Device and method for automatically controlling excavation depth

Abstract

The present invention relates to an automatic control device and method for digging depth, belonging to the field of information technology applied to agricultural machinery. The device includes a harvesting platform frame, an L-shaped forward-extending digging shovel, and a clamping conveyor chain. The harvesting platform frame is hingedly mounted with a depth-limiting wheel rocker arm at the rear end, and a depth-limiting wheel is mounted at the front end of the depth-limiting wheel rocker arm. The middle portion of the depth-limiting wheel rocker arm is hingedly connected to one end of a power push rod, and the other end of the power push rod is hingedly connected to the front lower end of the harvesting platform frame. The harvesting platform frame is equipped with an industrial camera facing the front garlic bulb area, and a displacement sensor is installed on one side of the power push rod. The signal output ends of the industrial camera and the displacement sensor are respectively connected to corresponding signal ports of a main control unit via a host computer and an analog / digital conversion module. The control output end of the main control unit is connected to the controlled end of the power push rod via a drive circuit. The present invention can intelligently, accurately, and automatically adjust the digging depth to an appropriate value, significantly improving the quality and efficiency of garlic harvesting, reducing energy waste, and lowering harvesting costs.

Description

Technical Field

[0001] The present invention relates to a digging depth control device for a garlic combine harvester, in particular to an automatic digging depth control device, and also relates to an automatic control method based on deep learning, belonging to the field of application of information technology to agricultural machinery. Background Art

[0002] Garlic is an important cash crop. Its production process includes tillage, sowing, field management, and harvesting, with harvesting being the most labor-intensive and labor-intensive. To improve harvesting efficiency and reduce labor costs, garlic combine harvesters have become a research and development priority. Properly controlling the digging depth of combine harvesters is crucial for ensuring harvest quality. During the harvesting process, if the digging shovel digs too deep, the machine's digging resistance increases, increasing the harvester's energy consumption and making digging more difficult. However, if the digging depth is too shallow, the garlic bulbs may be damaged or even missed, resulting in unnecessary economic losses.

[0003] A search shows that the Chinese invention patent document with application number 201911322799.5 shows the typical structure of a garlic combine harvester and discloses its depth-limiting excavation mechanism. The front end of the harvesting frame of this mechanism is equipped with a cutting platform supported on depth-limiting wheels on both sides, and a forward-extending digging shovel is installed under the cutting platform, and a straw separating device and a straw supporting device are installed at the front end; the digging shovel is hinged to the lower end of the front hanging rod and the vibration driving plate, and the upper ends of the front hanging rod and the vibration driving plate are respectively hinged to the lower part of the harvesting frame, the middle part of the vibration driving plate is hinged to the front end of the vibration driving rod, and the rear end of the vibration driving rod is connected to the eccentric vibration driving wheel driven by the power source; the depth-limiting wheel is installed at the lower end of the depth-limiting wheel frame tilted backward and downward, and is located behind the shovel handle of the vibrating digging shovel. During harvesting, the high-frequency vibrations of the vibrating shovel reduce soil breaking effort and effectively loosen the clay around the garlic bulbs. The depth-limiting wheel not only limits the digging depth but also holds down debris such as mulch film hanging from the shovel handle, effectively preventing it from getting tangled. However, this existing technology fails to automatically adjust the digging depth. Improper digging depth control can seriously impact the quality and efficiency of the garlic harvest. Digging too shallow will inevitably damage the bulbs, while digging too deep will not only increase energy consumption but also carry over a significant amount of soil. Therefore, harvesting still requires the operator's experience and the ability to manually adjust the digging depth as needed. Summary of the Invention

[0004] The purpose of the present invention is to address the shortcomings of the above existing technologies and to provide an automatic digging depth control device that adaptively adjusts the digging depth during the harvesting process. At the same time, a corresponding control method is provided, thereby avoiding damage to the bulbs and eliminating the subsequent root cutting process, thereby significantly improving the quality and efficiency of garlic harvesting.

