High-throughput laser slicing equipment and method for rice breeding

Abstract

The invention discloses high-throughput laser slicing equipment and a high-throughput laser slicing method for rice breeding. An axial rotation driving motor and a tail end negative pressure suction nozzle are installed at the tail end of the mechanical arm. The supply module comprises a seed disc and a first visual camera, the seed disc is arranged in the working range of the mechanical arm, and the first visual camera is used for recognizing the poses of rice seeds in the seed disc; the vibration type rice seed-metering device is used for supplementing rice seeds to the seed tray; the clamping and collecting module comprises a seed clamping position, a parent body collecting pipe and a sample collecting pipe, the seed clamping position and the parent body collecting pipe are both arranged in the working range of the mechanical arm, and an inlet of the sample collecting pipe is located in one side below the seed clamping position; the laser is used for cutting the rice seeds on the seed clamping position by using laser beams; the second visual camera is used for adjusting the laser cutting path of the laser. The working quality and efficiency of equipment are effectively improved, and the situations of invalid sampling, low-quality sampling and the like are avoided.

Description

TECHNICAL FIELD

[0001] The present application belongs to the field of agricultural engineering, and particularly relates to a high-throughput laser sectioning device and method for rice breeding. BACKGROUND

[0002] With the development of biotechnology, people gradually realize that genes are directly related to plant traits. As long as the genotype of a biological body is sequenced, the traits it exhibits can be calculated. Therefore, molecular breeding technology has been booming.

[0003] In the process of molecular breeding, the genotype of a large number of seeds needs to be detected, and then seeds meeting the breeding target are selected for breeding. At present, there are a small number of automatic rice breeding sectioning devices in China, which generally use blade cutting. However, there are some problems such as difficulty in cutting some materials (such as corn seeds), unclear cutting of shelled seed samples (such as rice seeds), seed infection and material genotype pollution caused by contact cutting of the blade, etc. Therefore, the present application provides a rice seed automatic laser non-contact sectioning device, which can effectively improve the breeding process and promote the revitalization of the seed industry. SUMMARY

[0004] The present application aims to solve the defects and deficiencies in the prior art, and provides a high-throughput laser sectioning device and method for rice breeding. The high-throughput laser sectioning device has the advantages of high throughput, high precision, micro-damage, and no pollution, and can realize non-contact automatic sampling operation of rice seeds.

[0005] The technical scheme adopted by the present application is as follows: One, a high-throughput laser sectioning device for rice breeding comprises: a mechanical arm, the end of which is provided with an end negative pressure suction nozzle, the suction nozzle of which can rotate around a vertical shaft; a supply module comprising a seed disc and a first vision camera, the seed disc being arranged within the working range of the mechanical arm, and the first vision camera being used to collect images of the rice seeds in the seed disc; a vibrating rice seed metering device for supplementing rice seeds to the seed disc; a clamping and collecting module comprising a seed clamping position, a parent collecting tube and a sample collecting tube, the seed clamping position and the parent collecting tube being arranged within the working range of the mechanical arm, and the inlet of the sample collecting tube being located on one side below the seed clamping position; a laser for cutting rice seeds in the seed clamping position with a laser beam; a second vision camera for collecting images of the rice seeds in the seed clamping position.

[0006] Specifically, the end suction hole of the suction nozzle is provided with a profiled ellipsoidal groove which is adapted to the profile curvature of the rice seeds.

[0007] Specifically, the feeding module further comprises a seed slide and a light supplementing lamp strip; the bottom of the seed disc is made of a transparent glass plane, and the side wall is made of an inclined curved surface; the seed slide is arranged above the seed disc, and the upper and lower ends of the seed slide are connected with the seed feeding end of the vibrating rice seed metering device and the inclined curved surface respectively; the first visual camera is arranged below the transparent glass plane, the lens of the first visual camera is arranged upward, and the light supplementing lamp strip is arranged between the lens and the transparent glass plane.

[0008] Specifically, the clamping and collecting module further comprises an outer clamping plate, an inner clamping plate, a pressure sensor and a driving motor; the outer clamping plate and the inner clamping plate are oppositely arranged on the two sides of the seed clamping position, the outer clamping plate is fixed, the inner clamping plate is drivingly connected with the driving motor, and can reciprocally move along the direction of approaching and moving away from the outer clamping plate under the driving of the driving motor, and the pressure sensor is arranged on the opposite side surface of the inner clamping plate and / or the outer clamping plate.

[0009] Specifically, the mechanical arm is a Cartesian type mechanical arm.

[0010] Further, the high-throughput sampling and slicing device further comprises a robot multi-node operating system. The robot multi-node operating system realizes control and communication through electrical connection or communication connection with the mechanical arm, the axial rotation driving motor, the first visual camera, the vibrating rice seed metering device, the laser, the second visual camera, the driving motor and the pressure sensor.

[0011] II. A high-throughput laser slicing method using the high-throughput laser slicing device comprising the following steps: S1, delivering rice seeds to the seed disc through the vibrating rice seed metering device.

[0012] S2, collecting the image of the rice seeds in the seed disc by using the first visual camera, and identifying the pose of the rice seeds in the seed disc according to the image.

