An automatic clamping control system for plant specimens

Abstract

This invention discloses an automatic clamping control system for plant specimens, belonging to the field of automatic control technology. The invention acquires high-resolution color images of plant specimens through a data acquisition module; a model building module constructs a color point cloud model based on the high-resolution color images; a pressure monitoring module acquires real-time pressure distribution data between the clamping actuator and the plant specimen; a data processing module analyzes the color point cloud model and pressure distribution data to obtain a fused feature sequence, and identifies candidate abnormal regions and their types based on multimodal joint analysis; a decision execution module generates differentiated clamping control strategies based on the abnormal region types. This system realizes multi-source perception and collaborative decision-making during the plant specimen clamping process, enabling early identification of pre-damage risks and dynamic optimization of clamping behavior, effectively avoiding damage to plant specimens and significantly improving the safety of digital processing of plant specimens.

Description

Technical Field

[0001] This invention relates to the field of automatic control technology, specifically an automatic clamping and control system for plant specimens. Background Technology

[0002] In the digital processing and automated management of plant specimens, automated clamping is a key step in achieving efficient operation. However, plant specimens are generally characterized by dryness, brittleness, irregular shape, and fragile tissues, making them highly susceptible to irreversible damage using traditional clamping methods. Current automated clamping systems mostly rely on pressure or torque sensors to construct closed-loop control, limiting clamping strength by setting a uniform force threshold. However, this method has significant drawbacks: macroscopic force values ​​cannot reflect local stress states; furthermore, existing vision systems are mainly used for target identification and positioning before clamping, and their image data is not involved in the dynamic control of the clamping process; although high-speed cameras or microscope lenses can capture surface details, there is no technology to use them for real-time monitoring of microscopic deformation in the clamping contact area, and these microscopic changes are often precursors to damage—the "pre-damage stage." Existing control systems typically only respond after a sudden change in force or fracture occurs, at which point the damage is already irreversible; more importantly, force and visual data are disconnected, lacking a fusion analysis mechanism. How to provide an automatic clamping control system that can integrate visual and force information in real time and dynamically monitor and respond to the microscopic deformation of the specimen contact area, so as to achieve non-destructive adaptive clamping of fragile plant specimens, is the problem we need to solve. To this end, we now provide an automatic clamping control system for plant specimens. Summary of the Invention

[0003] The purpose of this invention is to provide an automatic clamping and control system for plant specimens.

[0004] The objective of this invention can be achieved through the following technical solution: an automatic clamping and control system for plant specimens, comprising the following: The data acquisition module is used to acquire high-resolution color images of plant specimens; The model building module is used to build a color point cloud model based on the acquired high-resolution color images; The pressure monitoring module is used to acquire real-time pressure distribution data between the clamping actuator and the plant specimen; The data processing module is used to process the color point cloud model and pressure distribution data to obtain the fused feature sequence, and to perform multimodal joint analysis based on the fused feature sequence; The decision execution module is used to generate corresponding clamping control strategies based on the results of multimodal joint analysis.

[0005] Furthermore, the plant specimen includes a mounting paper for supporting the specimen, a plant specimen fixed on the mounting paper, and an information label corresponding to the plant specimen; Fix the industrial color camera to a rigid bracket that can be adjusted in three dimensions, and adjust the vertical distance between the industrial color camera and the worktable. Embed evenly distributed reference marks on the worktable surface; During the shooting process of the industrial color camera, the worktable remains stationary; An industrial color camera moves along a preset trajectory, acquiring images from multiple angles during the movement to obtain several high-resolution color images taken from different angles.

[0006] Furthermore, the data acquisition module is also used for: When the clamping actuator moves to the planned clamping point and contacts the paper table, record the current position and pose parameters of the gripper in the global spatial coordinate system.