[0005] To achieve the above-mentioned objectives, the basic technical solution of the automatic digging depth control device of the present invention is as follows: it includes a harvesting table frame with a rear end hingedly mounted on the front of the harvester, an L-shaped forward-extending digging shovel fixedly connected to the front lower part of the harvesting table frame, and a clamping conveyor chain starting from above the digging shovel and extending from the bottom to the top along the harvesting table frame;

[0006] The rear end of the harvesting platform frame is hingedly mounted with a depth-limiting wheel rocker arm, the front end of which is provided with a rotatable depth-limiting wheel, the middle portion of the depth-limiting wheel rocker arm is hingedly mounted to one end of a power push rod, and the other end of the power push rod is hingedly mounted to the front lower end of the harvesting platform frame;

[0007] The harvesting platform frame is equipped with an industrial camera facing the front garlic bulb area, and a displacement sensor is installed on one side of the power push rod; the signal output ends of the industrial camera and the displacement sensor are respectively connected to the corresponding signal ports of the main control unit through the host computer and the analog / digital conversion module, and the control output end of the main control unit is connected to the controlled end of the power push rod through the drive circuit.

[0008] The mechanical structure and sensor control settings of the device enable it to work. When the upper computer obtains the actual root length based on the image processing results, compares it with the set expected root length, calculates the expected excavation depth and sends the result to the main control unit. The main control unit drives the power push rod to extend and retract according to the expected excavation depth, thereby ensuring precise control of the excavation depth.

[0009] After adopting the above device, the host computer automatically controls the excavation depth, including the following basic steps:

[0010] The first step is to set the minimum excavation depth, expected root length, and detection flag as initial values;

[0011] The second step is to collect image information and set the target area in the rectangular detection area;

[0012] Step 3: determine whether the image information detects a bulb, if not, proceed to step 11; if yes, proceed to the next step;

[0013] Step 4: Determine whether the pixel coordinates of the upper left corner of the bulb image are within the target area. If so, set the detection flag position to the current value and proceed to step 11; if not, proceed to the next step.

[0014] Step 5: Determine whether the detection flag is the current value. If so, proceed to the next step; if not, proceed to step 11;

[0015] Step 6: Determine whether roots are detected in the image information. If so, calculate the actual root length and proceed to the next step; if not, use the expected root length as the excavation depth adjustment value and proceed to step 8;

[0016] Step 7: Determine whether the absolute value of the difference between the expected root length and the actual root length is greater than a predetermined value. If so, proceed to the next step; otherwise, proceed to the tenth step.

[0017] Step 8: Using the difference between the expected root length and the actual root length as the excavation depth adjustment value, and revising the expected excavation depth based on the current excavation depth and the excavation depth adjustment value;

[0018] Step 9: Send the desired excavation depth to the main control unit to adjust the excavation depth;

[0019] Step 10: Set the detection mark position to the initial value;

[0020] Step 11: Determine whether parking control information is received. If yes, proceed to the next step; if not, return to step 2.

[0021] Step 12: End automatic depth limit.

[0022] After adopting the present invention, the digging depth can be adjusted to an appropriate value intelligently, accurately and automatically, thereby avoiding damage to the bulbs and eliminating the need for subsequent root cutting, significantly improving the quality and efficiency of garlic harvesting, reducing energy waste and lowering harvesting costs.

[0023] A further improvement of the present invention is that the process of calculating the actual root length in the sixth step is: judging the position status of the bulb and the roots according to the received pixel coordinates of the upper left corner and the lower right corner of the bulb image and the pixel coordinates of the upper left corner and the lower right corner of the root image;

[0024] When the bulb and root images are judged to be non-overlapping, non-overlapping, or offset-overlapping, the actual root length is calculated using the following formulas:

[0025] h1=(yr2-yr1) ×ρ×d / f

[0026] h2=(yr2-yr1+yr2-yg2) ×ρ×d / 2f

[0027] h3=cosω×(yr2-yr1+yr2-yg2) ×ρ×d / 2f

[0028] In the above formula

[0029] h1——actual root length where the bulb and root images do not overlap, in mm;

[0030] h2——actual root length when the bulb and root images are not offset and overlapped, in mm;

[0031] h3——actual root length when bulb and root images are offset and overlapped, in mm;

[0032] yr1——the vertical coordinate of the upper left corner pixel of the root image, in pixels;

[0033] yr2——the vertical coordinate of the pixel in the lower right corner of the root image, in pixels;

[0034] yg2——the vertical coordinate of the pixel in the lower right corner of the bulb image, in pixels;

[0035] ρ——pixel size of the camera, in mm / pixel;

[0036] d——object distance of the camera, in mm;

[0037] f——focal length of the camera, in mm;

[0038] ω is the inclination angle of the diagonal line of the overlapping part of the bulb image and the root image, in degrees.