[0013] Specifically, the step S2 is: collecting the image of the rice seeds in the seed disc by using the first visual camera, processing the image by using a key point algorithm to obtain the parent box of the rice seeds and two key points, the two key points representing the embryo side and the endosperm side respectively; calculating the midpoint coordinates of the two key points as the suction point coordinates, and calculating the included angle between the vector formed by the two key points and the reference direction in the image coordinate system as the attitude angle, the suction point coordinates and the attitude angle forming the pose of the rice seeds in the seed disc.

[0014] S3. Based on the position of the rice seeds, control the robotic arm to move the end negative pressure suction nozzle to the rice seeds, and use the end negative pressure suction nozzle to pick up the rice seeds. After picking up the rice seeds, move the picked-up rice seeds to the seed posture secondary calibration point on the seed tray.

[0015] Step S3 is as follows: After performing inverse kinematics calculation based on the current coordinate values ​​of the end of the robotic arm and the coordinate values ​​of the suction point, interpolation is performed to obtain the predicted motion trajectory. The robotic arm is controlled to move to the target position according to the predicted motion trajectory, so that the suction port at the end of the suction nozzle is aligned with the suction point in the horizontal plane and in contact with the upper surface of the rice seed in the vertical direction. The negative pressure suction nozzle at the end is used to suck up the rice seeds on the seed tray. According to the coordinates of the pre-selected seed posture secondary calibration point on the seed tray, the robotic arm is controlled to move the sucked rice seeds to the seed posture secondary calibration point.

[0016] S4. Use the first vision camera to acquire images of rice seeds at the secondary calibration point of seed posture, identify the current posture of rice seeds at the secondary calibration point of seed posture based on the images, and combine the current posture and the preset posture to perform feedback control on the axial rotation drive motor, and rotate the rice seeds to the preset posture through the axial rotation drive motor.

[0017] S5. Using a robotic arm, the picked-up rice seeds are moved from the seed posture secondary calibration point to above the preset clamping point of the seed clamping position. The second vision camera is used to collect images. After identifying the total length of the rice seeds based on the images, the robotic arm moves the picked-up rice seeds to the preset clamping point.

[0018] S6. After releasing the rice seeds using the end negative pressure nozzle, clamp the rice seeds tightly using the inner and outer clamps.

[0019] S7. Use a second vision camera to acquire images, identify the exposed length of the rice seed endosperm side based on the images, calculate the target cutting length based on the total length, determine the cutting path based on the target cutting length and the exposed length of the endosperm side, and use a laser beam to cut the rice seed according to the cutting path to obtain the seed mother and the cut sample. The cut sample falls into the sample collection tube, and a robotic arm moves the seed mother to the entrance of the mother collection tube so that the seed mother falls into the mother collection tube.

[0020] In step S7, determining the cutting path based on the exposed length and the target cutting length means: determining the cutting point based on the exposed length and the target cutting length, and then generating the cutting path corresponding to the cutting point; the cutting point is located at one-third of the total length from the seed endosperm side.

[0021] S8. Repeat steps S1 to S7 until the job is finished.

[0022] Specifically, in steps S2 to S7, a key point algorithm is used to perform image recognition processing, and the key point algorithm is the Yolov11-pose algorithm.

[0023] The beneficial effects of this invention are as follows: 1. Compared with existing blade cutting technology, the rice breeding high-throughput slicing equipment of this invention uses laser non-contact cutting of samples, which has higher precision, smoother cutting surface, and will not cause gene contamination or disease transmission between seeds. 2. Compared with existing technologies, the secondary calibration point scheme of the high-throughput rice slicing equipment and control method of this invention effectively avoids the seed posture deviation problem caused by the instantaneous negative pressure when the nozzle picks up seeds, thus improving the robustness of the system. Furthermore, the use of fixed-coordinate secondary calibration points effectively reduces data copying between nodes of the robot's multi-node operating system (ROS), improving the system response speed.

[0024] 3. Compared with the prior art, the high-throughput slicing equipment and control method for rice breeding of the present invention compares the length of the seed before clamping with the length of the exposed clamping point after clamping, and dynamically adjusts the laser trajectory through visual feedback, thereby improving the uniformity of the cut sample. Attached Figure Description

[0025] Figure 1 A schematic diagram of the structure of the high-throughput slicing device for rice breeding provided by the present invention; Figure 2 A schematic diagram of the terminal negative pressure suction nozzle in the high-throughput rice breeding slicing device provided by the present invention; Figure 3 A schematic diagram of the supply module in the high-throughput slicing device for rice breeding provided by the present invention; (a) is a side view and a cross-sectional view, and (b) is a top view of the seed tray; Figure 4 The diagram shows the structure of the clamping and collecting module in the high-throughput slicing device for rice breeding provided by the present invention; (a) is an overall schematic diagram, and (b) is a partial schematic diagram around the seed clamping position. Figure 5 The diagram shows the structure of the laser in the high-throughput slicing device for rice breeding provided by the present invention; (a) is an overall schematic diagram, and (b) is a partial schematic diagram of the galvanometer shell and the galvanometer part. Figure 6 The control flowchart of the high-throughput slicing method for rice breeding provided by the present invention is shown.

