A machine vision-based Chinese cabbage dynamic root cutting device and control method

Abstract

This invention provides a machine vision-based dynamic cabbage root-cutting device and control method, including a root-clamping mechanism, a vision detection module, a clamping and conveying mechanism, a dynamic cabbage root-cutting mechanism, an edge computing module, and a motion control module. The root-clamping mechanism and the clamping and conveying mechanism work together to convey the cabbage backward and maintain its stable posture. The vision detection module is mounted on the side frame of the clamping and conveying mechanism, with its image acquisition direction facing the conveying channel of the clamping and conveying mechanism. It is used to acquire images of the cabbage and output them to the edge computing module. The edge computing module is electrically connected to both the vision detection module and the motion control module. It processes the received images to obtain the spatial coordinates of the optimal cutting point and sends this coordinate information to the motion control module. The motion control module is electrically connected to the dynamic cabbage root-cutting mechanism and controls the dynamic cabbage root-cutting mechanism to drive the cutter to move axially to the target cutting position based on the received cutting point coordinates.

Description

Technical Field

[0001] This invention belongs to the field of agricultural machinery technology, and in particular relates to a dynamic root-cutting device and control method for cabbage based on machine vision. Background Technology

[0002] my country has a vast vegetable planting area, reaching 40 million mu (approximately 6.67 million hectares). Chinese cabbage, as an important vegetable category, accounts for about 12% of the total planting area, playing a crucial role in ensuring the supply of vegetables for urban and rural residents and stabilizing the vegetable market. In the entire process of Chinese cabbage production, the harvesting stage is labor-intensive and time-consuming, accounting for about 40% of the entire production process. The root and stem cutting process, in particular, still heavily relies on manual operation. The fixed cutting devices commonly used in current Chinese cabbage harvesting machinery, due to their simple structure and poor adaptability, often result in "overcutting" and "undercutting" in actual operation. Overcutting damages the edible parts of the cabbage, causing resource waste; undercutting leaves root residue, affecting appearance and subsequent processing efficiency. Both situations force a significant amount of manual labor to perform secondary trimming after harvesting, which is not only inefficient but also increases production costs and labor burden. Summary of the Invention

[0003] To address the aforementioned technical problems, this invention provides a machine vision-based dynamic root-cutting device and control method for Chinese cabbage, which helps improve the cutting accuracy during harvesting, reduce the damage rate during harvesting, and remove the outer leaves together without the need for secondary manual cutting, thereby improving harvesting efficiency.

[0004] This invention relates to a cabbage conveying mechanism for transporting harvested cabbages backwards. A vision detection module is mounted on the side frame of the cabbage conveying mechanism to acquire images of the cabbage and transmit them to an edge computing module. The edge computing module is connected to both the vision detection module and a motion control module. After processing the images transmitted by the vision detection module, the edge computing module obtains the optimal cutting point coordinates and outputs them to the motion control module. The motion control module then controls the dynamic cabbage root-cutting mechanism to adjust the cutter position. This invention can reduce damage rates and improve harvesting efficiency by adjusting the cutter position in real time.

[0005] The present invention achieves the above-mentioned technical objectives through the following technical means.

[0006] A machine vision-based dynamic cabbage root-cutting device includes a root-clamping mechanism, a vision detection module, a clamping and conveying mechanism, a dynamic cabbage root-cutting mechanism, an edge computing module, and a motion control module.

[0007] Both the root clamping mechanism and the clamping and conveying mechanism are mounted on the frame and are used together to convey the cabbage backward and keep the cabbage in a stable posture.

[0008] The root clamping mechanism is installed below the clamping and conveying mechanism, and the cabbage dynamic root cutting mechanism is also installed below the clamping and conveying mechanism and located behind the root clamping mechanism.

[0009] The visual inspection module is installed on the side frame of the clamping and conveying mechanism. The image acquisition direction of the visual inspection module is facing the conveying channel of the clamping and conveying mechanism. It is used to acquire cabbage images and output them to the edge computing module.

[0010] The edge computing module is electrically connected to the visual detection module and the motion control module respectively. It is used to process the received image, obtain the spatial coordinates of the optimal cutting point, and send the coordinate information to the motion control module.

[0011] The motion control module is electrically connected to the cabbage dynamic root cutting mechanism and is used to control the cabbage dynamic root cutting mechanism to drive the cutter to move axially to the target cutting position according to the received cutting point coordinates.

[0012] In the above scheme, the clamping mechanism includes symmetrically arranged clamping chains; the clamping chains are provided with bent steel plates; the upper end of the bent steel plates is connected to the frame of the clamping and conveying mechanism, and the clamping mechanism is suspended and installed below the clamping and conveying mechanism through the bent steel plates.

[0013] The bent steel plate is provided with a clamping sprocket motor mounting plate. The clamping sprocket motor mounting plate is used to install the clamping motor. The output shaft of the clamping motor is connected to the active clamping sprocket to drive the clamping chain to rotate.

[0014] A sensor mounting base is installed on the clamping sprocket motor mounting plate;

[0015] A speed sensor is installed on the sensor mounting base. The speed sensor is used to detect the running speed of the clamp chain in real time. The speed sensor is electrically connected to the motion control module and transmits the detected speed signal to the motion control module.

