A machine vision-based self-adaptive feeding and pulling device for Chinese cabbage harvesters and a control method thereof

Abstract

The present application provides a kind of based on machine vision's Chinese cabbage harvester self-adapting feed and pull device and control method, device includes lead and pull mechanism, pull mechanism, clamping and conveying mechanism, visual detection module, edge computing module and motion control module.Visual detection module gathers Chinese cabbage head image and is transmitted to edge computing module, and the diameter of Chinese cabbage is identified and calculated;Motion control module drives rudder in pull mechanism according to diameter data, adjusts the opening of feed inlet, so that it is self-adapted matching with the diameter of Chinese cabbage.The present application avoids jam, slip and mechanical damage by adjusting the size of feed inlet in real time, improves the smoothness of Chinese cabbage harvesting, operation efficiency and whole machine reliability, and is suitable for intelligent harvesting of different varieties and size Chinese cabbage.

Description

Technical Field

[0001] This invention belongs to the field of intelligent agricultural machinery technology, and in particular relates to an adaptive feeding and pulling device and control method for a cabbage harvester based on machine vision. Background Technology

[0002] To reduce labor costs and improve operational efficiency, various cabbage harvesters have emerged in recent years. Among them, in harvesting methods centered on "feeding and pulling," the matching degree between the feeding inlet structure and the pulling mechanism directly affects the continuity of the operation, damage rate, and missed pull rate, making it a key factor in the overall reliability and harvest quality. Most existing cabbage harvesters use fixed feeding and pulling devices, with the opening width of the feeding inlet or clamping mechanism typically being a fixed value. During operation, the cabbage is pulled from the soil by the front-end pulling mechanism and then conveyed backward through the feeding inlet by the clamping and conveying mechanism. However, due to the wide variety of cabbage varieties and varying growing conditions, the diameter and shape of cabbages in the field differ significantly, making it difficult for fixed feeding inlets to accommodate cabbages of different sizes. When the cabbage diameter is too large, it can cause jamming and machine stoppage; when the cabbage diameter is too small, the clamping force is insufficient, easily leading to slippage. Uneven squeezing pressure on cabbage due to fixed clamping spacing can easily cause mechanical damage such as leaf breakage and root breakage, increasing the loss rate of cabbage. Although some existing harvesters are equipped with manual adjustment or spring passive adjustment mechanisms, the adjustment process is time-consuming and laborious, and they cannot make dynamic adjustments in real time according to the size of the cabbage. They have low intelligence and poor overall adaptability. Summary of the Invention

[0003] To address the aforementioned technical problems, this invention provides an adaptive feeding and pulling device and control method for a cabbage harvester based on machine vision, which is beneficial for improving the cabbage harvester's ability to continuously, non-destructively, and efficiently harvest different varieties and sizes of cabbage.

[0004] Note that the description of these objectives does not preclude the existence of other objectives. One aspect of the invention does not require achieving all of the above objectives. Objectives other than those described above can be extracted from the description, drawings, and claims.

[0005] This invention includes a pulling mechanism, a picking mechanism, a clamping and conveying mechanism, a vision detection module, an edge computing module, and a motion control module. The vision detection module acquires an image of the cabbage head and transmits it to the edge computing module to identify and calculate the cabbage diameter. The motion control module drives a servo motor in the picking mechanism based on the diameter data to adjust the feed inlet opening to adaptively match the cabbage diameter. This invention, by adjusting the feed inlet size in real time, avoids jamming, slippage, and mechanical damage, improving the smoothness of cabbage harvesting, operational efficiency, and overall machine reliability. It is suitable for the intelligent harvesting of different varieties and sizes of cabbage.

[0006] The present invention achieves the above-mentioned technical objectives through the following technical means.

[0007] An adaptive feeding and pulling device for a cabbage harvester based on machine vision includes a pulling mechanism, a pulling mechanism, a clamping and conveying mechanism, a vision detection module, an edge computing module, and a motion control module.

[0008] The pulling mechanism and the picking mechanism are installed below the front end of the clamping and conveying mechanism, with the pulling mechanism located in front of the picking mechanism; the vision detection module is installed above the clamping and conveying mechanism; the edge computing module is electrically connected to the vision detection module and the motion control module respectively; the motion control module is electrically connected to the picking mechanism.

[0009] The guiding mechanism is used to guide and straighten the cabbage, so that the cabbage enters the working area of ​​the picking mechanism in a predetermined posture; the clamping and conveying mechanism is used to transport the picked cabbage backward; the vision detection module is used to acquire images of the cabbage and transmit them to the edge computing module; the edge computing module is used to process the received images, obtain the diameter of the cabbage and output it to the motion control module; the motion control module is used to control the picking mechanism to adjust the size of the feeding inlet according to the cabbage diameter output by the edge computing module.

[0010] In the above scheme, the pulling mechanism includes a T-shaped reinforcing rod, a first guide rod, a second guide rod, a third guide rod, an upper fixing plate for the pulling rod, a lower fixing plate for the pulling rod, a clamping mechanism frame, and angle steel;

[0011] The upper fixing plate of the pull rod is installed above the frame of the clamping and conveying mechanism, and one end of the first guide rod is connected to the upper fixing plate of the pull rod.

[0012] The lower fixing plate of the pull rod is installed below the frame of the clamping and conveying mechanism, and one end of the second guide rod is connected to the lower fixing plate of the pull rod.

