Intelligent separation and high-precision measurement device and method for stems and leaves

Abstract

The application discloses a kind of stem and leaf intelligent separation and high-precision measuring device and method, and it is related to plant phenotype analysis technical field.The device includes clamping module, weighing module, visual identification module, intelligent separation module, stem processing module, leaf processing module, chopping module, drying and weighing module and central control system.After clamping module fixes plant sample, weighing module obtains initial fresh weight;Visual identification module constructs stem and leaf point cloud model by three-dimensional scanning;Intelligent separation module plans cutting path based on model and executes separation;Stem processing module measures stem diameter and fresh stem weight;Leaf processing module flattens leaf and measures leaf area;After stem and leaf are chopped, send into drying and weighing module, and determine dry weight stable point by automatic weighing.The application integrates separation, measurement, drying function in one, realizes full-process automation, solves the problem of low efficiency, sample easy to damage, poor data consistency of traditional method, significantly improves the precision and efficiency of plant phenotype parameter acquisition.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of plant phenotype analysis, in particular to a stem-leaf intelligent separation and high-precision measurement device and method. BACKGROUND

[0002] In the fields of plant physiology, crop genetics and breeding, phenomics and precision agriculture, the accurate measurement of stem-leaf separation and related trait parameters (such as stem diameter, leaf area, fresh weight and dry weight) is a basic and key technical link. The accuracy of this technology directly affects the reliability of plant growth and development research, the efficiency of crop variety selection, and the scientific nature of quality evaluation.

[0003] Currently, the mainstream solutions for plant stem-leaf separation and measurement mainly fall into two categories: traditional manual operation and special mechanical separation devices.

[0004] The traditional manual operation method requires researchers to manually separate the stem and leaf, then use a ruler, vernier caliper, etc. to measure the stem diameter, obtain leaf area data through grid paper method or scanner, and finally obtain dry matter weight after drying and weighing. This operation mode has many drawbacks:

[0005] Manual separation is inefficient, and it usually takes several minutes to process a single plant sample, which cannot meet the needs of high-throughput phenotype analysis;

[0006] Manual operation inevitably causes sample damage, especially for plant species with thin leaves or brittle stems, affecting the authenticity of the data;

[0007] Human error is significant during the measurement process, and there is a large variation in measurement results between different operators, which seriously affects the comparability and reliability of the data;

[0008] During the transfer of samples between different devices (such as from the separation table to the scanner and then to the oven), the water loss caused by sample exposure significantly changes the tissue state, making trait parameters such as fresh weight and leaf area unable to reflect the initial state;

[0009] The long period from sample collection to the final complete data acquisition causes differences in the initial state of the samples, resulting in poor data consistency.

[0010] A variety of stem-leaf separation machines are also disclosed in the prior art, for example, patent document CN2018106472517A discloses a "deep and shallow two-time alternating cutting rosemary stem-leaf separation device", which cuts through alternating rotating cutters and separates stems and leaves by using a suction system. Patent document CN2021105451040A discloses a "stem-leaf separation machine", which realizes stem-leaf separation by cooperating a separation hole with an adaptive diameter adjustment with a traction mechanism. Other similar devices (such as CN2021109060648A, CN2021114449124A, etc.) also use mechanical cutting, brushing or pulling.

[0011] However, these existing mechanical separation schemes generally have the following limitations:

[0012] Lack of intelligent recognition and adaptive ability: existing devices cannot identify and adaptively adjust according to the specific morphology of plants (such as leaf type, stem thickness, branching structure), and the separation strategy is single, which can easily cause the sample integrity to be damaged and the damage rate to be high when facing different species or complex morphological plants.

[0013] Fragmented functions, poor systematic data: separation, measurement, drying, etc. are usually completed by different equipment, and the sample needs to be transferred between multiple devices. This not only is cumbersome and inefficient, but more importantly, the sample state (especially moisture) changes during the transfer process, resulting in the final measurement of multiple parameters (such as fresh weight, leaf area and dry weight) not being derived from the same sample in the same state, the internal relationship of the data is broken, and the system and comparability are poor.

[0014] Low degree of automation and integration: most existing solutions focus on the separation function itself and fail to integrate subsequent key parameter measurement and data processing, failing to achieve full-process automation from "sample input" to "data output", and still rely heavily on human intervention.

[0015] Therefore, there is an urgent need in the art for an integrated device that can realize intelligent identification, non-destructive separation, multi-parameter in-situ synchronous measurement and automatic data processing, to solve the problems of low efficiency, sample damage, inaccurate data and poor system in the prior art. SUMMARY

[0016] The purpose of the present application is to provide a stem-leaf intelligent separation and high-precision measurement device and method to solve the problems existing in the prior art, which is efficient, accurate and consistent.

[0017] To achieve the above purpose, the present application provides the following solutions:

[0018] The application provides a stem and leaf intelligent separation and high-precision measurement device, which comprises a clamping module, a weighing module, a first conveying module, a visual recognition module, an intelligent separation module, a stem processing module, a leaf processing module, a second conveying module, a chopping module, a drying and weighing module and a central control system. The clamping module is used for clamping samples; the weighing module is used for weighing the clamping module with samples and obtaining the initial fresh weight of the samples; the first conveying module is used for conveying the weighed samples; the visual recognition module is used for obtaining a three-dimensional point cloud model of the samples conveyed by the first conveying module and identifying the stem and leaf structure; the intelligent separation module is in communication connection with the visual recognition module and is used for planning a cutting path based on the three-dimensional point cloud model and physically separating the plant stems and leaves; the stem processing module is used for weighing and stem diameter measurement of the stems after the leaves are separated; the leaf processing module is used for flattening and measuring the leaf area of the separated leaves; the second conveying module is used for conveying the leaves and stems respectively; the chopping module is used for chopping the leaves and stems conveyed by the second conveying module respectively; the drying and weighing module is used for receiving the leaves and stems chopped by the chopping module, drying and automatically weighing until the stable dry weight is obtained; and the central control system is in communication connection with the clamping module, the weighing module, the first conveying module, the visual recognition module, the intelligent separation module, the stem processing module, the leaf processing module, the second conveying module, the chopping module and the drying and weighing module respectively, and is used for coordinating the work of the modules and performing data management.