[0016] In the above scheme, the cabbage dynamic root cutting mechanism also includes a bending base, a motor input shaft, a motor mounting plate, a gearbox, a motor, a cutter flange, and a motor output shaft;

[0017] The bending base is mounted on the frame of the clamping and conveying mechanism; the motor mounting plate is located below the bending base; the motor is mounted on the motor mounting plate; the gearbox is fixed to the bending base by a gearbox fixing plate, and the gearbox contains two meshing bevel gears with round holes; the motor input shaft is connected to the output end of the motor and is connected to the bevel gear with round holes; the motor output shaft is connected to the bevel gear with round holes and is rotatably supported on the bending base by a bearing seat;

[0018] The cutter flange is mounted on the motor output shaft and is driven to rotate by the motor output shaft;

[0019] The cutter is mounted on the cutter flange; the two cutters are arranged symmetrically vertically and partially overlap.

[0020] Furthermore, it also includes the cutter flange gasket;

[0021] The cutter flange gasket is set on one side of the cutter flange, so that the two cutters are arranged symmetrically from top to bottom and partially overlap.

[0022] Furthermore, this also includes shift forks and servo electric cylinders;

[0023] The shift fork is connected to the cutter flange and is used to drive the cutter flange to move axially;

[0024] The servo electric cylinder is mounted on the bending base, and the lead screw of the servo electric cylinder is connected to the shift fork.

[0025] Furthermore, the outer circumference of the cutter flange is provided with an annular shift fork groove;

[0026] The front end of the fork is a U-shaped structure. The two claws of the U-shaped fork are engaged in the annular fork groove, and there is a degree of freedom of circumferential relative movement between the fork and the annular fork groove, so as to realize that the cutter flange can move axially independently while rotating.

[0027] The rear end of the shift fork is connected to one end of the shift fork connecting plate one, and the other end of the shift fork connecting plate one is connected to the shift fork connecting plate two.

[0028] The servo electric cylinder is mounted on the bending base via an L-shaped fixed base, and the L-shaped fixed base is connected to the bending base.

[0029] The lead screw of the servo electric cylinder is provided with a limit groove at its top, and the shift fork connecting plate is connected to the limit groove.

[0030] In the above solution, the visual inspection module includes a depth camera, an angle adjustment mounting plate, and a camera mounting base;

[0031] The depth camera is mounted on an angle adjustment mounting plate, which is connected to the camera mounting base via fastening bolts. The camera mounting base is mounted on a bent steel plate of a clamping mechanism. The depth camera is connected to an edge computing module.

[0032] A control method for the machine vision-based dynamic cabbage root-cutting device includes the following steps:

[0033] Step S1: The visual detection module acquires an image of the cabbage and transmits it to the edge computing module;

[0034] Step S2: The edge computing module processes the image from step S1, identifies the image coordinates of the cabbage cutting point, and calculates the position of the cutting point in the conveying direction using the calibration relationship. and the target position along the cutting axis The output is sent to the motion control module;

[0035] Step S3: The motion control module receives the speed of the clamping chain measured by the speed sensor of the clamping mechanism. According to the preset speed ratio Calculate conveyor belt speed And adjust the speed ratio to stabilize the cabbage's posture;

[0036] Step S4: The motion control module determines the distance between the depth camera and the cutter based on the fixed distance between them. , position of the cutting point conveying direction And the speed of cabbage delivery Predict the remaining time for the cutting point to reach the cutting line position of the cutter. ,in satisfy , The moment of image acquisition;

[0037] Step S5: The motion control module determines the current axial position of the cutter. and target location Determine the axial movement distance, i.e., the position error. and combined with the remaining time Calculate the axial movement speed of the cutter ;

[0038] Step S6: The motion control module will adjust the position error. If the error is less than a preset threshold, the cutter is kept in its current position; if the error is greater than the threshold, the cabbage dynamic root-cutting mechanism is controlled to drive the cutter at a certain speed. Move axially to the target position.

[0039] In the above scheme, in step S2, the edge computing module uses a deep learning model to process the image. The deep learning model is the YOLOv8n-pose model. The YOLOv8n-pose model takes the distortion-corrected whole frame image as input and synchronously outputs the target bounding box and the coordinates of the cutting point key points of the cabbage.

[0040] In the above scheme, step S2, the calibration relationship conversion includes using the calibration parameters and depth values ​​of the depth camera. The pixel coordinates of the cutting point are determined based on the perspective projection model. Convert to 3D coordinates in the camera coordinate system. Then, through rigid body transformation, the coordinates are converted to coordinates in the world coordinate system. ), of which, the world coordinate system The axis indicates the conveying direction. The axis is the axial direction of the cutter's movement, thus obtaining , .

[0041] Compared with the prior art, the beneficial effects of the present invention are:

[0042] This invention utilizes a machine vision-based dynamic root-cutting device for cabbage to achieve low-loss harvesting, improve harvesting efficiency, and reduce manual intervention. The vision detection module of this invention can acquire the optimal cutting point's position information along the conveying direction and the cutter's axis, enabling online identification and coordinate output of the cabbage cutting point, thus improving cutting positioning accuracy. The dynamic root-cutting mechanism of this invention uses an electric cylinder to drive the cutter to complete dynamic root cutting, enabling precise root cutting and removal of excess outer leaves for cabbages of different varieties and sizes.

[0043] This invention stabilizes the cabbage's posture during transport through the synergistic action of the clamping and conveying mechanism and the root clamping mechanism, laying the foundation for precise cutting. The visual detection module identifies the cutting point position in real time, and the edge computing module and motion control module work together to control the dynamic root cutting mechanism of the cabbage to dynamically adjust the position of the cutter. This achieves full automation of the entire process from "visual perception → coordinate calculation → dynamic adjustment → precise cutting," effectively solving the common problems of "overcutting" and "missed cutting" in traditional fixed cutting devices. It significantly reduces the harvest damage rate, improves harvest efficiency, and allows the process that originally required secondary manual trimming to be completed in one go during the harvest.