[0039] In this way, not only can the error caused by sample marking be reduced, but also the impact of the tilt of the garlic plant on the root measurement results can be reduced.

[0040] A further improvement of the present invention is that, in the ninth step, the main control unit controls the excavation depth by:

[0041] Step 1: Read the current extension of the power push rod and calculate the current digging depth based on the current extension of the power push rod;

[0042] Step 2: Determine whether the difference between the expected excavation depth and the current excavation depth is greater than a positive threshold. If so, control the power push rod to extend and return to the first step; otherwise, proceed to the next step.

[0043] Step 3: Determine whether the difference between the expected excavation depth and the current excavation depth is less than a negative threshold. If so, control the power push rod to retract and return to the first step; otherwise, proceed to the next step.

[0044] Step 4: Control the power push rod to stop and end this control. BRIEF DESCRIPTION OF THE DRAWINGS

[0045] The present invention will be further described below with reference to the accompanying drawings.

[0046] Figure 1 It is a structural diagram of an embodiment of the present invention.

[0047] Figure 2 yes Figure 1 Schematic diagram of the mechanism of the middle excavation and depth control part.

[0048] Figure 3 yes Figure 1 Schematic diagram of the working status structure of the embodiment.

[0049] Figure 4 yes Figure 1 Block diagram of the electronic control principle of the embodiment.

[0050] FIG5(a) to FIG5(d) are schematic diagrams of various digging positions of garlic.

[0051] Figure 6 yes Figure 1 Flowchart of host computer equipment inspection in an embodiment.

[0052] Figure 7 yes Figure 1 Flowchart of the main control unit power push rod control in the embodiment.

[0053] Figure 8 yes Figure 1 Flowchart of monitoring actual root length by the host computer in the embodiment. DETAILED DESCRIPTION

[0054] Example 1

[0055] This embodiment is a device for automatically controlling the mining depth based on deep learning. Figure 1 、 Figure 2 and Figure 3 As shown, the harvesting platform frame 8 comprises a front upright 8-1, an upper crossbar 8-3, a rear upright 8-5, and a lower crossbar 8-2, forming a support frame. The lower end of the frame is hinged to the front of the harvester at position O via a clamp 10. The front and rear uprights 8-1 and 8-5 are connected to a conveyor beam 8-4, which extends forward and backward, with a lower front and higher rear position. The front of the conveyor beam 8-4 is connected to a vertical sleeve 7-1 with a tightening screw 7-2, which is used to mount the L-shaped, forward-extending digging shovel 7 at the desired height.

[0056] The conveying beam 8-4 is equipped with a clamping conveyor chain 15 starting from the top of the digging shovel 7 and rising from low to high. The front end of the clamping conveyor chain 15 is equipped with a forward extending guide rod 3, and the input end of the sorting and alignment device 9 is installed in the middle and lower part. The output end of the sorting and alignment device 9 is installed with a cutting disc 11.

[0057] A transmission box 14 driven by a hydraulic motor 13 is installed at the rear end of the upper crossbar 8-3 of the harvesting platform frame 8. The transmission shaft 12 of the transmission box 14 is in transmission connection with the drive wheel at the rear end of the clamping conveyor chain 15. A forward-extending support rod 8-6 is fixedly connected to the middle part of the front upright 8-1 of the harvesting platform frame 8. The rear end of the depth-governing wheel rocker arm 6 is hingedly mounted at the middle part D of the support rod 8-6. The front end A of the depth-governing wheel rocker arm 6 is equipped with a rotatable depth-governing wheel 2 and a forward-extending rice seedling separating rod 1. The middle part B of the depth-governing wheel rocker arm 6 is hinged to one end of the power push rod 5. The other end of the power push rod 5 is hinged to the front lower end of the harvesting platform frame 8.

