Open field vegetable swing type intelligent inter-plant weeding device and control method thereof

Abstract

The invention relates to the technical field of agricultural machinery, and provides an open field vegetable swing type intelligent inter-plant weeding device and a control method thereof, and the device comprises a plurality of weeding monomers, a sensor mechanism and a control mechanism. The sensor mechanism comprises depth cameras, laser sensors and absolute value encoders, the depth cameras are installed on the front portions of the multiple weeding single bodies, the visual field covers multiple rows of crops, and the laser sensors and the absolute value encoders are arranged on the weeding single bodies; the control mechanism is connected with the depth camera, the laser sensor and the absolute value encoder and used for processing data of the sensor mechanism, calculating the fusion height through a fusion algorithm and controlling and driving the lifting mechanism, the transverse moving mechanism and the weeding knife opening and closing mechanism to act based on the error between the fusion height and the target depth so as to achieve accurate weeding and seedling avoiding. The device can quickly avoid crops while effectively removing weeds between rows and plants, and finally realizes mechanical weed removal of open field vegetable crops which are large in bare soil proportion and easy to generate weeds.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of agricultural machinery, in particular to a swing type intelligent inter-plant weeding device for open field vegetables and a control method thereof. BACKGROUND

[0002] Weeds compete with crops for sunlight, water, space and nutrients, leading to reduced crop yield and quality, causing huge losses to the agricultural economy. For crops such as corn, Chinese cabbage, broccoli, lettuce and cabbage, especially in the early and middle stages of their growth, the soil bare area is large, the weed density is high and the damage is serious.

[0003] Currently, there are mainly two ways of weeding: mechanical weeding and chemical weeding. Chemical weeding often uses a continuous application method, which is easy to cause excessive crop pesticide residues, affecting the normal growth of crops and posing a threat to the safety of agricultural products. Mechanical weeding is considered to be the most environmentally friendly weeding method in the world, with the advantages of high efficiency and no pollution. However, mechanical weeding still faces the problems of poor terrain and agronomic adaptability, high energy consumption, low work efficiency, and inability to eradicate some types of weeds. SUMMARY

[0004] The present application provides a swing type intelligent inter-plant weeding device for open field vegetables and a control method thereof, to solve the problems of poor terrain and agronomic adaptability, high energy consumption, low work efficiency, and inability to eradicate some types of weeds in current mechanical weeding.

[0005] The present application provides a swing type intelligent inter-plant weeding device for open field vegetables, comprising: a plurality of weeding units, each weeding unit comprising: a weeding knife opening and closing mechanism, a continuous weeding mechanism, a horizontal moving mechanism and a lifting mechanism; the weeding knife opening and closing mechanism and the continuous weeding mechanism are both arranged on the lifting mechanism, the lifting mechanism is used to adjust the soil penetration depth of the weeding knife opening and closing mechanism and the continuous weeding mechanism; the lifting mechanism is arranged on the horizontal moving mechanism to adjust the horizontal position of the weeding knife opening and closing mechanism and the continuous weeding mechanism through the horizontal moving mechanism; a sensor mechanism comprising a depth camera, a laser sensor and an absolute value encoder, the depth camera is installed at the front of the plurality of weeding units, covering multiple rows of crops in the field of view, the laser sensor and the absolute value encoder are arranged on each weeding unit; a control mechanism connected with the depth camera, the laser sensor and the absolute value encoder, used to process the data of the sensor mechanism and calculate the fusion height through a fusion algorithm, and control the actions of the lifting mechanism, the horizontal moving mechanism and the weeding knife opening and closing mechanism based on the error between the fusion height and the target depth, to achieve precise weeding and seedling avoidance.

[0006] According to the open field vegetable swing type intelligent inter-plant weeding device, the weeding knife opening and closing mechanism comprises: A first driving assembly; A transmission assembly comprising a driving gear, a driven gear, a first connecting shaft and a second connecting shaft; The driving gear is in transmission connection with the first driving assembly, the driving gear and the driven gear are in meshing with each other, and the first connecting shaft and the second connecting shaft are connected with the driving gear and the driven gear respectively; A first swing weeding knife and a second swing weeding knife, the first swing weeding knife and the second swing weeding knife are connected with the first connecting shaft and the second connecting shaft respectively.

[0007] According to the open field vegetable swing type intelligent inter-plant weeding device, the continuous weeding mechanism comprises: A fixing assembly arranged on the lifting mechanism; A first continuous weeding knife and a second continuous weeding knife, both of which are fixed on the fixing assembly and located at the rear part of the weeding knife opening and closing mechanism.

[0008] According to the open field vegetable swing type intelligent inter-plant weeding device, the lifting mechanism comprises a second driving assembly, a lifting sliding table, a lead screw, a lead screw nut and an L-shaped plate. The lead screw is in transmission connection with the second driving assembly, the lead screw is arranged on the lifting sliding table in a vertical direction, the lead screw nut is rotatably arranged on the lead screw, the L-shaped plate is connected with the lead screw nut, and the weeding knife opening and closing mechanism and the continuous weeding mechanism are arranged on the L-shaped plate; the bottom of the L-shaped plate is connected with a profiling wheel, and the absolute value encoder and the laser sensor are arranged on the profiling wheel.