[0044] The visual inspection module of this invention adopts a YOLOv8n-pose integrated model based on deep learning to perform target detection and key point regression on the distortion-corrected whole frame image, and simultaneously outputs the target bounding box of the Chinese cabbage and the pixel coordinates of the cutting point. Combining the depth value of the depth camera and the camera calibration parameters, the pixel coordinates are accurately converted into the conveying direction position and the axial target position of the cutter in the device's world coordinate system through perspective projection model and rigid body transformation, realizing online identification and high-precision spatial positioning of the Chinese cabbage cutting point, and providing an accurate position reference for dynamic root cutting.

[0045] The dynamic root-cutting mechanism for cabbage of this invention employs a cooperative structure between a fork and a ring-shaped fork groove on the cutter flange. The fork engages in the ring groove, allowing for relative circumferential movement, thus decoupling the high-speed rotation and axial movement of the cutter. The rotation is independently driven by a motor via bevel gear transmission, while the axial movement is driven by a servo electric cylinder via a fork connecting plate; the two do not interfere with each other. This structure, combined with a dynamic speed control method based on remaining time prediction, enables the cutter to adjust its axial position in real time according to visual detection results during cabbage transport, achieving precise root cutting for cabbages of different varieties and sizes, while simultaneously removing excess outer leaves, significantly improving cutting adaptability and consistency. Attached Figure Description

[0046] Figure 1 This is a schematic diagram of the relative positions between the cabbage conveying and alignment mechanism and the cabbage dynamic root cutting mechanism according to an embodiment of the present invention.

[0047] Figure 2 This is a schematic diagram of the structure of the root clamping mechanism, the cabbage floating root cutting mechanism, the visual detection module, the edge computing module, and the motion control module according to an embodiment of the present invention.

[0048] Figure 3 This is a schematic diagram of the speed sensor installation according to one embodiment of the present invention.

[0049] Figure 4 This is a schematic diagram of the overall structure of a floating root-cutting mechanism for cabbage according to an embodiment of the present invention.

[0050] Figure 5 This is a rear view structural schematic diagram of a cabbage floating root-cutting mechanism according to an embodiment of the present invention.

[0051] Figure 6 This is a bottom view of the cabbage floating root-cutting mechanism according to an embodiment of the present invention.

[0052] Figure 7 This is a schematic diagram of the cutter flange structure according to one embodiment of the present invention.

[0053] Figure 8 This is a schematic diagram of a shift fork structure according to an embodiment of the present invention.

[0054] Figure 9 This is a schematic diagram of a bending base according to an embodiment of the present invention.

[0055] Figure 10 This is a schematic flowchart of the control method for a machine vision-based dynamic cabbage root-cutting device according to an embodiment of the present invention.

[0056] Figure 11 This is a schematic diagram of the conveying direction of the cabbage and the moving direction of the cutting knife according to an embodiment of the present invention.

[0057] In the diagram, 1. Root clamping mechanism; 101. Root clamping chain; 102. Bending steel plate; 103. Root clamping sprocket motor mounting plate; 104. Sensor mounting base; 105. Speed ​​sensor; 2. Vision inspection module; 201. Depth camera; 202. Angle adjustment mounting plate; 203. Camera mounting base; 3. Clamping and conveying mechanism; 4. Cabbage floating root cutting mechanism; 401. Bending base; 402. Dust cover; 403. Gearbox fixing plate; 404. Motor input shaft; 405. Gearbox; 406. Motor mounting plate; 407. Motor; 408. Cutter blade; 409. Cutter blade flange; 410. Shift fork; 411. Motor output shaft; 412. Shift fork connecting plate one; 413. Shift fork connecting plate two; 414. L-shaped fixed base; 415. Servo electric cylinder; 416. Lead screw; 417. Limiting groove; 418. Cutter blade flange gasket; 419. Bearing seat; 420. Circular hole bevel gear one; 421. Circular hole bevel gear two; 422. Annular Shift fork groove; 5. Edge computing module; 6. Motion control module. Detailed Implementation

[0058] Embodiments of the present invention are described in detail below, examples of which are illustrated in the accompanying drawings, wherein the same or similar reference numerals denote the same or similar elements or elements having the same or similar functions throughout. The embodiments described below with reference to the accompanying drawings are exemplary and intended to explain the present invention, and should not be construed as limiting the present invention.

[0059] In the description of this invention, it should be understood that the terms "center," "longitudinal," "lateral," "length," "width," "thickness," "front," "rear," "left," "right," "upper," "lower," "axial," "radial," "vertical," "horizontal," "inner," and "outer," etc., indicating orientation or positional relationships based on the orientation or positional relationships shown in the accompanying drawings, are only for the convenience of describing the invention and simplifying the description, and do not indicate or imply that the device or element referred to must have a specific orientation, or be constructed and operated in a specific orientation, and therefore should not be construed as a limitation of the invention. Furthermore, the terms "first" and "second" are used for descriptive purposes only and should not be construed as indicating or implying relative importance or implicitly specifying the number of indicated technical features. Thus, a feature defined with "first" or "second" may explicitly or implicitly include one or more of that feature. In the description of this invention, "a plurality of" means two or more, unless otherwise explicitly specified.