[0007] Furthermore, the process by which the model building module constructs a color point cloud model based on the acquired high-resolution color image includes: Each image is subjected to distortion correction and spatial positioning calibration using reference markers to eliminate lens distortion and camera parameter errors; Feature points are extracted from each distortion-corrected high-resolution color image to obtain a set of feature point matching pairs; Based on the camera parameters and the set of feature point matching pairs, the coordinates of each feature point in three-dimensional space are obtained; Based on the coordinates of each feature point in three-dimensional space, a sparse three-dimensional point cloud is obtained; Based on the sparse 3D point cloud and camera parameters, a dense 3D point cloud model is obtained; The color information of the acquired high-resolution color image is mapped to each point of the dense 3D point cloud model to form a textured color point cloud model, where each point contains spatial coordinates and RGB color values.

[0008] Furthermore, the process by which the pressure monitoring module acquires real-time pressure distribution data between the clamping actuator and the plant specimen includes: The clamping actuator includes a pair of symmetrically arranged mechanical fingers, the inner contact surfaces of which are covered with a flexible thin-film pressure sensor array. The pressure distribution data is acquired through a flexible thin-film pressure sensor array; The flexible thin-film pressure sensor array includes several pressure sensing units, and obtains the position coordinates of the pressure sensing units on the contact surface of the gripper, the effective sensing area, and the pressure value corresponding to the pressure sensing units. Based on the pressure value and effective sensing area of ​​the pressure sensing unit, the normal force on each pressure sensing unit is obtained. The total clamping force, pressure center coordinates, and pressure distribution uniformity index of the clamping actuator are obtained based on the normal force experienced by all pressure sensing units. The pressure distribution uniformity index is obtained based on the pressure values ​​of each pressure sensing unit.

[0009] Furthermore, the data processing module processes the color point cloud model and pressure distribution data to obtain the fused feature sequence, including: Establish a global spatial coordinate system, map the reference marker points to the global spatial coordinate system, and determine the global spatial coordinates of each point in the color point cloud model in the global spatial coordinate system through the reference marker points; Convert the position coordinates of each pressure sensing unit into global spatial coordinates; The corresponding sensing range is set with the global spatial coordinates of each pressure sensing unit as the center, and a subset of local point clouds within the sensing range is obtained. Using the reference plane of the table paper as the reference plane, obtain the directed distances of all points in the local point cloud subset perpendicular to the table paper plane and the thickness value at that location; Eigenvalue decomposition is performed on all points in the local point cloud subset to obtain curvature eigenvalues; Each point in the local point cloud subset is processed to obtain a corresponding two-dimensional grayscale image block; The contrast of a two-dimensional grayscale image patch is used as the texture density feature value; Based on the obtained pressure value, thickness value, curvature feature value, and texture density feature value, a fused feature sequence corresponding to the corresponding pressure sensing unit is constructed.

[0010] Furthermore, the data processing module performs multimodal joint analysis based on the fused feature sequences, including: Candidate abnormal units are determined based on the pressure values ​​of all pressure sensing units; The area covered by spatially adjacent cells in the candidate anomaly cells is defined as the candidate anomaly region; Multimodal joint analysis is performed on the pressure value, thickness value, curvature feature value, and texture density feature value within the candidate anomaly region to determine the anomaly type; The thickness values ​​of each pressure sensing unit in the candidate anomaly region are compared with the preset reference thickness of the paper tray: If the thickness value of each pressure sensing unit is less than or equal to the preset base paper thickness, it is judged as the first type, and the base paper structure is abnormal. When the curvature feature value is greater than the preset curvature threshold and the texture density feature value is less than or equal to the preset texture threshold, it is determined to be a pressure area of ​​paper wrinkles. If the thickness value of each pressure sensing unit is greater than the preset base paper thickness, it is judged as the second type, plant specimen related anomaly; When the curvature feature value is greater than the preset curvature threshold and the texture density feature value is greater than the preset texture threshold, it is determined to be the edge of the plant specimen or the area where the leaf vein is compressed. When the curvature feature value is less than the preset curvature threshold and the texture density feature value is greater than the preset texture threshold, it is determined to be the pressure area of ​​the main body of the plant specimen; When the curvature feature value is less than the preset curvature threshold and the texture density feature value is less than or equal to the preset texture threshold, it is determined to be a label pressure area.