[0009] According to the open field vegetable swing type intelligent inter-plant weeding device, the transverse moving mechanism comprises a stepped gear, a rack, a third driving assembly and a support structure; the rack extends in a horizontal direction, the stepped gear is in meshing with the rack, the third driving assembly is in transmission connection with the rack through the stepped gear, and the support structure is provided with a guide rail, and the lifting sliding table is slidably arranged on the guide rail in a horizontal direction.

[0010] According to the open field vegetable swing type intelligent inter-plant weeding device, the control mechanism comprises a plurality of PID controllers.

[0011] The application also provides a control method of the open field vegetable swing type intelligent inter-plant weeding device, which comprises the following steps: The ground surface height data of each crop row is acquired by a depth camera, a laser sensor and an absolute value encoder in a sensor mechanism, and a fused height is calculated using a weighted fusion algorithm ; wherein, is the fused height of the i-th crop row, , , are height measurement values of the absolute value encoder, the laser sensor and the depth camera, respectively, , , are weights of the absolute value encoder, the laser sensor and the depth camera; based on the fused height and a target depth error , a control signal is generated by an independent control mechanism to drive an actuator to adjust the depth of the weeding knife into the soil; The crop row image is captured by the depth camera, and the field of view is partitioned to process the center line in each row. When a missing row is detected, the center line in the missing row is predicted by the coordinates of the left and right adjacent rows and , and the horizontal movement mechanism is driven to realize cooperative row adjustment.

[0012] The control method of the open-field vegetable swing-type intelligent inter-plant weeding device provided by the application further comprises: According to the displacement distance Δd, a speed strategy is dynamically selected; When ∆d≦ , the maximum speed is set to , and the maximum acceleration is set to is a speed scaling factor, an acceleration scaling factor; When ∆d< < , the maximum speed is set to , and the maximum acceleration is set to , and are intermediate values; When ∆d> , the maximum speed is set to , and the maximum acceleration is set to , is a speed scaling factor, an acceleration scaling factor.

[0013] The control method of the open-field vegetable swing-type intelligent inter-plant weeding device provided by the application, the multi-sensor data fusion step further comprises a data preprocessing sub-step: The point cloud data acquired by the depth camera is filtered and denoised, and the average ground height of each crop row is extracted by partitioning. The change in the profile wheel angle measured by the absolute encoder is calculated and converted into a contact ground height. Real-time calibration of laser sensor data provides high-precision non-contact height information; The dynamic weight adjustment is based on the real-time confidence levels of the depth camera, laser sensor, and absolute encoder. When the data from a certain sensor is abnormal, the weights of other sensors are increased to maintain control continuity.

[0014] The control method for the swing-type intelligent inter-plant weeding device for open-field vegetables provided by the present invention further includes: The system continuously monitors the data streams from the absolute encoder, laser sensor, and depth camera. If an anomaly is detected, the input from the abnormal sensor is paused, and interpolation compensation is performed using historical data or data from adjacent sensors. Parameter K of the control mechanism p K i K d Adaptive optimization is performed based on historical operational data to improve the accuracy of response to different soil conditions.

[0015] This invention also provides a swing-type intelligent inter-row weeding device and its control method for open-field vegetables. Through real-time control of the lateral movement and lifting mechanisms, the weeding blade opening and closing mechanism swings towards the inter-row area or back to the row area. The continuous weeding mechanism can supplement weeding in areas between rows that have not yet been weeded. This allows the device to effectively remove weeds between rows and plants while quickly avoiding crops, ultimately achieving mechanical weed removal for open-field vegetable crops with a large proportion of bare soil and prone to weed growth. Furthermore, this invention also features crop row tracking, which can address seedling damage caused by planting deviations during transplanting and robot movement deviations. It can also adapt to different agronomical variations in crop row spacing and automatically increase or decrease the number of weeding units based on the number of rows. Attached Figure Description

[0016] To more clearly illustrate the technical solutions in this invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are some embodiments of this invention. For those skilled in the art, other drawings can be obtained from these drawings without creative effort.

[0017] Figure 1 This is a front view of the swing-type intelligent inter-plant weeding device for open-field vegetables provided by the present invention.

[0018] Figure 2This is a three-dimensional structural diagram of the swing-type intelligent inter-plant weeding device for open-field vegetables provided by the present invention.

[0019] Figure 3 This is a top view of the swing-type intelligent inter-plant weeding device for open-field vegetables provided by the present invention.

[0020] Figure 4 This is a side view of the swing-type intelligent inter-plant weeding device for open-field vegetables provided by the present invention.

[0021] Figure 5 This is a schematic diagram of the operation of the swing-type intelligent inter-plant weeding device for open-field vegetables provided by the present invention.

[0022] Figure 6 This is a flowchart illustrating the control method of the swing-type intelligent inter-plant weeding device for open-field vegetables provided by the present invention.