[0060] In this invention, unless otherwise explicitly specified and limited, the terms "installation," "connection," "linking," and "fixing," etc., should be interpreted broadly. For example, they can refer to a fixed connection, a detachable connection, or an integral connection; they can refer to a mechanical connection or an electrical connection; they can refer to a direct connection or an indirect connection through an intermediate medium; and they can refer to the internal connection of two components. Those skilled in the art can understand the specific meaning of the above terms in this invention according to the specific circumstances.

[0061] Figure 1 and 2 The image shows a preferred embodiment of the machine vision-based dynamic cabbage root-cutting device. The device includes a root-clamping mechanism 1, a vision detection module 2, a clamping and conveying mechanism 3, a dynamic cabbage root-cutting mechanism 4, an edge computing module 5, and a motion control module 6. The root-clamping mechanism 1 and the clamping and conveying mechanism 3 are both mounted on a frame and work together to convey the cabbage backward and maintain its stable posture. The root-clamping mechanism 1 is installed below the clamping and conveying mechanism 3, and the dynamic cabbage root-cutting mechanism 4 is also installed below the clamping and conveying mechanism 3 and located behind the root-clamping mechanism 1. The vision detection module 2... Mounted on the side frame of the clamping and conveying mechanism 3, the image acquisition direction of the vision detection module 2 faces the conveying channel of the clamping and conveying mechanism 3, and is used to acquire cabbage images and output them to the edge computing module 5; the edge computing module 5 is electrically connected to the vision detection module 2 and the motion control module 6 respectively, and is used to process the received images to obtain the spatial coordinates of the optimal cutting point, and send the coordinate information to the motion control module 6; the motion control module 6 is electrically connected to the cabbage dynamic root cutting mechanism 4, and is used to control the cabbage dynamic root cutting mechanism 4 to drive the cutter 408 to move up and down along the axis to the target cutting position according to the received cutting point coordinates.

[0062] The clamping mechanism 1 includes symmetrically arranged clamping chains 101; a bent steel plate 102 is provided on the clamping chain 101; the upper end of the bent steel plate 102 is connected to the frame of the clamping and conveying mechanism 3, and the clamping mechanism 1 is suspended below the clamping and conveying mechanism 3 via the bent steel plate 102; a clamping sprocket motor mounting plate 103 is provided below the bent steel plate 102, the clamping sprocket motor mounting plate 103 is used to install the clamping motor, the output shaft of the clamping motor is connected to the active clamping sprocket, and drives the clamping chain 101 to rotate; a sensor mounting base 104 is installed on the clamping sprocket motor mounting plate 103; a speed sensor 105 is installed on the sensor mounting base 104, the speed sensor 105 is used to detect the running speed of the clamping chain 101 in real time, the speed sensor 105 is electrically connected to the motion control module 6, and transmits the detected speed signal to the motion control module 6.

[0063] The root clamping mechanism 1 ensures the stability of the cabbage's posture during harvesting.

[0064] like Figure 2 and Figure 3 As shown, the cabbage floating root cutting mechanism 4 is installed below the frame of the clamping and conveying mechanism 3 and is used to adjust the cutting blade 408 to move up and down along the axial direction. The speed sensor 105 is installed on the sensor mounting base 104 by bolts, and the sensor mounting base 104 is fixed to the root clamping sprocket motor mounting plate 103 by bolts.

[0065] like Figure 4-9 As shown, the cabbage dynamic root cutting mechanism 4 includes a bending base 401, a dust cover 402, a gearbox fixing plate 403, a motor input shaft 404, a gearbox 405, a motor mounting plate 406, a motor 407, a cutter 408, a cutter flange 409, a shift fork 410, a motor output shaft 411, a shift fork connecting plate one 412, a shift fork connecting plate two 413, an L-shaped fixing base 414, a servo electric cylinder 415, a lead screw 416, a limit groove 417, a cutter flange pad 418, a bearing seat 419, a round hole bevel gear one 420, and a round hole bevel gear two 421;

[0066] The bending base 401 is fixedly mounted on the frame of the clamping and conveying mechanism 3; the motor mounting plate 406 is welded to the bottom of the bending base 401, and the motor 407 is mounted on the motor mounting plate 406 by bolts; the gearbox 405 is fixed to the bending base 401 by the gearbox fixing plate 403, and the gearbox 405 contains meshing bevel gear 1 420 and bevel gear 2 421; the output end of the motor 407 is connected to the motor input shaft 404, and the motor input shaft 404 is connected to bevel gear 2 421; the motor output shaft 411 is connected to bevel gear 1 420, and the motor output shaft 411 is supported by the bearing seat 419, which is fixed to the bending base 401 by bolts; the dust cover 402 is fixed above the bending base 401 by bolts to seal the upper part of the gearbox 405.

[0067] The cutter flange 409 is mounted on the motor output shaft 411 and is driven to rotate by the motor output shaft 411; two cutters 408 are provided and are respectively mounted on the cutter flange 409. A cutter flange gasket 418 is installed on one of the cutter flanges 409, so that the two cutters 408 are arranged symmetrically above and below and partially overlap.

