Seedling grafting robot flexible seedling clamping device and control method thereof

Abstract

This invention relates to a flexible clamping and seedling-picking device for a seedling grafting robot and its control method, belonging to the field of agricultural machinery technology. It addresses the problems of inaccurate positioning, high damage rate, and poor adaptability in the seedling picking process of seedling grafting robots. The clamping and seedling-picking device includes a mounting base, a clamping actuator, a force sensing module, a displacement detection module, a vision detection module, and a control module. The mounting base is detachably connected to the end of the robotic arm of the seedling grafting robot via a modular quick-change interface. The clamping actuator includes a first clamping arm and a second clamping arm symmetrically arranged and movably connected to the mounting base. The force sensing module includes miniature thin-film pressure sensors mounted on the first and second clamping arms. The displacement detection module includes displacement sensors mounted on the first and second clamping arms. The vision detection module includes a depth camera mounted above the mounting base. This invention achieves precise detection and positioning of the spatial position of the seedling stem.

Description

Technical Field

[0001] This invention relates to the field of agricultural machinery technology, specifically to a flexible clamping and seedling picking device for seedling grafting robots and its control method. Background Technology

[0002] Seedling grafting is a crucial technical step in protected horticulture production, and the quality of grafted seedlings directly affects crop yield and quality. With the rapid development of factory-scale seedling production, seedling grafting robots have become core equipment replacing traditional manual grafting operations. In the operation process of seedling grafting robots, the seedling picking and transportation stage is the primary and critical step. The robotic arm needs to remove the seedlings from the seedling trays and transport them to the cutting station for subsequent grafting operations. However, due to individual differences in the cultivation posture of seedlings in the seedling trays, the seedling stems may tilt or bend in different directions. In addition, there is a certain degree of repetitive positioning deviation in the placement of the seedling trays, resulting in a deviation between the actual spatial position of the seedling stem and the preset position. Existing seedling picking devices mostly adopt a fixed-position picking method, where the robotic arm moves to a fixed coordinate position according to a preset trajectory to grip and pick up the seedling. This method cannot detect and adapt to the above-mentioned positional deviations, often resulting in the gripper missing or misaligning, which seriously affects the success rate of seedling picking and operational efficiency.

[0003] Meanwhile, existing seedling clamping devices suffer from two prominent problems at the clamping execution level: First, the clamping mechanism has poor adaptability to seedling stems, easily causing mechanical damage. The rigid grippers or cylinder-driven clamping mechanisms widely used in existing technologies have fixed clamping opening sizes and non-adjustable clamping angles, making it difficult to adapt to individual differences in stem diameter among different seedling species. When the clamping opening is smaller than the seedling stem diameter, the grippers forcibly squeeze the stem, easily causing damage to the stem epidermis or internal tissues; when the clamping opening is too large, effective clamping cannot be achieved, affecting the stability of subsequent handling and grafting operations. Second, the clamping force cannot be accurately sensed and adjusted in real time, lacking a force feedback control mechanism and precise control logic for clamping displacement. Traditional clamping devices mostly use simple cylinder on / off control, with the clamping force determined by the air source pressure, making it impossible to differentiate according to the seedling type and diameter. More importantly, the existing clamping control logic lacks closed-loop detection and control of the gripper displacement. The control module cannot know how much the gripper has actually closed or whether the flexible gripper fingers have made contact with the seedling stem. It relies solely on open-loop control based on time or air pressure. This blind clamping method either results in inadequate clamping due to insufficient closure or damage to the seedling due to excessive closure. Furthermore, different types of seedlings have significantly different compressive strengths. Using a uniform clamping force can easily cause irreversible mechanical damage to seedlings with weak compressive strength, affecting the grafting survival rate.

[0004] In summary, there is an urgent need for a seedling picking device and its control method that can accurately locate and adaptively hold seedlings of different positions, postures, types, and sizes with low damage, in order to solve the technical problems of inaccurate positioning, high damage rate, and poor adaptability in the seedling picking process of seedling grafting robots. Summary of the Invention

[0005] The purpose of this invention is to provide a flexible clamping and seedling picking device for a seedling grafting robot and its control method, which can achieve precise positioning and low-damage adaptive clamping of seedlings of different positions, postures, types and sizes, so as to solve the problems of inaccurate positioning, high damage rate and poor adaptability in the seedling picking process of the seedling grafting robot.

[0006] The technical solution adopted by this invention to solve its technical problem is: a flexible clamping and seedling picking device for a seedling grafting robot, including a mounting base, a clamping execution mechanism, a force sensing module, a displacement detection module, a vision detection module, and a control module. The mounting base is detachably connected to the end of the robotic arm of the seedling grafting robot through a modular quick-change interface. The clamping execution mechanism includes a first clamping arm and a second clamping arm symmetrically arranged and movably connected to the mounting base. The mounting base has a clamping arm driving mechanism for driving the opening and closing movements of the first and second clamping arms. The force sensing module includes a miniature thin-film pressure sensor disposed on the first and second clamping arms. A thin-film pressure sensor is used to acquire the contact pressure signal applied to the seedling stem during clamping in real time; the displacement detection module includes displacement sensors disposed on the first and second clamping arms, used to detect the displacement of the first and second clamping arms in the opening and closing direction in real time; the vision detection module includes a depth camera disposed above the mounting base, used to acquire image information and depth information of the seedlings to be taken; the control module is signal-connected to the clamping arm drive mechanism, the micro-thin-film pressure sensor, the displacement sensor, and the depth camera; the detection signals of the micro-thin-film pressure sensor, the displacement sensor, and the depth camera are uploaded to the control module, and the control module sends control commands to the clamping arm drive mechanism.

[0007] Furthermore, the modular quick-change interface includes a claw and a quick-change drive mechanism. The quick-change drive mechanism is disposed inside the mounting base, and the claw is disposed at the output end of the quick-change drive mechanism. The quick-change drive mechanism drives the extension and retraction of the claw. The end of the robotic arm is provided with an annular positioning groove. After the quick-change drive mechanism drives the claw to extend into the annular positioning groove, the mounting base and the end of the robotic arm are locked. After the quick-change drive mechanism drives the claw to extend out of the annular positioning groove, the mounting base and the end of the robotic arm are disassembled.