[0019] Preferably, the visual recognition module comprises:

[0020] An identification station is used for receiving the samples conveyed by the first conveying module;

[0021] A main 3D camera is installed above the identification station;

[0022] A plurality of circumferential auxiliary 3D cameras are arranged around the identification station to form a multi-view imaging array;

[0023] A structured light projector is used for projecting near-infrared structured light to the identification station;

[0024] A ring-shaped LED light supplementing system is used for providing uniform illumination for imaging;

[0025] The main 3D camera, the circumferential auxiliary 3D cameras and the structured light projector work cooperatively to construct a high-precision three-dimensional point cloud model of the plant stems and leaves.

[0026] Preferably, the intelligent separation module comprises:

[0027] A three-axis servo driving system;

[0028] At least one group of hard alloy blades is driven by the three-axis servo driving system;

[0029] The central control system is configured to generate a plurality of candidate cutting paths based on the three-dimensional point cloud model through a machine learning algorithm, and control the blade to cut according to the selected path.

[0030] Preferably, three groups of cemented carbide blades are provided; the three groups of cemented carbide blades are respectively driven to move by three sub-three-axis servo driving systems; the three sub-three-axis servo driving systems are all driven by the three-axis servo driving system; the three sub-three-axis servo driving systems serve as three independent actuators, which can process different spatial cutting points on the stem in parallel to improve efficiency, or cooperatively cut at the same complex part.

[0031] Preferably, the second conveying module comprises a conveyor belt; the identification station is located directly above part of the conveyor belt, and the intelligent separation module is used for separating stems and leaves of the sample at the identification station; the leaf processing module comprises micro-protrusion structures arranged in an array on the surface of the conveyor belt, a plurality of suction holes opened on the conveyor belt, a vacuum suction system, and a leaf area measurement unit; the vacuum suction system generates micro-negative pressure on the surface of the conveyor belt through the suction holes on the surface of the conveyor belt, and cooperates with the micro-protrusion structures to prevent the leaf from curling during the transmission process; the leaf area measurement unit is arranged on the path of the conveyor belt and is used for measuring the leaf area of the flattened leaf.

[0032] Preferably, the clamping module is arranged at the weighing end of the weighing module, and the weighing module is arranged on a support; the first conveying module can convey the support, the weighing module, and the clamping module clamping the sample into the visual identification module; the stem processing module comprises the weighing module and uses the weighing module to weigh the stem after the leaf separation; the stem processing module further comprises an optical micrometer, which can move up and down and is used for scanning the stem to obtain the stem diameter and transmit the stem diameter information to the central control system, and the central control system processes the stem diameter information to obtain the maximum stem diameter value and saves and / or outputs the data.

[0033] Preferably, the drying and weighing module comprises:

[0034] A plurality of drying units, each of which has a separately arranged heater and a temperature sensor inside, for realizing independent temperature control;

[0035] An independent miniature high-precision weighing sensor is arranged below each of the drying units, for automatically and periodically weighing the sample weight during the drying process;

[0036] A plurality of containers are arranged, each of which is used for receiving the leaf or stem cut by the chopping module;

[0037] Six-axis robot for transporting containers with samples into the drying unit.

[0038] Preferably, the chopping module comprises a fixed blade and a rotating blade arranged at the end of the conveyor belt, the fixed blade is arranged at the bottom of the rotating blade, and empty containers can be transported by the third conveying module to the lower part of the chopping module to receive the leaves or stems falling from the chopping module after being chopped.

[0039] Preferably, the central control system comprises:

[0040] The host computer runs on the Ubuntu and ROS architecture, and is responsible for visual processing, path planning algorithm and data management;

[0041] The lower computer is based on STM32 series MCU, and is responsible for controlling the motion mechanism, motor and sensor;

[0042] The host computer and the lower computer communicate through the EtherCAT bus.

[0043] The application also provides a method for separating and measuring stems and leaves by using the intelligent stem and leaf separation and high-precision measurement device, comprising the following steps: after the sample is clamped by the clamping module and the initial fresh weight is recorded, the sample is sent to the visual recognition module for three-dimensional modeling;

[0044] The intelligent separation module plans a path according to the model and performs stem and leaf separation;

[0045] The separated leaves enter the leaf processing module, are measured for leaf area after being flattened, are chopped and transferred to the drying and weighing module;

[0046] The stems left in the stem processing module after separation are weighed for fresh stem weight and measured for stem diameter, and then are chopped and transferred to the drying and weighing module;

[0047] The drying and weighing module dries the stem and leaf sample, determines the dry weight stabilization point through automatic weighing, and records the final dry weight;

[0048] Preferably, the fresh weight of the leaves is calculated by subtracting the fresh stem weight from the initial fresh weight.

[0049] The application has the following technical effects compared with the prior art:

[0050] The application integrates clamping, weighing, conveying, visual recognition, intelligent separation, stem treatment, leaf treatment, chopping, drying and weighing, and central control modules in a coherent system to build a complete plant phenotype data analysis pipeline. The core advantage is that the necessity of sample transfer between different independent devices is eliminated. Based on this, it can be concluded that: first, all data from initial weighing to final drying of the sample are derived from the same spatio-temporal state of the entity, effectively avoiding water loss or state changes caused by sample transfer and exposure, thereby fundamentally ensuring the systematicness and internal consistency of data such as fresh weight, leaf area and dry weight; second, the entire process is automatically scheduled by a central control system without human intervention in each operation link, which not only minimizes human operation errors, but also greatly improves processing efficiency, making high-throughput phenotype analysis possible; finally, this integrated design realizes full automation from "sample input" to "data output", providing a fundamental technical solution to the problems of fragmented functions, low efficiency and inaccurate data in the prior art. BRIEF DESCRIPTION OF DRAWINGS

[0051] In order to more clearly illustrate the technical solutions in the embodiments of the present application or the prior art, the drawings needed in the embodiments will be briefly introduced below. Obviously, the drawings in the following description are only some embodiments of the present application, and other drawings can be obtained by those skilled in the art without creative labor.