[0026] The components include: 1. Robotic arm; 2. End-effector negative pressure nozzle; 21. Negative pressure channel; 22. Axial rotation drive motor; 23. Motor housing; 24. Nozzle air inlet; 25. End-effector air intake; 26. Nozzle; 3. Seed tray; 31. Seed slide; 32. Inclined curved surface; 322. Rice seeds; 33. Stand connector; 34. Supplemental lighting strip; 35. Transparent glass plane; 36. Lens; 37. Camera lens housing; 38. First-person vision camera; 3220. Attitude secondary calibration point; 4. Vibrating rice seed metering device. 5. Stand; 6. Clamping and collecting module; 61. Outer clamping plate; 62. Sample collecting tube; 62. Seed clamping position; 63. Inner clamping plate; 64. Pressure sensor; 65. Connecting plate; 66. Lead screw sleeve plate; 67. Mother collection tube; 68. Lead screw; 69. Drive motor; 610. Manual adjuster; 7. Second vision camera; 8. Laser; 81. Galvanometer housing and galvanometer; 82. Laser beam outlet; 821. Laser beam; 83. Fixing groove; 84. Heat dissipation hole; 85. Laser tube and galvanometer connector. Detailed Implementation

[0027] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0028] This invention provides a high-throughput laser slicing device for rice breeding.

[0029] like Figure 1 As shown, the high-throughput laser slicing device of the present invention includes: The robotic arm 1 has a negative pressure suction nozzle 2 installed at its end. The suction nozzle 26 of the negative pressure suction nozzle 2 can rotate 360 ​​degrees around the vertical axis. The supply module includes a seed tray 3 and a first vision camera 38. The seed tray 3 is arranged within the working range of the robotic arm 1, and the first vision camera 38 is used to acquire images of rice seeds 322 in the seed tray 3. Vibrating rice seed metering device 4 is used to replenish single rice seeds 322 to seed tray 3; The clamping and collecting module 6 includes a seed clamping position 622, a mother collection tube 67, and a sample collection tube 62. The seed clamping position 622 and the mother collection tube 67 are both arranged within the working range of the robotic arm 1. The inlet of the sample collection tube 62 is located on one side below the seed clamping position 622. Laser 8 is used to cut rice seeds 322 fixed on seed clamping position 622 using a laser beam; The second vision camera 7, with its lens facing the seed clamping position 622, is used to acquire images of the rice seeds 322 on the seed clamping position 622.

[0030] Preferably, such as Figure 1 As shown, robotic arm 1 is a Cartesian type robotic arm.

[0031] Optionally, the second vision camera 7 can be fixedly mounted on the stand 5 using a bracket or other structure.

[0032] Optionally, the second vision camera 7 is arranged around the seed clamping position 622, preferably on the opposite side of the sample collection tube 62.

[0033] Preferably, the parent collection tube 67 is arranged inside the sample collection tube 62, that is, on the side close to the seed disk 3.

[0034] like Figure 2 As shown, the end negative pressure suction nozzle 2 includes an axial rotation drive motor 22 and a suction nozzle 26. The motor housing 23 of the axial rotation drive motor 22 is installed at the end of the robotic arm 1, and the output shaft of the axial rotation drive motor 22 is connected to the suction nozzle 26. The axial rotation drive motor 22 can drive the suction nozzle 26 to rotate 360 ​​degrees around the vertical axis, so that the rice seed 322 can be placed at the preset clamping point on the seed clamping position 622 in a preset posture. After the rice seed 322 at the preset clamping point is clamped and fixed, the endosperm segment of the rice seed 322 can be placed outside to one side of the seed clamping position 622, so that the cut sample obtained after laser 7 cutting can fall directly into the sample collection tube 62. A negative pressure channel 21 is provided inside the axial rotation drive motor 22, and the upper and lower ends of the suction nozzle 26 are respectively provided with interconnected suction nozzle air inlets 24 and end suction holes 25. The inlet of the negative pressure channel 21 is connected to an external air pump, and the outlet is connected to the suction nozzle air inlet 24 in sequence.

[0035] Preferably, the suction port 25 at the end of the suction nozzle 26 is provided with a contoured ellipsoidal groove, the contour curvature of which is adapted to the contour curvature of the rice seed 322.

[0036] Specifically, the center of the contoured ellipsoidal groove is located below the air intake hole at the end 25 of the nozzle.

[0037] like Figure 3 As shown, the supply module also includes a seed chute 31 and a supplementary lighting strip 34. (As shown...) Figure 3 As shown in (b), the bottom of the seed disk 3 is made of a transparent glass plane 35, and the sidewalls are made of annular inclined curved surfaces 32. Figure 3As shown in (a), a seed chute 31 is arranged above the seed tray 3. The upper and lower ends of the seed chute 31 are connected to the seed delivery end (i.e., seed outlet) of the vibrating rice seed metering device 4 and the inclined curved surface 32, respectively. The vibrating rice seed metering device 4 replenishes single rice seeds 322 to the seed tray 3 through the seed chute 31. A first vision camera 38 is arranged below the transparent glass plane 35. The lens 36 of the first vision camera 38 faces upward and is directly opposite the transparent glass plane 35. A supplementary lighting strip 34 is arranged between the lens 36 and the transparent glass plane 35.

[0038] Furthermore, such as Figure 3 As shown in (a), the supply module also includes a platform connector 33. The platform connector 33 has a through hole, an annular inclined curved surface 32 is arranged above the through hole, and a transparent glass plane 35 is arranged below it. The inclined curved surface 32 is connected to the upper surface of the platform connector 33, and the transparent glass plane 35 is connected to the lower surface of the platform connector 33. The inclined curved surface 32 and the transparent glass plane 35 form a seed disk 3. The camera lens housing 37 of the first vision camera 38 is arranged below the platform connector 33 and connected to the bottom of the platform connector 33. The platform connector 33 is mounted on the table surface of the platform 5.

