computer composed of an ARM architecture and a SoC as its core, employing a novel method of combined laser and chemical weeding to remove weeds from farmland. This avoids chemical pollution caused by excessive pesticide use and significantly optimizes the limitations of laser weeding, thereby improving agricultural production efficiency.

Description

Technical Field

[0001] This invention relates to the technical fields of artificial intelligence and agricultural robots, and more specifically, to a laser-chemical synergistic weeding method and system based on target detection. Background Technology

[0002] Agriculture is one of the most important sectors for human survival and development, ensuring basic food needs while also generating significant economic benefits. However, weeds in farmland greatly reduce crop yield and quality, increase management labor and production costs, and seriously threaten agricultural production.

[0003] The extensive use of single herbicides in crop cultivation has exacerbated the problem of herbicide resistance in weeds, prompting a shift in chemical herbicides from single-agent to compound formulations. While compound formulations have, to some extent, curbed the spread of resistant weeds and ensured food security, the emergence of cross-resistant and multi-resistant weeds has shortened their lifespan, increasing the weeding burden on agricultural producers. In particular, resistance to hormone-based herbicides with unique modes of action has also emerged, posing unprecedented challenges to effective weed control.

[0004] With the rapid development of deep learning technology, weed detection and removal based on deep learning have become an important direction for agricultural development. Currently, deep learning-based weeding methods mainly fall into two categories: one is precise chemical spraying of weeds, but while this method reduces harm to crops, it still increases weed resistance during the weeding process, which is detrimental to long-term development; the other is a physical method using laser weeding, which can greatly reduce harm to crops and the environment. Several problems exist with laser weeding: first, it is less effective in areas with dense weed growth; second, it requires high precision in positioning, making it difficult to implement in complex farmland environments; and third, the essence of laser weeding is the rapid evaporation of moisture from weeds, which is not very efficient in practical work.

[0005] In the existing technology, some patents disclose devices that simultaneously set up laser and spraying devices. However, in practice, the laser device is used to remove weeds, and the spraying device is used to spray crops. These devices are still used independently, and the problems mentioned above still exist. Summary of the Invention

[0006] To address the various problems associated with simple chemical weeding and simple laser weeding as mentioned in the background art, this invention provides a laser-chemical synergistic weeding method based on target detection. This invention not only overcomes the harm of increasing weed resistance during chemical weeding, but also makes up for the problems of high precision positioning, difficulty in weeding in high-density weed areas, and low work efficiency in laser weeding.

[0007] To achieve the above objectives, this invention proposes a laser-chemical synergistic weeding method and system based on target detection, as follows:

[0008] Firstly, this invention proposes a laser-chemical synergistic weeding method based on target detection, comprising a control system and weeding equipment. The control system is at its core, using a target detection algorithm to locate the coordinates of weeds. A laser emitting device and a chemical spraying device work together to remove the weeds, thus achieving weed control in farmland. Specifically, the method includes the following steps:

[0009] Acquire image data of the fields;

[0010] Filter the image data to remove image noise;

[0011] The image data is packaged and segmented so that each image block has a corresponding address. The controller reads the image data according to the corresponding address, parses the received image data, and sends it to the target detection model. The target detection model obtains the category and image coordinates of the weed target. The image coordinates of the weeds are mapped to obtain the actual geographic coordinates. Based on the actual geographic coordinates, the laser emitting device is controlled to emit lasers to strike the main body of the weeds. After the laser emitting device has finished working, the chemical spraying device sprays herbicide on the laser-treated weed wounds, thereby removing the weeds.

[0012] As a further technical solution, image data is acquired through cameras, and each camera is labeled, as well as the image data collected by each camera, in order to distinguish the image data collected by different cameras.

[0013] As a further technical solution, the bilateral filter used for image denoising encapsulates the filter algorithm into a function, and the input data of each camera is fed into this function for denoising processing.

[0014] As a further technical solution, the image data is packaged and segmented as follows: the controller performs raster segmentation on the image data transmitted from each camera, and each piece of image data is sent to the corresponding address in the controller. The control system can quickly obtain the image data by reading the fixed address of the controller.

[0015] As a further technical solution, the target detection model is based on the FPN+PAN architecture. After training on the server, the optimal model parameters are obtained, and finally it is ported to the control system for application.

[0016] As a further technical solution, the mapping between image coordinates and actual geographic coordinates is achieved by taking one of the cameras as the origin of the spatial coordinate system, with downward as the positive z-axis, rightward as the positive x-axis, and forward as the y-axis, thus mapping the image coordinates to the actual geographic coordinates.