[0013] The angle steel is installed below the frame of the clamping mechanism, and one end of the third guide rod is connected to the angle steel;

[0014] The other ends of the first guide rod, the second guide rod, and the third guide rod all extend forward along the forward direction of the cabbage harvester and are respectively connected to the T-shaped reinforcing rod. The first guide rod is connected to the upper surface of the T-shaped reinforcing rod, and the second guide rod and the third guide rod are respectively connected to the two side surfaces of the T-shaped reinforcing rod.

[0015] In the above scheme, the extraction mechanism includes multiple extraction tension wheels, an upper mounting plate for the tension wheels, a lower mounting plate for the tension wheels, a rotating shaft, a rotating shaft sleeve, and a servo motor;

[0016] The tensioning puller is installed between the upper mounting plate and the lower mounting plate of the tensioning puller.

[0017] The tensioning wheel mounting plate is installed below the rotating shaft;

[0018] The rotating shaft is connected to the shaft sleeve via bearings and retaining rings;

[0019] The rotating shaft cylinder is mounted on the frame of the clamping and conveying mechanism;

[0020] The upper part of the rotating shaft is connected to the servo motor via a coupling;

[0021] The servo motor is mounted on a servo motor mounting base, and the servo motor mounting base is mounted on the front end of the clamping and conveying mechanism;

[0022] The servo motor is electrically connected to the motion control module. The motion control module drives the servo motor, which drives the rotating shaft to swing back and forth around its own axis through a coupling. This causes the upper mounting plate of the tension wheel, the lower mounting plate of the tension wheel, and the tension wheel fixed on the rotating shaft to swing synchronously.

[0023] Furthermore, the visual inspection module includes a depth camera, a camera mounting base, a camera mounting bracket, a photoelectric sensor, and a sensor mounting bracket;

[0024] The depth camera is mounted on a camera mounting base, the camera mounting base is mounted on a camera mounting bracket, and the camera mounting bracket is mounted on the frame of the clamping and conveying mechanism.

[0025] The photoelectric sensor is mounted on a sensor mounting bracket, which is mounted on the frame of the clamping and conveying mechanism.

[0026] The depth camera is located above the cabbage conveying path, with its lens facing downwards, for taking a top-down view of the top of the cabbage.

[0027] The photoelectric sensor is used to detect whether the cabbage has entered the detection area and provides trigger signals to the edge computing module and the motion control module respectively. The trigger signal is used to trigger the edge computing module to start image processing and diameter calculation, and serves as the starting reference for the timing control of the motion control module.

[0028] Furthermore, the edge computing module is an embedded computing board, the input end of the edge computing module is electrically connected to the depth camera and photoelectric sensor of the vision detection module, and the output end of the edge computing module is electrically connected to the motion control module;

[0029] The input terminal of the motion control module is also electrically connected to the photoelectric sensor, and the output terminal of the motion control module is electrically connected to the servo motor.

[0030] The edge computing module is used to complete image processing and cabbage diameter calculation, and the motion control module is used to complete photoelectric sensor signal acquisition, servo motor target angle calculation, and servo motor drive control.

[0031] In the above scheme, the geometric matching model between the size of the feeding inlet of the extraction mechanism and the diameter of the cabbage is as follows:

[0032]

[0033]

[0034] Limit conditions: , ,

[0035] in, The feeding inlet needs to be opened at an additional angle. To provide the minimum safe angle for the feed inlet, The physical diameter of the cabbage To be the minimum safe diameter of the feed inlet, The straight-line distance from the center of the shaft of the servo motor and the extraction mechanism to the equivalent contact point of the feed inlet is denoted as . The angle between one side of the clamping arm of the extraction mechanism and the vertical center line. The angle that the servo motor needs to rotate. This is the maximum permissible rotation angle of the servo motor.

[0036] Furthermore, the timing and speed control model of the servo motor is as follows:

[0037]

[0038]

[0039] in, The maximum time allowed for the servo to complete one adjustment. The spacing between the cabbages The forward speed of the cabbage harvester. This is the total delay from when the photoelectric sensor triggers the action to when the servo motor completes its movement. The angular velocity of the servo motor;

[0040] When the adjustment time is insufficient, the motion control module controls the pulling mechanism to adjust to a preset safe angle.

[0041] In the above scheme, the clamping and conveying mechanism includes a conveyor belt, a driven wheel, and a tensioning wheel;

[0042] The driven wheel is mounted on a driven wheel mounting plate, which is mounted on a square tube. The square tube is mounted on the front end of the frame. The position of the driven wheel is adjusted by adjusting bolts and locking bolts.

[0043] The tensioning wheel is mounted on the frame of the clamping and conveying mechanism and abuts against the working stroke section of the conveyor belt to apply tension to the conveyor belt.

[0044] A control method for an adaptive feeding and pulling device for a cabbage harvester based on machine vision includes the following steps:

[0045] Step S1: The visual detection module acquires an image of the cabbage and transmits it to the edge computing module;

[0046] Step S2: The edge computing module processes the image, identifies the image coordinates of the cabbage's diameter, and calculates the physical diameter of the cabbage through calibration relationships, then outputs the result to the motion control module.

[0047] Step S3: The motion control module receives the photoelectric sensor signal and the diameter data output by the edge computing module, and converts the cabbage diameter into the target angle of the servo motor;

[0048] Step S4: The motion control module drives the servo motor according to the target angle, and drives the rotating shaft to swing back and forth around its own axis through the coupling, thereby driving the components connected to the rotating shaft in the picking mechanism to swing synchronously, so as to change the distance between the left and right sides of the picking mechanism, and adjust the feeding inlet opening to the target opening that matches the diameter of the cabbage.