[0052] Fig. 1 Structure diagram of the stem-leaf intelligent separation and high-precision measurement device provided for the first embodiment of the present application;

[0053] Fig. 2 Flowchart of the stem-leaf intelligent separation and high-precision measurement method provided for the second embodiment of the present application;

[0054] In the figure: 1 - support; 2 - clamping module; 3 - weighing module; 4 - first conveying module; 5 - visual recognition module; 6 - intelligent separation module; 7 - optical micrometer; 8 - circumferential auxiliary 3D camera; 9 - structured light projector; 10 - main 3D camera; 11 - second conveying module; 12 - chopping module; 13 - drying and weighing module; 14 - six-axis mechanical arm; 15 - container; 16 - third conveying module; 17 - container storage warehouse door; 18 - leaf area measurement unit; 19 - central control system; 20 - feed inlet; 21 - first automatic door; 22 - second automatic door. DETAILED DESCRIPTION

[0055] Clearly and completely describe the technical solutions in the embodiments of the present application in conjunction with the drawings in the embodiments of the present application. Obviously, the described embodiments are only a part of the embodiments of the present application, rather than all. Based on the embodiments in the present application, all other embodiments obtained by those of ordinary skill in the art without creative labor fall within the scope of the present application.

[0056] In order to make the above-mentioned purposes, features and advantages of the present application more obvious and easy to understand, the present application will be further described in detail below in conjunction with the drawings and specific embodiments.

[0057] The purpose of the scheme provided in the specification is to output leaf fresh weight, stem fresh weight, overall sample fresh weight, stem dry weight, leaf dry weight, leaf area, and stem maximum diameter, which are measured by respective execution components, and data processing and data output are performed by a central control system.

[0058] The embodiments of the present application will be described below in conjunction with the drawings. Figs. 1-2

[0059] Embodiment One

[0060] The present application provides a stem and leaf intelligent separation and high-precision measurement device, comprising: a clamping module 2, a weighing module 3, a first conveying module 4, a visual recognition module 5, an intelligent separation module 6, a stem processing module, a leaf processing module, a second conveying module 11, a chopping module 12, a drying and weighing module 13, and a central control system 19. The clamping module 2 is used for clamping samples; the weighing module 3 weighs the samples and obtains the initial fresh weight of the samples; the first conveying module 4 is used for conveying the weighed samples; the visual recognition module 5 is used for obtaining a three-dimensional point cloud model of the samples conveyed by the first conveying module 4 and identifying the stem and leaf structure; the intelligent separation module 6 is in communication connection with the visual recognition module 5 and is used for planning a cutting path based on the three-dimensional point cloud model and physically separating the plant stems and leaves; the stem processing module is used for weighing and stem diameter measurement of the stems after leaf separation; the leaf processing module is used for flattening and measuring the leaf area of the separated leaves; the second conveying module 11 is used for conveying the leaves and stems respectively; the chopping module 12 is used for chopping the leaves and stems conveyed by the second conveying module 11 respectively; the drying and weighing module 13 is used for receiving the chopped leaves and stems by the chopping module 12, drying and automatically weighing until the stable dry weight is obtained; and the central control system 19 is in communication connection with the clamping module 2, the weighing module 3, the first conveying module 4, the visual recognition module 5, the intelligent separation module 6, the stem processing module, the leaf processing module, the second conveying module 11, the chopping module 12, and the drying and weighing module 13 respectively, and is used for coordinating the work of the modules and performing data management.

[0061] ​The embodiment integrates clamping, weighing, conveying, visual recognition, intelligent separation, stem treatment, leaf treatment, chopping, drying and weighing, and central control, etc. modules in a coherent system to build a complete plant phenotype data analysis pipeline. The core advantage is to eliminate the need for sample transfer between different independent devices. Based on this, it can be concluded that: first, all data from the initial weighing to the final drying of the sample are derived from the same spatio-temporal state of the entity, effectively avoiding water loss or state changes caused by sample transfer and exposure, thereby fundamentally ensuring the systematicness and internal consistency of data such as fresh weight, leaf area and dry weight; second, the entire process is automatically scheduled by the central control system 19 without the need for manual intervention in each operation link, and the detected data is output by the central control system 19, which not only minimizes human operation errors, but also greatly improves processing efficiency, making high-throughput phenotype analysis possible; finally, this integrated design realizes full automation from "sample input" to "data output", providing a fundamental technical solution to the functional fragmentation, low efficiency and data inaccuracy problems in the prior art.

[0062] In some embodiments, the visual recognition module 5 includes a recognition station, a plurality of circumferential auxiliary 3D cameras 8, a main 3D camera 10, a structured light projector 9, and a ring-shaped LED light supplement system; the recognition station is used to receive samples conveyed by the first conveying module 4; the main 3D camera 10 is installed above the recognition station; the plurality of circumferential auxiliary 3D cameras 8 are arranged around the recognition station to form a multi-view imaging array; the structured light projector 9 is used to project near-infrared structured light to the recognition station; the ring-shaped LED light supplement system is used to provide uniform illumination for imaging; the 3D camera is also called an RGB-D camera; wherein the main 3D camera 10, the circumferential auxiliary 3D camera 8 and the structured light projector 9 work together to construct a high-precision plant stem and leaf three-dimensional point cloud model, with a point cloud resolution of more than 200 points per square millimeter and a reconstruction accuracy of ±0.1 mm.

[0063] The structured light projector 9 preferably uses near-infrared light of 850 nm wavelength, which can reduce the interference of natural light in the environment and help obtain more delicate surface contours.

[0064] The heights of all circumferential auxiliary 3D cameras 8 are not completely the same and can be located obliquely above and below the sample, which helps to more accurately capture the contour information of the sample.