[0017] Secondly, based on the laser-chemical synergistic weeding method of target detection, the present invention proposes a system, including a mobile platform, a control system, a camera, a motor, a slide rail, a platform, a laser emitting device, and a chemical spraying device.

[0018] The control system is placed on top of the movable platform;

[0019] The camera is fixed to the bottom of the movable platform, and there is a set distance between two adjacent cameras;

[0020] The motor and slide rail are placed adjacent to each other at the bottom of the movable platform, with the overall position located behind the camera. A movable platform is mounted on the slide rail, and an edge switch is installed at each end of the slide rail. A laser emitting device and a chemical spraying device are placed on the platform.

[0021] As a further technical solution, the laser emitting device and the chemical spraying device are mounted on a drive platform. The drive platform includes a rotating platform, a guide rail platform, a stepper motor, and a support. The rotating platform is connected to the top of the guide rail platform and can drive the guide rail platform to rotate. A stepper motor is installed on the guide rail platform, and the stepper motor drives the connecting piece to move linearly along the guide rail platform through a lead screw transmission device. A support is installed on the connecting piece, and the laser emitting device and the chemical spraying device are installed on the support. The rotating platform and the stepper motor together drive the laser emitting device and the chemical spraying device to perform circular motion with the axis of the rotating platform as a reference point, with the radius decreasing and then increasing again.

[0022] As a further technical solution, the camera is configured with one row and multiple columns.

[0023] As a further technical solution, the control system is an embedded computer composed of ARM and chip system.

[0024] The technical effects of this invention are as follows:

[0025] This invention provides a laser-chemical synergistic weeding method based on target detection. First, images of weeds are acquired using three cameras. Then, a target detection algorithm based on a SoC (System-on-a-Chip) is used to detect and locate the weeds' coordinates. A laser-emitting device strikes the main body of the weed based on the located coordinates, creating obvious wounds. Next, a chemical spraying device sprays herbicide onto the wounded weeds. Compared to simple chemical weeding, this method requires only a small amount of chemical to remove the weeds. In other words, this invention reduces the amount of chemical used and improves weeding efficiency compared to simple laser weeding, achieving effective weed removal.

[0026] This invention uses a composite embedded computer composed of ARM+SoC as its core and a novel method of weeding by combining laser and chemical agents to remove weeds in farmland. This method avoids chemical pollution caused by excessive pesticides, greatly optimizes the limitations of laser weeding, improves agricultural production efficiency, and solves the problems of high environmental pollution and low weeding efficiency in related technologies.

[0027] The present invention enables continuous laser operation from the outer edge of the weeds to the center of the weeds through the combined movement of the rotating platform and the guide rail platform. First, the scattered leaves of the weeds are removed, and then the laser operation is carried out on the center position where the roots and stems of the weeds are located to further remove the weeds. This method can not only make up for the shortcomings of the current target detection model that cannot achieve accurate positioning due to the low recognition accuracy, but also increase the stability of the laser in the process of removing weeds. Attached Figure Description

[0028] Figure 1 This is a flowchart of the method of the present invention;

[0029] Figure 2 This is a schematic diagram of the device distribution at the bottom of the mobile platform of the present invention;

[0030] Figure 3 This is a schematic diagram of the device distribution at the top of the mobile platform of the present invention;

[0031] Figure 4 This is a schematic diagram of the stage of the present invention;

[0032] In the diagram: 1. Rotary platform; 2. Guide rail platform; 3. Stepper motor; 4. Support frame; 5. Laser emitting device; 6. Screw drive device. Detailed Implementation

[0033] It should be noted that, unless otherwise specified, the embodiments and features described in this application can be combined with each other. The present invention will now be described in detail with reference to the accompanying drawings and embodiments.

[0034] To enable those skilled in the art to better understand the present invention, the technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings of the embodiments. 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 should fall within the scope of protection of the present invention.

[0035] It should be noted that the terms "first," "second," etc., in the specification, claims, and accompanying drawings of this invention are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence. It should be understood that such data can be interchanged where appropriate for the embodiments of the invention described herein. Furthermore, the terms "comprising" and "having," and any variations thereof, are intended to cover non-exclusive inclusion; for example, a process, method, system, product, or apparatus that comprises a series of steps or units is not necessarily limited to those steps or units explicitly listed but may include other steps or units not explicitly listed or inherent to such processes, methods, products, or apparatus.