[0023] Figure label: 1. Second servo motor; 2. Second reducer; 3. First servo motor; 4. First reducer; 5. First rigid coupling; 6. First bearing; 7. Drive gear; 8. Second connecting shaft; 9. Second bearing; 10. Driven gear; 11. Weeding reducer fixing plate; 12. Gearbox; 13. First connecting shaft; 14. First oscillating weeding blade fixing plate; 15. Second oscillating weeding blade fixing plate; 16. First oscillating weeding blade; 17. Second oscillating weeding blade; 18. Lifting slide; 19. Second rigid coupling; 20. Third reducer; 21. Third servo motor; 22. First square tube; 23. First slider; 24. Second square tube; 25. First guide rail; 26. Third third tube; 27. Second slider; 28. Second guide rail; 29. ​​First 30. Fixed square tube sleeve; 31. Lead screw; 32. Encoder fixing plate; 33. Absolute encoder; 34. Flexible coupling; 35. Third connecting shaft; 36. Contouring wheel fixing plate; 37. Contouring wheel; 38. Fourth square tube; 39. Second fixed square tube sleeve; 40. Third fixed square tube sleeve; 41. Fifth fixed square tube sleeve; 42. First continuous weeding blade fixing plate; 43. Second continuous weeding blade fixing plate; 44. First continuous weeding blade; 45. Second continuous weeding blade; 46. Third reducer fixing plate; 47. Stepped gear; 48. Rack; 49. Fifth square tube; 50. First L-shaped plate; 51. Second L-shaped plate; 52. Sixth square tube; 53. Seventh square tube; 54. First hook point; 55. Second hook point. Detailed Implementation

[0024] To make the objectives, technical solutions, and advantages of this invention clearer, the technical solutions of this invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some, not all, of the embodiments of this invention. All other embodiments obtained by those skilled in the art based on the embodiments of this invention without creative effort are within the scope of protection of this invention.

[0025] In some embodiments, such as Figures 1 to 5 As shown, this open-field vegetable swing-type intelligent inter-plant weeding device includes: multiple weeding units, a sensor mechanism, and a control mechanism. Each weeding unit includes: a weeding blade opening and closing mechanism, a continuous weeding mechanism, a lateral movement mechanism, and a lifting mechanism. The weeding blade opening and closing mechanism and the continuous weeding mechanism are both mounted on the lifting mechanism, which is used to adjust the soil penetration depth of the weeding blade opening and closing mechanism and the continuous weeding mechanism. The lifting mechanism is mounted on the lateral movement mechanism, which can be used to adjust the lateral position of the weeding blade opening and closing mechanism and the continuous weeding mechanism. The sensor mechanism includes a depth camera, a laser sensor, and an absolute encoder 32. The depth camera is installed at the front of multiple weeding units, covering multiple rows of crops. The laser sensor and the absolute encoder 32 are mounted on each weeding unit. The control mechanism is connected to the depth camera, laser sensor, and absolute encoder 32. It processes the sensor data and calculates the fusion height through a fusion algorithm. Based on the error control between the fusion height and the target depth, it drives the lifting mechanism, the lateral movement mechanism, and the weeding blade opening and closing mechanism to achieve precise weeding and seedling avoidance.

[0026] In this embodiment, the absolute encoder 32 is used for contact-based ground height monitoring. A laser sensor is used to detect the non-contact ground height of the crops. A depth camera is located at the front of the device, with its field of view divided into four regions, each corresponding to one row of crops, extracting ground depth information for each of the four rows. The depth camera provides 3D terrain data as a supplement to the global depth of each crop row.

[0027] During operation, the horizontal movement mechanism and the lifting mechanism are controlled in real time to allow the weeding blade opening and closing mechanism to swing towards the inter-plant area or back to the inter-row area. The continuous weeding mechanism can supplement weeding in the inter-row area where weeding has not been done. This allows it to effectively remove weeds between rows and plants while quickly avoiding crops, ultimately achieving mechanical weed removal for open-field vegetable crops with a large proportion of bare soil and easy weed growth.

[0028] For each crop row, the weed control unit acquires ground height data from the corresponding areas of the absolute encoder 32, laser sensor, and depth camera. A weighted fusion algorithm is used to calculate the fused height.

[0029] ; in, It is the fusion height of the i-th crop row. , , These are the height measurements from the absolute encoder, laser sensor, and depth camera, respectively. , , Weights for the absolute encoder, laser sensor, and depth camera.

[0030] Furthermore, the real-time depth control algorithm is based on the fusion height. Depth of target error Control signals are generated through an independent control mechanism. The actuator adjusts the depth of the weeding blade in the soil.

[0031] When there are missing seedlings in a single row of crops, a navigation line cannot be formed. The position of the missing seedling row center line can be predicted by the center lines of the two adjacent rows, so as to achieve multi-single-plant coordinated row alignment.

[0032] Specific implementation plan: A depth camera will be used to cover the entire ridge (4 rows of crops) for crop row detection and coordinated control in case of missing seedlings. The specific plan is as follows: The depth camera is mounted at the front and can cover all four rows of crops in the ridge. After the depth camera captures an image, it is partitioned, dividing the field of view into four regions, each corresponding to the detection of one row of crops.

[0033] Image partitioning and row centerline extraction: Using deep learning image processing algorithms, the row centerlines of crops are extracted separately for each partition. Normally, the depth camera detects the row centerlines of each crop row and generates navigation lines for each row individually.

[0034] Feature extraction for each row of crops is based on the distribution of green vegetation in the image and the density of feature points along the row centerline. If the number of detected feature points in a row is below a set threshold, that row is considered a missing row. A depth camera, acting as a global sensor covering the entire ridge, can simultaneously monitor all four rows of crops. If a row is missing crops or the crop distribution is uneven, the missing area will be detected and marked.