[0039] like Figure 4 As shown, the clamping and collecting module 6 also includes an outer clamping plate 61, an inner clamping plate 63, a pressure sensor 64, and a drive motor 69. The outer clamping plate 61 and the inner clamping plate 63 are arranged opposite each other on both sides of the seed clamping position 622. The outer clamping plate 61 is fixed, and the inner clamping plate 63 is connected to the drive motor 69 and can reciprocate in the direction of approaching and moving away from the outer clamping plate 61 under the drive of the drive motor 69. The pressure sensor 64 is provided on the opposing side of the inner clamping plate 63 and / or the outer clamping plate 61.

[0040] Furthermore, such as Figure 4 As shown in (b), the seed clamping position 622 is connected to a sample slide on one side of the sample collection tube 62. The upper end of the sample slide is connected to the seed clamping position 622, and the lower end is connected to or communicates with the inlet of the sample collection tube 62, so that the cut sample can pass through the sample slide and fall into the sample collection tube 62 by free fall.

[0041] Specifically, the outer clamp 61 can be fixed by directly connecting it to the table surface of the frame 5, or by connecting it to the table surface of the frame 5 through a connector.

[0042] Preferably, the inner clamping plate 63 and the drive motor 69 are connected by a lead screw transmission assembly.

[0043] Preferably, the lead screw drive assembly includes a connecting plate 65, a lead screw sleeve 66, and a lead screw 68. The lead screw 68 is connected to the output shaft of the drive motor 69. The lead screw 68 is threaded into the center hole of the lead screw sleeve 66. The lead screw sleeve 66 is connected to the inner clamping plate 63 through the connecting plate 65.

[0044] Furthermore, the high-throughput sampling and slicing equipment also includes a robot multi-node operating system. The robot multi-node operating system achieves control and communication through electrical or communication connections with the robotic arm 1, axial rotation drive motor 22, first vision camera 38, vibrating rice seed metering device 4, laser 8, second vision camera 7, drive motor 69, and pressure sensor 64.

[0045] Furthermore, the robot's multi-node operating system is equipped with a pre-trained keypoint algorithm for performing image recognition.

[0046] like Figure 6 As shown, the high-throughput sampling and slicing method of the high-throughput sampling and slicing device of the present invention includes the following steps: S1. Single rice seeds 322 are continuously fed to the seed tray 3 by a vibrating rice seed metering device 4.

[0047] S2. Use the first vision camera 38 to acquire images of rice seeds 322 in the seed tray 3, and identify the pose of rice seeds 322 in the seed tray 3 based on the images. Step S2 specifically involves: using the first vision camera 38 to acquire images of rice seeds 322 within the seed tray 3; the robot's multi-node operating system receiving and processing the images using a keypoint algorithm to obtain the parent bounding box of the rice seed 322 and two keypoints, which represent the embryo side and endosperm side, respectively; calculating the midpoint coordinates of the two keypoints as the acquisition point coordinates; and using the angle between the vector formed by the two keypoints and the reference direction in the image coordinate system as the attitude angle. The acquisition point coordinates and the attitude angle together form the pose of the rice seed 322 within the seed tray 3. The image coordinate system is the two-dimensional coordinate system corresponding to the image acquired by the first vision camera 38.

[0048] S3. Based on the position of the rice seed 322 identified in step S2, control the robotic arm 1 to move the end negative pressure suction nozzle 2 to the rice seed 322. After using the end negative pressure suction nozzle 2 to suck up the rice seed 322, move the sucked rice seed 322 to the seed posture secondary calibration point 3220 of the seed tray 3. Step S3 is as follows: After performing inverse kinematics calculation based on the current coordinate values ​​of the end of the robotic arm 1 and the coordinate values ​​of the suction point, interpolation processing is performed to obtain the predicted motion trajectory. The robotic arm 1 is controlled to move to the target position according to the predicted motion trajectory, so that the suction port 25 at the end of the suction nozzle 26 is aligned with the suction point in the horizontal plane and in contact with the upper surface of the rice seed 322 in the vertical direction. The air pump provides negative pressure to the end negative pressure suction nozzle 2, and the end negative pressure suction nozzle 2 is used to pick up the single rice seed 322 on the seed tray 3. According to the coordinates of the pre-selected seed posture secondary calibration point 3220 on the seed tray 3, the robotic arm 1 is controlled to move the picked-up rice seed 322 to the seed posture secondary calibration point 3220.

[0049] S4. Use the first vision camera 38 to acquire the image of rice seed 322 at the seed posture secondary calibration point 3220. Based on the image, identify the current posture of rice seed 322 at the seed posture secondary calibration point 3220. Combine the current posture and the preset posture to perform feedback control on the axial rotation drive motor 22. The axial rotation drive motor 22 rotates the rice seed 322 adsorbed by the end negative pressure suction nozzle 2 to the preset posture. In step S4, both the current attitude and the preset attitude can be represented by attitude angles.

[0050] S5. Using the robotic arm 1, the rice seed 322 is moved from the seed posture secondary calibration point 3220 to above the preset clamping point of the seed clamping position 622. The second vision camera 7 is used to collect images. After identifying the total length of the rice seed 322 based on the images, the robotic arm 1 moves the rice seed 322 to the preset clamping point. In step S5, before identifying the length of the rice seed 322, it is possible to first identify whether the negative pressure nozzle 2 at the end has adsorbed the rice seed 322. If not, return to step S1; if so, continue to identify the length of the rice seed 322.