[0008] Furthermore, the clamping arm drive mechanism includes a servo motor and a reducer. The input end of the reducer is fixedly connected to the output shaft of the servo motor, and the output end of the reducer is fixedly connected to the rotating shaft of the first clamping arm and the second clamping arm.

[0009] Furthermore, the reducer incorporates a worm gear transmission mechanism, which includes a meshing worm and a worm wheel. The worm is fixedly connected to the output shaft of the servo motor, and the worm wheel is fixedly connected to the rotating shaft of the first clamping arm and the second clamping arm. Alternatively, the reducer incorporates a gear transmission mechanism, which includes a meshing driving gear and a driven gear. The driving gear is fixedly connected to the output shaft of the servo motor, and the driven gear is fixedly connected to the rotating shaft of the first clamping arm and the second clamping arm.

[0010] Furthermore, both the first clamping arm and the second clamping arm include a rigid arm body and flexible clamping fingers. The upper end of the rigid arm body is hinged to the mounting base via a pivot, and the flexible clamping fingers are disposed at the lower end of the rigid arm body.

[0011] Furthermore, a hollow sleeve connection part is provided at the lower end of the rigid arm body. The flexible clamping finger is made of silicone material by injection molding or 3D printing. During the molding process, the root of the flexible clamping finger directly covers and is embedded in the sleeve connection part.

[0012] Furthermore, the flexible clamping finger has a V-shaped groove extending along the clamping direction on its clamping surface, the included angle of the V-shaped groove being 60°-120°, and the inner surface of the V-shaped groove having an anti-slip texture; the flexible clamping finger has multiple elastic deformation chambers arranged along the clamping direction embedded inside, and each elastic deformation chamber is connected to an external air source through an independent micro air passage; when gas at different pressures is filled into each elastic deformation chamber, different areas of the flexible clamping finger will undergo different degrees of expansion and deformation.

[0013] Furthermore, the rigid arm body has a shallow groove mounting position on the end face facing the flexible gripper finger, and the micro-thin film pressure sensor is embedded in the shallow groove mounting position, with its sensing surface closely fitting the root end face of the flexible gripper finger.

[0014] Furthermore, the displacement detection module includes a displacement sensor disposed on the first clamping arm and the second clamping arm. The displacement sensor is a high-precision grating ruler displacement sensor, with its reading head fixed on the rigid arm body and the grating ruler fixed on the guide rail of the mounting base, for real-time detection of the displacement of the clamping arm in the opening and closing direction.

[0015] Furthermore, the image information includes the stem diameter, stem tilt angle, and leaf distribution of the seedling, while the depth information is the position coordinates of the seedling stem in three-dimensional space.

[0016] The present invention also provides a control method for a flexible gripping and seedling-retrieving device for a seedling grafting robot, comprising the following steps:

[0017] S1. Detection of the spatial position and posture of seedling stems.

[0018] First, the image and depth information of the seedlings to be collected are obtained through the visual detection module. After image processing and 3D reconstruction, the position coordinates (X, Y, Z) of the seedling stem in 3D space are extracted; the diameter d of the seedling stem and the inclination angle θ of the stem are also extracted.

[0019] S2. Seedling retrieval path planning.

[0020] Based on the position coordinates (X, Y, Z) and the current spatial pose of the clamping actuator, calculate the displacement control amount required for the gripper to move from the initial position to the target clamping position; plan a smooth motion trajectory from the initial position to the target position based on the target position and the current pose.

[0021] S3. Calculation of target clamping force.

[0022] Based on the stem diameter d obtained in step S1 and the preset seedling type information, the seedling diameter-clamping force mapping model stored in the control module is invoked to calculate the target clamping force F of the clamping actuator. target .

[0023] S4. Move to the target location.

[0024] According to the trajectory planned in step S2, the installation base and clamping actuator are moved to the target clamping position, so that the flexible clamping fingers are aligned with the part of the seedling stem to be clamped.

[0025] S5. Flexible finger stiffness adaptive adjustment.

[0026] The target clamping force F calculated in step S3 target Based on the stem inclination angle θ, the target air pressure value P of each elastic deformation chamber is calculated. i , i=1,2,...,n, and inflate each elastic deformation chamber to make the flexible clamping fingers achieve the stiffness and contour shape that match the seedling stem.

[0027] S6. Gripping and grasping operation.

[0028] S6.1. Position control stage - pre-clamping.

[0029] The first and second clamping arms are driven to close towards the seedling stem at a first speed of 50 mm / s-100 mm / s, while the displacement S of the clamping arms collected by the displacement sensor is read in real time. actual When S actual Reaching the preset displacement threshold S thresholdAt that time, the position control phase stops.

[0030] S6.2. Force control stage - precision clamping.

[0031] The first and second clamping arms are driven to close slowly at a second speed of 5 mm / s - 20 mm / s, while the actual clamping force value F collected by the micro-film pressure sensor is read in real time. actual ; F actual Clamping force F with the target target Compare; when F actual <F target When F continues to increase the clamping stroke; actual ≥F target When the clamping action stops and the current clamping force is maintained, the clamping operation is completed.

[0032] S6.3. Displacement-force coordination judgment and anomaly handling.