[0011] Furthermore, the process by which the decision execution module generates the corresponding clamping control strategy based on the multimodal joint analysis results includes: If the area is determined to be the main part of the plant specimen that is under pressure, maintain the current clamping state and continuously monitor changes in pressure value; If the area is determined to be the edge of a plant specimen or a leaf vein under pressure, a slight retraction command for the gripper is triggered, and the gripper retracts step by step until the pressure value drops to a preset safe range. If the area is determined to be a wrinkled and pressure-bearing area of ​​the paper, a gripper posture fine-tuning command is triggered to adjust the gripper to make the pressure value evenly distributed. If the area is determined to be under pressure, a local decompression strategy is triggered, allowing the pressure value to be slightly lower than the target clamping force while maintaining overall clamping stability.

[0012] Compared with existing technologies, the beneficial effects of this invention are as follows: This invention acquires high-resolution color images of plant specimens through a data acquisition module; a model building module constructs a color point cloud model based on the high-resolution color images, achieving high-precision digital representation of the geometric morphology and color texture of the plant specimen surface, supporting accurate extraction of spatial features; a pressure monitoring module acquires real-time pressure distribution data between the clamping actuator and the plant specimen, effectively reflecting the local contact state and avoiding the risk of local stress concentration being masked by a normal overall clamping force, thus achieving accurate perception of the clamping contact surface; a data processing module analyzes the color point cloud model and pressure distribution data to obtain a fusion feature sequence. The system identifies candidate abnormal regions and their types based on multimodal joint analysis, significantly improving the accuracy of different regions on plant specimens. The decision execution module generates differentiated clamping control strategies according to the abnormal region types, realizing adaptive response strategies for regions with different materials and structures, balancing the stability of clamping plant specimens with the non-destructive nature of the specimens. The system realizes multi-source perception and collaborative decision-making in the process of clamping plant specimens. It can identify pre-damage risks in advance and dynamically optimize clamping behavior without relying on the judgment of total clamping force exceeding the limit, effectively avoiding hidden or obvious damage to the specimens, and significantly improving the safety and automation level of digital processing of plant specimens. Attached Figure Description

[0013] To more clearly illustrate the technical solutions in the embodiments of this application or the prior art, the drawings used in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments recorded in this invention. For those skilled in the art, other drawings can be obtained based on these drawings.

[0014] Figure 1 This is a schematic diagram of the present invention. Detailed Implementation

[0015] like Figure 1 As shown, an automatic clamping and control system for plant specimens includes a data acquisition module, a model building module, a pressure monitoring module, a data processing module, and a decision execution module. The data acquisition module is used to acquire high-resolution color images of plant specimens; The model building module is used to build a color point cloud model based on the acquired high-resolution color image; The pressure monitoring module is used to acquire real-time pressure distribution data between the clamping actuator and the plant specimen; The data processing module is used to process the color point cloud model and pressure distribution data to obtain a fused feature sequence, and to perform multimodal joint analysis based on the fused feature sequence; The decision execution module is used to generate corresponding clamping control strategies based on the results of multimodal joint analysis.