[0058] The lower part of the front upright 8-1 of the harvesting table frame 8 is equipped with an industrial camera 16 facing the front garlic G bulb part, and a displacement sensor 4 is installed on one side of the power push rod 5. Figure 4 The onboard 24V battery serves as the vehicle's power supply, powering the main control unit and the actuator's power pushrod drive circuit via a 12V step-down module. Industrial camera 16 is connected to the corresponding signal port of host computer 17, which communicates with the main control unit in electrical control box 18 via RS232. Displacement sensor 4 is connected to the corresponding signal port of the main control unit via an analog-to-digital conversion module, transmitting image and displacement signals to the main control unit. The main control unit's control output is connected to the controlled end of the actuator via the actuator's power pushrod drive circuit.

[0059] The industrial camera in this embodiment is the MV-SUA131GC-T color industrial camera from Maideweishi. The main control unit in the electrical control box uses a control board based on the STM32F103VET6 chip and integrates an acquisition board. The control board is responsible for outputting control signals, while the acquisition board is responsible for receiving information collected by the sensors. Because the captured images are pre-processed using the state-of-the-art YOLO v5 (or other similar deep learning software algorithms) for target detection, the target area can be accurately extracted. In this embodiment, the industrial camera was used to capture a total of 1,152 images of the scene, annotating two targets: bulbs and roots. 864 images were selected as the training set, and 288 images were used as the validation set. After training, the average accuracy of bulbs and roots was 99.5%. The detection results using the trained YOLO v5s model are shown in Figures 5(a) to 5(d). The bulb area is represented by the solid line box, and the root area is represented by the dashed line box. This shows that the trained model can detect the corresponding targets with relatively high accuracy.

[0060] When working, see Figure 6 , the host computer first controls the equipment inspection according to the following steps:

[0061] Step 1: Send communication instructions to the main control unit;

[0062] Step 2: Determine whether the communication instruction fed back by the main control unit is consistent with the communication instruction sent. If not, issue a corresponding fault elimination warning signal and return to the first step; if yes, proceed to the next step;

[0063] Step 3: Send sensor signal acquisition instructions to the main control unit;

[0064] Step 4: Determine whether the displacement sensor is working normally according to its monitoring interface. If not, issue a corresponding fault warning signal and return to the first step; if yes, proceed to the next step.

[0065] Step 5: Complete the equipment inspection and conduct actual root length measurement and control.

[0066] See also Figure 8The specific process of the host computer automatically controlling the actual root length is as follows:

[0067] Step 1: Set the minimum digging depth to prevent missed digging according to the field conditions;

[0068] Step 2: Set the desired root length according to actual needs;

[0069] Step 3: Set the detection flag to 0;

[0070] Step 4: Enable the displacement sensor;

[0071] Step 5: Determine whether the depth limiting wheel has touched the ground, i.e., whether the excavating device is in place, based on the depth limiting wheel switch control signal. If so, proceed to the next step; otherwise, return to the previous step and continue waiting.

[0072] Step 6: Start the automatic depth limit system;

[0073] Step 7: Collect 800×600 pixel images trained with the YOLO v5 algorithm.

[0074] Step 8. Set the rectangular detection area with the upper left pixel coordinates (100,0) and the lower right pixel coordinates (550,600) according to the shooting range of the industrial camera, and use the rectangular area with the pixel coordinates (100,0) and the pixel coordinates (200,600) as the target area of ​​the detection area. The purpose of setting the detection area is to prevent the ground and the excavation device bracket from affecting the root posture, ensuring that the roots are in a natural drooping state during image processing.

[0075] Step 9: Determine whether a bulb is detected based on the collected image information. If not, proceed to step 18; if yes, proceed to the next step.

[0076] Step 10: Determine whether the pixel coordinates of the upper left corner of the bulb image are within the target area of ​​the detection area. If so, set the detection flag position to 1 and proceed to step 18; otherwise, proceed to the next step. The purpose of setting the target area and the detection flag position is to prevent the system from repeatedly controlling the same garlic plant.