[0049] In the above scheme, in step S2, the edge computing module uses a deep learning-based keypoint detection model to identify the center keypoint and edge keypoints of the cabbage; it uses histogram statistical filtering on the depth image to remove abnormal radius values, and takes the average value of the remaining candidate pixel radii to obtain the pixel radius; and it calibrates the conversion factor. The formula for converting pixel radius to physical diameter is:

[0050]

[0051]

[0052] in, The pixel distance between the two points with the largest depth values ​​in the RGB image. The physical Euclidean distance between the two points with the largest depth values. The filtered pixel radius, The physical diameter of the cabbage;

[0053] During calibration, a circular calibration plate of known diameter is placed within the visual detection area that the cabbage actually passes through. Multiple sets of data are collected at different heights or depths to establish... The fitting relationship between the depth value and the depth value.

[0054] Compared with the prior art, the beneficial effects of the present invention are:

[0055] This invention utilizes a machine vision-based adaptive feeding and pulling device for a cabbage harvester to achieve continuously adjustable feeding inlet width. This allows the same device to adapt to cabbages of different diameters and varieties, improving versatility. The invention uses a visual detection device to detect the cabbage diameter, which is then converted in real-time by a motion control module into a target servo motor angle. This achieves precise matching between the feeding inlet opening and the cabbage diameter, improving harvesting efficiency. The visual detection device employs a fusion sensing scheme combining a depth camera, a key point detection algorithm, and a photoelectric sensor. This effectively suppresses false positives and false negatives caused by factors such as lighting and occlusion, improving system robustness.

[0056] This invention uses a visual detection module to acquire real-time images of cabbage, an edge computing module to calculate the cabbage diameter, and a motion control module to drive the picking mechanism to adjust the opening of the feeding inlet, so that the width of the feeding inlet dynamically matches the current cabbage diameter. This effectively avoids jamming and machine stoppage caused by cabbage diameter being too large, as well as insufficient clamping force and slippage caused by diameter being too small. It helps to reduce damage to cabbage leaves and root breakage, and improves the continuity, reliability, and quality of cabbage harvesting.

[0057] The present invention employs a frame structure of three guide rods and T-shaped reinforcing rods to guide and straighten the cabbage, which facilitates the stable entry of the cabbage into the working area of ​​the extraction mechanism, reduces the risk of jamming, and improves the smoothness of feeding.

[0058] The servo motor of the extraction mechanism of this invention drives the rotating shaft to swing around its own axis through a coupling, thereby driving the tension wheel and the mounting plate to swing synchronously, realizing continuous adjustment of the feed inlet opening, and has a compact structure and fast response.

[0059] The edge computing module of this invention has a clear division of labor between image processing and real-time control, which ensures both the computing power requirements of the deep learning algorithm and the real-time requirements of the servo control.

[0060] This invention establishes a geometric matching model and limiting conditions between the feed inlet opening and the cabbage diameter, accurately converting the cabbage diameter into the target rotation angle of the servo motor, ensuring the accuracy and safety of the adjustment, and preventing the mechanical limit from being exceeded or damaging the cabbage.

[0061] This invention establishes a servo timing and speed control model, calculates the allowable adjustment time based on the cabbage spacing and the harvester's forward speed, and considers the total system delay. When the time is insufficient, the system adjusts to a preset safe angle, ensuring that the adjustment action is completed in a timely manner when adjacent cabbages pass by.

[0062] The clamping and conveying mechanism of this invention adopts an adjustable driven wheel and tension wheel structure. The position of the driven wheel can be flexibly adjusted by adjusting bolts and locking bolts, and the tension wheel applies a stable tension force to the conveyor belt, ensuring the stable clamping and backward conveying of the cabbage.

[0063] This invention employs a deep learning-based keypoint detection model to identify the center and edge keypoints of a cabbage, combines histogram statistical filtering to remove abnormal radius values, and establishes a multi-depth calibration system. The fitting relationship with depth values ​​significantly improves the accuracy and environmental adaptability of diameter measurement.

[0064] Note that the description of these effects does not preclude the existence of other effects. One aspect of the invention does not necessarily have to have all of the above.

[0065] The effects described above are obvious from the description, drawings, claims, etc. Attached Figure Description

[0066] Figure 1 This is a schematic diagram showing the relative positions between the cabbage harvester feeding device and the cabbage harvester according to an embodiment of the present invention.

[0067] Figure 2 This is a schematic diagram of the structure of the pulling mechanism, the picking mechanism, the clamping and conveying mechanism, the vision detection module, the edge computing module, and the motion control module according to an embodiment of the present invention.

[0068] Figure 3 This is a schematic diagram of the cabbage pulling mechanism according to one embodiment of the present invention.

[0069] Figure 4 This is a schematic diagram of the cabbage-picking mechanism according to one embodiment of the present invention.

[0070] Figure 5 This is a schematic diagram of the installation structure of the visual inspection module and photoelectric sensor according to one embodiment of the present invention.

[0071] Figure 6 This is a schematic flowchart of the control method for an adaptive feeding and pulling device for a cabbage harvester based on machine vision, according to one embodiment of the present invention.

[0072] Figure 7 This is a schematic diagram of the operation of a feeding and pulling device according to an embodiment of the present invention.