[0016] It should be further explained that, in the specific implementation process, the process of the data acquisition module acquiring high-resolution color images of plant specimens includes: In the actual work scenario of specimen digitization, the object to be processed is a plant specimen that has been pressed and dried. The plant specimen usually includes a mounting paper for the specimen, the plant specimen fixed on the mounting paper, and information tags corresponding to the plant specimen. The mounting paper refers to a standard paper base used to support and fix plant specimens, and is attached with information labels containing the corresponding plant specimens; An industrial color camera combined with a multi-angle shooting method is used to achieve multi-angle image acquisition; An industrial color camera is fixed on a rigid bracket that can be adjusted in three dimensions. The rigid bracket has three-axis fine-tuning knobs for X, Y, and Z axes and an angle pitch adjustment mechanism. Adjust the vertical distance between the industrial color camera and the worktable to ensure that a single image can completely cover the entire worktable paper, and the image resolution reaches 0.3mm-0.5mm, which is sufficient to distinguish the boundaries of the worktable paper, information labels, the direction of the plant stem and the outline of the leaves. The workbench is used to place plant specimens and is made of a low-reflectivity material to create a moderate contrast with the white table paper, so as to avoid affecting the image acquisition quality due to reflection or color interference. The workbench surface is embedded with evenly distributed reference markers. The reference markers are a black circular array with a diameter of 3mm and a spacing of 50mm, distributed in a grid pattern, which serves as input data for 3D modeling. During the shooting process with the industrial color camera, the worktable remains stationary, the plant specimens are fixed on the table paper and do not move, and the light source uses multi-angle LED lighting arranged around the worktable to ensure uniform illumination of the specimen surface when shooting from all angles, avoiding the loss of plant specimen details due to lighting from a single direction. A rigid bracket drives an industrial color camera to move around the workbench along a preset trajectory to take pictures of plant specimens from multiple angles. When each image is acquired, the spatial position coordinates, rotation angle, camera parameters and shooting timestamp of the current industrial color camera are recorded, and the coordinates of the reference marker points visible in the image are also recorded. The images are high-resolution color images acquired from different angles. After completing the multi-angle image acquisition, a set of high-resolution color images taken from different angles is obtained. Each high-resolution color image contains a table paper, plant specimens, information labels, and reference markers evenly distributed on the workbench surface. When the gripping actuator moves to the planned gripping point and contacts the paper table, the current position parameters of the gripper in the global spatial coordinate system are recorded. These position parameters include the coordinates of the gripper's center position. , , and gripper posture angle ( , , ).

[0017] It should be further explained that, in the specific implementation process, the model building module constructs a color point cloud model based on the acquired high-resolution color image, including: Receives high-resolution color images from the data acquisition unit, camera parameters from the industrial color camera, and reference markers; Each image is subjected to distortion correction and spatial positioning calibration using reference markers to eliminate lens distortion and camera parameter errors; It should be noted that performing distortion correction and spatial position calibration on each image to eliminate lens distortion and camera parameter errors is a common technique used by those skilled in the art, and will not be elaborated here. Feature points of scale-invariant feature transformation are extracted from each distortion-corrected high-resolution color image. The number of feature points is controlled at 2000-3000 per image to ensure coverage of plant specimens. The correspondence between feature points in the images is established through the nearest neighbor matching algorithm to generate a set of feature point matching pairs. Based on camera parameters and a set of matching feature points, the coordinates of each feature point in three-dimensional space are calculated, and the camera parameters are optimized to generate a sparse three-dimensional point cloud. The sparse three-dimensional point cloud contains the main structural feature points of the plant specimen, including the tabletop boundary, information tags, the orientation of the main stem of the plant, and the leaf outline. Based on sparse 3D point cloud and camera parameters, a multi-view stereo vision algorithm is used for dense matching. The 3D coordinates of each pixel are calculated to generate a dense 3D point cloud model. The color information of the acquired high-resolution color image is mapped to each point of the dense 3D point cloud model to form a textured color point cloud model, where each point contains (X, Y, Z) spatial coordinates and RGB color values.

[0018] It should be further explained that, in the specific implementation process, the process by which the pressure monitoring module acquires real-time pressure distribution data between the clamping actuator and the plant specimen includes: The clamping actuator includes a pair of symmetrically arranged mechanical fingers, the inner contact surfaces of which are covered with a flexible thin-film pressure sensor array. The flexible thin-film pressure sensor array integrates multiple pressure sensing units. Each pressure sensing unit has an effective sensing area of ​​2mm×2mm, a center-to-center distance of 2.5mm between adjacent units, and an overall coverage area of ​​25mm×30mm, which can completely cover the edge area of ​​the paper table under the standard clamping width. The pressure distribution data is acquired through a flexible thin-film pressure sensor array deployed on the gripper. This array contains n pressure sensing units, each labeled as i, where i = 1, 2, ..., n. The coordinates of the i-th pressure sensing unit on the gripper contact surface are denoted as (...). , The corresponding pressure value detected is recorded as follows: ; Let the effective sensing area of ​​the i-th pressure sensing unit be denoted as . The normal force generated by the pressure sensing unit is denoted as ,Right now: ; The total clamping force is obtained by summing the normal forces acting on all pressure sensing units, and is denoted as . ,Right now: ; The coordinates of the pressure center are obtained by calculating the weighted average of the normal forces acting on each pressure sensing unit, i.e.: ; in, The x-coordinate represents the center of pressure. Represents the vertical coordinate of the pressure center; The pressure distribution uniformity index is obtained by dividing the standard deviation of the pressure values ​​of each pressure sensing unit by the arithmetic mean of the pressure values, and is denoted as . ,Right now: ; in, This represents the arithmetic mean of the pressure values ​​from all pressure sensing units. This represents the standard deviation of the pressure values ​​of all pressure sensing units.