[0033] During the clamping process, the displacement S collected by the displacement sensor is monitored simultaneously. actual The force value F collected by the miniature thin-film pressure sensor actual The following collaborative judgment logic is executed: Case 1: Abnormalities such as excessively thick stems or obstruction by foreign objects; if S actual The preset maximum safe displacement S has been reached. max And F actual F has not yet been reached target If the system detects an abnormality in the seedling stem size or the presence of a foreign object obstructing the flow, an alarm signal will be issued and the clamping operation will be terminated to avoid damage to the clamping mechanism or the seedling; the maximum safe displacement S max =W0-d min d min The minimum statistical diameter of the stem of this species of seedling, W0 is the initial opening width; Case 2: Abnormalities such as excessively thin stems or missing seedlings; if F actual F has been reached target And S actual The expected lower limit of displacement range S was not reached. min If the seedling stem diameter is smaller than the expected value or the seedling is missing, a warning signal will be issued and the abnormal state will be recorded for subsequent sorting or manual handling; where the lower limit of the expected displacement range S min = W0-d max -ε,d max The maximum statistical diameter of the stem of this type of seedling is given by ε, where ε is the allowable error margin; Case 3: Normal; If S min ≤S actual ≤S max And F actual ≥F target If the clamping is normal, the seedling removal operation is completed.

[0034] Further, in step S1, color images and depth images of the seedling area are acquired by a depth camera. The U-Net semantic segmentation algorithm is used to extract the pixel mask of the stem area from the color image. The pixel coordinates of the stem area are converted into a three-dimensional spatial point cloud by combining the depth image and camera intrinsic parameters. The point cloud is clustered and fitted with a center line. The intersection of the fitted center line and the upper surface of the seedling substrate is used as the clamping target point, and its spatial coordinates are (X, Y, Z).

[0035] Furthermore, in step S2, (X, Y, Z) is transformed to the robot base coordinate system to obtain the target pose matrix T. target Read the current pose matrix T of the current clamping actuator. current ; Calculate the pose deviation ΔT=T between the two. target ×T current -1 The angular or linear displacement of the joint formed by the relative swing of the first clamping arm 2 and the second clamping arm 3 is solved by inverse kinematics and used as the displacement control quantity.

[0036] Further, in step S4, firstly, based on the displacement control amount calculated in step S2 and the planned smooth motion trajectory, a corresponding drive command is generated to drive the mounting base to move the entire clamping actuator smoothly along the planned path towards the target clamping position. During the entire movement, the control module reads the feedback values ​​of the joint encoder or displacement sensor in real time to form a position closed-loop control. When the clamping actuator approaches the target position, the control module further fine-tunes the posture of the mounting base based on the three-dimensional spatial coordinates (X, Y, Z) and tilt angle θ of the seedling stem previously obtained by the vision detection module, so that the center line of the V-shaped groove of the two flexible clamping fingers is precisely aligned with the axial direction of the seedling stem. At the same time, the position of the clamping actuator in the height direction is adjusted so that the clamping surface of the flexible clamping fingers is facing the predetermined clamping part, that is, the suitable stem segment located below the cotyledon node or above the grafting cut. Finally, the mounting base stops at the target clamping position.

[0037] The beneficial effects of this invention are as follows: This invention achieves precise detection and positioning of the seedling stem's spatial position, solving the problems of missed gripping and biased gripping. By accurately detecting the seedling stem's position coordinates (X, Y, Z) and tilt angle θ in three-dimensional space, and converting the target position into a precise displacement control quantity for the clamping actuator through inverse kinematics, precise closed-loop control of the clamping process is achieved by introducing displacement-force dual-stage collaborative control logic. Adaptive flexible clamping of the seedling stem is realized, significantly reducing the clamping damage rate. Flexible clamping fingers made of silicone material can form a wrapping contact with the seedling stem during clamping, ensuring uniform distribution of clamping force along the stem's circumference and avoiding stress concentration. The diameter of the seedling stem is obtained through a depth camera, and the target clamping force is calculated using a pre-stored seedling diameter-clamping force mapping model, overcoming the technical limitations of fixed clamping force in traditional clamping devices. A progressive clamping strategy and force / position hybrid control balance efficiency and safety. Attached Figure Description

[0038] Figure 1 This is a schematic diagram of the overall structure of the flexible clamping and seedling-taking device of the present invention;

[0039] Figure 2 This is a three-dimensional diagram of the flexible finger clamp of the present invention;

[0040] Figure 3 This is a diagram showing the distribution of the flexible finger clip's internal elastic deformation chamber and independent micro air passages in this invention.

[0041] Figure 4 This is a control system diagram of the present invention;

[0042] Figure 5 This is a flowchart of the control method of the present invention;

[0043] Figure 6 This is a schematic diagram of the displacement-force coordinated control curve of the present invention;

[0044] In the figure: 1 Mounting base; 2 First clamping arm; 3 Second clamping arm; 4 Rigid arm body; 5 Flexible clamping finger; 51 V-groove; 52 Elastic deformation chamber; 53 Independent micro air passage; 6 Micro thin film pressure sensor; 7 Depth camera; 8 Modular quick-change interface; 9 Displacement sensor. Detailed Implementation

[0045] like Figure 1 and Figure 2 As shown, the flexible clamping and seedling picking device of the seedling grafting robot of the present invention includes a mounting base 1, a clamping execution mechanism, a force sensing module, a displacement detection module, a vision detection module and a control module. The structure and working principle of the clamping and seedling picking device of the present invention will be described below with reference to the accompanying drawings.