[0036] Glossary: ​​The "synergy" mentioned in this invention refers to the synergy of three technologies. Specifically, firstly, high-definition cameras collect real-time field image data, and the target detection algorithm of the chip system realizes the target detection and coordinate positioning of weeds; secondly, the laser emitting device first strikes the main body of the weeds according to the positioning coordinates, causing obvious wounds on the weeds; thirdly, a chemical spraying device sprays herbicide onto the wounds of the laser-treated weeds to achieve complete removal of the weeds; that is, the synergy between image acquisition, laser emission, and chemical agents; it should be noted that the laser weeding and chemical spraying of this invention are not independent weeding methods. The chemical agent is sprayed on the laser-treated weeds, truly realizing the integration and synergy of laser weeding and chemical spraying.

[0037] As described in the background section, existing deep learning-based weeding methods mainly fall into two categories. One category involves precise chemical spraying of weeds, but while this method reduces harm to crops, it still increases weed resistance during the weeding process, which is detrimental to long-term development. The other category uses a physical method of laser weeding, which can significantly reduce harm to crops and the environment. Several problems exist with laser weeding: firstly, it is less effective in areas with dense weed growth; secondly, it requires high precision in positioning, making it difficult to implement in complex farmland environments; and thirdly, the essence of laser weeding is rapid evaporation of moisture from weeds, resulting in low efficiency in practical applications. To address these issues, this embodiment discloses a laser-chemical synergistic weeding method and system based on target detection, including a control system and weeding equipment. The method uses an embedded computer composed of an ARM architecture processor (ARM) and a system-on-a-chip (SoC) as the control system. It acquires real-time field image data through a high-definition camera, and the SoC's target detection algorithm achieves target detection and coordinate positioning of weeds. The laser-emitting device first strikes the main body of the weeds based on the positioning coordinates, causing obvious damage. Then, a chemical spraying device sprays herbicides to remove the weeds. This invention uses a composite embedded computer composed of an ARM+ chip system as its core, and employs a novel method of combined laser and chemical weed control to remove weeds in farmland. This avoids chemical pollution caused by excessive pesticides, greatly optimizes the limitations of laser weed control, and improves agricultural production efficiency.

[0038] Specifically, this embodiment provides a laser-chemical synergistic weeding method based on target detection, including acquiring image data of the field; filtering the image data to remove image noise; packaging and segmenting the image data so that each image has a corresponding address; the controller reads the image data according to the corresponding address; parses the received image data and sends it to the target detection model, which obtains the category and image coordinates of the weed target; maps the image coordinates of the weeds to obtain their actual geographic coordinates; controls a laser emitting device to emit lasers to strike the main body of the weeds according to the actual geographic coordinates; and after the laser emitting device has finished working, a chemical spraying device sprays herbicide onto the laser-treated weed wounds to remove the weeds.

[0039] Specifically, Figure 1 This is a flowchart of a weeding method according to an embodiment of the present invention, such as... Figure 1 As shown, the flowchart includes the following steps:

[0040] Step 102: The ARM controller simultaneously captures image data of the field using three cameras;

[0041] Step 104: The acquired image data will first be filtered in the ARM to remove image noise;

[0042] Step 106: The ARM performs packet segmentation processing on the image data and transmits it to the chip system via bidirectional RAM;

[0043] Step 108: The chip system parses the received image data and then sends it into the target detection model;

[0044] Step 110: The target detection model obtains the category and image coordinates of the weed targets;

[0045] Step 112: The chip system maps the image coordinates of the weeds to obtain their actual geographic coordinates;

[0046] Step 114: The chip system transmits the actual geographic coordinates back to the ARM via IIC communication;

[0047] Step 116: The ARM drives the stepper motor corresponding to the actual geographical coordinates, and the stepper motor drives the stage to move.

[0048] Step 118: After the platform moves to the target position, the laser emitting device emits a laser to strike the main body of the weeds.

[0049] Step 120: After the laser emitting device has finished working, the chemical spraying device sprays herbicide on the weeds to remove them.

[0050] Furthermore, in step 102, each camera is labeled as camera 01, camera 02, and camera 03, and the image data collected by each camera is also labeled to distinguish the image data collected by different cameras.

[0051] Furthermore, in step 104, the filter used for image denoising is a bilateral filter. The filter algorithm is encapsulated into a function using C language, and the input data of each camera is connected to this function for denoising processing.

[0052] Furthermore, in step 106, the ARM performs grid segmentation on the image data transmitted from each camera, and each piece of image data is sent to the corresponding address in RAM. The chip system can quickly obtain the image data by reading the fixed address in RAM. This method can achieve high-speed and accurate data transmission through ping-pong operation.