[0019] It should be further explained that, in the specific implementation process, the data processing module processes the color point cloud model and pressure distribution data to obtain the fused feature sequence, including: Establish a global spatial coordinate system, map the reference marker points to the global spatial coordinate system, and determine the global spatial coordinates of each point in the color point cloud model in the global spatial coordinate system through the reference marker points; The position coordinates of each pressure sensing unit ( , Convert to global space coordinates ( , , This achieves a spatial correspondence between pressure distribution data and a color point cloud model, with the following conversion relationship: ; in, The rotation matrix represents the gripper's posture; Centered on the global spatial coordinates of each pressure sensing unit, a local point cloud subset with a radius of 2 mm is extracted from the color point cloud model. Each local point cloud subset corresponds to the sensing range of a pressure sensing unit. Using the reference plane of the paper substrate as the reference surface, calculate the directed distances from all points in the local point cloud subset to the paper substrate plane, and take the average value as the thickness value at that location, denoted as . ; Eigenvalue decomposition is performed on all points in the local point cloud subset to obtain curvature eigenvalues, denoted as . ,Right now: ; in, The largest eigenvalue, For intermediate eigenvalues, For the smallest eigenvalue, the relation satisfies ; When the curvature characteristic value is closer to 0, it indicates that the local surface is flatter; when the curvature characteristic value is larger, it indicates that the local surface is more undulating. Perform grayscale processing on each point in the local point cloud subset to obtain the corresponding two-dimensional grayscale image block; A gray-level co-occurrence matrix is ​​constructed based on two-dimensional gray-level image patches, and the contrast of the two-dimensional gray-level image patches is calculated as the texture density feature value, denoted as . ,Right now: ; Where N is the total number of grayscale levels, m is the row index of the grayscale co-occurrence matrix, and n is the column index of the grayscale co-occurrence matrix. This is the joint probability distribution of pixel pairs with gray levels m and n under a specified spatial relationship. The larger the value of this texture density feature, the more complex and detailed the local surface texture is, which is used to distinguish different material areas such as table paper, plant specimens and information labels; A fused feature sequence is constructed based on the pressure value, thickness value, curvature feature value, and texture density feature value of the i-th pressure sensing unit, denoted as . ,Right now: .

[0020] It should be further explained that, in the specific implementation process, the data processing module performs multimodal joint analysis based on the fused feature sequences, including: Candidate abnormal units are determined based on the pressure values ​​of all pressure sensing units; Specifically: Condition 1: The pressure value is greater than 1.5 times the arithmetic mean of the pressure values ​​of all pressure sensing units, that is: ; Condition 2: The pressure difference between the pressure value of the pressure sensing unit and the pressure of the adjacent pressure sensing unit exceeds the preset pressure gradient threshold. The area covered by spatially adjacent cells in the candidate anomaly cells is defined as the candidate anomaly region; Multimodal joint analysis is performed on the pressure value, thickness value, curvature feature value, and texture density feature value within the candidate anomaly region to determine the anomaly type; The thickness values ​​of each pressure sensing unit in the candidate anomaly region are compared with the preset reference thickness of the paper tray: If the thickness value of each pressure sensing unit is less than or equal to the preset base paper thickness, it is judged as the first type, and the base paper structure is abnormal. When the curvature feature value is greater than the preset curvature threshold and the texture density feature value is less than or equal to the preset texture threshold, it is determined to be a pressure area of ​​paper wrinkles. If the thickness value of each pressure sensing unit is greater than the preset base paper thickness, it is judged as the second type, plant specimen related anomaly; When the curvature feature value is greater than the preset curvature threshold and the texture density feature value is greater than the preset texture threshold, it is determined to be the edge of the plant specimen or the area where the leaf vein is compressed. When the curvature feature value is less than the preset curvature threshold and the texture density feature value is greater than the preset texture threshold, it is determined to be the pressure area of ​​the main body of the plant specimen; When the curvature feature value is less than the preset curvature threshold and the texture density feature value is less than or equal to the preset texture threshold, it is determined to be a label pressure area.