[0073] In the diagram, 1. Pulling mechanism; 101. T-shaped reinforcing rod; 102. First guide rod; 103. Third guide rod; 104. Upper fixing plate of the pulling rod; 105. Angle steel; 106. Second guide rod; 107. Lower fixing plate of the pulling rod; 108. Root clamping mechanism frame; 2. Pulling mechanism; 201. Pulling tension wheel; 202. Mounting plate of the tension wheel; 203. 1. Rotary shaft; 204. Rotary shaft; 205. Tensioner wheel mounting plate; 206. Servo mount; 207. Servo; 208. Coupling; 3. Vision inspection module; 301. Photoelectric sensor; 302. Sensor mounting bracket; 303. Camera mounting bracket; 304. Camera mounting base; 305. Depth camera; 4. Clamping and conveying mechanism; 401. Conveyor belt; 402. Square tube; 403. Driven wheel mounting plate; 404. Driven wheel; 405. Tensioner wheel; 406. Frame; 5. Edge computing module; 6. Motion control module. Detailed Implementation

[0074] 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.

[0075] 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.

[0076] 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.

[0077] Figure 1 The image shows a preferred embodiment of the adaptive feeding and pulling device for a cabbage harvester based on machine vision. The adaptive feeding and pulling device for a cabbage harvester based on machine vision includes a pulling mechanism 1, a pulling mechanism 2, a clamping and conveying mechanism 4, a vision detection module 3, an edge computing module 5, and a motion control module 6.

[0078] The pulling mechanism 1 and the picking mechanism 2 are installed below the front end of the clamping and conveying mechanism 4, with the pulling mechanism 1 located in front of the picking mechanism 2; the vision detection module 3 is installed above the clamping and conveying mechanism 4; the edge computing module 5 is electrically connected to the vision detection module 3 and the motion control module 6 respectively; the motion control module 6 is electrically connected to the picking mechanism 2.

[0079] The pulling mechanism 1 is used to guide and straighten the cabbage, so that the cabbage enters the working area of ​​the pulling mechanism 2 in a predetermined posture; the clamping and conveying mechanism 4 is used to convey the pulled cabbage backward; the vision detection module 3 is used to acquire images of the cabbage and transmit them to the edge computing module 5; the edge computing module 5 is used to process the received images, obtain the diameter of the cabbage and output it to the motion control module 6; the motion control module 6 is used to control the pulling mechanism 2 to adjust the size of the feeding inlet according to the cabbage diameter output by the edge computing module 5.

[0080] The guiding mechanism 1 is used to reduce the clamping force required by the pulling mechanism 2 to pull the cabbage, guide and straighten the cabbage in front, so that the cabbage enters the working area of ​​the pulling mechanism 2 in a predetermined posture, achieving continuous feeding and reducing jamming and damage. Figure 2As shown, the pulling mechanism 1 includes a T-shaped reinforcing rod 101, a first guide rod 102, a second guide rod 106, a third guide rod 103, an upper fixing plate 104 for the pulling rod, a lower fixing plate 107 for the pulling rod, a clamping mechanism frame 108, and an angle steel 105. The upper fixing plate 104 for the pulling rod is installed above the frame 406 of the clamping and conveying mechanism 4. One end of the first guide rod 102 is connected to the upper fixing plate 104 for the pulling rod. The lower fixing plate 107 for the pulling rod is installed below the frame 406 of the clamping and conveying mechanism 4. The second guide rod 102 is connected to the upper fixing plate 104 for the pulling rod. One end of the 6 is connected to the lower fixing plate 107 of the pull rod; the angle steel 105 is installed below the frame 108 of the root clamping mechanism, and one end of the third guide rod 103 is connected to the angle steel 105; the other ends of the first guide rod 102, the second guide rod 106 and the third guide rod 103 all extend forward along the forward direction of the cabbage harvester and are respectively connected to the T-shaped reinforcing rod 101, and the first guide rod 102 is connected to the upper surface of the T-shaped reinforcing rod 101, and the second guide rod 106 and the third guide rod 103 are respectively connected to the two side surfaces of the T-shaped reinforcing rod 101.

[0081] In a specific embodiment of the present invention, the first guide rod 102 is fixed to the upper fixing plate 104 of the pull rod by a threaded connection, and the upper fixing plate 104 of the pull rod is fixed above the frame 406 of the clamping and conveying mechanism 4 by welding; the second guide rod 106 is fixed to the lower fixing plate 107 of the pull rod by a threaded connection, and the lower fixing plate 107 of the pull rod is fixed below the frame 406 of the clamping and conveying mechanism 4 by welding; the third guide rod 103 is fixed to the angle steel 105 by welding, and the angle steel 105 is fixed below the frame 108 of the clamping mechanism by welding; the front ends of the first guide rod 102, the second guide rod 106, and the third guide rod 103 are fixed to the T-shaped reinforcing rod 101 by welding.

[0082] like Figure 3 and 4As shown, the extraction mechanism 2 includes multiple extraction tension rollers 201, an upper tension roller mounting plate 202, a lower tension roller mounting plate 205, a rotating shaft 204, a rotating shaft sleeve 203, and a servo motor 207; the extraction tension rollers 201 are installed between the upper tension roller mounting plate 202 and the lower tension roller mounting plate 205; the lower tension roller mounting plate 205 is installed below the rotating shaft 204; the rotating shaft 204 is connected to the rotating shaft sleeve 203 via bearings and retaining rings; the rotating shaft sleeve 203 is installed on the frame 406 of the clamping and conveying mechanism 4; the rotating shaft 204... The upper part is connected to the servo motor 207 via a coupling 208; the servo motor 207 is mounted on the servo motor mounting base 206, which is mounted on the front end of the clamping and conveying mechanism 4; the servo motor 207 is electrically connected to the motion control module 6, which drives the servo motor 207, and the servo motor 207 drives the rotating shaft 204 to swing back and forth around its own axis via the coupling 208, thereby driving the tension wheel upper mounting plate 202, tension wheel lower mounting plate 205 and pull-out tension wheel 201 fixed on the rotating shaft 204 to swing synchronously.