[0051] S6. After using the end negative pressure suction nozzle 2 to release the adsorption and release the rice seeds 322, the inner clamping plate 63 is moved closer to the outer clamping plate 61 by the drive motor 69, and the rice seeds 322 are clamped by the inner clamping plate 63 and the outer clamping plate 61. S7. Using the second vision camera 7, images are acquired. Based on the images, the exposed length of the endosperm side of the rice seed 322 (i.e., the length extending beyond the two clamps) is identified. The target cutting length is calculated based on the total length. The cutting path is determined based on the target cutting length and the exposed length of the endosperm side. The laser 8 cuts the rice seed 322 using a laser beam according to the cutting path, obtaining the seed mother body clamped between the outer clamp 61 and the inner clamp 63 and the cut sample that is not clamped. The cut sample falls into the sample collection tube 62. The robotic arm 1 moves the seed mother body to the entrance of the mother body collection tube 67, so that the seed mother body falls into the mother body collection tube 67. The cut sample is used for analyzing genetic information, and the seed mother body is used for breeding. In step S7, determining the cutting path based on the exposed length and the target cutting length means: determining the cutting point based on the exposed length and the target cutting length, and then generating the cutting path corresponding to the cutting point; the cutting point is located at one-third of the total length from the endosperm side of the seed, that is, at the position where the total length of the seed extends from the endosperm side key point to the embryo side key point, and the cutting path is a straight line passing through the cutting point and perpendicular to the long axis.

[0052] Specifically: The total seed length L is obtained by the second vision camera 7 in step S5, and the exposed seed length l is obtained by the second vision camera 7 in step S7. The cutting point is L / 3 on the seed endosperm side. If l is greater than L / 3, the robot's multi-node operating system modifies the laser cutting path to L / 3 on the seed endosperm side, i.e., the outer side of the clamping table (the laser cutting area, with air below the seed). If l is less than L / 3, the robot's multi-node operating system modifies the laser cutting path to L / 3 on the seed endosperm side, i.e., the inner side of the clamping table (the laser cutting area, with the clamping table below the seed). After confirming the cutting path, the robot's multi-node operating system sends the cutting path and cutting command to the host computer of the laser cutter to complete one cutting operation.

[0053] Without the aforementioned visual feedback to dynamically adjust the laser trajectory, meaning the laser cutting path is fixed, the cut samples may be uneven due to issues such as the suction nozzle not picking up the seeds at the exact midpoint and differences in seed length.

[0054] S8. Repeat steps S1 to S7 until the job is finished.

[0055] Furthermore, in steps S2 to S7, the key point algorithm is used to perform image recognition processing, that is, to identify the pose, posture and length of rice seed 322.

[0056] Preferably, the keypoint algorithm is the Yolov11-pose algorithm.

[0057] Specific embodiments of the present invention are as follows: Example 1 like Figures 1 to 5 As shown in the figure, this embodiment of the invention provides a high-throughput slicing device and control method for rice breeding.

[0058] The high-throughput sampling and slicing device for rice breeding provided in this embodiment includes a frame 5. A robotic arm 1 is mounted on the platform of the frame 5. A negative pressure suction nozzle 2 is mounted at the end of the robotic arm 1. The negative pressure suction nozzle 2 can move above the seed tray 3 under the drive of the robotic arm 1 to pick up seeds, and then move to the seed clamping position 622 in the clamping and collecting device 6. The image acquired by the vision camera 7 is used to dynamically adjust the laser cutting path. The collection tube 62 is located on one side below the seed clamping position 622 and is used to collect the cut samples that fall during the cutting process. The mother collection tube 67 is located inside the sample collection tube 62, that is, on the side close to the seed tray 3. The seed delivery end of the vibrating rice seed metering device 4 is located above the seed tray 3. Single rice seeds are added to the seed tray 3 through the slide 31. The vision camera 38 is placed directly below the seed tray 3. The image acquired by the vision camera 38 is used to identify and locate the seed pose and the suction point after processing by the key point algorithm.

[0059] like Figure 2 As shown, in this embodiment, the main body of the end negative pressure suction nozzle mechanism 2 adopts an axial rotation drive motor 22 (model HNKR-XPXZ35DJ-001, Hunan Kerry Precision Technology Co., Ltd.). The axial rotation drive motor 22 has an internal negative pressure channel 21. One end of the negative pressure channel 21 is connected to a pneumatic hose, and the other end is connected to the suction nozzle 26. The output end of the axial rotation drive motor 22 can rotate 360 ​​degrees around the Z-axis to control the 360-degree circular motion of the suction nozzle 26, which is further used for rotation after seed posture recognition to adapt to the seed clamping position 622, placing the rice endosperm segment externally, facilitating direct feeding into the sample collection tube 62 after laser cutting. When the end negative pressure suction nozzle mechanism 2 picks up seeds, the end suction hole 25 contacts the seed, and the suction nozzle air inlet 24 is connected to the negative pressure channel 21, thereby attracting the seed to overcome gravity through the negative pressure provided by the air pump.

[0060] The bottom of the end suction hole 25 is provided with a contoured ellipsoidal groove. The contour curvature of the contoured ellipsoidal groove is consistent with that of rice seeds, which makes the grasping force on the seeds more uniform during the suction process and keeps the seed posture more stable.