[0046] like Figure 1As shown, the mounting base 1 is the basic component of this invention. The mounting base 1 is detachably connected to the end of the robotic arm of the seedling grafting robot via a modular quick-change interface 8. The modular quick-change interface 8 includes a gripper and a quick-change drive mechanism. The quick-change drive mechanism is located inside the mounting base 1, and the gripper is located at the output end of the quick-change drive mechanism. The quick-change drive mechanism drives the extension and retraction of the gripper. The end of the robotic arm is provided with an annular positioning groove. After the quick-change drive mechanism drives the gripper to extend into the annular positioning groove, the mounting base 1 and the end of the robotic arm are locked. After the quick-change drive mechanism drives the gripper to extend out of the annular positioning groove, the mounting base 1 is detached from the end of the robotic arm. The clamping actuator includes a first clamping arm 2 and a second clamping arm 3. The first clamping arm 2 and the second clamping arm 3 are symmetrically arranged on the mounting base 1, and their structures are identical. The first clamping arm 2 and the second clamping arm 3 are driven by the clamping arm drive mechanism to achieve opening and closing actions. Specifically, the first clamping arm 2 and the second clamping arm 3 are both hinged to the mounting base 1 via rotating shafts. A clamping arm drive mechanism drives the first clamping arm 2 and the second clamping arm 3 to rotate around their respective rotating shafts, thereby achieving symmetrical opening and closing movements. The clamping arm drive mechanism includes a servo motor and a reducer. The input end of the reducer is fixedly connected to the output shaft of the servo motor, and the output end of the reducer is fixedly connected to the rotating shafts of the first clamping arm 2 and the second clamping arm 3. The reducer incorporates a worm gear transmission mechanism or a gear transmission mechanism. When using a worm gear transmission mechanism, the output shaft of the servo motor is fixedly connected to the worm, and the worm meshes with the worm wheel. The output shaft of the servo motor drives the worm wheel to rotate through the worm. The worm wheel is coaxially and fixedly connected to the rotating shaft of the clamping arm, thus transmitting the rotational motion of the servo motor as the oscillation of the clamping arm around its rotating shaft. By controlling the rotation direction and angle of the servo motor, the opening angle and clamping displacement of the clamping arm can be precisely controlled. The encoder built into the servo motor provides real-time position information to the control module to achieve closed-loop position control. When a gear reducer is used, the gear transmission mechanism includes a driving gear and a driven gear. The output shaft of the servo motor is fixedly connected to the driving gear, and the driving gear meshes with the driven gear. The output shaft of the servo motor drives the driven gear to rotate through the driving gear in the gear transmission mechanism. The driven gear is coaxially and fixedly connected to the rotating shaft of the clamping arm, thereby transmitting the rotational motion of the servo motor as the oscillation of the clamping arm around the rotating shaft. Here, the clamping arm refers to the first clamping arm 2 and the second clamping arm 3. The first clamping arm 2 and the second clamping arm 3 can also be fixedly connected to a set of corresponding linear motion modules, and the clamping arm drive mechanism drives the clamping arm to move in opposite or opposite directions along the linear guide rail on the mounting base 1 to achieve symmetrical opening and closing motion.

[0047] Both the first clamping arm 2 and the second clamping arm 3 include a rigid arm body 4 and flexible clamping fingers 5. The upper end of the rigid arm body 4 is hinged to the mounting base 1 via a pivot, and the flexible clamping fingers 5 are located at the lower end of the rigid arm body 4. Specifically, a hollow sleeve connecting part is provided at the end of the rigid arm body 4. The flexible clamping fingers 5 are made of silicone material with a Shore hardness of 20A-50A by injection molding or 3D printing. During the molding process, the root of the flexible clamping fingers 5 directly covers and embeds into the sleeve connecting part. After molding, the flexible clamping fingers 5 and the rigid arm body 4 are integrated at the material level. The clamping surface of the flexible clamping fingers 5 is provided with a V-shaped groove 51 extending along the clamping direction. The included angle of the V-shaped groove 51 is 60°-120°, preferably 90°. The inner surface of the V-shaped groove 51 is provided with anti-slip texture 511 to increase the friction between the groove and the seedling stem and prevent the seedling from slipping during clamping. Figure 2 As shown, the flexible gripper 5 has multiple elastic deformation chambers 52 embedded inside, arranged along the gripping direction, such as... Figure 3 As shown, each elastic deformation chamber 52 is connected to an external air source through an independent micro-air passage 53. By controlling the inflation pressure of each elastic deformation chamber 52, the stiffness of the flexible clamping finger 5 and the contour of the clamping surface can be adaptively adjusted. When different pressures of gas are injected into each elastic deformation chamber 52, different areas of the flexible clamping finger 5 will undergo different degrees of expansion and deformation, thereby changing the curvature and contour of the clamping surface, making it fit the surface shape of the seedling stem more closely.

[0048] The force sensing module includes a miniature thin-film pressure sensor 6 disposed on the first clamping arm 2 and the second clamping arm 3. Specifically, the miniature thin-film pressure sensor 6 is disposed between the rigid arm body 4 and the flexible clamping finger 5, and is used to collect the contact pressure signal applied by the flexible clamping finger 5 to the seedling stem in real time during the clamping process. Specifically, the end face of the rigid arm body 4 facing the flexible clamping finger 5 has a shallow groove mounting position, and the miniature thin-film pressure sensor 6 is embedded in the mounting position, with its sensing surface closely fitting the root end face of the flexible clamping finger 5. The depth of the shallow groove mounting position is configured to be slightly less than the thickness of the miniature thin-film pressure sensor 6, so that when the flexible clamping finger 5 is fixedly connected to the rigid arm body 4 by fasteners, the root of the flexible clamping finger 5 presses the miniature thin-film pressure sensor 6 tightly into the shallow groove mounting position, realizing the stable installation of the miniature thin-film pressure sensor 6 and reliable pressure transmission.

[0049] The displacement detection module includes a displacement sensor 9 mounted on the first clamping arm 2 and the second clamping arm 3. The displacement sensor 9 is a high-precision grating ruler displacement sensor, with its reading head fixed to the rigid arm body 4 and the grating ruler fixed to the guide rail of the mounting base 1. It is used to detect the displacement of the clamping arms in the opening and closing direction in real time. The displacement sensor 9 can also be a linear variable differential transformer or a magnetostrictive displacement sensor. The preferred resolution of the displacement sensor 9 is 0.001 mm, and the preferred measurement range is 0 mm to 50 mm to meet the accurate detection requirements of the seedling stem diameter range.

[0050] The visual inspection module includes a depth camera 7 mounted on the mounting base 1, which is used to collect image information and depth information of the seedlings to be collected. The image information includes the stem diameter, stem tilt angle and leaf distribution of the seedlings, and the depth information is the position coordinates of the seedling stem in three-dimensional space.