[0065] The embodiment is a multi-source information fusion high-precision three-dimensional visual recognition system formed by arranging a main 3D camera 10, a plurality of circumferential auxiliary 3D cameras 8, a structured light projector 9, and a ring-shaped light supplementing system. The circumferentially distributed cameras form a multi-view imaging array, which can collect 360° non-dead-angle samples, greatly reducing the missing of three-dimensional point cloud data caused by the overlap and shielding of the leaves themselves. Combined with the active optical texture provided by the structured light projector 9, the feature points on the surface of the object can be enhanced, thereby significantly improving the three-dimensional reconstruction model, especially the accuracy and completeness of the complex profile at the leaf edge and leaf vein. A more complete and more accurate three-dimensional model is the basis for subsequent intelligent separation module 6 to carry out precise path planning, so the embodiment directly provides the primary visual guarantee for the separation non-destructiveness and measurement accuracy of the entire device.

[0066] In some examples, the visual sensing scheme can be replaced under the premise of ensuring basic functions. For example, a laser scanning radar can be used instead of a structured light scheme for three-dimensional modeling, although there may be advantages and disadvantages in cost or point cloud density, but it can also obtain three-dimensional space information. Or, in applications with extremely high requirements for color information, some 3D cameras can be replaced by high-resolution multispectral cameras to collect spectral information in specific wavebands while obtaining three-dimensional shapes, thereby expanding the functionality of the device and enabling simultaneous estimation of physiological parameters such as chlorophyll content.

[0067] In some examples, in order to further improve the model adaptability and recognition robustness, the height and horizontal position of the sample at the recognition station can be adjusted to adjust the distance between the sample and the camera lens.

[0068] In some embodiments, the intelligent separation module 6 includes a three-axis servo drive system and at least one set of hard alloy blades. The hard alloy blades are driven by the three-axis servo drive system; wherein the central control system 19 is configured to generate a plurality of candidate cutting paths based on the three-dimensional point cloud model through a machine learning algorithm, and control the blades to cut according to the selected path.

[0069] The core of the embodiment is the introduction of an intelligent separation strategy based on a three-dimensional point cloud model and a machine learning algorithm to plan the path. The effect can be derived from the superiority of this strategy over fixed program control: the machine learning algorithm (such as CNN) can intelligently identify key parts such as stems, leaves, and branch points from the three-dimensional model and understand their spatial topological relationships, thereby planning personalized cutting paths that adapt to the unique shape of each plant. This is in sharp contrast to the "one-size-fits-all" mechanical approach in existing technologies, which enables the blades to actively avoid tough stems and fragile branch points, achieving a leap from "blind cutting" to "precise cutting", thereby significantly reducing sample damage rates and improving leaf separation completeness at the root cause.

[0070] In some examples, the intelligent algorithm for path planning is not unique, and in addition to CNN, other advanced deep learning models such as Transformer architecture can also be used for part recognition and relationship modeling, which may perform better in the processing of large-scale complex plants. In addition, as a simplified alternative, instead of using complex machine learning models, stem and leaf segmentation and path planning can be performed based on the geometric features (such as curvature, normal vector) of the three-dimensional point cloud through traditional image processing algorithms (such as region growing, edge detection), although the intelligence level and adaptability may be reduced, but for simple structure plants still can play a certain effect.

[0071] In some examples, leaf node detection is based on a CNN algorithm, path planning is based on an RRT algorithm, and real-time interaction is achieved with the intelligent separation module 6.

[0072] In some embodiments, three groups of hard alloy blades are provided; the three groups of hard alloy blades are driven to move by three sub-three-axis servo drive systems; the three sub-three-axis servo drive systems are driven by a three-axis servo drive system; the three sub-three-axis servo drive systems serve as three independent actuators, which can handle cutting points in different spaces on the stem in parallel to improve efficiency, or cooperatively cut at the same complex part.

[0073] This embodiment introduces parallel and cooperative working mechanism by setting three groups of blades controlled by independent sub-three-axis servo drive systems. Its effect can be deduced from parallel processing capability: when processing complex plants with multiple branches or a large number of leaves, the three groups of blades can simultaneously aim at cutting points in different spatial positions for work, which is equivalent to changing serial processing into parallel processing, thereby doubling the total time required for separation work and greatly improving the processing throughput of the device. In addition, when encountering a particularly complex structure (such as closely growing leaf bases), multiple blades can be fed cooperatively from different angles to perform "relay" cutting, which can more safely and effectively handle complex parts that a single blade cannot cope with, further ensuring the thoroughness and non-destructiveness of separation.

[0074] It can be understood that the number of blades can be adjusted according to cost and performance requirements. For example, in a low-cost version facing simple crops, only one or two groups of blades can be configured. Alternatively, in order to pursue extreme efficiency and flexibility, four or more groups of blades can be configured. In addition, the driving mode of the blades is not limited to the three-axis servo system, and for specific applications, SCARA robots or six-axis joint robots can also be considered to hold the blades to provide a larger working space and more flexible cutting angles.

[0075] In some embodiments, the second conveying module 11 comprises a conveyor belt; the identification station is located directly above a part of the conveyor belt, and the intelligent separation module 6 is used for stem-leaf separation of the sample at the identification station; the leaf processing module comprises: a micro-protrusion structure arranged in an array on the surface of the conveyor belt, a plurality of suction holes opened on the conveyor belt, a vacuum suction system, and a leaf area measurement unit 18; the vacuum suction system generates a micro-negative pressure on the surface of the conveyor belt through the suction holes on the surface of the conveyor belt, and cooperates with the micro-protrusion structure to prevent the leaf from curling during transmission; the leaf area measurement unit 18 is arranged on the path of the conveyor belt and is used for measuring the leaf area of the flattened leaf.