[0068] like Figure 7 As shown, the outer circumference of the cutter flange 409 is provided with an annular shift fork groove 422; as Figure 8As shown, the shift fork 410 has a U-shaped structure, with its two claws engaging the annular shift fork groove 422 of the cutter flange 409, allowing circumferential relative movement between the shift fork 410 and the annular groove; the shift fork 410 is welded to shift fork connecting plate one 412, and shift fork connecting plate one 412 is welded to shift fork connecting plate two 413; the L-shaped fixed base 414 is fixed to the bending base 401 by bolts passing through the dust cover 402; the servo electric cylinder 415 is mounted on the L-shaped fixed base 41 by bolts. 4. The top end of the lead screw 416 of the servo electric cylinder 415 is provided with a limiting groove 417, and the second shift fork connecting plate 413 is connected to the limiting groove 417 by bolts; the servo electric cylinder 415 drives the lead screw 416 to extend and retract, and drives the shift fork 410 to move axially through the second shift fork connecting plate 413 and the first shift fork connecting plate 412. The shift fork 410 drives the cutter flange 409 and the cutter 408 to move axially through the annular shift fork groove 422, while allowing the cutter flange 409 to maintain rotational movement.

[0069] The aforementioned dynamic cabbage root cutting mechanism 4 is mounted on the frame of the clamping and conveying mechanism 3 via a bending base 401; the cutter 408 is mounted on the cutter flange 409, with a cutter flange pad 418 mounted on one side to make the two cutters 408 symmetrically arranged vertically, with some overlapping areas; the cutter flange 409 is driven to rotate by the motor output shaft 411; the motor output shaft 411 is fixed by the bearing seat 419, which is fixed to the bending base 401 by bolts; the motor 407 is mounted on the motor mounting plate 406 by bolts, and the motor mounting plate 406 is welded to the underside of the bending base 401; the motor 407 drives the motor input shaft 404 to rotate, and the motor input shaft 404 is connected to the second round-hole bevel gear 421 in the gearbox 405; the second round-hole bevel gear 421 is fixed in the gearbox 405 by a retaining ring and connected to the first round-hole bevel gear 420; the first round-hole bevel gear 420 is connected to the motor output shaft 411, thereby realizing the rotation of the cutter 408.

[0070] The cutter flange 409 is provided with an annular shift fork groove on its outer circumference. The two claws of the shift fork 410 are engaged in the annular groove to apply axial force to the cutter flange 409, thereby driving the cutter flange 409 to move axially. The rotation of the cutter flange is provided by the motor output shaft 411. The shift fork 410 and the annular groove are allowed to move circumferentially relative to each other, thereby achieving motion decoupling between the high-speed rotation and axial adjustment of the cutter flange 409, ensuring that the rotational stability of the cutter 408 is not disturbed during axial adjustment. The shift fork 410 is welded to the shift fork connecting plate 1 412, and the shift fork connecting plate 1 412 is welded to the shift fork connecting plate 2 413. The shift fork connecting plate 2 413 is installed on the top of the lead screw 416 of the servo electric cylinder 415 by bolts and the limiting groove 417. The servo electric cylinder 415 is installed on the L-shaped fixed base 414 by bolts. The L-shaped fixed base 414 is fixed to the bending base 401 by bolts passing through the dust cover 402. The dust cover 402 is fixed above the bending base 401 by bolts.

[0071] The visual inspection module 2 includes a depth camera 201, an angle adjustment mounting plate 202, and a camera mounting base 203; for example Figure 2 As shown, the depth camera 201 is mounted on the angle adjustment mounting plate 202, which is connected to the camera mounting base 203 by fastening bolts. The camera mounting base 203 is mounted on the bent steel plate 102 by bolts. The depth camera 201 is connected to the edge computing module 5.

[0072] Preferably, the visual inspection module 2 is located inside the bent steel plate 102 of the clamping mechanism 1.

[0073] like Figure 10 As shown, a machine vision-based dynamic cabbage root-cutting device and control method includes the following steps:

[0074] Step S1: The visual detection module 2 acquires an image of the cabbage and transmits it to the edge calculation module 5;

[0075] Step S2: The edge computing module 5 processes the image from step S1, identifies the image coordinates of the cabbage cutting point, and calculates the position of the cutting point in the conveying direction through calibration relationships. and the target position along the cutting axis The output is sent to the motion control module 6;

[0076] Step S3: The motion control module 6 receives the speed of the clamping chain measured by the speed sensor 105 of the clamping mechanism 1. According to the preset speed ratio Calculate conveyor belt speed And adjust the speed ratio to stabilize the cabbage's posture;

[0077] Step S4: The motion control module 6 determines the fixed distance between the depth camera 201 and the cutter 408. , position of the cutting point conveying direction And the speed of cabbage delivery Predict the remaining time for the cutting point to reach the cutting line position of the cutter. ,in satisfy , The moment of image acquisition;

[0078] Step S5: The motion control module 6 determines the current axial position of the cutter 408 based on... and target location Determine the axial movement distance, i.e., the position error. and combined with the remaining time Calculate the axial movement speed of the cutter ;

[0079] Step S6: Motion control module 6 will adjust the position error. If the error is less than the preset threshold, the cutter 408 is kept in its current position; if the error is greater than the threshold, the cabbage dynamic root cutting mechanism 4 is controlled to drive the cutter 408 at a speed... Move axially to the target position.

[0080] This invention calculates the conveyor belt speed based on the speed of the clamping chain and a preset speed ratio, and adjusts the speed ratio to stabilize the cabbage posture. At the same time, it predicts the time it takes for the cutting point to reach the position of the cutting blade 408 based on the installation distance between the depth camera 201 and the cutting blade 408.

[0081] like Figure 11 As shown, the position error in the above scheme

[0082]

[0083] according to With the set allowable error threshold The comparison is made to determine whether the cutter movement needs to continue.