[0051] like Figure 4 As shown, the control module is connected to the independent micro air circuit 53 of the clamping actuator, the clamping arm drive mechanism, the micro thin-film pressure sensor 6, the displacement sensor 9, and the depth camera 7. The control module includes a force / position hybrid control unit, which adopts a dual closed-loop control architecture: the outer loop is the position control loop, which controls the movement position of the first clamping arm 2 and the second clamping arm 3; the inner loop is the force control loop, which performs PID adjustment based on the real-time feedback signal from the micro thin-film pressure sensor 6, so that the actual clamping force F of the clamping actuator is controlled. actual Approaching the target clamping force F of the clamping actuator target During the position control phase, the force / position hybrid control unit operates in position control mode, using the displacement threshold S. threshold To control the target, the gripper arm is driven to move according to a preset position trajectory. When displacement sensor 9 detects a displacement S... actual ≥S threshold When the target clamping force F is applied, the system automatically switches to force control mode. target Force closed-loop regulation is performed to control the target. The PID regulation formula for the force control loop is: ΔF = K p ×(F target -F actual )+K i ×∫(F target -F actual )d t +K d ×d(F target -F actual ) / d t Among them, d t K represents the time step size for each sampling / control cycle. p K is the proportionality coefficient. i K is the integral coefficient. dThese are the differential coefficients, determined experimentally based on the system characteristics. Preferably, K... p =2.0, K i =0.5, K d =0.1. The control module is used to execute the control method of this invention, such as... Figure 5 As shown, the control method includes the following steps.

[0052] S1. Detection of the spatial position and posture of seedling stems.

[0053] First, the image and depth information of the seedlings to be collected are obtained through the visual detection module. After image processing and 3D reconstruction, the position coordinates (X, Y, Z) of the seedling stems in 3D space are extracted.

[0054] Specifically, color and depth images of the seedling area are acquired using depth camera 7. The U-Net semantic segmentation algorithm is used to extract the pixel mask of the stem area from the color image. The pixel coordinates of the stem area are converted into a three-dimensional point cloud by combining the depth image and camera intrinsic parameters. The point cloud is clustered and fitted with a center line. The intersection of the fitted center line and the upper surface of the seedling substrate is used as the clamping target point, and its spatial coordinates are (X, Y, Z).

[0055] Then, the image information of the seedlings to be collected is obtained through the visual detection module, and the stem diameter d and stem tilt angle θ are extracted. Specifically, based on the obtained pixel mask of the stem region, the average width of the stem region is calculated as the pixel value of the stem diameter d, and combined with the depth information to convert it into the actual physical size; by analyzing the direction vector of the stem centerline, the tilt angle θ of the stem relative to the vertical direction is calculated.

[0056] S2. Seedling retrieval path planning.

[0057] Based on the position coordinates (X, Y, Z) and the current spatial pose of the clamping actuator, calculate the displacement control amount required for the gripper to move from the initial position to the target clamping position.

[0058] The specific method is as follows: transform (X, Y, Z) to the robot's base coordinate system to obtain the target pose matrix T. target Read the current pose matrix T of the current clamping actuator. current ; Calculate the pose deviation ΔT=T between the two. target ×T current -1 The angular or linear displacement of each joint is calculated using inverse kinematics and used as displacement control quantities. Simultaneously, a smooth motion trajectory from the initial position to the target position is planned based on the target position and current attitude, ensuring that the clamping actuator can smoothly and accurately reach the target clamping position.

[0059] S3. Calculation of target clamping force.

[0060] Based on the stem diameter d obtained in step S1 and the preset seedling type information, the seedling diameter-clamping force mapping model stored in the control module is invoked to calculate the target clamping force F of the clamping actuator. target .

[0061] The method for constructing the seedling diameter-clamping force mapping model is as follows: For different types of seedlings, the transverse compressive yield strength F of their stems under different diameter conditions is measured using a universal testing machine. yield (d). Applying the target clamping force F target Set to F target =α×F yield (d). Wherein, α is the safety factor, 0<α<1, and the preferred value range of α is 0.3-0.6.

[0062] S4. Move to the target location.

[0063] According to the trajectory planned in step S2, the installation base 1 and the clamping actuator are moved to the target clamping position, so that the flexible clamping finger 5 is aligned with the part of the seedling stem to be clamped.

[0064] Specifically, firstly, based on the displacement control amount calculated in step S2 and the planned smooth motion trajectory, a corresponding drive command is generated to drive the mounting base 1 to move the entire clamping actuator smoothly along the planned path towards the target clamping position. During the entire movement, the control module reads the feedback values ​​of each joint encoder or displacement sensor 9 in real time, forming a position closed-loop control. When the clamping actuator approaches the target position, the control module further fine-tunes the posture of the mounting base 1 based on the three-dimensional spatial coordinates (X, Y, Z) and tilt angle θ of the seedling stem previously obtained by the vision detection module, so that the center line of the V-shaped groove of the two flexible clamping fingers 5 is precisely aligned with the axial direction of the seedling stem. At the same time, the position of the clamping actuator in the height direction is adjusted so that the clamping surface of the flexible clamping fingers 5 is facing the predetermined clamping part, that is, the suitable stem segment located below the cotyledon node or above the grafting incision. Finally, the mounting base 1 precisely stops at the target clamping position. At this point, only a tiny gap remains between the flexible clamping fingers 5 and the surface of the seedling stem, determined by the displacement threshold in subsequent steps. This prepares the system for entering the position control stage and ensures that the force and displacement coordination control during the subsequent clamping process is not affected by the initial pose deviation. The control module issues drive commands, including commands sent to the grafting robot and commands sent to the clamping arm drive mechanism. The commands sent to the grafting robot drive the robotic arm to move the mounting base 1, while the commands sent to the clamping arm drive mechanism drive the clamping arm to move.

[0065] S5. Flexible finger stiffness adaptive adjustment.

[0066] The target clamping force F calculated in step S3 target Based on the stem inclination angle θ, the target air pressure value P of each elastic deformation chamber is calculated. i Where i=1,2,...,n, and air is supplied to each elastic deformation chamber through independent micro air channels, so that the flexible clamping finger 5 achieves the stiffness and contour shape that matches the seedling stem.