[0083] In one specific embodiment of the present invention, the tensioning pulley 201 is fixed between the upper tensioning pulley mounting plate 202 and the lower tensioning pulley mounting plate 205 by bolts; the lower tensioning pulley mounting plate 205 is fixed below the rotating shaft 204 by bolts; the rotating shaft cylinder 203 is fixed to the frame 406 of the clamping and conveying mechanism 4 by welding; and the servo motor 207 is fixed to the servo motor mounting base 206 by bolts.

[0084] The clamping and conveying mechanism 4 includes a conveyor belt 401, a driven wheel 404, and a tensioning wheel 405. The driven wheel 404 is mounted on a driven wheel mounting plate 403, which is mounted on a square tube 402. The square tube 402 is mounted on the front end of a frame 406. The position of the driven wheel 404 is adjusted by adjusting bolts and locking bolts. The tensioning wheel 405 is mounted on the frame 406 of the clamping and conveying mechanism 4 and abuts against the working stroke section of the conveyor belt 401 to apply tension to the conveyor belt 401.

[0085] The driven wheel 404 and several tensioning wheels 405 of the extraction mechanism 2 and the clamping and conveying mechanism 4 tension the conveyor belt 401, thereby stably clamping and frictionally extracting the cabbage and preventing slippage. The driven wheel 404 is installed at the front end of the frame 406 of the clamping and conveying mechanism 4 and is used to adjust the center distance for tensioning. The tensioning wheels 405 are installed in the working stroke section of the conveyor belt 401 and are used to apply tension to the conveyor belt 401. The driven wheel 404 is installed on the driven wheel mounting plate 403. The driven wheel mounting plate 403 is fixed to the square tube 402 by welding. The square tube 402 is installed at the front end of the frame 406 of the clamping and conveying mechanism 4. The position of the driven wheel 404 is adjusted by adjusting bolts and locking bolts.

[0086] like Figure 5 As shown, the visual detection module 3 includes a depth camera 305, a camera mounting base 304, a camera mounting bracket 303, a photoelectric sensor 301, and a sensor mounting bracket 302. The depth camera 305 is mounted on the camera mounting base 304, the camera mounting base 304 is mounted on the camera mounting bracket 303, and the camera mounting bracket 303 is mounted on the frame 406 of the clamping and conveying mechanism 4. The photoelectric sensor 301 is mounted on the sensor mounting bracket 302, and the sensor mounting bracket 302 is mounted on the frame 406 of the clamping and conveying mechanism 4. The depth camera 305 is located above the cabbage conveying path, with its lens facing downwards, for top-down imaging of the top of the cabbage. The photoelectric sensor 301 is used to detect whether the cabbage has entered the detection area and provides trigger signals to the edge computing module 5 and the motion control module 6, respectively. The trigger signals are used to trigger the edge computing module 5 to start image processing and diameter calculation, and serve as the starting reference for the timing control of the motion control module 6.

[0087] In one specific embodiment of the present invention, the depth camera 305 is mounted on the camera mounting base 304 by bolts; the camera mounting base 304 is fixed on the camera mounting bracket 303 by bolts; the camera mounting bracket 303 is fixed on the frame 406 of the clamping and conveying mechanism 4 by bolts; the photoelectric sensor 301 is mounted on the sensor mounting bracket 302; the sensor mounting bracket 302 is mounted on the frame 406 of the clamping and conveying mechanism 4 by bolts.

[0088] The photoelectric sensor 301 provides a signal that the cabbage has entered the detection area, avoiding false detections when the cabbage is not in the field of view or when its motion is blurred. The depth camera 305 is mounted above the cabbage conveying path, with its lens facing downwards, so that the camera's optical axis is perpendicular to the ground plane, and performs a top-down image of the top of the cabbage entering the detection area to obtain a depth image for diameter calculation.

[0089] The input terminal of the edge computing module 5 is electrically connected to the depth camera 305 and the photoelectric sensor 301 of the vision detection module 3, and the output terminal of the edge computing module 5 is electrically connected to the motion control module 6. The input terminal of the motion control module 6 is also electrically connected to the photoelectric sensor 301, and the output terminal of the motion control module 6 is electrically connected to the servo motor 207. The edge computing module 5 is used to complete image processing and cabbage diameter calculation, and the motion control module 6 is used to complete signal acquisition from the photoelectric sensor 301, target angle calculation of the servo motor 207, and drive control of the servo motor 207.

[0090] In one specific embodiment, the edge computing module 5 can be an embedded computing board with GPU, NPU, or other neural network inference acceleration capabilities, such as NVIDIA Jetson series embedded AI boards, RK3588 series embedded boards, or other industrial edge computing devices capable of running target detection and key point recognition models. The edge computing module 5 and the motion control module 6 adopt a separate structure. When using a separate structure, the edge computing module 5 mainly performs image processing and cabbage diameter calculation, while the motion control module 6 mainly performs signal acquisition from the photoelectric sensor 301, target angle calculation for the servo motor 207, and drive control of the servo motor 207.

[0091] like Figure 7 As shown, the geometric matching model between the size of the feeding inlet of the extraction mechanism 2 and the diameter of the cabbage is as follows:

[0092]

[0093]

[0094] Limit conditions: , , ;

[0095] in, The feeding inlet needs to be opened at an additional angle. To provide the minimum safe angle for the feed inlet, The physical diameter of the cabbage To be the minimum safe diameter of the feed inlet, The straight-line distance from the center of the rotating shaft of the servo motor 207 and the picking mechanism 2 to the equivalent contact point of the cabbage at the feeding inlet is defined as the position where the cabbage initially makes effective contact with the clamping conveyor belt when it enters the picking mechanism. The angle between one side of the clamping arm of the extraction mechanism 2 and the vertical center line. For the angle that the servo motor 207 needs to rotate, This is the maximum permissible rotation angle for the servo motor 207.