[0076] The innovation of the embodiment is to integrate the conveyor belt with micro-protrusion structure and vacuum suction. The effect can be deduced from the physical principle level: thin and soft leaves are prone to curling and wrinkling during transmission due to inertia, air flow or insufficient self-rigidity, which will seriously affect the accuracy of leaf area measurement. The micro-protrusion structure suppresses the transverse sliding and wrinkling tendency of the leaf by increasing the micro-friction with the leaf surface and mechanical limiting; and the vacuum suction provides uniform distributed suction force perpendicular to the surface of the conveyor belt, tightly "ironing" the leaf on the surface of the conveyor belt. The two work together to form a powerful "flattening force field", ensuring that the leaf is always in an ideal flat state when entering the measurement unit, thereby providing an important prerequisite guarantee for the subsequent leaf area measurement unit 18 to obtain high-precision data.

[0077] More specifically, the micro-protrusions on the surface of the conveyor belt are biomimetic mushroom head-shaped structures. The specific parameters of the structure are: regularly arrayed on the surface of the conveyor belt, with a center-to-center distance of 2 mm, a single body diameter of 1.2 mm, a height of 0.5 mm, and the top of the protrusion being disc-shaped with the edge slightly raised. The vacuum suction system is composed of vacuum chambers uniformly distributed under the conveyor belt, with a single vacuum chamber size of 100 × 50 mm and a spacing of 100 mm. Each vacuum chamber is connected to a group of negative pressure pumps that can provide a micro-negative pressure of 5 kPa, and is connected to the suction holes on the surface of the conveyor belt through a spiral suction hole pipeline with a hole diameter of 0.8 mm. The driving speed of the conveyor belt is stably controlled at 0.1 m / s, and at least 500 mm long stable flattening area is provided in front of the leaf area measurement unit 18.

[0078] Of course, the function of flattening the leaf is not limited to the above specific parameters. For example, the structure of the micro-protrusion can be replaced by other geometric shapes such as pyramid and cylinder. The vacuum suction system can not be divided into multiple independent chambers, but use a unified negative pressure chamber, and adjust the distribution density of the suction holes to control the uniformity of the suction force. In addition, when dealing with particularly thick and hard-to-curl leaves, a simplified scheme of using only biomimetic micro-protrusion structure or only low-power vacuum suction can be considered to reduce the cost.

[0079] In some embodiments, the clamping module 2 is arranged at the weighing end of the weighing module 3, and the weighing module 3 is arranged on a support 1; the first conveying module 4 can convey the support 1, the weighing module 3 and the clamping module 2 with the sample to the visual recognition module 5; the stem processing module includes the weighing module 3 and uses the weighing module 3 to weigh the stem after the leaf separation; the stem processing module further includes an optical micrometer 7, which can move up and down and is used to scan the stem to obtain the stem diameter and transmit the stem diameter information to the central control system 19, and the central control system 19 processes the stem diameter information to obtain the maximum stem diameter value and saves and / or outputs the data.

[0080] The present embodiment integrates the clamping module 2, the weighing module 3 and the support 1, and reuses the weighing module 3 for stem weighing, which embodies high functional integration and process optimization design. Specifically, the plant is clamped at the entrance to complete the initial fresh weight measurement, and the stem is always on the same weighing module 3 until the leaf separation is completed, which enables the measurement of fresh stem weight to be completed in situ without sample transfer. This design brings two core benefits: first, the leaf fresh weight can be accurately calculated by simple subtraction (initial fresh weight - fresh stem weight), without the need for separate weighing of loose leaves which are easily damaged and adhered, simplifying the process and avoiding errors; second, the number of clamping and releasing of the sample is minimized, which not only improves the efficiency, but also reduces the risk of stem damage or data connection error caused by repeated operations.

[0081] It can be understood that the measurement of the geometric parameters of the stem is not limited to the optical micrometer 7. For example, the high-precision three-dimensional point cloud model generated by the visual recognition module 5 can be directly used to automatically extract the diameters of each part of the stem through software algorithms, thereby eliminating the need for an independent hardware sensor. Alternatively, a contact diameter gauge can be used, but it may cause slight damage or interference due to physical contact with the sample.

[0082] In some embodiments, the drying and weighing module 13 includes a plurality of drying units, containers 15 and a six-axis mechanical arm 14. Each drying unit has a separate heater and temperature sensor arranged therein for independent temperature control; a separate micro high-precision weighing sensor is arranged below each drying unit for automatically and periodically weighing the sample during the drying process; the containers 15 are provided in multiple numbers, and each container 15 is used to receive the chopped leaves or stems from the chopping module 12; the six-axis mechanical arm 14 is used to convey the containers 15 containing the samples to the drying units.

[0083] The embodiment adopts a matrix distribution of independent drying and weighing units, and cooperates with a mechanical arm to schedule samples. Multiple independent drying units allow simultaneous drying of a large number of samples, and each unit is independently temperature-controlled and weighed, avoiding cross-influence between samples. The key is that the weighing sensor integrated below each unit can realize online and automatic weight monitoring. The system determines the drying endpoint by judging that "the difference between two consecutive weighings is less than a threshold value", which is a precise stopping method based on objective data, completely eliminating the problem of over-drying or under-drying caused by traditional drying relying on experience timing, ensuring the absolute reliability of dry weight data, and realizing unmanned value of the drying process.

[0084] In some examples, the container 15 is an aluminum box with an open top, which is convenient for the chopped samples to automatically fall into the aluminum box.

[0085] The size of the aluminum box is preferably 40 × 40 × 40 mm.

[0086] In some embodiments, the chopping module 12 includes a fixed blade and a rotating blade arranged at the end of the conveyor belt, the fixed blade is arranged at the bottom of the rotating blade, and the empty container 15 can be transported by the third conveying module 16 to the lower part of the chopping module 12 to receive the chopped leaves or stems falling from the chopping module 12.

[0087] The embodiment sets the chopping module 12 at the end of the conveyor belt and supplies empty containers 15 by the third conveying module 16. This design allows the measured whole leaves and complete stems to be immediately chopped in place, and the fragments directly fall into the prepared standard containers 15 below. On the one hand, this prepares the subsequent drying module with sample material that is uniform in shape and easy to dry uniformly, significantly improving drying efficiency and data accuracy; on the other hand, it regularizes the dispersed leaf and stem material into different containers 15 at the same location, greatly facilitating the grabbing, transferring and stacking of the mechanical arm, realizing the conversion from "bulk material" to "standard unit", and laying a solid foundation for full-process automation.