[0077] Step 11: Determine whether the detection flag is 1, if so, proceed to the next step; if not, proceed to step 18;

[0078] Step 12: Determine whether roots are detected based on the collected image information. If so, calculate the actual root length and proceed to the next step; if not, use the expected root length as the excavation depth adjustment value and proceed to step 15;

[0079] Since the detection effect of the trained yolov5s model actually has four situations as shown in Figure 5 (a) to Figure 5 (d), it is necessary to judge the position status of the bulb and root according to the pixel coordinates of the upper left corner (xg1, yg1) and the lower right corner (xg2, yg2) of the received bulb image and the pixel coordinates of the upper left corner (xr1, yr1) and the lower right corner (xr2, yr2) of the root image;

[0080] When the bulb and root images are judged as not overlapping as shown in Figure 5(a), not offset overlapping as shown in Figure 5(b), and offset overlapping as shown in Figure 5(c) and Figure 5(d), the actual root length is calculated according to the following formulas respectively;

[0081] h1=(yr2-yr1) ×ρ×d / f

[0082] h2=(yr2-yr1+yr2-yg2) ×ρ×d / 2f

[0083] h3=cosω×(yr2-yr1+yr2-yg2) ×ρ×d / 2f

[0084] In the above formula

[0085] h1——actual root length where the bulb and root images do not overlap, in mm;

[0086] h2——actual root length when the bulb and root images are not offset and overlapped, in mm;

[0087] h3——actual root length when bulb and root images are offset and overlapped, in mm;

[0088] yr1——the vertical coordinate of the upper left corner pixel of the root image, in pixels;

[0089] yr2——the vertical coordinate of the pixel in the lower right corner of the root image, in pixels;

[0090] yg2——the vertical coordinate of the pixel in the lower right corner of the bulb image, in pixels;

[0091] ρ——pixel size of the camera, in mm / pixel;

[0092] d——object distance of the camera, in mm;

[0093] f——focal length of the camera, in mm;

[0094] ω——the inclination angle of the diagonal line of the overlapping part of the bulb image and the root image, unit: °;

[0095] Step 13: Determine whether the absolute value of the difference between the expected root length and the actual root length is greater than 5 mm. If so, proceed to the next step; otherwise, proceed to step 17;

[0096] Step 14: The difference between the expected root length and the actual root length is used as the excavation depth adjustment value;

[0097] Step 15: Calculate the expected excavation depth by combining the current excavation depth and the excavation depth adjustment value;

[0098] Step 16: Send the desired excavation depth to the main control unit via RS232 communication to adjust the excavation depth;

[0099] Step 17: Set the detection flag to 0;

[0100] Step 18: Determine whether parking control information is received, if yes, proceed to the next step; if not, return to step 7;

[0101] Step 19: Turn off the automatic depth limit system.

[0102] In this embodiment, after the main control unit receives the desired excavation depth signal sent by the host computer, Figure 7 As shown in the figure, follow the specific steps below to adjust the digging depth by using the power push rod:

[0103] Step 1: Receive the desired excavation depth sent by the host computer via RS232 communication;

[0104] Step 2: Read the current extension of the power push rod through the displacement sensor;

[0105] Step 3: Calculate the current excavation depth based on the current power push rod extension (the relationship between the current power push rod extension and the current excavation depth can be calculated based on the current power push rod extension). Figure 1 The specific structure and dimensions of the mechanism shown are derived from the corresponding geometric formulas and are therefore not described in detail);

[0106] Step 4: Determine whether the difference between the expected excavation depth and the current excavation depth is greater than 5 mm. If so, control the power push rod to extend and return to step 2; if not, proceed to the next step;

[0107] Step 5: Determine whether the difference between the desired excavation depth and the current excavation depth is less than -5 mm. If so, control the power push rod to retract and return to step 2; if not, control the power push rod to stop and proceed to the next step;

[0108] Step 6: End this control.

[0109] In short, before turning on the automatic depth limit system, the equipment needs to be checked (see Figure 6), the host computer monitoring interface sends a communication command to the main control unit. If communication is normal, the main control unit returns the communication command to the host computer; otherwise, the fault needs to be corrected. After communication is normal, the displacement sensor is checked again. The host computer sends a data acquisition command to the main control unit to observe whether the displacement sensor's return value is the current hydraulic cylinder extension. If not, the fault needs to be corrected. If so, the equipment inspection ends and the excavation depth is automatically adjusted.