[0096] Further, after obtaining the target turning angle of servo motor 207 using the aforementioned adaptive matching model at the feed inlet, to ensure that servo motor 207 completes adjustment within the time window of adjacent cabbages passing through the feed inlet, a timing and speed control model for servo motor 207 is further established; the timing and speed control model for servo motor 207 is as follows:

[0097]

[0098]

[0099] in, The maximum time allowed for servo 207 to complete one adjustment. The spacing between the cabbages, The forward speed of the cabbage harvester. The total delay from the triggering of photoelectric sensor 301 to the completion of the action of servo motor 207. The rotational angular velocity of servo motor 207;

[0100] When the adjustment time is insufficient, the motion control module 6 controls the pulling mechanism 2 to adjust to a preset safe angle. This refers to the total delay between system detection triggering and control action taking effect. The distance between the photoelectric sensor 301 and the feed inlet affects the time it takes for the servo motor 207 to make its first adjustment. The further forward the photoelectric sensor 301 is, the more time the servo motor 207 has to complete the action. When harvesting continuously, the time it takes for the servo motor 207 to complete one adjustment is related to the spacing between the cabbages and the forward speed of the harvester.

[0101] The actual time required for the servo motor 207 to complete one adjustment includes: image processing time of the edge computing module 5, communication transmission time, calculation time of the motion control module 6, electrical response time of the servo motor 207, and mechanical action delay time of the pulling mechanism 2. These times can be measured through prototype testing and stored in the motion control module 6 as a total delay parameter.

[0102] In a specific embodiment of the present invention, preferably, for the identification of the diameter of cabbage, the edge computing module 5 uses a key point detection algorithm based on deep learning to construct a detection model; the key point detection model adopts the YOLOv8n-pose model, takes RGB three-channel images as input, and outputs target detection boxes and key point coordinates.

[0103] like Figure 6 As shown, a control method for an adaptive feeding and pulling device for a cabbage harvester based on machine vision includes the following steps:

[0104] Step S1: The visual detection module 3 acquires an image of the cabbage and transmits it to the edge calculation module 5;

[0105] Step S2: The edge computing module 5 processes the image, identifies the image coordinates of the cabbage diameter, and calculates the physical diameter of the cabbage through calibration relationship, and outputs the physical diameter to the motion control module 6;

[0106] Step S3: The motion control module 6 receives the signal from the photoelectric sensor 301 and the diameter data output by the edge computing module 5, and converts the cabbage diameter into the target angle of the servo motor 207;

[0107] Step S4: The motion control module 6 drives the servo motor 207 according to the target angle, and drives the rotating shaft 204 to swing back and forth around its own axis through the coupling 208, thereby driving the component connected to the rotating shaft 204 in the extraction mechanism 2 to swing synchronously, so as to change the distance between the left and right sides of the extraction mechanism 2, and adjust the feeding inlet opening to the target opening that matches the diameter of the cabbage.

[0108] To further achieve the conversion from pixel coordinates to physical coordinates, outliers are removed from the depth image through depth filtering, and the average value of the remaining effective radius values ​​is taken to obtain a reliable pixel radius; finally, the physical diameter is calculated based on the geometry and depth information of the depth camera.

[0109] Histogram-based statistical filtering is employed, first applying histogram filtering to pixel radii, and then to keypoint depth values. The edge computing module first uses a keypoint detection model to identify the central and edge keypoints of the cabbage, then matches the keypoint coordinates to the depth image to obtain the corresponding depth value for each keypoint. Next, the central keypoint is connected to each edge keypoint to form multiple radial line segments, and the pixel length of each radial line segment is calculated to obtain multiple pixel radii. Then, the pixel radii are sorted by length and a radius histogram is constructed. After filtering, the average of the remaining candidate pixel radii is used for subsequent calculation of the cabbage's physical diameter.

[0110] During field calibration, a circular calibration plate of known diameter is placed within the visual detection area traversed by the actual cabbage during harvest, ensuring its height and depth are similar to the actual head of the cabbage. To accommodate variations in cabbage height, camera installation height, ground undulations, and camera vibration during field operations, calibration objects can be placed at different heights or depths, and multiple sets of data can be repeatedly collected to establish a calibration system. The fitting relationship between the depth value and the depth value.

[0111] Specifically, in step S2, the edge computing module 5 uses a deep learning-based keypoint detection model to identify the center keypoint and edge keypoints of the cabbage; it uses histogram statistical filtering on the depth image to remove abnormal radius values, and takes the average of the remaining candidate pixel radii to obtain the pixel radius; and it calibrates the conversion factor. The formula for converting pixel radius to physical diameter is:

[0112]

[0113]

[0114] in, The conversion factor indicates how many millimeters each pixel represents in the current depth plane. The pixel distance between the two points with the largest depth values ​​in the RGB image. The physical Euclidean distance between the two points with the largest depth values. The filtered pixel radius, The physical diameter of the cabbage;

[0115] During calibration, a circular calibration plate of known diameter is placed within the visual detection area that the cabbage actually passes through. Multiple sets of data are collected at different heights or depths to establish... The fitting relationship between the depth value and the depth value.