[0061] like Figure 3As shown, the supply module includes a seed chute 31, an inclined curved surface 32, rice seeds 322, a frame connector 33, a supplementary lighting strip 34, a transparent glass plane 35, a lens 36, a camera lens housing 37, and a first-vision camera 38. The inclined curved surface 32 and the transparent glass plane 35 are connected to the upper and lower sides of the frame connector 33, respectively, forming the seed tray 3. The camera, with its lens, can capture the seed image on the transparent glass plane 35, and the supplementary lighting reduces interference from light and dark areas, improving recognition accuracy. It should be noted that... Figure 3 In (b), the rice seed 322 is only an example. In actual implementation, the transparent glass plane 35 may have multiple rice seeds 322.

[0062] like Figure 4 As shown, the clamping and collecting device 6 includes an outer clamping plate 61, a sample collection tube 62, a seed clamping position 622, an inner clamping plate 63, a pressure sensor 64, a connecting plate 65, a lead screw sleeve 66, a seed mother collection tube 67, a lead screw 68, and a drive motor 69. A clamping point is pre-selected on the seed clamping position 622. During the slicing process, the robotic arm 1 places the adsorbed seeds at the clamping point according to a preset posture.

[0063] A fixed outer clamping plate 61 and an inner clamping plate 63, indirectly driven by a drive motor 69, clamp the seed embryo side, thus fixing the seed at the clamping point. Simultaneously, the seed endosperm side is exposed to the laser 7. The cut sample falls freely into the sample collection tube 62 via a sample slide. After sample cutting is complete, the inner clamping plate 63 is released from the seed mother body by the drive motor 69, and the seed mother body is re-lifted by the suction nozzle, moved above the seed mother body collection tube 67, and released, completing the collection of the seed mother body.

[0064] Furthermore, the clamping and collecting device 6 may also include a manual adjuster 610, which can manually drive the lead screw 68 to rotate, thereby manually fixing the seed.

[0065] like Figure 5 As shown, the laser 8 includes a galvanometer housing and a galvanometer 81, a laser beam outlet 82 disposed on the galvanometer housing, a fixing groove 83 for mounting the laser 8, a heat dissipation hole 84 opened on the laser tube housing, and a laser tube and galvanometer connector 85. The laser beam 821 is emitted from the laser beam outlet 82.

[0066] In this embodiment, the keypoint algorithm is specifically the YOLOv11-pose algorithm. The YOLOv11-pose model is pre-trained according to the following process: First, multiple images of rice seeds are acquired to ensure the sample diversity, position, and pose diversity of the training set; a total of 400 images are acquired. Then, image enhancement methods such as contrast adjustment, image flipping, and Gaussian blur are used to generate a training set of 2000 images. Ultralytics (the YOLOv11-pose pre-trained weights provided by YOLO) is used, and the output layer parameters are modified to use the parent bounding box as the seed and include two keypoints (head and tail) in the output layer JSON file. The model is then trained using the training methods provided by YOLO.

[0067] Example 2 like Figure 6 As shown in the figure, this embodiment proposes a high-throughput sampling and slicing method for rice breeding.