[0067] Specifically, when the stem is thin, the air pressure in each elastic deformation chamber is reduced, making the flexible gripper 5 softer and more deformable, thus gently wrapping the thin stem. When the stem is thick, the air pressure in each elastic deformation chamber is increased, increasing the stiffness of the flexible gripper 5 and providing sufficient clamping force to prevent the seedling from slipping off. Furthermore, based on the stem's tilt angle θ, asymmetrical air pressure is applied to each elastic deformation chamber; higher air pressure is applied to one side of the tilt direction to increase support stiffness, while lower air pressure is applied to the opposite side to provide greater adaptive deformation, thereby allowing the flexible gripper 5 to better conform to the surface of the tilted stem.

[0068] S6. Gripping and grasping operation.

[0069] In this step, the displacement-force dual-stage coordinated control specifically includes the following sub-steps.

[0070] S6.1. Position control stage - pre-clamping.

[0071] The first clamping arm 2 and the second clamping arm 3 are driven to close towards the seedling stem at a first speed of 50mm / s-100mm / s, while the displacement S of the clamping arms collected by the displacement sensor 9 is read in real time. actual When S actual Reaching the preset displacement threshold S threshold At that time, the position control phase stops.

[0072] Displacement threshold S threshold S is determined based on the seedling stem diameter d and the initial opening width W0 of the flexible clamping finger 5. threshold =W0-d-δ. Where δ is the reserved gap, and the value of δ ranges from 0.5mm to 1.5mm. The setting of δ ensures that at the end of the position control stage, the flexible clamping finger 5 is just in contact with the surface of the seedling stem but has not yet applied a significant clamping force, thereby achieving a smooth switch from position control to force control and avoiding force impact.

[0073] S6.2. Force control stage - precision clamping.

[0074] The first clamping arm 2 and the second clamping arm 3 are driven to close slowly at a second speed of 5 mm / s-20 mm / s, while the actual clamping force value F collected by the micro-film pressure sensor 6 is read in real time. actual ; Factual Clamping force F with the target target Compare; when F actual <F target When F continues to increase the clamping stroke; actual ≥F target When the clamping action stops and the current clamping force is maintained, the clamping operation is completed.

[0075] S6.3. Displacement-force coordination judgment and anomaly handling.

[0076] During the clamping process, the displacement S collected by the displacement sensor is monitored simultaneously. actual The force value F collected by the miniature thin-film pressure sensor 6 actual Execute the following collaborative judgment logic.

[0077] Situation 1: Abnormalities such as excessively thick stems or obstruction by foreign objects; such as Figure 6 As shown, if S actual The preset maximum safe displacement S has been reached. max And F actual F has not yet been reached target If the system detects an abnormality in the seedling stem size or the presence of a foreign object obstructing the flow, an alarm signal will be issued and the clamping operation will be terminated to avoid damage to the clamping mechanism or the seedling. The maximum safe displacement S... max = W0-d min d min This is the minimum statistical diameter of the stem of this type of seedling.

[0078] Scenario 2: Abnormalities such as excessively thin stems or missing seedlings; if F actual F has been reached target And S actual The expected lower limit of displacement range S was not reached. min If the seedling stem diameter is smaller than the expected value or the seedling is missing, a warning signal will be issued and the abnormal state will be recorded for subsequent sorting or manual handling. The lower limit of the expected displacement range, S... min = W0-d max -ε,d max The maximum statistical diameter of the stem of this type of seedling is ε, which is the allowable error margin, ranging from 0.5 mm to 2.0 mm.

[0079] Scenario 3: Normal; if S min ≤S actual ≤S max And F actual ≥F target If the clamping is successful, the seedling removal operation is complete. control This represents the control displacement at time t1.

[0080] Compared with existing technologies, this invention achieves precise detection and positioning of the seedling stem's spatial position, solving the problems of missed grasping and misalignment. By combining a depth camera with semantic segmentation algorithms and 3D point cloud processing technology, it can accurately detect the seedling stem's position coordinates (X, Y, Z) and tilt angle θ in 3D space. Furthermore, through inverse kinematics solutions, the target position is converted into precise displacement control values ​​for the clamping actuator. Compared to the fixed-position seedling retrieval method in existing technologies, this invention can adapt to the actual growth position and posture of each seedling, ensuring that the gripper is always accurately aligned with the part of the stem to be clamped, significantly improving the success rate of seedling retrieval and operational reliability.

[0081] Compared with existing technologies, this invention achieves precise closed-loop control of the clamping process by introducing a displacement-force dual-stage collaborative control logic. A displacement sensor is integrated into the clamping actuator, constructing a dual-stage control strategy of position control pre-clamping → force control fine clamping. In the position control stage, a displacement threshold is used as the control target, enabling the grippers to close quickly and accurately to the position where they just make contact with the stem surface, avoiding the blind clamping problem caused by the lack of displacement feedback in traditional solutions. In the force control stage, the target clamping force is used as the control target, achieving precise force-controlled clamping. More importantly, a displacement-force collaborative judgment mechanism is established. By simultaneously monitoring displacement and force values, it can effectively identify abnormal working conditions such as abnormal stem size, missing seedlings, and foreign object obstruction, and take corresponding actions, significantly improving the robustness and safety of the system.

[0082] Compared with existing technologies, this invention achieves adaptive flexible clamping of seedling stems, significantly reducing the clamping damage rate. The flexible clamping fingers 5, made of silicone material, have a V-shaped groove structure on their clamping surface, enabling a wrapping contact with the seedling stem during clamping, ensuring uniform distribution of clamping force along the circumference of the stem and avoiding stress concentration. More importantly, multiple elastic deformation chambers embedded within the flexible clamping fingers 5 can achieve adaptive adjustment of stiffness and contour through independent air pressure control. For seedlings with thinner stems, reducing the chamber air pressure makes the flexible clamping fingers 5 softer and more deformable, achieving a gentle wrapping of the thin stem; for seedlings with thicker stems, increasing the chamber air pressure increases the stiffness of the flexible clamping fingers 5, providing sufficient clamping force. Compared to existing technologies that only use passive flexible materials on the surface of the clamping claws, the active stiffness adjustment mechanism of this invention can optimize clamping parameters in real time according to the diameter of the seedling stem, resulting in greater adaptability and a lower damage rate.