[0035] When a missing row is detected, the control mechanism extracts the row centerline data from the adjacent crop rows on the left and right. This is denoted as... and This represents the coordinates of the row centerlines for the left and right rows. The control mechanism calculates the row centerlines of the missing rows using linear interpolation or the geometric midpoint method. The formula is as follows: If the spacing between adjacent rows is inconsistent, the control mechanism can make fine adjustments based on the actual crop row spacing to ensure that the predicted centerline of the missing row is reasonable. In the area of ​​the missing row, the control mechanism activates a collaborative control mode. Although the centerline of the missing row cannot be directly obtained visually, the data of the adjacent rows acquired by the camera will serve as a navigation reference for that row.

[0036] In some embodiments, such as Figures 1 to 5 As shown, the weeding blade opening and closing mechanism includes: a first drive assembly, a transmission assembly, a first oscillating weeding blade 16, and a second oscillating weeding blade 17. The first drive assembly includes: a first servo motor 3 and a first reducer 4. The transmission assembly includes a drive gear 7, a driven gear 10, a first connecting shaft 13, and a second connecting shaft 8; the drive gear 7 is drively connected to the first drive assembly, and the drive gear 7 and the driven gear 10 mesh with each other. The first oscillating weeding blade 16 and the second oscillating weeding blade 17 are respectively connected to the first connecting shaft 13 and the second connecting shaft 8.

[0037] In this embodiment, the first servo motor 3 serves as the power source, controlling forward and reverse rotation via electrical signals. The first reducer 4 is directly connected to the first servo motor 3 to reduce speed and increase torque, ensuring stable and reliable output power. The first drive assembly is stably mounted on the gearbox 12 via the weeding reducer fixing plate 11 to prevent vibration and displacement during operation. The first reducer 4 is connected to the first rigid coupling 5, and the drive gear 7 is connected to the first drive assembly via the first rigid coupling 5, i.e., meshing with the output shaft of the first reducer 4 via the first rigid coupling 5. The driven gear 10 meshes with the drive gear 7 to form a gear pair. The first connecting shaft 13 and the second connecting shaft 8 are keyed or fixedly connected to the drive gear 7 and the driven gear 10, respectively, transmitting rotational motion to the weeding blade. The drive gear 7 is supported in the gearbox 12 by the first bearing 6, and the driven gear 10 is supported by the second bearing 9.

[0038] The first swinging weeding blade 16 is connected to the first connecting shaft 13 via the first swinging weeding blade fixing plate 14, and the second swinging weeding blade 17 is connected to the second connecting shaft 8 via the second swinging weeding blade fixing plate 15. The fixing plates are made of rigid material to ensure that the weeding blades do not deform or loosen during swinging. The weeding blades are made of wear-resistant alloy material, and the blades are designed with an arc shape to meet the cutting needs of weeds between plants.

[0039] The specific workflow is as follows: The control mechanism sends a command to the first servo motor 3, driving it to rotate forward or reverse. When rotating forward, the driving gear 7 rotates clockwise, driving the driven gear 10 to rotate counterclockwise. Through the connecting shaft, the first oscillating weeding blade 16 and the second oscillating weeding blade 17 swing open to both sides, entering the inter-plant weeding state. When rotating in reverse, the gears move in the opposite direction, and the weeding blades swing close towards the center, thus avoiding the crops.

[0040] In some embodiments, such as Figures 1 to 5 As shown, the continuous weeding mechanism includes: a fixing component, a first continuous weeding blade 44, and a second continuous weeding blade 45. The fixing component is mounted on the lifting mechanism; both the first continuous weeding blade 44 and the second continuous weeding blade 45 are fixed to the fixing component and are located at the rear of the weeding blade opening and closing mechanism.

[0041] Specifically, the fixing components include: a sixth square tube 52, a seventh square tube 53, a first continuous weeding blade fixing plate 42, a second continuous weeding blade fixing plate 43, a second fixing square tube sleeve 38, a third fixing square tube sleeve 39, a fourth fixing square tube sleeve 40, and a fifth fixing square tube sleeve 41.

[0042] The seventh square tube 53 is connected to the lifting mechanism via the sixth square tube 52. The second fixed square tube sleeve 38 and the third fixed square tube sleeve 39 are fitted onto the seventh square tube 53, the fourth fixed square tube sleeve 40 is fitted onto the first continuous weeding blade fixing plate 42, and the fifth fixed square tube sleeve 41 is fitted onto the second continuous weeding blade fixing plate 43. The first continuous weeding blade 44 is fixed onto the first continuous weeding blade fixing plate 42. The second continuous weeding blade 45 is fixed onto the second continuous weeding blade fixing plate 43. The square tube sleeves are designed with standardized interfaces, allowing for quick installation and disassembly to adapt to different operational needs. During operation, as the device moves forward, the first continuous weeding blade 44 and the second continuous weeding blade 45 continuously cut the soil between rows, removing weed roots. In conjunction with the oscillating weeding blade (weeding blade opening and closing mechanism), a forward-swinging and backward-filling operation mode is formed: the oscillating weeding blade handles complex areas between plants, while the continuous blade handles open areas between rows, significantly improving weeding efficiency.