[0068] Specifically, the following steps are included: 1) First, based on the nine-point calibration method, the first vision camera 38 is used to calibrate the robotic arm 1 in the seed disk workspace, and the correspondence between the vision camera pixel value and the motor pulse is obtained, as well as the absolute coordinate value of the seed posture secondary calibration point 3220 in the camera coordinate system. 2) Using a vibrating seed metering device 4, single rice seeds 322 are fed into the seed tray 3. The rice seeds 322 fall onto the transparent glass plane 35 through the inclined curved surface 32. The first vision camera 38 captures the seed image through the transparent glass plane 35. 3) The Robot Multi-Node Operating System (ROS) uses the Yolov11-pose algorithm to identify the parent frame of rice seed 322, key points representing the embryo side, and key points representing the endosperm side. The midpoint coordinates of two key points are calculated as the sampling point coordinates, and the angle between the vector formed by the two key points and the positive x-axis in the image coordinate system is used as the attitude deflection angle. 4) The Robot Multi-Node Operating System (ROS) performs inverse kinematics calculation based on the current coordinates of the end of the robotic arm 1 and the coordinates of the rice seed 322 suction point, and interpolates to obtain the required motion trajectory. When the rice seed 322 is being sucked up, the suction port of the negative pressure suction nozzle 2 installed on the end of the robotic arm 1 coincides with the suction point on the horizontal plane and is in contact with the upper surface of the rice seed 322 in the vertical direction. At this time, the air pump provides negative pressure, and the negative pressure suction nozzle 2 will apply an attractive force to the rice seed 322, thereby overcoming the gravity of the rice seed 322 and picking up the rice seed 322. 5) After the rice seed 322 is grasped, the robot arm 1 is driven by the robot multi-node operating system (ROS) to first move the grasped rice seed 322 to the posture secondary calibration point 3220, and then drive the axial rotation drive motor 22 to rotate the seed according to the posture angle information. After the rotation is completed, the image feedback collected by the first vision camera 38 is used to determine whether the current seed posture is within a reasonable range. If the range is not reasonable, the rotation is repeated to prevent posture deviation that may occur when picking up the rice seed 322 due to instantaneous negative pressure and to improve the robustness of the system. 6) After the seed's secondary posture calibration is passed, it moves from the secondary calibration point 3220 to above the seed clamping position 622. The robot's multi-node operating system (ROS) receives and identifies whether there is a rice seed 322 at the current end air inlet 25 based on the image collected by the second vision camera 7, thereby determining whether to perform a cutting operation. If there is a rice seed 322 at the end air inlet 25, the current seed length is recorded and the rice seed 322 is placed at the seed clamping position 622. 7) After the rice seed 322 reaches the seed clamping position 622, the air pump provides positive pressure to the end negative pressure suction nozzle 2. At this time, the rice seed 322 separates from the end suction hole 25. The inner clamping plate 63 is moved to apply pressure to the rice seed 322 and press the rice seed 322 tightly and fix it. 8) After rice seed 322 is fixed, the robot's multi-node operating system (ROS) receives and identifies the length of the seed endosperm exposed on the clamping stage based on the image acquired by the second vision camera 7. It calculates the length of the sample to be cut by the laser based on the seed length, compares it with the exposed length identified by the second vision camera 7, and modifies the laser cutting path file to ensure that the cut sample reaches 1 / 3 of the total seed length, enabling genetic information analysis. The remaining seed mother body length reaches 2 / 3, reducing the impact on seed activity, and the cutting accuracy reaches 0.2 mm. After modifying the laser cutting path, the robot's multi-node operating system (ROS) sends instructions to the laser 8 via TCP. The laser 8 executes the instructions to complete the laser cutting. 9) The sample after the seed is cut will fall into the sample collection tube 62 under the action of gravity due to the lack of support, while the parent body will remain on the cutting stage 622. 10) Separate the inner clamping plate 63 from the seed mother body. The seed mother body returns to the free state from the previous fixed state. At this time, the end suction port 25 installed on the robotic arm 1 controlled by the robot multi-node operating system (ROS) coincides with the suction point after cutting in the horizontal direction and just contacts the upper surface of the seed in the vertical direction. At this time, the air pump provides negative pressure, and the end negative pressure suction nozzle 2 will apply attraction to the seed, thereby overcoming the seed's own gravity and picking up the seed. 11) The robotic arm 1 moves to the coordinate position of the mother seed collection tube 67. When the mother seed is above the mother seed collection tube 67, the air pump provides positive pressure to the end negative pressure nozzle. At this time, the mother seed separates from the air nozzle and the seed falls into the mother seed collection tube 67.

[0069] In steps 3, 4, and 5 above, when multiple rice seeds 322 may appear on the transparent glass plane 35, the same image captured by the first vision camera 38 will contain multiple rice seeds 322. When traversing the center point coordinates of all rice seeds 322, the coordinates of the seed center points near the secondary posture calibration point 3220 are prioritized. This ensures that if the robotic arm 1 picks up a rice seed 322 and moves it to the secondary calibration point to prepare for angle calibration, the calibration angle will definitely be that of the seed currently picked up by the robotic arm. Through this setting, in the robot multi-node operating system (ROS), the communication between the robotic arm 1 and the first vision camera 38 only transmits the coordinates and posture angle of one point at a time, reducing memory copying and improving system response speed.

[0070] In steps 6, 7, and 8 above, because pressure is applied to the seed for fixation in step 7, the length of the seed protruding from the clamping stage may shift, making it unreliable to use the absolute position of the robotic arm's movement as a reference for the cutting path. Therefore, the current seed length is first saved, then the sample cutting length is calculated, and finally, based on the length of the seed endosperm protruding from the clamping stage, the laser cutting path is flexibly adjusted to complete the effective cutting of the sample and improve sample uniformity.

[0071] Although embodiments of the invention have been shown and described, it will be understood by those skilled in the art that various changes, modifications, substitutions and alterations can be made to these embodiments without departing from the principles and spirit of the invention, the scope of which is defined by the appended claims and their equivalents.

Claims

1. A high-throughput laser slicing device for rice breeding, characterized in that, include: The robotic arm (1) is equipped with a negative pressure suction nozzle (2) at its end, and the nozzle (26) of the negative pressure suction nozzle (2) can rotate around a vertical axis; The supply module includes a seed tray (3) and a first vision camera (38). The seed tray (3) is arranged within the working range of the robotic arm (1). The first vision camera (38) is used to collect images of rice seeds (322) in the seed tray (3). A vibrating rice seed metering device (4) is used to replenish rice seeds (322) to the seed tray (3); The clamping and collecting module (6) includes a seed clamping position (622), a mother body collecting tube (67) and a sample collecting tube (62). The seed clamping position (622) and the mother body collecting tube (67) are both arranged within the working range of the robotic arm (1). The inlet of the sample collecting tube (62) is located on one side below the seed clamping position (622). A laser (8) is used to cut rice seeds (322) on a seed holder (622) using a laser beam. A second vision camera (7) is used to acquire images of rice seeds (322) on the seed clamping position (622).

2. The high-throughput laser slicing device for rice breeding according to claim 1, characterized in that: The suction port (25) at the end of the suction nozzle (26) is provided with a contoured ellipsoidal groove, which is adapted to the contour curvature of the rice seed (322).