[0083] Compared with existing technologies, this invention achieves precise force control tailored to each seedling by using a visual detection-based seedling diameter-clamping force mapping model. By acquiring the seedling stem diameter using a depth camera and combining it with a pre-stored seedling diameter-clamping force mapping model to calculate the target clamping force, it overcomes the technical limitations of traditional clamping devices that use a fixed clamping force. This mapping model is constructed based on transverse compressive yield strength test data for different types of seedlings, ensuring that the clamping force is always below the critical value for seedling damage. Through real-time closed-loop feedback from a force sensor, the clamping force control accuracy can reach ±0.1N, achieving true soft-contact clamping.

[0084] Compared with existing technologies, this invention employs a progressive clamping strategy and force / position hybrid control, balancing efficiency and safety. The invention utilizes a three-stage progressive clamping strategy: rapid approach → slow contact → force-controlled clamping. This ensures efficiency in seedling removal while avoiding impact damage through slow contact. The force / position hybrid control architecture enables seamless switching between position and force control, allowing for rapid positioning before contact and precise force control after contact, thus balancing operational efficiency and clamping safety.

[0085] Compared with existing technologies, this invention features a modular quick-change design, which enhances the versatility and applicability of the device. Through the modular quick-change interface, the gripping head module with corresponding parameters can be quickly replaced according to the morphology and mechanical properties of different types of seedlings, achieving multi-functionality and significantly improving the applicability and operational flexibility of the seedling grafting robot.

Claims

1. A flexible clamping and seedling-grafting device for a seedling grafting robot, characterized in that, The system includes a mounting base, a clamping actuator, a force sensing module, a displacement detection module, a vision detection module, and a control module. The mounting base is detachably connected to the end effector of the seedling grafting robot's robotic arm via a modular quick-change interface. The clamping actuator includes a first clamping arm and a second clamping arm symmetrically arranged and movably connected to the mounting base. The mounting base has a clamping arm drive mechanism for driving the opening and closing movements of the first and second clamping arms. The force sensing module includes miniature thin-film pressure sensors mounted on the first and second clamping arms for real-time data acquisition during the clamping process. The system includes a contact pressure signal applied to the seedling stem; a displacement detection module comprising displacement sensors mounted on the first and second clamping arms for real-time detection of the displacement of the first and second clamping arms in the opening and closing direction; a vision detection module comprising a depth camera mounted above the mounting base for acquiring image and depth information of the seedlings to be harvested; a control module connected to the clamping arm drive mechanism, a micro-film pressure sensor, a displacement sensor, and a depth camera; detection signals from the micro-film pressure sensor, displacement sensor, and depth camera are uploaded to the control module; and the control module sends control commands to the clamping arm drive mechanism.

2. The flexible clamping and seedling-retrieving device for seedling grafting robots according to claim 1, characterized in that, The modular quick-change interface includes a claw and a quick-change drive mechanism. The quick-change drive mechanism is located inside the mounting base, and the claw is located at the output end of the quick-change drive mechanism. The quick-change drive mechanism drives the extension and retraction of the claw. The end of the robotic arm is provided with an annular positioning groove. After the quick-change drive mechanism drives the claw to extend into the annular positioning groove, the mounting base and the end of the robotic arm are locked. After the quick-change drive mechanism drives the claw to extend out of the annular positioning groove, the mounting base and the end of the robotic arm are disassembled.

3. The flexible clamping and seedling-retrieving device for seedling grafting robots according to claim 1, characterized in that, The clamping arm drive mechanism includes a servo motor and a reducer. The input end of the reducer is fixedly connected to the output shaft of the servo motor, and the output end of the reducer is fixedly connected to the rotating shaft of the first clamping arm and the second clamping arm.

4. The flexible clamping and seedling-retrieving device for seedling grafting robots according to claim 3, characterized in that, The reducer has a built-in worm gear transmission mechanism, which includes a meshing worm and a worm wheel. The worm is fixedly connected to the output shaft of the servo motor, and the worm wheel is fixedly connected to the rotating shaft of the first clamping arm and the second clamping arm. Alternatively, the reducer has a built-in gear transmission mechanism, which includes a meshing driving gear and a driven gear. The driving gear is fixedly connected to the output shaft of the servo motor, and the driven gear is fixedly connected to the rotating shaft of the first clamping arm and the second clamping arm.

5. The flexible clamping and seedling-retrieving device for seedling grafting robots according to claim 1, characterized in that, Both the first clamping arm and the second clamping arm include a rigid arm body and flexible clamping fingers. The upper end of the rigid arm body is hinged to the mounting base via a pivot, and the flexible clamping fingers are disposed at the lower end of the rigid arm body.

6. The flexible clamping and seedling-retrieving device for seedling grafting robots according to claim 5, characterized in that, A hollow sleeve connecting part is provided at the lower end of the rigid arm body. The flexible clamping finger is made of silicone material by injection molding or 3D printing. During the molding process, the root of the flexible clamping finger is directly covered and embedded in the sleeve connecting part. The end face of the rigid arm body facing the flexible clamping finger is provided with a shallow groove mounting position. The micro-film pressure sensor is embedded in the shallow groove mounting position, and the sensing surface of the micro-film pressure sensor is closely attached to the root end face of the flexible clamping finger.

7. The flexible clamping and seedling-retrieving device for seedling grafting robots according to claim 6, characterized in that, The flexible clamping finger has a V-shaped groove extending along the clamping direction on its clamping surface. The included angle of the V-shaped groove is 60°-120°, and the inner surface of the V-shaped groove is provided with anti-slip texture. The flexible clamping finger has multiple elastic deformation chambers arranged along the clamping direction embedded inside. Each elastic deformation chamber is connected to an external air source through an independent micro air passage. When gas at different pressures is filled into each elastic deformation chamber, different areas of the flexible clamping finger will expand and deform to different degrees.