[0043] In some embodiments, such as Figures 1 to 5 As shown, the lifting mechanism includes: a second drive assembly, a lifting slide 18, a lead screw 30, a lead screw 30 nut, and an L-shaped plate. The lead screw 30 is connected to the second drive assembly and is vertically mounted on the lifting slide 18. The lead screw 30 nut is rotatably mounted on the lead screw 30. The L-shaped plate is connected to the lead screw 30 nut. The weeding blade opening and closing mechanism and the continuous weeding mechanism are both mounted on the L-shaped plate. A contour wheel 36 is connected to the bottom of the L-shaped plate, and an absolute encoder 32 and a laser sensor are mounted on the contour wheel 36.

[0044] Specifically, to avoid interference between adjacent weeding units, multiple weeding units employ L-shaped plates of different lengths, namely a first L-shaped plate 50 and a second L-shaped plate 51. The length of the second L-shaped plate 51 is greater than that of the first L-shaped plate 50, thus placing the weeding blade opening and closing mechanism and the continuous weeding mechanism on the first L-shaped plate 50 and the second L-shaped plate 51 in different positions. This design ensures that the weeding mechanisms of adjacent weeding units are positioned differently, avoiding mechanical interference during operation. The second drive assembly includes: a second servo motor 1 and a second reducer 2. The lifting mechanism also includes: a fourth square tube 37, a fifth square tube 49, a second rigid coupling 19, an encoder fixing plate 31, a flexible coupling 33, a third connecting shaft 34, a contour wheel fixing plate 35, a contour wheel 36, and the fourth square tube 37. The second servo motor 1 achieves forward and reverse rotation through control signals. The second reducer 2 is connected to the output shaft of the second servo motor 1 to reduce the speed and increase the torque, ensuring smooth output power. The second rigid coupling 19 is mounted on the shaft of the second reducer 2. Generally, the second rigid coupling 19 is connected to the lead screw, thereby controlling the rotation of the lead screw. The lead screw 30 is vertically mounted on the lifting slide 18. The lead screw 30 nut is rotatably mounted on the lead screw 30 and is driven to rotate by a servo motor, causing the lead screw 30 nut to move up and down along the lead screw 30. The L-shaped plate is fixedly connected to the lead screw 30 nut. The weeding blade opening and closing mechanism and the continuous weeding mechanism are both mounted on the L-shaped plate. The first fixed square tube sleeve 29 is mounted on the fourth square tube 37. The fifth square tube 49 is horizontally connected to the L-shaped plate, and the fourth square tube 37 is vertically connected to the fifth square tube 49 via the first fixed square tube sleeve 29. The absolute encoder 32 is connected to the contour wheel 36 via a flexible coupling 33 and a third connecting shaft 34, monitoring the angle change of the contour wheel 36 in real time for contact-type ground height detection. The laser sensor is mounted next to the contour wheel 36, providing non-contact height data. The contour wheel 36 is connected to the fourth square tube 37 via the contour wheel fixing plate 35, and rolls with the terrain undulations to transmit the height change to the absolute encoder 32.

[0045] In some embodiments, such as Figures 1 to 5 As shown, the transverse mechanism includes: a stepped gear 47, a rack 48, a third drive assembly, and a support structure; the rack 48 extends in the horizontal direction, the stepped gear 47 meshes with the rack 48, the third drive assembly is connected to the rack 48 through the stepped gear 47, the support structure is provided with a guide rail, and the lifting slide 18 is slidably arranged on the guide rail in the horizontal direction.

[0046] In this embodiment, the third drive component includes a third reducer 20 and a third servo motor 21. The third reducer 20 is mounted on a third reducer mounting plate 46. The third servo motor 21 serves as a power source, achieving forward and reverse rotation via control signals. The third reducer 20 is connected to the output shaft of the servo motor to reduce speed and increase torque, ensuring smooth output power. The support structure includes a first square tube 22, a second square tube 24, and a third third tube 26 arranged sequentially from top to bottom along the height direction. The first square tube 22, the second square tube 24, and the third third tube 26 all extend horizontally. A rack 48 is arranged horizontally on the first square tube 22. A first guide rail 25 is provided on the second square tube 24, and a second guide rail 28 is provided on the third third tube 26. A first slider 23 and a second slider 27 are connected to the lifting mechanism. The first slider 23 is slidably mounted on the first guide rail 25, and the second slider 27 is slidably mounted on the second guide rail 28, forming a double guide rail support, allowing the lifting mechanism to move smoothly horizontally in conjunction with the weeding blade opening and closing mechanism and the continuous weeding mechanism above.

[0047] In addition, the open-field vegetable swing-type intelligent inter-plant weeding device is also equipped with a first attachment point 54 and a second attachment point 55 for connecting with external equipment to ensure operational stability.

[0048] In some embodiments, the control mechanism includes multiple PID controllers. Each weeding unit is equipped with an independent PID controller, allowing data from different crop rows to be processed in parallel. For example, in a four-row crop system, four PID controllers operate simultaneously, each processing sensor data for its corresponding row and driving the actuators. This design avoids inter-row interference and improves the system's response speed and reliability. Independence is also reflected in parameter customization—soil conditions in different rows may vary significantly (e.g., the left row has loose soil, while the right row has firm soil), and the parameters of each controller can be independently adjusted to adapt to the local environment.