[0110] For example, according to the field conditions, the minimum excavation depth is set to 40mm, the expected root length is set to 10mm, the displacement sensor is enabled, and the automatic depth limit system is turned on after the depth limit wheel is completely on the ground. The image resolution is 800×600, and the detection area is set to the rectangular area with the upper left pixel coordinates (100,0) and the lower right pixel coordinates (550,600). The target area of ​​the detection area is the rectangular area with the upper left pixel coordinates (100,0) and the lower right pixel coordinates (200,600). The YOLOv5 algorithm detects the upper left pixel coordinates (236,214) and lower right pixel coordinates (466,382) of the bulb image and the upper left pixel coordinates (291,378) and lower right pixel coordinates (435,449) of the root image. At this time, the flag bit is 1. It is known that the pixel size ρ of the industrial camera is 6×10 -3 mm / pixel, focal length f is 7.18mm, and object distance d is 350mm.

[0111] According to the pixel coordinates, the bulb and roots are in an unshifted overlapping state, so the actual root length h is calculated using the h2 formula (2), and the result is 20.18 mm. exp is 10 mm, calculate the expected root length h exp The difference Δh from the actual root length h is -10.18mm, and Δh is used as the digging depth adjustment value. The displacement sensor reads the current hydraulic cylinder extension of 22.33mm. Based on the hydraulic cylinder extension and the geometric relationship, the current digging depth d is calculated to be 66.08mm. The expected digging depth d is exp According to the current excavation depth and the excavation depth adjustment value, which is 55.90 mm, the upper computer sends the expected excavation depth of 55.90 mm to the main control unit through RS232 communication. Then, the detection flag position is set to 0, and the switch value is used to determine whether to stop. If so, the automatic depth limit system is turned off. Otherwise, the next frame of image is captured.

[0112] The main control unit receives the desired digging depth of 55.90mm from the host computer, reads the current hydraulic cylinder extension through the displacement sensor, calculates the current digging depth as 66.08mm based on the hydraulic cylinder extension, and controls the hydraulic cylinder to retract until the current digging depth is within the range of 55.90±5mm, controlling the hydraulic cylinder to stop.

[0113] Experiments show that the present embodiment can effectively avoid damaging the bulbs by shoveling, and eliminates the need for root cutting, thereby significantly improving the quality and efficiency of garlic harvesting.

[0114] In addition to the above embodiments, the present invention may also have other implementations. Any technical solution formed by equivalent replacement or equivalent transformation falls within the scope of protection required by the present invention.