[0053] Furthermore, in step 108, the target detection model is based on the FPN+PAN architecture. After training on the server, the optimal model parameters are obtained, and finally, it is ported to the chip system for application. Furthermore, in step six, the mapping between image coordinates and actual geographic coordinates is performed with camera 02 as the origin of the spatial coordinate system, downward as the positive z-axis, rightward as the positive x-axis, and forward as the y-axis, mapping the image coordinates to the actual geographic coordinates.

[0054] Furthermore, in step 116, the stepper motor controls the slide rail through the transmission structure. The slide rail is connected to the stage, so the stepper motor can drive the stage to move to the target coordinates. An edge switch is placed at each end of the slide rail. The stage is a thin iron sheet on which the laser emitting device and the chemical spraying device are fixed. The center of the laser emitting device and the chemical spraying device is located on the axis of the stage.

[0055] Furthermore, in steps 118 and 120, the target coordinates to which the platform is moved are the coordinates corresponding to the laser emitting device. The laser emitting device performs rapid operation and can cause effective damage to the main body or leaf surface of the weeds within 100 milliseconds. Then, the chemical spraying device sprays a small amount of herbicide to cover the wounds of the weeds, thereby achieving targeted weed removal.

[0056] Through the above steps, the coordinates of weeds are obtained based on the target detection algorithm. The ARM controls the laser emitting device and the chemical spraying device to achieve weed removal. This not only makes up for the problems of low efficiency and difficulty in accurate target positioning of laser weed removal, but also greatly reduces the serious environmental damage caused by the use of chemical weed removal alone. It provides a new weed removal method and improves the efficiency of agricultural production.

[0057] Furthermore, the control system and weeding equipment required by this weeding method are both mounted on a movable platform, and the arrangement of the various devices is as follows: Figure 2 As shown, three cameras are fixed at the bottom of the platform, arranged in a row of three columns, with a 60-centimeter interval between adjacent cameras. The motor and slide rail are placed adjacent to each other at the bottom of the platform, with the overall position located behind the cameras. Each slide rail is 40 centimeters long and has a movable platform mounted on it. An edge switch is installed at each end of the slide rail. A laser emitting device and a chemical spraying device are placed on the platform.

[0058] Furthermore, the specific structure of the stage is described in [reference needed]. Figure 4It may include a rotating platform 1, a guide rail platform 2, a stepper motor 3, and a bracket 4; the rotating platform 1 is connected to the top of the guide rail platform 2, and the rotating platform 1 can drive the guide rail platform 2 to rotate; the stepper motor 3 is installed on the guide rail platform 2, and the stepper motor 3 drives the connecting part to move linearly along the guide rail platform 2 through the lead screw transmission device 6; the bracket is installed on the connecting part, and the laser emitting device and the chemical agent spraying device are installed on the bracket; the rotating platform and the stepper motor together drive the laser emitting device and the chemical agent spraying device to make circular motion with the axis of the rotating platform as the reference point, with the radius decreasing and then increasing again. This method increases the tolerance for inaccurate weed positioning and the stability of laser weed removal during operation. The device utilizes the combined movement of a rotating platform and a guide rail platform to achieve continuous laser operation from the outer edge to the center of the weeds. First, it removes the scattered leaves, then lasers the central area where the weed roots and stems are located. Simultaneously, after the laser operation, a chemical spraying device sprays herbicides to further remove the weeds. This method not only overcomes the shortcomings of current target detection models that lack precise positioning due to low recognition accuracy, but also increases the stability of laser weed removal.

[0059] In one specific embodiment, technicians collect a certain amount of dataset as needed, annotate the data, and then divide the dataset into training, testing, and validation sets. The main purpose of the training set is to train and optimize model parameters; the testing set is used to evaluate the final model's performance after training; and the validation set is used to select hyperparameters that are more suitable for the model. After training is complete, the PyTorch-type parameter files are converted to ONNX format parameter files for better portability to the chip system. Once these steps are completed, the weeding device can perform efficient weeding operations under the control of the control system. First, a camera captures image data of the field, which is then sent to an ARM processor for image denoising. The image data is then transmitted via bidirectional RAM to the chip system for parsing. After passing through a target detection model, the image data yields corresponding weed classifications and image coordinates. These coordinates are then mapped to their actual geographic coordinates, which the chip system sends back to the ARM via IIC communication. The ARM processor, upon receiving the geographic coordinates, drives a stepper motor to move the platform to a fixed position. A laser emitter then fires a laser to strike the main weeds, while a chemical spraying device on the same platform sprays pesticides to remove the weeds. By distinguishing between crops and weeds using pre-defined target detection categories, the system can effectively target and locate identified weeds during operation. The weeding equipment efficiently removes weeds without human intervention, solving the problems of high pollution and low efficiency in existing technologies and achieving excellent results.