[0084] The mathematical model for the speed of the axial movement of the cutter in the cabbage dynamic root-cutting mechanism 4 is as follows:

[0085]

[0086]

[0087]

[0088]

[0089] in, It is a positional error. This is the current position of the cutting tool 408. This is the target location for the 408 cutting tool. D is the fixed distance from depth camera 201 to cutter 408, and D is the remaining transport distance from the currently detected cutting point to the cutting line of cutter 408. The edge computing module 5 outputs the position of the cutting point relative to the transport direction of the depth camera 201. The speed of the clamp chain is measured by speed sensor 105. It is the speed ratio used for attitude stabilization settings. It is the conveyor belt speed. This is the time point when depth camera 201 acquires the image. It is the time it takes for the cutting point position output by edge computing module 5 to reach the cutting line position of cutter 408. This is the current position of the cutting tool 408. This is the target location for the 408 cutting tool. That is, the distance that the cutter 408 moves to the target position. It is the speed at which the 408 cutting blade moves to the target position.

[0090] In step S2, the edge computing module 5 uses a deep learning model to process the image. The deep learning model is the YOLOv8n-pose model. The YOLOv8n-pose model takes the distortion-corrected whole frame image as input and synchronously outputs the target bounding box and the coordinates of the cutting point key points of the cabbage.

[0091] In a specific embodiment of the present invention, preferably, the edge computing module 5 employs a one-stage integrated algorithm for target detection and keypoint regression based on deep learning to construct a model for cabbage target detection and cutting point keypoint detection; wherein, the model adopts the YOLOv8n-pose model. The full-frame image acquired by the visual detection module 2 is first subjected to distortion correction processing to eliminate the influence of lens distortion on positioning accuracy. Then, the corrected image is directly input into the YOLOv8n-pose model for inference. The model simultaneously outputs the target bounding box, category, and detection confidence of the cabbage to be cut, and also outputs the coordinates of the cutting point keypoints.

[0092] In step S2, the calibration relationship conversion includes using the calibration parameters and depth values ​​of the depth camera 201. The pixel coordinates of the cutting point are determined based on the perspective projection model. Convert to 3D coordinates in the camera coordinate system. Then, through rigid body transformation, the coordinates are converted to coordinates in the world coordinate system. ), of which, the world coordinate system The axis indicates the conveying direction. The axis is the axial direction of the cutter's movement, thus obtaining , .

[0093] Specifically, to achieve the conversion of the detected cabbage cutting point from image coordinates to device motion control coordinates, a geometric model for the conversion from image coordinates to world coordinates is established based on the perspective projection model and the rigid body transformation principle. After the device is installed, the depth camera 201 is calibrated to obtain the camera's intrinsic and extrinsic parameters. The edge computing module 5 identifies the pixel coordinates of the cutting point. Then, the cutting point is back-projected using the depth value d at that pixel provided by the depth camera 201. Based on the perspective projection model, the distortion-free pixel coordinates and depth value are converted into three-dimensional coordinates in the camera coordinate system. Subsequently, the 3D point was transformed from the camera coordinate system to the world coordinate system using rigid body transformation relationships. The world coordinate system is fixed to the conveying and cutting mechanism, and the conveying direction is defined as... The axis is defined as the axial movement direction of the cutter 408. Axis, corresponding to the axial speed of the cutting tool Direction. The above position data and confidence information are then output as the cutting point position information to motion control module 6. The coordinate transformation relationship is as follows:

[0094] (1)

[0095] (2)

[0096] (3)

[0097] (4)

[0098] Where K is the camera intrinsic parameter matrix, , To normalize the focal length, , Principal point coordinates These are coordinates in the camera coordinate system, corresponding to image pixel coordinates. R is The rotation matrix is ​​given, where t is a 3×1 translation vector and d is the depth value provided by the depth camera 201. It is a world coordinate system.

[0099] Preferably, the motion control module 6 receives the cutting point position information output by the edge computing module 5, and simultaneously receives the root chain speed signal collected by the speed sensor 105. Based on the speed ratio relationship between the root chain and the conveyor belt, it obtains the conveyor belt running speed and adjusts the speed ratio between the two to stabilize the cabbage posture. Under the premise of stable posture, the motion control module 6 predicts the remaining time for the cutting point to reach the cutting position of the cutter 408 based on the relative installation position of the depth camera 201 and the cutter 408 and the position data of the cutting point in the conveying direction. It also calculates the target adjustment amount and adjustment time of the axial direction of the cutter 408 by combining the current axial position of the cutter 408, the response time of the servo cylinder 415 and the maximum allowable speed.

[0100] The visual detection module 2 collects field images and transmits them to the edge computing module 5; the edge computing module 5 processes the images collected by the visual detection module 2, obtains the coordinates of the cutting point, and outputs them to the motion control module 6.

[0101] The motion control module 6 sets a threshold for the axial target adjustment of the cutter. When the axial position error of the cutter 408 is within the allowable error range, the motion control module 6 controls the cabbage dynamic root cutting mechanism 4 to keep the axial position of the cutter 408 unchanged. When the axial position error of the cutter 408 exceeds the allowable error range and the remaining time meets the requirement for the servo cylinder 415 to complete the adjustment, the motion control module 6 controls the servo cylinder 415 to drive the cutter 408 to move axially, so that the cutter 408 moves to the target position.