[0088] More specifically, after all the leaf area measurements are completed (no data appears within 10s), the container 15 carrying the leaves is driven to rise, and the container 15 is automatically transferred by the six-axis robot arm 14 into an empty drying unit (the empty drying unit number is determined by the weighing module 3 in the drying unit, i.e., each drying unit has a unique number, and the central control system 19 records the number and position information of each drying unit to facilitate the control of the six-axis robot arm 14 to deliver the container 15 into the controlled drying unit). The heater in the working state usually maintains a constant temperature of 80±1°C, of course, it can also be adjusted according to the needs. Automatic weighing is triggered every 15 minutes. If the difference between two consecutive weighings is less than 0.1g, the system automatically determines that the dry weight is stable, records the final dry weight, and completes the whole process of the sample. When the leaves are placed in the drying unit, the process of delivering the stem sample into the container 15 can be started. At the same time, step one of loading and positioning can be restarted to place a new sample. When the container 15 is not enough, an alarm signal is sent, and a new container 15 can be placed on the third conveying module 16 from the hatch to supply new empty containers 15 below the chopping module 12, which can prevent 10 empty containers 15 at a time. When a container 15 is taken away by the six-axis robot arm 14, the next empty container 15 can be automatically delivered to the bottom of the chopping module 12.

[0089] The form of the cutter in the chopping module 12 can be varied. For example, a high-speed rotating multi-blade cutter can be used instead of the combination of fixed blades and rotating blades to obtain a finer sample.

[0090] In some embodiments, the central control system 19 includes an upper computer and a lower computer. The upper computer runs on the Ubuntu and ROS architecture, is responsible for visual processing, path planning algorithm and data management; the lower computer is based on the STM32 series MCU, is responsible for controlling the motion mechanism, motor and sensor; wherein the upper computer and the lower computer communicate through the EtherCAT bus.

[0091] The embodiment adopts a distributed control system architecture of "upper computer (Ubuntu+ROS) + lower computer (STM32) + EtherCAT bus". The upper computer is responsible for complex algorithms such as visual processing and path planning, and a large amount of data operation, which requires powerful general computing power and rich software ecological support. The combination of Ubuntu and ROS perfectly meets this demand. The lower computer focuses on tasks with extremely high real-time requirements such as motion control and sensor reading, and the STM32 series MCU performs well in this regard. The two are connected through the high-performance industrial real-time bus EtherCAT, ensuring that massive data such as point clouds and control instructions can be transmitted between the upper and lower computers and between the numerous axis stations (servo drives and sensors) with microsecond-level synchronization and deterministic transmission. This architecture integrates the flexibility of general computing platforms and the real-time reliability of special control systems, and is the core "nerve center" of the stable, precise and collaborative operation of the entire complex device.

[0092] In some embodiments, when planning the cutting path based on the three-dimensional point cloud model, the central control system 19 is configured to generate three candidate paths simultaneously, and the generation principles include:

[0093] Path A (shortest path method): The total travel distance of the blade is the primary optimization target, aiming to improve processing efficiency.

[0094] Path B (smoothest tangent method): The cutting path is as smooth as possible with the stem surface, and actively avoids complex structure areas such as stem nodes and bifurcation angles, aiming to improve the completeness rate of the leaves.

[0095] Path C (overlapping leaf avoidance method): A buffer cutting curve is set for the overlapping area of the leaves identified in the point cloud to avoid the blade being cut into the area, aiming to prevent the blade from being stuck or the leaves from not being completely detached.

[0096] The system visualizes the three candidate paths through the human-computer interaction interface, allowing the operator to perform fine-tuning operations such as ±5° rotation, ±3mm translation, or path fusion. During the cutting execution phase, the system enables a closed-loop real-time monitoring mechanism, including: current feedback: real-time monitoring of the current of the cutting motor, if the instantaneous current surge exceeds 10%, it is judged as abnormal blade sticking, and the deceleration and repositioning program is triggered immediately.

[0097] Image difference feedback: through the camera, the front and rear frame images are compared to identify the leaves still attached to the stem after cutting, if the un-detached area is found, the blade is automatically dispatched for secondary cutting.

[0098] The embodiment introduces multi-criteria path planning and human-machine collaborative decision-making, and combines multi-modal real-time feedback control. The beneficial effects can be deduced from decision optimization and system robustness: three candidate paths based on different optimization objectives (efficiency, integrity, safety) are provided, and the final decision is left to the operator. The combination of algorithmic computing power and human expertise in the field enables optimal decision-making in complex and unstructured scenarios, which is more flexible and reliable than the path selected by a single algorithm. Furthermore, during execution, through current and image dual feedback, the system has the ability to perceive environmental uncertainty (such as local hardness abnormalities of plant tissues) and evaluate its own operation results (whether the cutting is successful). This makes the cutting process change from "open-loop execution" to "closed-loop adaptive control", which can dynamically respond to various unexpected situations in actual operation, effectively avoiding equipment jamming, sample damage and incomplete separation, and greatly improving the intelligent level and operation success rate of the entire system.

[0099] It can be understood that the strategy of path planning can be simplified, for example, the system can only generate a recommended path with the highest comprehensive score, without providing multiple path options. The real-time feedback mechanism can also be costed according to the cost, for example, only using current feedback for basic equipment protection, or only using image differential feedback to ensure separation effect.