[0021] It should be further explained that, in the specific implementation process, the process by which the decision execution module generates the corresponding clamping control strategy based on the multimodal joint analysis results includes: If the area is determined to be the main part of the plant specimen that is under pressure, maintain the current clamping state and continuously monitor changes in pressure value; If the area is determined to be the edge of a plant specimen or a leaf vein under pressure, a slight retraction command for the gripper is triggered, and the gripper retracts step by step until the pressure value drops to a preset safe range. If the area is determined to be a wrinkled and pressure-bearing area of ​​the paper, a gripper posture fine-tuning command is triggered to adjust the gripper to make the pressure value evenly distributed. If the area is determined to be under pressure, a local decompression strategy is triggered, allowing the pressure value to be slightly lower than the target clamping force while maintaining overall clamping stability.

[0022] The above description is merely a preferred embodiment of the present invention and is not intended to limit the present invention in any way. Although the present invention has been disclosed above with reference to preferred embodiments, it is not intended to limit the present invention. Any person skilled in the art can make some modifications or alterations to the above-disclosed technical content to create equivalent embodiments without departing from the scope of the present invention. Any modifications or equivalent substitutions made to the above embodiments based on the technical essence of the present invention without departing from the scope of the present invention shall still fall within the scope of the present invention.

Claims

1. An automatic clamping and control system for plant specimens, characterized in that, include: The data acquisition module is used to acquire high-resolution color images of plant specimens; The model building module is used to build a color point cloud model based on the acquired high-resolution color images; The pressure monitoring module is used to acquire real-time pressure distribution data between the clamping actuator and the plant specimen; The data processing module is used to process the color point cloud model and pressure distribution data to obtain the fused feature sequence, and to perform multimodal joint analysis based on the fused feature sequence; The decision execution module is used to generate corresponding clamping control strategies based on the results of multimodal joint analysis. The process by which the pressure monitoring module acquires real-time pressure distribution data between the clamping actuator and the plant specimen includes: The clamping actuator includes a pair of symmetrically arranged mechanical fingers, the inner contact surfaces of which are covered with a flexible thin-film pressure sensor array. The pressure distribution data is acquired through a flexible thin-film pressure sensor array; The flexible thin-film pressure sensor array includes several pressure sensing units, and obtains the position coordinates of the pressure sensing units on the contact surface of the gripper, the effective sensing area, and the pressure value corresponding to the pressure sensing units. Based on the pressure value and effective sensing area of ​​the pressure sensing unit, the normal force on each pressure sensing unit is obtained. The total clamping force, pressure center coordinates, and pressure distribution uniformity index of the clamping actuator are obtained based on the normal force experienced by all pressure sensing units. The pressure distribution uniformity index is obtained based on the pressure values ​​of each pressure sensing unit. The data processing module processes the color point cloud model and pressure distribution data to obtain the fused feature sequence, including: Establish a global spatial coordinate system, map the reference marker points to the global spatial coordinate system, and determine the global spatial coordinates of each point in the color point cloud model in the global spatial coordinate system through the reference marker points; Convert the position coordinates of each pressure sensing unit into global spatial coordinates; The corresponding sensing range is set with the global spatial coordinates of each pressure sensing unit as the center, and a subset of local point clouds within the sensing range is obtained. Using the reference plane of the table paper as the reference plane, obtain the directed distances of all points in the local point cloud subset perpendicular to the table paper plane and the thickness value at that location; Eigenvalue decomposition is performed on all points in the local point cloud subset to obtain curvature eigenvalues; Each point in the local point cloud subset is processed to obtain a corresponding two-dimensional grayscale image block; The contrast of a two-dimensional grayscale image patch is used as the texture density feature value; Based on the obtained pressure value, thickness value, curvature feature value, and texture density feature value, a fusion feature sequence corresponding to the corresponding pressure sensing unit is constructed.