[0102] This invention stabilizes the cabbage's posture during transport by the synergistic action of the clamping and conveying mechanism 3 and the root clamping mechanism 1, laying the foundation for precise cutting. The visual detection module 2 identifies the cutting point position in real time, and the edge calculation module 5 and the motion control module 6 work together to control the dynamic root cutting mechanism 4 to dynamically adjust the cutter position. This achieves full automation of the entire process from "visual perception → coordinate calculation → dynamic adjustment → precise cutting," effectively solving the common problems of "overcutting" and "missed cutting" in traditional fixed cutting devices. It significantly reduces the harvest damage rate, improves harvest efficiency, and allows the process that originally required secondary manual trimming to be completed in one go during the harvest.

[0103] The visual detection module 2 of this invention employs a YOLOv8n-pose integrated model based on deep learning to perform target detection and keypoint regression on the distortion-corrected whole-frame image, simultaneously outputting the target bounding box and the pixel coordinates of the cutting point. Combining the depth values ​​from the depth camera 201 with the camera calibration parameters, the pixel coordinates are precisely converted into the transport direction position in the device's world coordinate system through a perspective projection model and rigid body transformation. and the target position of the cutting tool axis This technology enables online identification and high-precision spatial positioning of the cutting points of Chinese cabbage, providing an accurate positional reference for dynamic root cutting.

[0104] The dynamic cabbage root-cutting mechanism 4 of this invention employs a mating structure between a fork 410 and an annular fork groove 422 on a cutter flange 409. The fork engages in the annular groove, allowing relative circumferential movement, thus decoupling the high-speed rotation and axial movement of the cutter 408. The rotation is independently driven by a motor 407 via a bevel gear transmission, while the axial movement is driven by a servo cylinder 415 via a fork connecting plate; the two movements do not interfere with each other. This structure, combined with a dynamic speed control method based on remaining time prediction, further enhances the performance. This allows the cutter to adjust its axial position in real time based on visual inspection results during the cabbage transport process, enabling precise root cutting for cabbages of different varieties and sizes, while removing excess outer leaves, significantly improving cutting adaptability and consistency.

[0105] This invention addresses the problems of high damage rates and poor adaptability of traditional fixed cutting devices in complex field environments. The described dynamic root-cutting mechanism for Chinese cabbage can identify the optimal cutting point online, achieving low-loss harvesting and improving harvesting efficiency.

[0106] It should be understood that although this specification is described according to various embodiments, not every embodiment contains only one independent technical solution. This way of describing the specification is only for clarity. Those skilled in the art should regard the specification as a whole. The technical solutions in each embodiment can also be appropriately combined to form other implementation methods that can be understood by those skilled in the art.

[0107] The detailed descriptions listed above are merely specific illustrations of feasible embodiments of the present invention and are not intended to limit the scope of protection of the present invention. All equivalent embodiments or modifications made without departing from the spirit of the present invention should be included within the scope of protection of the present invention.

Claims

1. A machine vision-based dynamic root cutting device for Chinese cabbage, characterized by, It includes a root clamping mechanism (1), a vision detection module (2), a clamping and conveying mechanism (3), a cabbage dynamic root cutting mechanism (4), an edge computing module (5), and a motion control module (6). The root clamping mechanism (1) and the clamping and conveying mechanism (3) are both installed on the frame and are used to jointly convey the cabbage backward and keep the cabbage in a stable posture. The root clamping mechanism (1) is installed below the clamping and conveying mechanism (3), and the cabbage dynamic root cutting mechanism (4) is also installed below the clamping and conveying mechanism (3) and located behind the root clamping mechanism (1). The visual inspection module (2) is installed on the side frame of the clamping and conveying mechanism (3). The image acquisition direction of the visual inspection module (2) is facing the conveying channel of the clamping and conveying mechanism (3), and is used to acquire cabbage images and output them to the edge computing module (5). The edge computing module (5) is electrically connected to the visual detection module (2) and the motion control module (6) respectively, and is used to process the received image, obtain the spatial coordinates of the optimal cutting point, and send the coordinate information to the motion control module (6). The motion control module (6) is electrically connected to the cabbage dynamic root cutting mechanism (4) and is used to control the cabbage dynamic root cutting mechanism (4) to drive the cutter (408) to move axially to the target cutting position according to the received cutting point coordinates.

2. The machine vision-based dynamic cabbage root-cutting device according to claim 1, characterized in that, The clamping mechanism (1) includes symmetrically arranged clamping chains (101); the clamping chains (101) are provided with bent steel plates (102); the upper end of the bent steel plates (102) is connected to the frame of the clamping and conveying mechanism (3), and the clamping mechanism (1) is suspended and installed below the clamping and conveying mechanism (3) through the bent steel plates (102); The bent steel plate (102) is provided with a clamping sprocket motor mounting plate (103) below it. The clamping sprocket motor mounting plate (103) is used to install the clamping motor. The output shaft of the clamping motor is connected to the active clamping sprocket to drive the clamping chain (101) to rotate. A sensor mounting base (104) is mounted on the clamping sprocket motor mounting plate (103). A speed sensor (105) is installed on the sensor mounting base (104). The speed sensor (105) is used to detect the running speed of the clamping chain (101) in real time. The speed sensor (105) is electrically connected to the motion control module (6) and transmits the detected speed signal to the motion control module (6).