[0060] This embodiment proposes a laser-chemical synergistic weeding method and system based on target detection, which truly realizes the synergy between the laser emitting device and the chemical spraying device. The laser emitting device first strikes the main body of the weeds according to the positioning coordinates, causing obvious damage to the weeds. Then, the chemical spraying device sprays herbicides to remove the weeds. This new method of weeding by combining laser and chemical agents avoids chemical pollution caused by excessive pesticides, greatly optimizes the limitations of laser weeding, and improves agricultural production efficiency.

[0061] The above description is merely a preferred embodiment of the present invention and is not intended to limit the invention. Various modifications and variations can be made to the present invention by those skilled in the art. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the scope of protection of the present invention.

Claims

1. A laser-chemical synergistic weeding method based on target detection, characterized in that, as follows: Acquire image data of the fields; Filter the image data to remove image noise; Image data is packaged and segmented to assign a corresponding address to each image block. The controller reads the image data based on the corresponding address, parses the received image data, and sends it to the target detection model. The target detection model obtains the category and image coordinates of the weed target. The image coordinates of the weeds are mapped to obtain their actual geographic coordinates. Based on the actual geographic coordinates, the laser emitting device is controlled to emit lasers to strike the main body of the weeds. After the laser emitting device finishes its work, the chemical spraying device sprays herbicide onto the laser-treated weed wounds, thereby removing the weeds. Laser weeding and chemical spraying are not independent weeding methods; the chemical spraying is applied to the laser-treated weeds, achieving the integration and synergy of laser weeding and chemical spraying. The laser emitting device and the chemical spraying device are mounted on a drive platform, which includes a rotating platform, a guide rail platform, a stepper motor, and a support. The rotating platform is connected to the top of the guide rail platform and can drive the guide rail platform to rotate. A stepper motor is installed on the guide rail platform, and the stepper motor drives a connecting piece to move linearly along the guide rail platform through a lead screw transmission device. A support is installed on the connecting piece, and the laser emitting device and the chemical spraying device are installed on the support. The rotating platform and the stepper motor together drive the laser emitting device and the chemical spraying device to perform circular motion with the axis of the rotating platform as a reference point, with the radius decreasing and then increasing again.

2. The laser-chemical synergistic weeding method based on target detection as described in claim 1, characterized in that, Image data is acquired through cameras, and each camera is labeled, as well as the image data collected by each camera, in order to distinguish the image data collected by different cameras.

3. The laser-chemical synergistic weeding method based on target detection as described in claim 1, characterized in that, The bilateral filter used for image denoising encapsulates the filter algorithm into a function, and the input data from each camera is fed into this function for denoising processing.

4. The laser-chemical synergistic weeding method based on target detection as described in claim 1, characterized in that, The process of packaging and segmenting image data is as follows: the controller performs raster segmentation on the image data transmitted from each camera, and each piece of image data is sent to the corresponding address in the controller. The control system can quickly obtain the image data by reading the fixed address of the controller.

5. The laser-chemical synergistic weeding method based on target detection as described in claim 1, characterized in that, The target detection model is based on the FPN+PAN architecture. After training on the server, the optimal model parameters are obtained, and finally it is ported to the control system for application.

6. The laser-chemical synergistic weeding method based on target detection as described in claim 1, characterized in that, The mapping between image coordinates and actual geographic coordinates is achieved by taking one of the cameras as the origin of the spatial coordinate system, with downward as the positive z-axis, rightward as the positive x-axis, and forward as the y-axis, thus mapping the image coordinates to the actual geographic coordinates.

7. The system used in the laser-chemical synergistic weeding method based on target detection as described in any one of claims 1-6, characterized in that, It includes a mobile platform, control system, camera, motor, slide rail, platform, laser emitting device, and chemical spraying device; The control system is placed on top of the movable platform; The camera is fixed to the bottom of the movable platform, and there is a set distance between two adjacent cameras; The motor and slide rail are placed adjacent to each other at the bottom of the movable platform, with the overall position located behind the camera. A movable platform is mounted on the slide rail, and an edge switch is installed at each end of the slide rail. A laser emitting device and a chemical spraying device are placed on the platform.

8. The system as described in claim 7, characterized in that, The camera is configured with one row and multiple columns.

9. The system as described in claim 7, characterized in that, The control system is an embedded computer composed of ARM and chip system.

Images