[0049] Figure 6 This is a flowchart illustrating the control method of the swing-type intelligent inter-plant weeding device for open-field vegetables provided by the present invention, as shown below. Figure 6 As shown, the control method includes the following steps: Step S601: Obtain the ground height data of each crop row using the depth camera, laser sensor and absolute encoder in the sensor mechanism, and calculate the fused height using a weighted fusion algorithm.

[0050] Step S602: Based on the error between the fusion height and the target depth, a control signal is generated through an independent control mechanism to drive the actuator to adjust the penetration depth of the weeding blade.

[0051] Step S603: Capture crop row images using a depth camera, divide the field of view into sections to extract the center lines of each row, and when a missing row is detected, predict the center line of the missing row using the coordinates of the center lines of the left and right adjacent rows, and drive the lateral movement mechanism to achieve coordinated row alignment.

[0052] Specifically, by controlling the lateral movement mechanism and the lifting mechanism in real time, the weeding blade opening and closing mechanism can be moved towards the inter-plant area or back to the inter-row area. The continuous weeding mechanism can supplement the weeding in the inter-row area where weeding has not been done, so that it can effectively remove weeds between rows and plants while quickly avoiding crops. Ultimately, it can achieve mechanical removal of weeds for open-field vegetable crops with a large proportion of bare soil and easy weed growth.

[0053] For each crop row, the weed control unit acquires ground height data from the corresponding areas of the absolute encoder, laser sensor, and depth camera. A weighted fusion algorithm is used to calculate the fused height.

[0054] ; in, It is the fusion height of the i-th crop row. , , These are the height measurements from the absolute encoder, laser sensor, and depth camera, respectively. , , Weights for the absolute encoder, laser sensor, and depth camera.

[0055] Furthermore, the real-time depth control algorithm is based on the fusion height. Depth of target error Control signals are generated through an independent control mechanism. The actuator adjusts the depth of the weeding blade in the soil.

[0056] When there are missing seedlings in a single row of crops, a navigation line cannot be formed. The position of the missing seedling row center line can be predicted by the center lines of the two adjacent rows, so as to achieve multi-single-plant coordinated row alignment.

[0057] Specific implementation plan: A depth camera will be used to cover the entire ridge (4 rows of crops) for crop row detection and coordinated control in case of missing seedlings. The specific plan is as follows: The depth camera is mounted at the front and can cover all four rows of crops in the ridge. After the depth camera captures an image, it is partitioned, dividing the field of view into four regions, each corresponding to the detection of one row of crops.

[0058] Image partitioning and row centerline extraction: Using deep learning image processing algorithms, the row centerlines of crops are extracted separately for each partition. Normally, the depth camera detects the row centerlines of each crop row and generates navigation lines for each row individually.

[0059] Feature extraction for each row of crops is based on the distribution of green vegetation in the image and the density of feature points along the row centerline. If the number of detected feature points in a row is below a set threshold, that row is considered a missing row. A depth camera, acting as a global sensor covering the entire ridge, can simultaneously monitor all four rows of crops. If a row is missing crops or the crop distribution is uneven, the missing area will be detected and marked.

[0060] When a missing row is detected, the control mechanism extracts the row centerline data from the adjacent crop rows on the left and right. This is denoted as... and This represents the coordinates of the row centerlines for the left and right rows. The control mechanism calculates the row centerlines of the missing rows using linear interpolation or the geometric midpoint method. The formula is as follows: If the spacing between adjacent rows is inconsistent, the control mechanism can make fine adjustments based on the actual crop row spacing to ensure that the predicted centerline of the missing row is reasonable. In the area of ​​the missing row, the control mechanism activates a collaborative control mode. Although the centerline of the missing row cannot be directly obtained visually, the data of the adjacent rows acquired by the camera will serve as a navigation reference for that row.

[0061] To meet the system's requirements for different displacement distances and to ensure that the servo motor can avoid overshoot and vibration while maintaining precise control over short distances and rapid response over long distances, the control strategy is as follows: Based on different displacement distances, a dynamic speed and acceleration / deceleration control strategy is adopted to ensure both rapid response and avoidance of overshoot at short distances.

[0062] Short-distance displacement (∆d≦) ): Control strategy: Use S-shaped speed control to ensure smooth acceleration and deceleration, and avoid system overshoot and vibration.

[0063] Speed ​​setting: based on displacement distance The maximum speed is set to: in It is a speed scaling factor that ensures that speed is proportional to distance and avoids excessively fast movement.

[0064] Acceleration setting: Maximum acceleration It should be lower to ensure a smooth acceleration process: in An acceleration scaling factor makes acceleration correlated with displacement distance.

[0065] Control process: Acceleration phase: The motor starts from zero acceleration and gradually accelerates to the target speed.

[0066] Constant speed phase: Adjust whether to enter constant speed operation based on the displacement distance. If the distance is very short, it may directly enter deceleration.

[0067] Deceleration phase: The acceleration gradually decreases, eventually decelerating to the target position.