[0116] This invention identifies the diameter information of the cabbage through a visual detection module 3 and uses a photoelectric sensor 301 to detect in real time whether the cabbage has reached the feeding area. Simultaneously, the physical diameter parameters of the cabbage obtained by the edge computing module 5 are input to the motion control module 6. The motion control module 6 calculates based on its internal mathematical model and then outputs servo control commands. When the calculated servo target rotation angle... When the limit conditions are met, the motion control module 6 drives the servo motor 207 to rotate to the target angle, achieving adaptive matching between the feeding inlet opening and the cabbage diameter. The limit conditions prevent the servo motor 207 from exceeding its mechanical limit and causing the mechanism to jam, while also preventing the feeding inlet opening from being too small and damaging the cabbage. When the limit conditions are not met, the extraction mechanism 2 maintains its current position to avoid damaging the cabbage or exceeding the mechanical limit, ensuring the reliability of the feeding process.

[0117] This invention maintains a real-time match between the feeding inlet opening and individual differences in Chinese cabbage, ultimately achieving precise and damage-free feeding and harvesting. This invention boasts advantages such as high adaptive matching accuracy, simple and reliable structure, and high operational efficiency. It is suitable for the harvesting needs of different varieties of Chinese cabbage and can further improve the level of automation and intelligence in harvesting.

[0118] This invention uses machine vision combined with a geometric mathematical model to complete the conversion of cabbage diameter and control of the feeding inlet size, which has the advantages of strong robustness and strong environmental adaptability.

[0119] 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.

[0120] 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. An adaptive feeding and pulling device for a cabbage harvester based on machine vision, characterized in that, It includes a pulling mechanism (1), a picking mechanism (2), a clamping and conveying mechanism (4), a vision detection module (3), an edge computing module (5), and a motion control module (6). The pulling mechanism (1) and the picking mechanism (2) are installed below the front end of the clamping and conveying mechanism (4), and the pulling mechanism (1) is located in front of the picking mechanism (2); the vision detection module (3) is installed above the clamping and conveying mechanism (4); the edge computing module (5) is electrically connected to the vision detection module (3) and the motion control module (6) respectively; the motion control module (6) is electrically connected to the picking mechanism (2); The pulling mechanism (1) is used to guide and straighten the cabbage so that it enters the working area of ​​the pulling mechanism (2) in a predetermined posture; the clamping and conveying mechanism (4) is used to convey the pulled cabbage backward; the vision detection module (3) is used to collect images of the cabbage and transmit them to the edge computing module (5); the edge computing module (5) is used to process the received images, obtain the cabbage diameter and output it to the motion control module (6); the motion control module (6) is used to control the pulling mechanism (2) to adjust the size of the feeding inlet according to the cabbage diameter output by the edge computing module (5).

2. The adaptive feeding and pulling device for a cabbage harvester based on machine vision according to claim 1, characterized in that, The pulling mechanism (1) includes a T-shaped reinforcing rod (101), a first guide rod (102), a second guide rod (106), a third guide rod (103), an upper fixing plate (104) for the pulling rod, a lower fixing plate (107) for the pulling rod, a clamping mechanism frame (108) and an angle steel (105). The upper fixing plate (104) of the pull rod is installed above the frame (406) of the clamping and conveying mechanism (4), and one end of the first guide rod (102) is connected to the upper fixing plate (104) of the pull rod. The lower fixing plate (107) of the pull rod is installed below the frame (406) of the clamping and conveying mechanism (4), and one end of the second guide rod (106) is connected to the lower fixing plate (107). The angle steel (105) is installed below the clamping mechanism frame (108), and one end of the third guide rod (103) is connected to the angle steel (105); The other ends of the first guide rod (102), the second guide rod (106) and the third guide rod (103) extend forward along the forward direction of the cabbage harvester and are respectively connected to the T-shaped reinforcing rod (101). The first guide rod (102) is connected to the upper surface of the T-shaped reinforcing rod (101), and the second guide rod (106) and the third guide rod (103) are respectively connected to the two side surfaces of the T-shaped reinforcing rod (101).

3. The adaptive feeding and pulling device for a cabbage harvester based on machine vision according to claim 1, characterized in that, The extraction mechanism (2) includes multiple extraction tension wheels (201), an upper mounting plate (202) for the tension wheels, a lower mounting plate (205) for the tension wheels, a rotating shaft (204), a rotating shaft sleeve (203), and a servo motor (207). The tensioning pulley (201) is installed between the upper mounting plate (202) and the lower mounting plate (205) of the tensioning pulley; The tensioning wheel mounting plate (205) is installed below the rotating shaft (204); The rotating shaft (204) is connected to the rotating shaft sleeve (203) via bearings and retaining rings; The rotating shaft sleeve (203) is mounted on the frame (406) of the clamping and conveying mechanism (4); The upper part of the rotating shaft (204) is connected to the servo motor (207) via a coupling (208); The servo motor (207) is mounted on the servo motor mounting base (206), and the servo motor mounting base (206) is mounted on the front end of the clamping and conveying mechanism (4); The servo motor (207) is electrically connected to the motion control module (6). The motion control module (6) drives the servo motor (207). The servo motor (207) drives the rotating shaft (204) to swing back and forth around its own axis through the coupling (208), thereby driving the tension wheel upper mounting plate (202), tension wheel lower mounting plate (205) and pull-out tension wheel (201) fixed on the rotating shaft (204) to swing synchronously.