[0100] In some embodiments, the clamping module 2 is arranged in the first housing, the visual recognition module 5 and the intelligent separation module 6 are arranged in the second housing, the first housing and the second housing are arranged side by side, and an automatic door is arranged between the two housings. The support 1 of the clamping module 2 is detachably arranged on a lifting module, which is preferably a linear electric cylinder. The bottom of the support 1 and the free end of the linear electric cylinder are detachably connected through an electromagnetic lock connector. The electromagnetic lock connector is divided into two parts, one part is fixedly connected with the bottom of the support 1, and the other part is fixedly connected with the free end of the linear electric cylinder. The first conveying module 4 is a linear driving mechanism capable of driving the clamping module 2 to move linearly from the first housing to the second housing through the automatic door, for example, a linear module using a screw nut for transmission. The support 1 and the free end of the linear module are detachably connected, specifically through electromagnetic adsorption principle for adsorption connection and separation. More specifically, hooks are arranged on the support 1, and the free end of the linear module has a pin shaft. The hooks can be hooked on the pin shaft by sliding down, and then the support 1 is adsorbed and fixed by the magnetic attraction device arranged at the free end of the linear module.

[0101] The detachable separation structure of the embodiment realizes the purpose of sharing one clamping module 2 and weighing module 3 to clamp the initially placed sample, weigh the fresh weight of the whole sample, clamp the sample for subsequent blade cutting processing, and weigh the fresh weight of the stem.

[0102] It can be understood that after the sample is processed, the central control system 19 can control the first conveying module 4 to drive the support 1 to carry the weighing module 3 and the clamping module 2 to reset, so as to facilitate the automatic processing and measurement of the next sample.

[0103] In some examples, the top of the first shell is provided with a feeding port 20, and a sample is manually placed on the clamping module 2 through the feeding port 20 for clamping.

[0104] In some examples, a third shell is further included, and the leaflet processing module is arranged in the third shell. The third shell, the second shell and the first shell are sequentially arranged. The first automatic door 21 is arranged between the first shell and the second shell, and the second automatic door 22 is arranged between the second shell and the third shell. The automatic doors are controlled to open and close by the central control system 19.

[0105] In some examples, a circumferential auxiliary 3D camera 8 is arranged below the support 1, that is, the circumferential auxiliary 3D camera 8 is fixedly connected with the support 1. Since the transmission belt is below the identification station, it is not provided with a circumferential auxiliary 3D camera 8 directly below the identification station. Based on this, the circumferential auxiliary 3D camera 8 is arranged on the support 1 so as to move with the support 1.

[0106] More specifically, the number of circumferential auxiliary 3D cameras 8 is preferably 9, that is, four are arranged at the four corners above the identification station, four are arranged at the four corners below the identification station, and one is arranged directly below the identification station. Of course, it is not limited to this number.

[0107] In some examples, in order to adjust the height of the sample at the identification station, the first conveying module 4 is arranged on a mechanism capable of driving the first conveying module 4 to ascend and descend. The mechanism can drive the first conveying module 4 to ascend and descend with the clamping module 2 and the sample thereon. The horizontal position of the sample can be adjusted by the first conveying module 4.

[0108] In some examples, the clamping module 2 is a pair of opposed pneumatic clamps or electric clamps. The surface of the clamp for contacting the sample is provided with a flexible silica gel layer, and the surface is provided with a component for detecting whether the sample is in place, such as a limit switch. When the component is triggered, the surface of the sample is in place, and the central control system 19 controls the start of the next action.

[0109] More specifically, in order to avoid damage to the sample caused by the clamping module 2, a plurality of bifurcated guide grooves are arranged on the surface of the flexible silica gel layer. When the clamping jaw is closed, the branches at the bifurcation in the sample will contact and press the slope of the groove edge. Under the action of clamping force, the soft silica gel will locally elastically deform and automatically form a "customized" wrapping space around the branches. It is equivalent to wrapping an irregular object with a piece of soft clay, which will adaptively fill all gaps.

[0110] In some examples, a container storage compartment is also included, which stores a plurality of empty containers. The container storage compartment has a container storage compartment door 17.

[0111] Embodiment two

[0112] The present application also provides a method for separating and measuring stems and leaves using the stem and leaf intelligent separation and high-precision measurement device of embodiment one, comprising: after the sample is clamped by the clamping module 2 and the initial fresh weight is recorded, it is sent to the visual recognition module 5 for three-dimensional modeling;

[0113] The intelligent separation module 6 plans the path according to the model and performs stem and leaf separation;

[0114] After separation, the leaves enter the leaf processing module, are flattened to measure the leaf area, are then chopped and transferred to the drying and weighing module 13;

[0115] After separation, the stems left in the stem processing module are weighed for fresh stem weight and the stem diameter is measured, and then are chopped and transferred to the drying and weighing module 13;

[0116] The drying and weighing module 13 dries the stem and leaf sample, and determines the dry weight stability point by automatic weighing to record the final dry weight;

[0117] Wherein, the fresh weight of the leaves is calculated by subtracting the fresh stem weight from the initial fresh weight.

[0118] This embodiment defines the working method of the device, the core of which is to construct an automatic closed-loop process driven by data. Each step of the method produces key data and drives the execution of the next step: the initial fresh weight and the three-dimensional model drive intelligent separation, the state after separation triggers the parallel measurement process of stems and leaves, and the measured sample enters the intelligent drying and dry weight determination process after being standardized and chopped. The beneficial effect of this method is that it integrates a series of discrete operations into a seamless and organic whole that is connected by data flow throughout. Through this method, not only is the whole process automated, but more importantly, it ensures that all measurement parameters (stem diameter, leaf area, fresh weight, dry weight) are derived from the continuous state changes of the same sample in a short period of time in a closed environment, thereby obtaining highly consistent, traceable and mutually verifiable comprehensive phenotype data, greatly improving the quality and value of scientific research data.

[0119] Of course, this embodiment is an improvement based on embodiment one, which has all the advantages of embodiment one, and will not be repeated here.

[0120] The principles and implementation manners of the present application are described by using specific examples in the present application. The above description of the examples is only used to help understand the method of the present application and its core idea. Meanwhile, for those skilled in the art, according to the idea of the present application, the specific implementation manners and application ranges will be changed. In summary, the content of the present description should not be understood as a limitation of the present application.