[0068] Mid-distance displacement (∆d < < ): Control strategy: Based on trapezoidal velocity control, maintain a high response speed while preventing excessive vibration.

[0069] Speed ​​setting: Maximum speed Set it to a medium value to ensure a fast enough response while preventing overshoot: Acceleration setting: Acceleration Choose a moderate value to ensure that the acceleration and deceleration processes are both smooth and do not prolong the process: Control process: Acceleration phase: The motor rapidly accelerates to the set maximum speed. And maintain a constant speed for a period of time.

[0070] Deceleration phase: Gradually decelerate as you approach the target position to ensure a smooth stop.

[0071] Long-distance displacement (∆d> ): Control strategy: Use trapezoidal speed control to allow the system to quickly reach maximum speed and respond as soon as possible.

[0072] Speed ​​setting: Maximum speed It should be close to the system's speed limit to quickly complete a large lateral movement: in It is a speed scaling factor that ensures that speed is proportional to distance and avoids excessively fast movement.

[0073] Acceleration setting: Maximum acceleration It should be lower to ensure a smooth acceleration process: in An acceleration scaling factor makes acceleration correlated with displacement distance.

[0074] Control process: Acceleration phase: Rapidly accelerate to the set maximum speed Constant speed phase: Maintain high speed over long distances to minimize time.

[0075] Deceleration phase: Gradually decelerate as you approach the target to avoid impacting its position.

[0076] In some embodiments, the multi-sensor data fusion step further includes a data preprocessing sub-step: Step S701: Filter and denoise the point cloud data acquired by the depth camera, and extract the average ground height of each crop row by region.

[0077] Step S702: Calculate the change in the profile wheel angle measured by the absolute encoder and convert it into a contact ground height.

[0078] Step S703: Perform real-time calibration of the laser sensor data to provide high-precision non-contact height information.

[0079] In this embodiment, a depth camera is mounted at the front of the device, covering the entire ridge (e.g., 4 rows of crops) and divided into multiple regions (each region corresponding to one row of crops). The point cloud data is denoised using statistical filtering algorithms (e.g., radius outlier removal) to remove outliers caused by ambient light, dust, or flying insects. Subsequently, the point cloud data of each region is clustered, the average point cloud density is calculated, and the average ground height of each crop row is extracted.

[0080] An absolute encoder is installed on the contour wheel of each weeding unit to monitor the angular changes of the contour wheel in real time. The ground height at the contact point is calculated using a geometric model: Let the radius of the contour wheel be R, and the angular change be Δθ, then the height change ΔH = R × sin(Δθ). Combined with the initial calibration height, this is converted into an absolute ground height value. Encoder data is sampled at a high frequency (e.g., 100Hz) to ensure fast response.

[0081] A laser sensor, positioned next to the contour wheel, measures non-contact ground elevation using the Time-of-Flight (ToF) principle. Data calibration includes temperature compensation and background light filtering. The calibration algorithm is based on linear regression of historical data, adjusting the offset in real time. For example, a reference surface calibration is performed before operation to ensure initial accuracy.

[0082] The weights are dynamically adjusted based on real-time confidence levels from the depth camera, laser sensor, and absolute encoder. When data from one sensor becomes abnormal, the weights of other sensors are increased to maintain control continuity. Furthermore, adaptive control algorithms, combined with historical operational data, can optimize the control strategy, continuously improving response speed and accuracy under different crop and soil conditions.

[0083] In some embodiments, during the control process, the data streams of the absolute encoder, laser sensor, and depth camera are continuously monitored. If an anomaly is detected, the input of the abnormal sensor is paused, and interpolation compensation is performed using historical data or data from adjacent sensors. The parameter K of the control mechanism (generally a PID controller) is... p K i K d Adaptive optimization is performed based on historical operational data to improve the accuracy of response to different soil conditions.

[0084] In summary, this invention achieves precise control over the penetration depth of the weeding blade by integrating data from a depth camera, laser sensor, and absolute encoder, avoiding energy waste caused by excessive penetration or ineffective weeding due to insufficient penetration. Simultaneously, the depth camera provides full coverage monitoring of all four rows of crops in the work area, and centerline data from adjacent rows is used to predict the centerline of missing rows, ensuring accurate row-specific operations even in cases of missing seedlings. This invention uses a servo motor to control the oscillation of the weeding blade in real time, achieving precise weeding between plants and rows, and possesses rapid seedling avoidance capabilities, making it suitable for various agronomic requirements and different crop planting methods. Based on the working environment and crop row spacing, the number of weeding units and operating parameters are adaptively adjusted, and S-shaped and trapezoidal speed control algorithms are employed to ensure rapid response of the servo motor at different distances, avoiding overshoot and improving operational accuracy and efficiency.