4. The adaptive feeding and pulling device for a cabbage harvester based on machine vision according to claim 3, characterized in that, The visual inspection module (3) includes a depth camera (305), a camera mounting base (304), a camera mounting bracket (303), a photoelectric sensor (301), and a sensor mounting bracket (302). The depth camera (305) is mounted on the camera mounting base (304), the camera mounting base (304) is mounted on the camera mounting bracket (303), and the camera mounting bracket (303) is mounted on the frame (406) of the clamping and conveying mechanism (4); The photoelectric sensor (301) is mounted on the sensor mounting bracket (302), and the sensor mounting bracket (302) is mounted on the frame (406) of the clamping and conveying mechanism (4); The depth camera (305) is located above the cabbage conveying path, with its lens facing downwards, and is used to perform top-down imaging of the top of the cabbage; The photoelectric sensor (301) is used to detect whether the cabbage has entered the detection area and provides trigger signals to the edge computing module (5) and the motion control module (6) respectively. The trigger signal is used to trigger the edge computing module (5) to start image processing and diameter calculation, and serves as the starting reference for the timing control of the motion control module (6).

5. The adaptive feeding and pulling device for a cabbage harvester based on machine vision according to claim 4, characterized in that, The edge computing module (5) is an embedded computing board. The input end of the edge computing module (5) is electrically connected to the depth camera (305) and photoelectric sensor (301) of the vision detection module (3). The output end of the edge computing module (5) is electrically connected to the motion control module (6). The input terminal of the motion control module (6) is also electrically connected to the photoelectric sensor (301), and the output terminal of the motion control module (6) is electrically connected to the servo motor (207); The edge computing module (5) is used to complete image processing and cabbage diameter calculation, and the motion control module (6) is used to complete the signal acquisition of photoelectric sensor (301), target angle calculation of servo motor (207) and drive control of servo motor (207).

6. The adaptive feeding and pulling device for a cabbage harvester based on machine vision according to claim 3, characterized in that, The geometric matching model between the feeding inlet size of the extraction mechanism (2) and the diameter of the cabbage is as follows: Limit conditions: , , in, The feeding inlet needs to be opened at an additional angle. To provide the minimum safe angle for the feed inlet, The physical diameter of the cabbage To be the minimum safe diameter of the feed inlet, The straight-line distance from the center of the rotating shaft of the servo motor (207) and the extraction mechanism (2) to the equivalent contact point of the feed inlet is denoted as . The angle between one side of the clamping arm of the extraction mechanism (2) and the vertical center line. For the angle that the servo motor (207) needs to rotate, The maximum permissible rotation angle of the servo motor (207).

7. The adaptive feeding and pulling device for a cabbage harvester based on machine vision according to claim 6, characterized in that, The timing and speed control model of the servo motor (207) is as follows: in, The maximum time allowed for the servo motor (207) to complete one adjustment. The spacing between the cabbages The forward speed of the cabbage harvester. The total delay from triggering by the photoelectric sensor (301) to the completion of the servo motor (207) action is given. The rotational angular velocity of the servo motor (207); When the adjustment time is insufficient, the motion control module (6) controls the pulling mechanism (2) to adjust to a preset safe angle.

8. The adaptive feeding and pulling device for a cabbage harvester based on machine vision according to claim 1, characterized in that, The clamping and conveying mechanism (4) includes a conveyor belt (401), a driven wheel (404), and a tensioning wheel (405). The driven wheel (404) is mounted on the driven wheel mounting plate (403), the driven wheel mounting plate (403) is mounted on the square tube (402), the square tube (402) is mounted on the front end of the frame (406), and the position of the driven wheel (404) is adjusted by adjusting bolts and locking bolts; The tensioning wheel (405) is mounted on the frame (406) of the clamping and conveying mechanism (4) and abuts against the working stroke section of the conveyor belt (401) to apply tension to the conveyor belt (401).

9. A control method for an adaptive feeding and pulling device for a cabbage harvester based on machine vision, as described in any one of claims 1 to 8, characterized in that, Includes the following steps: Step S1: The visual detection module (3) acquires the cabbage image and transmits it to the edge computing module (5); Step S2: The edge computing module (5) processes the image, identifies the image coordinates of the cabbage diameter, and calculates the physical diameter of the cabbage through the calibration relationship, and outputs it to the motion control module (6). Step S3: The motion control module (6) receives the signal from the photoelectric sensor (301) and the diameter data output by the edge computing module (5), and converts the cabbage diameter into the target angle of the servo motor (207); Step S4: The motion control module (6) drives the servo motor (207) according to the target angle, and drives the rotating shaft (204) to swing back and forth around its own axis through the coupling (208), thereby driving the component connected to the rotating shaft (204) in the extraction mechanism (2) to swing synchronously, so as to change the distance between the left and right sides of the extraction mechanism (2) and adjust the opening of the feeding inlet to the target opening that matches the diameter of the cabbage.

10. The control method for the adaptive feeding and pulling device of a cabbage harvester based on machine vision according to claim 9, characterized in that, In step S2, the edge computing module (5) uses a deep learning-based key point detection model to identify the center key point and edge key points of the cabbage; it uses histogram statistical filtering on the depth image to remove abnormal radius values, and takes the average value of the remaining candidate pixel radii to obtain the pixel radius; it then calibrates the conversion factor. The formula for converting pixel radius to physical diameter is: in, The pixel distance between the two points with the largest depth values ​​in the RGB image. The physical Euclidean distance between the two points with the largest depth values. The filtered pixel radius, The physical diameter of the cabbage; During calibration, a circular calibration plate of known diameter is placed within the visual detection area that the cabbage actually passes through. Multiple sets of data are collected at different heights or depths to establish... The fitting relationship between the depth value and the depth value.

Images