3. The high-throughput laser slicing device for rice breeding according to claim 1, characterized in that: The supply module also includes a seed chute (31) and a supplementary light strip (34); the bottom of the seed tray (3) is made of a transparent glass plane (35), and the side wall is made of an inclined curved surface (32); a seed chute (31) is arranged above the seed tray (3), and the upper and lower ends of the seed chute (31) are connected to the seed delivery end of the vibrating rice seed metering device (4) and the inclined curved surface (32), respectively; a first vision camera (38) is arranged below the transparent glass plane (35), the lens (36) of the first vision camera (38) is arranged facing upward, and a supplementary light strip (34) is arranged between the lens (36) and the transparent glass plane (35).

4. The high-throughput laser slicing device for rice breeding according to claim 1, characterized in that: The clamping and collecting module (6) further includes an outer clamping plate (61), an inner clamping plate (63), a pressure sensor (64), and a drive motor (69); the outer clamping plate (61) and the inner clamping plate (63) are arranged opposite to each other on both sides of the seed clamping position (622). The outer clamping plate (61) is fixed, and the inner clamping plate (63) is connected to the drive motor (69) and can move back and forth along the direction of approaching and moving away from the outer clamping plate (61) under the drive of the drive motor (69). The inner clamping plate (63) and / or the outer clamping plate (61) are provided with pressure sensors (64) on their opposite sides.

5. The high-throughput laser slicing device for rice breeding according to claim 1, characterized in that: The robotic arm (1) is a Cartesian robotic arm.

6. A high-throughput sampling and slicing method using the high-throughput sampling and slicing equipment as described in any one of claims 1 to 5, characterized in that, Includes the following steps: S1. Rice seeds (322) are delivered to the seed tray (3) via a vibrating rice seed metering device (4). S2. Use the first vision camera (38) to acquire images of rice seeds (322) in the seed tray (3), and identify the pose of rice seeds (322) in the seed tray (3) based on the images; S3. Based on the position of the rice seed (322), control the robotic arm (1) to move the end negative pressure suction nozzle (2) to the rice seed (322), and use the end negative pressure suction nozzle (2) to suck up the rice seed (322). Then move the sucked rice seed (322) to the seed posture secondary calibration point (3220) of the seed tray (3). S4. Use the first vision camera (38) to acquire the image of the rice seed (322) at the seed posture secondary calibration point (3220), identify the current posture of the rice seed (322) at the seed posture secondary calibration point (3220) based on the image, and combine the current posture and the preset posture to perform feedback control on the axial rotation drive motor (22), and rotate the rice seed (322) to the preset posture through the axial rotation drive motor (22); S5. Using the robotic arm (1), the rice seed (322) is moved from the seed posture secondary calibration point (3220) to the preset clamping point of the seed clamping position (622). The second vision camera (7) is used to collect images. After identifying the total length of the rice seed (322) based on the images, the robotic arm (1) moves the rice seed (322) to the preset clamping point. S6. After releasing the rice seeds (322) using the end negative pressure suction nozzle (2), clamp the rice seeds (322) using the inner clamp (63) and the outer clamp (61). S7. Use the second vision camera (7) to collect images, identify the exposed length of the endosperm side of the rice seed (322) based on the image, calculate the target cutting length based on the total length, determine the cutting path based on the target cutting length and the exposed length of the endosperm side, and use the laser (8) to cut the rice seed (322) according to the cutting path, to obtain the seed mother and the cut sample. The cut sample falls into the sample collection tube (62), and use the robotic arm (1) to move the seed mother to the entrance of the mother collection tube (67) so that the seed mother falls into the mother collection tube (67). S8. Repeat steps S1 to S7 until the job is finished.

7. The high-throughput sampling and slicing method according to claim 6, characterized in that: The specific steps of step S2 are as follows: using a first vision camera (38) to acquire images of rice seeds (322) in the seed disk (3), processing the images using a key point algorithm to obtain the parent frame of the rice seeds (322) and two key points, the two key points representing the embryo side and the endosperm side respectively; calculating the midpoint coordinates of the two key points as the sampling point coordinates, and the angle between the vector formed by the two key points and the reference direction in the image coordinate system as the attitude angle, the sampling point coordinates and the attitude angle constitute the pose of the rice seeds (322) in the seed disk (3).

8. The high-throughput sampling and slicing method according to claim 7, characterized in that: Step S3 is as follows: After performing inverse kinematics calculation based on the current coordinate value of the end of the robotic arm (1) and the coordinate value of the suction point, interpolation is performed to obtain the predicted motion trajectory. The robotic arm (1) is controlled to move to the target position according to the predicted motion trajectory, so that the suction hole (25) at the end of the suction nozzle (26) is aligned with the suction point in the horizontal plane and in contact with the upper surface of the rice seed (322) in the vertical direction. The negative pressure suction nozzle (2) at the end is used to suck up the rice seed (322) on the seed tray (3). According to the coordinates of the seed posture secondary calibration point (3220) pre-selected on the seed tray (3), the robotic arm (1) is controlled to move the sucked rice seed (322) to the seed posture secondary calibration point (3220).

9. The high-throughput sampling and slicing method according to claim 6, characterized in that: In step S7, determining the cutting path based on the exposed length and the target cutting length means: determining the cutting point based on the exposed length and the target cutting length, and then generating the cutting path corresponding to the cutting point; the cutting point is located at one-third of the total length from the seed endosperm side.

10. The high-throughput sampling and slicing method according to claim 6, characterized in that: Image recognition is performed using a keypoint algorithm, specifically the Yolov11-pose algorithm.

Images