3. The machine vision-based dynamic cabbage root-cutting device according to claim 1, characterized in that, The cabbage dynamic root cutting mechanism (4) also includes a bending base (401), a motor input shaft (404), a motor mounting plate (406), a gearbox (405), a motor (407), a cutter flange (409), and a motor output shaft (411). The bending base (401) is mounted on the frame of the clamping and conveying mechanism (3); the motor mounting plate (406) is located below the bending base (401); the motor (407) is mounted on the motor mounting plate (406); the gearbox (405) is fixed on the bending base (401) by the gearbox fixing plate (403), and the gearbox (405) is provided with a meshing round hole bevel gear one (420) and a round hole bevel gear two (421); the motor input shaft (404) is connected to the output end of the motor (407) and is connected to the round hole bevel gear two (421); the motor output shaft (411) is connected to the round hole bevel gear one (420) and is rotatably supported on the bending base (401) by the bearing seat (419); The cutter flange (409) is mounted on the motor output shaft (411) and is driven to rotate by the motor output shaft (411); The cutter (408) is mounted on the cutter flange (409); the two cutters (408) are arranged symmetrically above and below and partially overlap.

4. The machine vision-based dynamic cabbage root-cutting device according to claim 3, characterized in that, It also includes the cutter flange gasket (418); The cutter flange gasket (418) is set on one of the cutter flanges (409), so that the two cutters (408) are arranged symmetrically above and below and partially overlap.

5. The machine vision-based dynamic cabbage root-cutting device according to claim 3, characterized in that, It also includes a shift fork (410) and a servo electric cylinder (415); The fork (410) is connected to the cutter flange (409) and is used to drive the cutter flange (409) to move axially; The servo electric cylinder (415) is mounted on the bending base (401), and the lead screw (416) of the servo electric cylinder (415) is connected to the shift fork (410).

6. The machine vision-based dynamic cabbage root-cutting device according to claim 5, characterized in that, The outer circle of the cutter flange (409) is provided with an annular fork groove (422). The front end of the fork (410) is a U-shaped structure. The two claws of the U-shaped structure of the fork (410) are inserted into the annular fork groove (422). The fork (410) and the annular fork groove (422) have a degree of freedom of relative circumferential movement, so as to realize that the cutter flange (409) can move axially independently while rotating. The rear end of the shift fork (410) is connected to one end of the shift fork connecting plate one (412), and the other end of the shift fork connecting plate one (412) is connected to the shift fork connecting plate two (413). The servo electric cylinder (415) is mounted on the bending base (401) via an L-shaped fixed base (414), and the L-shaped fixed base (414) is connected to the bending base (401). The lead screw (416) of the servo electric cylinder (415) is provided with a limiting groove (417) at its top end, and the shift fork connecting plate (413) is connected to the limiting groove (417).

7. The machine vision-based dynamic cabbage root-cutting device according to claim 1, characterized in that, The visual inspection module (2) includes a depth camera (201), an angle adjustment mounting plate (202), and a camera mounting base (203). The depth camera (201) is mounted on the angle adjustment mounting plate (202), which is connected to the camera mounting base (203) by fastening bolts. The camera mounting base (203) is mounted on the bent steel plate (102) of the clamping mechanism (1). The depth camera (201) is connected to the edge computing module (5).

8. A control method for a machine vision-based dynamic cabbage root-cutting device according to any one of claims 1 to 7, characterized in that, Includes the following steps: Step S1: The visual detection module (2) acquires the cabbage image and transmits it to the edge computing module (5); Step S2: The edge computing module (5) processes the image from step S1, identifies the image coordinates of the cabbage cutting point, and calculates the position of the cutting point in the conveying direction through calibration relationship. and the target position along the cutting axis , and output to the motion control module (6); Step S3: The motion control module (6) receives the speed of the clamping chain measured by the speed sensor (105) of the clamping mechanism (1). According to the preset speed ratio Calculate conveyor belt speed And adjust the speed ratio to stabilize the cabbage's posture; Step S4: The motion control module (6) determines the fixed distance between the depth camera (201) and the cutter (408). , position of the cutting point conveying direction And the speed of cabbage delivery Predict the remaining time for the cutting point to reach the cutting line position of the cutter. ,in satisfy , The moment of image acquisition; Step S5: The motion control module (6) determines the current axial position of the cutter (408). and target location Determine the axial movement distance, i.e., the position error. and combined with the remaining time Calculate the axial movement speed of the cutter ; Step S6: The motion control module (6) will adjust the position error. Compared with a preset threshold, if the error is less than the threshold, the cutter (408) is controlled to maintain its current position; if the error is greater than the threshold, the cabbage dynamic root cutting mechanism (4) is controlled to drive the cutter (408) at a speed of Move axially to the target position.

9. The control method for the machine vision-based dynamic root-cutting device for cabbage according to claim 8, characterized in that, In step S2, the edge computing module (5) uses a deep learning model to process the image. The deep learning model is the YOLOv8n-pose model. The YOLOv8n-pose model takes the distortion-corrected whole frame image as input and outputs the target bounding box and the coordinates of the cutting point key points of the cabbage in a synchronous manner.

10. The control method for the machine vision-based dynamic cabbage root-cutting device according to claim 8, characterized in that, In step S2, the calibration relationship conversion includes using the calibration parameters and depth values ​​of the depth camera (201). ), based on the perspective projection model, the pixel coordinates of the cutting point ( Convert to 3D coordinates in the camera coordinate system. Then, through rigid body transformation, the coordinates are converted to coordinates in the world coordinate system. ), of which, the world coordinate system The axis indicates the conveying direction. The axis is the axial direction of the cutter's movement, thus obtaining , .

Images