2. The automatic clamping and control system for plant specimens according to claim 1, characterized in that, The plant specimen includes a mounting paper for the specimen, the plant specimen fixed on the mounting paper, and information tags corresponding to the plant specimen; Fix the industrial color camera to a rigid bracket that can be adjusted in three dimensions, and adjust the vertical distance between the industrial color camera and the worktable. Embed evenly distributed reference marks on the worktable surface; During the shooting process of the industrial color camera, the worktable remains stationary; An industrial color camera moves along a preset trajectory, acquiring images from multiple angles during the movement to obtain several high-resolution color images taken from different angles.

3. The automatic clamping and control system for plant specimens according to claim 1, characterized in that, The data acquisition module is also used for: When the clamping actuator moves to the planned clamping point and contacts the paper table, record the current position and pose parameters of the gripper in the global spatial coordinate system.

4. An automatic clamping and control system for plant specimens according to claim 2, characterized in that, The model building module constructs a color point cloud model based on the acquired high-resolution color images, including the following steps: Each image is subjected to distortion correction and spatial positioning calibration using reference markers to eliminate lens distortion and camera parameter errors; Feature points are extracted from each distortion-corrected high-resolution color image to obtain a set of feature point matching pairs; Based on the camera parameters and the set of feature point matching pairs, the coordinates of each feature point in three-dimensional space are obtained; Based on the coordinates of each feature point in three-dimensional space, a sparse three-dimensional point cloud is obtained; Based on the sparse 3D point cloud and camera parameters, a dense 3D point cloud model is obtained; The color information of the acquired high-resolution color image is mapped to each point of the dense 3D point cloud model to form a textured color point cloud model, where each point contains spatial coordinates and RGB color values.

5. An automatic clamping and control system for plant specimens according to claim 1, characterized in that, The data processing module performs multimodal joint analysis based on the fused feature sequences, including: Candidate abnormal units are determined based on the pressure values ​​of all pressure sensing units; The area covered by spatially adjacent cells in the candidate anomaly cells is defined as the candidate anomaly region; Multimodal joint analysis is performed on the pressure value, thickness value, curvature feature value, and texture density feature value within the candidate anomaly region to determine the anomaly type; The thickness values ​​of each pressure sensing unit in the candidate anomaly region are compared with the preset reference thickness of the paper tray: If the thickness value of each pressure sensing unit is less than or equal to the preset base paper thickness, it is judged as the first type, and the base paper structure is abnormal. When the curvature feature value is greater than the preset curvature threshold and the texture density feature value is less than or equal to the preset texture threshold, it is determined to be a pressure area of ​​paper wrinkles. If the thickness value of each pressure sensing unit is greater than the preset base paper thickness, it is judged as the second type, plant specimen related anomaly; When the curvature feature value is greater than the preset curvature threshold and the texture density feature value is greater than the preset texture threshold, it is determined to be the edge of the plant specimen or the area where the leaf vein is compressed. When the curvature feature value is less than the preset curvature threshold and the texture density feature value is greater than the preset texture threshold, it is determined to be the pressure area of ​​the main body of the plant specimen. When the curvature feature value is less than the preset curvature threshold and the texture density feature value is less than or equal to the preset texture threshold, it is determined to be a label pressure area.

6. An automatic clamping and control system for plant specimens according to claim 5, characterized in that, The process by which the decision execution module generates the corresponding clamping control strategy based on the multimodal joint analysis results includes: If the area is determined to be the main part of the plant specimen under pressure, maintain the current clamping state and continuously monitor changes in pressure value; If the area is determined to be the edge of a plant specimen or a leaf vein under pressure, a slight retraction command for the gripper is triggered, and the gripper retracts step by step until the pressure value drops to a preset safe range. If the area is determined to be a wrinkled and pressure-bearing area of ​​the paper, a gripper posture fine-tuning command is triggered to adjust the gripper to make the pressure value evenly distributed. If the area is determined to be under pressure, a local decompression strategy is triggered, allowing the pressure value to be slightly lower than the target clamping force while maintaining overall clamping stability.

Images