[0085] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, and not to limit them; although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features; and these modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A swing-type intelligent inter-plant weeding device for open-field vegetables, characterized in that, include: Multiple weeding units, each of which includes: a weeding blade opening and closing mechanism, a continuous weeding mechanism, a traversing mechanism, and a lifting mechanism; Both the weeding blade opening and closing mechanism and the continuous weeding mechanism are mounted on the lifting mechanism. The lifting mechanism is used to adjust the soil penetration depth of the weeding blade opening and closing mechanism and the continuous weeding mechanism. The lifting mechanism is mounted on the lateral movement mechanism to adjust the lateral position of the weeding blade opening and closing mechanism and the continuous weeding mechanism. The sensor mechanism includes a depth camera, a laser sensor, and an absolute encoder. The depth camera is mounted on the front of the plurality of weeding units, covering multiple rows of crops. The laser sensor and the absolute encoder are disposed on each weeding unit. The control mechanism, connected to the depth camera, the laser sensor, and the absolute encoder, processes the sensor data and calculates the fusion height using a fusion algorithm. Based on the error between the fusion height and the target depth, it controls and drives the lifting mechanism, the lateral movement mechanism, and the weeding blade opening and closing mechanism to achieve precise weeding and seedling avoidance.

2. The swing-type intelligent inter-plant weeding device for open-field vegetables according to claim 1, characterized in that, The weeding blade opening and closing mechanism includes: First driving component; The transmission assembly includes a driving gear, a driven gear, a first connecting shaft, and a second connecting shaft; The driving gear is connected to the first driving component, the driving gear and the driven gear mesh with each other, and the first connecting shaft and the second connecting shaft are respectively connected to the driving gear and the driven gear; A first oscillating weed cutter and a second oscillating weed cutter are respectively connected to the first connecting shaft and the second connecting shaft.

3. The swing-type intelligent inter-plant weeding device for open-field vegetables according to claim 1, characterized in that, The continuous weeding mechanism includes: A fixed component is mounted on the lifting mechanism; The first and second continuous weeding blades are both fixed to the fixing assembly and located at the rear of the weeding blade opening and closing mechanism.

4. The swing-type intelligent inter-plant weeding device for open-field vegetables according to claim 1, characterized in that, The lifting mechanism includes: a second drive assembly, a lifting slide, a lead screw, a lead screw nut, and an L-shaped plate; The lead screw is connected to the second drive assembly for transmission. The lead screw is vertically mounted on the lifting slide. The lead screw nut is rotatably mounted on the lead screw. The L-shaped plate is connected to the lead screw nut. The weeding blade opening and closing mechanism and the continuous weeding mechanism are both mounted on the L-shaped plate. A contour wheel is connected to the bottom of the L-shaped plate. The absolute encoder and the laser sensor are mounted on the contour wheel.

5. The swing-type intelligent inter-plant weeding device for open-field vegetables according to claim 4, characterized in that, The lateral movement mechanism includes: a stepped gear, a rack, a third drive assembly, and a support structure; the rack extends horizontally, the stepped gear meshes with the rack, the third drive assembly is connected to the rack via the stepped gear, the support structure is provided with a guide rail, and the lifting slide is slidably mounted on the guide rail in the horizontal direction.

6. The swing-type intelligent inter-plant weeding device for open-field vegetables according to claim 4, characterized in that, The control mechanism includes multiple PID controllers.

7. A control method for an open-field vegetable swing-type intelligent inter-plant weeding device based on any one of claims 1-6, characterized in that, Includes the following steps: The ground elevation data for each crop row is acquired using a depth camera, laser sensor, and absolute encoder in the sensor array, and the fused elevation is calculated using a weighted fusion algorithm. ;in, It is the fusion height of the i-th crop row. , , These are the height measurements from the absolute encoder, laser sensor, and depth camera, respectively. , , Weights for the absolute encoder, laser sensor, and depth camera; Based on fusion height Depth of target error Control signals are generated through an independent control mechanism. The actuator is driven to adjust the depth of the weeding blade in the soil; Crop row images are captured using a depth camera, and the field of view is divided into sections to extract the centerlines of each row. When a missing row is detected, the coordinates of the centerlines of the left and right adjacent rows are used to identify it. and Predicting the midline of the row with missing seedlings, And drive the lateral movement mechanism to achieve coordinated alignment.

8. The control method according to claim 7, characterized in that, Also includes: Dynamically select the velocity strategy based on the displacement distance Δd; When ∆d≦ At that time, the maximum speed is set to The maximum acceleration is set to It is the speed scaling factor. Acceleration scaling factor; When ∆d < < At that time, the maximum speed is set to The maximum acceleration is set to , and All are median values; When ∆d> At that time, the maximum speed is set to The maximum acceleration is set to , It is the speed scaling factor. Acceleration scaling factor.

9. The control method according to claim 7, characterized in that, The multi-sensor data fusion process also includes a data preprocessing sub-step: The point cloud data acquired by the depth camera is filtered and denoised, and the average ground height of each crop row is extracted by partitioning. The change in the profile wheel angle measured by the absolute encoder is calculated and converted into a contact-type change in ground height; Real-time calibration of laser sensor data provides high-precision, non-contact height information; The dynamic weight adjustment is based on the real-time confidence levels of the depth camera, laser sensor, and absolute encoder. When the data from a certain sensor is abnormal, the weights of other sensors are increased to maintain control continuity.

10. The control method according to claim 7, characterized in that, Also includes: The system continuously monitors the data streams from the absolute encoder, laser sensor, and depth camera. If an anomaly is detected, the input from the abnormal sensor is paused, and interpolation compensation is performed using historical data or data from adjacent sensors. Parameter K of the control mechanism p K i K d Adaptive optimization is performed based on historical operational data to improve the accuracy of response to different soil conditions.

Images