Claims

1. A control method for an automatic excavation depth control device, the device comprising a harvesting platform frame (8) whose rear end is hingedly mounted to the front of a harvester, an L-shaped forward-extending excavating shovel (7) fixedly connected to the front lower portion of the harvesting platform frame, and a clamping conveying chain (15) starting from above the excavating shovel and extending from the bottom to the top along the harvesting platform frame; characterized in that: The rear end of the depth-limiting wheel rocker arm (6) is hingedly mounted on the harvesting platform frame, and a rotatable depth-limiting wheel (2) is mounted on the front end of the depth-limiting wheel rocker arm. The middle portion of the depth-limiting wheel rocker arm is hingedly connected to one end of a power push rod (5), and the other end of the power push rod is hingedly connected to the front lower end of the harvesting platform frame. The harvesting platform frame is equipped with an industrial camera (16) facing the front garlic (G) bulb part, and a displacement sensor (4) is installed on one side of the power push rod; the signal output ends of the industrial camera and the displacement sensor are respectively connected to the corresponding signal ports of the main control unit through the host computer and the analog / digital conversion module, and the control output end of the main control unit is connected to the controlled end of the power push rod through the drive circuit; The steps of the method are: The first step is to set the minimum excavation depth, expected root length, and detection flag as initial values; The second step is to collect image information and set the target area in the rectangular detection area; Step 3: determine whether the image information detects a bulb, if not, proceed to step 11; if yes, proceed to the next step; Step 4: Determine whether the pixel coordinates of the upper left corner of the bulb image are within the target area. If so, set the detection flag position to the current value and proceed to step 11. If not, proceed to the next step; Step 5: Determine whether the detection flag is the current value. If so, proceed to the next step; if not, proceed to step 11; Step 6: Determine whether roots are detected in the image information. If so, calculate the actual root length and proceed to the next step; if not, use the expected root length as the excavation depth adjustment value and proceed to step 8; Step 7: Determine whether the absolute value of the difference between the expected root length and the actual root length is greater than a predetermined value. If so, proceed to the next step; otherwise, proceed to the tenth step. Step 8: Using the difference between the expected root length and the actual root length as the excavation depth adjustment value, and revising the expected excavation depth based on the current excavation depth and the excavation depth adjustment value; Step 9: Send the desired excavation depth to the main control unit to adjust the excavation depth; Step 10: Set the detection mark position to the initial value; Step 11: Determine whether parking control information is received. If yes, proceed to the next step; if not, return to step 2. Step 12: End automatic depth limit; The process of calculating the actual root length in the sixth step is: Determine the position status of the bulb and the root according to the received pixel coordinates of the upper left corner and the lower right corner of the bulb image and the pixel coordinates of the upper left corner and the lower right corner of the root image; When the bulb and root images are judged to be non-overlapping, non-overlapping, or overlapping, the actual root length is calculated using the following formulas: h1=(yr2-yr1) ×ρ×d / f h2=(yr2-yr1+yr2-yg2) ×ρ×d / 2f h3=cosω×(yr2-yr1+yr2-yg2) ×ρ×d / 2f In the above formula h1——actual root length where the bulb and root images do not overlap, in mm; h2——actual root length when the bulb and root images are not offset and overlapped, in mm; h3——actual root length when bulb and root images are offset and overlapped, in mm; yr1——the vertical coordinate of the upper left corner pixel of the root image, in pixels; yr2——the vertical coordinate of the pixel in the lower right corner of the root image, in pixels; yg2——the vertical coordinate of the pixel in the lower right corner of the bulb image, in pixels; ρ——pixel size of the camera, in mm / pixel; d——object distance of the camera, in mm; f——focal length of the camera, in mm; ω is the inclination angle of the diagonal line of the overlapping part of the bulb image and the root image, in degrees.

2. The control method of the automatic excavation depth control device according to claim 1, characterized in that: The harvesting table frame includes a support frame formed by a front vertical pole, an upper horizontal pole, a rear vertical pole, and a lower horizontal pole; the front vertical pole and the rear vertical pole are fixedly connected to a conveying beam that is low in the front and high in the back and extends front and back; the front part of the conveying beam is fixedly connected to a vertical sleeve for installing and fixing an L-shaped forward-extending digging shovel at the required height.

3. The control method of the automatic excavation depth control device according to claim 2, characterized in that: The front end of the clamping conveyor chain is equipped with a forward extending guide rod, the input end of a sorting and aligning device is installed in the middle and lower part, and the output end of the sorting and aligning device is installed with a seedling cutting knife disc.

4. The control method of the excavation depth automatic control device according to claim 3, characterized in that: A transmission box driven by a motor is installed at the rear end of the upper cross bar of the harvesting platform frame, and a transmission shaft of the transmission box is transmission-connected to a driving wheel at the rear end of the clamping conveyor chain.

5. The control method of the excavation depth automatic control device according to claim 4, characterized in that: The middle part of the front upright pole of the harvesting platform frame is fixedly connected with a forward supporting rod, the middle part of the supporting rod is hinged with the rear end of a depth-limiting wheel rocker arm, and the front end of the depth-limiting wheel rocker arm is equipped with a rotatable depth-limiting wheel and a forward-extending seedling-dividing rod.

6. The control method of the automatic control device for excavation depth according to claim 5, characterized in that In the ninth step, the main control unit controls the excavation depth as follows: Step 1: Read the current extension of the power push rod and calculate the current digging depth based on the current extension of the power push rod; Step 2: Determine whether the difference between the expected excavation depth and the current excavation depth is greater than a positive threshold. If so, control the power push rod to extend and return to the first step; otherwise, proceed to the next step. Step 3: Determine whether the difference between the expected excavation depth and the current excavation depth is less than a negative threshold. If so, control the power push rod to retract and return to the first step; otherwise, proceed to the next step. Step 4: Control the power push rod to stop and end this control.

Images