8. The flexible clamping and seedling-retrieving device for seedling grafting robots according to claim 7, characterized in that, The displacement detection module includes displacement sensors mounted on the first and second clamping arms. The displacement sensors are high-precision grating ruler displacement sensors, with their reading heads fixed to the rigid arm body and the grating ruler fixed to the guide rail of the mounting base, for real-time detection of the displacement of the clamping arms in the opening and closing direction.

9. The control method for the flexible clamping and seedling-retrieving device of the seedling grafting robot according to claim 8, characterized in that, Includes the following steps: S1. Detection of the spatial position and posture of seedling stems; First, the image and depth information of the seedlings to be collected are obtained through the visual detection module. After image processing and 3D reconstruction, the position coordinates (X, Y, Z) of the seedling stem in 3D space are extracted; the diameter d of the seedling stem and the inclination angle θ of the stem are extracted. S2. Seedling retrieval path planning; Based on the position coordinates (X, Y, Z) and the current spatial pose of the gripping actuator, calculate the displacement control amount required for the gripper to move from the initial position to the target gripping position; plan a smooth motion trajectory from the initial position to the target position based on the target position and the current pose. S3. Calculation of target clamping force; Based on the stem diameter d obtained in step S1 and the preset seedling type information, the seedling diameter-clamping force mapping model stored in the control module is invoked to calculate the target clamping force F of the clamping actuator. target ; S4. Move to the target location; According to the trajectory planned in step S2, the installation base and clamping actuator are moved to the target clamping position, so that the flexible clamping fingers are aligned with the part of the seedling stem to be clamped. S5. Flexible finger stiffness adaptive adjustment; The target clamping force F calculated in step S3 target Based on the stem inclination angle θ, the target air pressure value P of each elastic deformation chamber is calculated. i , i=1,2,...,n, and inflate each elastic deformation chamber with air so that the flexible clamping fingers achieve the stiffness and contour shape that match the seedling stem; S6. Clamping and grasping operation; S6.

1. Position control stage - pre-clamping; The first and second clamping arms are driven to close towards the seedling stem at a first speed, while the displacement S of the clamping arms collected by the displacement sensor is read in real time. actual When S actual Reaching the preset displacement threshold S threshold At this time, the position control phase stops; S6.

2. Force Control Stage - Precision Clamping; The first and second clamping arms are driven to close slowly at a second speed, which is less than the first speed, while the actual clamping force value F collected by the micro-film pressure sensor is read in real time. actual ; F actual Clamping force F with the target target Compare; when F actual <F target When F continues to increase the clamping stroke; actual ≥F target When the clamping action stops and the current clamping force is maintained, the clamping operation is completed. S6.

3. Displacement-force coordination judgment and anomaly handling; During the clamping process, the displacement S collected by the displacement sensor is monitored simultaneously. actual The force value F collected by the miniature thin-film pressure sensor actual The following collaborative judgment logic is executed: Case 1: Abnormalities such as excessively thick stems or obstruction by foreign objects; if S actual The preset maximum safe displacement S has been reached. max And F actual F has not yet been reached target If the seedling stem size is abnormal or there is a foreign object obstructing the flow, an alarm signal will be issued and the clamping operation will be terminated; Situation 2: Abnormalities such as excessively thin stems or missing seedlings; If F actual F has been reached target And S actual The expected lower limit of displacement range S was not reached. min If the seedling stem diameter is smaller than expected or the seedling is missing, a warning signal is issued and the abnormal state is recorded for subsequent sorting or manual processing; Scenario 3: Normal; If S min ≤S actual ≤S max And F actual ≥F target If the clamping is normal, the seedling removal operation is completed.

10. The flexible clamping and seedling-retrieving device for seedling grafting robots according to claim 9, characterized in that, In step S1, color and depth images of the seedling area are acquired using a depth camera. The U-Net semantic segmentation algorithm is used to extract the pixel mask of the stem area from the color image. The pixel coordinates of the stem area are converted into a three-dimensional point cloud by combining the depth image and camera intrinsic parameters. The point cloud is clustered and fitted with a center line. The intersection of the fitted center line and the upper surface of the seedling substrate is used as the clamping target point, and its spatial coordinates are (X, Y, Z).

11. The flexible clamping and seedling-retrieving device for seedling grafting robots according to claim 10, characterized in that, In step S2, (X, Y, Z) is transformed to the robot base coordinate system to obtain the target pose matrix T. target Read the current pose matrix T of the current clamping actuator. current ; Calculate the pose deviation ΔT=T between the two. target ×T current -1 The angular or linear displacement of the joint formed by the relative swing of the first and second clamping arms is solved by inverse kinematics and used as the displacement control quantity.

12. The flexible clamping and seedling-retrieving device for seedling grafting robots according to claim 11, characterized in that, In step S4, firstly, based on the displacement control amount calculated in step S2 and the planned smooth motion trajectory, a corresponding drive command is generated to drive the mounting base to move the entire clamping actuator smoothly along the planned path towards the target clamping position. During the entire movement, the control module reads the feedback values ​​from the joint encoder or displacement sensor in real time to form a position closed-loop control. When the clamping actuator approaches the target position, the control module further fine-tunes the posture of the mounting base based on the three-dimensional spatial coordinates (X, Y, Z) and tilt angle θ of the seedling stem previously obtained by the vision detection module, so that the center line of the V-shaped groove of the two flexible clamping fingers is precisely aligned with the axial direction of the seedling stem. At the same time, the position of the clamping actuator in the height direction is adjusted so that the clamping surface of the flexible clamping fingers is facing the predetermined clamping part, that is, the suitable stem segment located below the cotyledon node or above the grafting incision. Finally, the mounting base stops at the target clamping position.

Images