Claims

1. A stem-leaf intelligent separation and high-precision measurement device, characterized in that, The application relates to a plant sample processing system, comprising: a clamping module for clamping a sample; a weighing module for obtaining the initial fresh weight of the sample; a first conveying module for conveying the weighed sample to the next link; a visual recognition module for obtaining a three-dimensional point cloud model of the sample conveyed by the first conveying module and recognizing stem and leaf structures; an intelligent separation module in communication connection with the visual recognition module, for planning a cutting path based on the three-dimensional point cloud model and physically separating plant stems and leaves; a stem processing module for weighing and measuring the stem diameter of the stem after leaf separation; a leaf processing module for flattening and measuring the area of the separated leaves; a second conveying module for conveying the leaves and stems to the next link, respectively; a chopping module for chopping the leaves and stems conveyed by the second conveying module, respectively; a drying and weighing module for receiving the chopped leaves and stems, drying and automatically weighing until the stable dry weight is obtained; a central control system in communication connection with the clamping module, the weighing module, the first conveying module, the visual recognition module, the intelligent separation module, the stem processing module, the leaf processing module, the second conveying module, the chopping module and the drying and weighing module, for coordinating the work of the modules and performing data management.

2. The stem-leaf intelligent separation and high-precision measurement device according to claim 1, characterized in that: The visual recognition module comprises: an identification station for receiving the sample conveyed by the first conveying module; a main 3D camera installed above the identification station; a plurality of circumferential auxiliary 3D cameras arranged around the identification station to form a multi-view imaging array; a structured light projector for projecting near-infrared structured light onto the identification station; a ring-shaped LED light supplementing system for providing uniform illumination for imaging; wherein the main 3D camera, the circumferential auxiliary 3D cameras and the structured light projector work cooperatively to construct a high-precision three-dimensional point cloud model of plant stems and leaves.

3. The stem-leaf intelligent separation and high-precision measurement device according to claim 1, characterized in that: The intelligent separation module comprises: a three-axis servo driving system; at least one group of hard alloy blades driven by the three-axis servo driving system; wherein the central control system is configured to generate a plurality of candidate cutting paths based on the three-dimensional point cloud model through a machine learning algorithm, and control the blades to cut according to the selected path.

4. The stem-leaf intelligent separation and high-precision measurement device according to claim 3, characterized in that: There are three groups of hard alloy blades; the three groups of hard alloy blades are respectively driven to move by three sub-three-axis servo driving systems; the three sub-three-axis servo driving systems are all driven by the three-axis servo driving system; the three sub-three-axis servo driving systems serve as three independent actuators, which can process cutting points at different spaces of the stem in parallel to improve efficiency, or cooperatively cut at the same complex part.

5. The stem-leaf intelligent separation and high-precision measurement device according to claim 2, characterized in that: The second conveying module comprises a conveying belt; the identification station is located directly above part of the conveying belt, and the intelligent separation module is used for separating stems and leaves of the sample at the identification station; the leaf processing module comprises micro-protrusion structures arranged in an array on the surface of the conveying belt, a plurality of suction holes opened on the conveying belt, a vacuum suction system, and a leaf area measurement unit; the vacuum suction system generates micro-negative pressure on the surface of the conveying belt through the suction holes on the surface of the conveying belt, and cooperates with the micro-protrusion structures to prevent the leaf from curling during the conveying process; the leaf area measurement unit is arranged on the path of the conveying belt and is used for measuring the leaf area of the flattened leaf.

6. The stem-leaf intelligent separation and high-precision measurement device according to claim 1, characterized in that: The clamping module is arranged at the weighing end of the weighing module, and the weighing module is arranged on a support; the first conveying module can convey the support, the weighing module, and the clamping module clamping the sample into the visual identification module; the stem processing module comprises the weighing module and uses the weighing module to weigh the stem after the leaf separation; the stem processing module further comprises an optical micrometer, which can move up and down and is used for scanning the stem to obtain the stem diameter and transmitting the stem diameter information to the central control system, and the central control system processes the stem diameter information to obtain the maximum stem diameter value and performs data saving and / or output.

7. The stem-leaf intelligent separation and high-precision measurement device according to claim 5, characterized in that: The drying and weighing module comprises a plurality of drying units, each of which is provided with a separate heater and a temperature sensor for realizing independent temperature control; Each of the drying units is provided with an independent micro high-precision weighing sensor for automatically and periodically weighing the sample weight during the drying process; A plurality of containers are arranged, and each of the containers is used for receiving the leaf or stem cut by the cutting module; A six-axis mechanical arm is used for conveying the container containing the sample into the drying unit.

8. The stem-leaf intelligent separation and high-precision measurement device according to claim 7, characterized in that: The cutting module comprises a fixed blade and a rotating blade arranged at the end of the conveying belt, the fixed blade is arranged at the bottom of the rotating blade, and an empty container can be conveyed by the third conveying module to the lower part of the cutting module to receive the leaf or stem cut and dropped from the cutting module.

9. The stem-leaf intelligent separation and high-precision measurement device according to claim 1, characterized in that: The central control system comprises: An upper computer running on the Ubuntu and ROS architecture is responsible for visual processing, path planning algorithm and data management; A lower computer based on the STM32 series MCU is responsible for controlling the motion mechanism, motor and sensor; The upper computer and the lower computer communicate through the EtherCAT bus.

10. A method for separating and measuring stems and leaves by using the stem and leaf intelligent separation and high-precision measurement device according to any one of claims 1-9, characterized in that: The sample is clamped by the clamping module and the initial fresh weight is recorded, and then is sent to the visual identification module for three-dimensional modeling; The intelligent separation module plans a path according to the model and performs stem and leaf separation; The separated leaf enters the leaf processing module, is measured for leaf area after being flattened, is then cut and transferred to the drying and weighing module; The stem left in the stem processing module after separation is weighed for fresh stem weight and measured for stem diameter, and then is cut and transferred to the drying and weighing module; The drying and weighing module dries the stem and leaf sample, determines the dry weight stabilization point through automatic weighing, and records the final dry weight. ​ wherein the fresh weight of the leaf is calculated from the initial fresh weight minus the fresh stem weight. wherein the fresh weight of the leaf is calculated from the initial fresh weight minus the fresh stem weight.

Images