A method and system for detecting the position of a servo gripper

Abstract

The application provides a kind of servo gripper position detection method and system, it is related to automated production line technical field, and each part is set with a part grabbing and placing element robot on the both sides of production line, and it is relatively arranged, there is servo gripper and end vision subsystem on each robot, there is positioning pin for positioning on servo gripper, comprising: obtaining the position information of each positioning pin of debugging qualified as template position information;Part grabbing and placing element robot passes through vision subsystem and collects the current position information of opposite positioning pin according to preset time interval;The current position information of positioning pin is compared with the template position information of the positioning pin, and the comparison result is obtained;If the comparison result exceeds the first threshold range, then determine that the current positioning pin position is abnormal.The application detects the position error of positioning pin by periodically online using the vision subsystem at the end of opposite part grabbing and placing element robot, realizes the closed-loop detection of online servo gripper positioning position, and improves production process control ability.

Description

Technical Field

[0001] This invention relates to the field of automated production line technology, and in particular to a method and system for detecting the position of a servo gripper. Background Technology

[0002] The positioning function of a servo gripper refers to its ability to accurately determine its own position and orientation during automated production processes in order to precisely grasp, place, or manipulate target objects. Servo grippers are widely used in many fields of industrial production. The positioning function of existing servo grippers is generally guaranteed by the position of the positioning pin. The servo gripper is usually directly installed at the end of the servo sliding mechanism. The positioning pin is connected to the servo sliding mechanism through a mechanical mechanism. The positional accuracy measurement of the positioning pin often requires offline coordinate measuring equipment to model, repeatedly measure and adjust to meet the positioning requirements (measurement error less than 0.15mm) in order to confirm the positioning position of the servo sliding axis. This positioning position is controlled by servo closed loop.

[0003] However, after long-term operation, the wear of the gears in the servo mechanism transmission leads to a decrease in rigidity, resulting in a decrease in repeatability and absolute position accuracy. Or, due to slight deformation caused by the pinning of parts, it is not easy to detect that the actual position has changed, resulting in a decrease in the batch size qualification rate. Often, it is necessary to use a CMM (Coordinate Measuring Machine) to spot check and find the problem before the servo gripper's positioning position is remeasured offline. Offline measurement often causes line stoppages and production capacity losses, as well as a lot of rework costs.

[0004] Therefore, there is an urgent need to provide a method for detecting the position of servo grippers, which can not only enable timely detection of the positioning accuracy of servo grippers, but also avoid a decrease in the pass rate of parts and rework, thereby improving the control capability of the production process. Summary of the Invention

[0005] To address the aforementioned issues, this invention provides a method and system for detecting the position of a servo gripper. By utilizing the position measurement function of the same-station gripping vision subsystem, the vision subsystems of the end effector of the opposite-side gripper robot that grips and places the same part are periodically used online to mutually detect the position error of the positioning pin of the opposite-side servo gripper, thereby achieving closed-loop detection of the online servo gripper positioning position and improving the production process control capability.

[0006] To achieve the above objectives, the present invention provides a method for detecting the position of a servo gripper. A part-grabbing robot is positioned on each side of a production line, with the two robots facing each other. Each robot is equipped with a servo gripper and a vision subsystem. The servo gripper is equipped with a positioning pin for positioning. The detection method includes: Obtain the position information of each locating pin that has passed the debugging process, and use it as template position information; The parts gripping and placing robot collects the current position information of the opposite positioning pin through the vision subsystem at preset time intervals; The current position information of the positioning pin is compared with the template position information of the positioning pin to obtain the comparison result; If the comparison result exceeds the first threshold range, the current positioning pin position is determined to be abnormal.

[0007] As a further improvement of the present invention, when it is determined that the position of the positioning pin is abnormal, the production line is stopped and the position of the positioning pin is adjusted according to the comparison results.

[0008] As a further improvement of the present invention, adjusting the position of the positioning pin according to the comparison result includes: Based on the comparison results, a positioning pin position compensation command is generated, including the target position; Send the target position compensation command to the servo system; The servo system drives the motor to move the positioning pin along the guide rail or lead screw according to the target position, thereby achieving positioning pin position compensation.

[0009] As a further improvement of the present invention Obtain the circular base outline of each locating pin that has passed the debugging process, and calculate the standard center point position data of the circular base outline as the template position information; The parts gripping and placing robot acquires the current circular base outline of the opposite positioning pin through the vision subsystem at preset time intervals, calculates the current center point position data of the circular base outline, and uses it as the current position information. Compare the current center point location data with the standard center point location data to obtain the comparison results.

[0010] As a further improvement of the present invention, the processes of calculating the standard center point position data and calculating the current center point position data both include: Complete point cloud data of the opposite positioning pin is acquired through the vision subsystem; The point cloud is subjected to downsampling filtering, noise reduction filtering, and region segmentation to extract the point cloud clusters of the circular base. Calculate the best-fit plane and normal vector of the point cloud; The point cloud is projected onto the fitting plane, and the coordinates of the two-dimensional circle center and radius are fitted using the least squares method. By back-projecting the two-dimensional center of the circle onto 3D space, the three-dimensional coordinates of the standard center point or the current center point can be obtained. Output the 3D coordinates and verify the radius deviation.

[0011] As a further improvement of the present invention, obtaining the comparison results includes: Based on the current center point coordinates (x1, y1, z1) and the standard center point coordinates (x0, y0, z0), calculate the positional deviation (Δx, Δy, Δz) and the comprehensive displacement deviation Δd between them, as the comparison result. The formula for the comprehensive displacement deviation Δd is as follows: Δx = x1 - x0 Δy = y1 - y0 Δz = z1 - z0 The first threshold range is a preset comprehensive displacement threshold range. If the comprehensive displacement deviation Δd is greater than the maximum value of the preset comprehensive displacement threshold range, that is, the comparison result exceeds the first threshold range, the current positioning pin position is determined to be abnormal.

[0012] As a further improvement of the present invention, the part gripping and placing robot acquires a complete image of the qualified opposite side positioning pin through the vision subsystem; The complete image is preprocessed to retain the appearance information of the circular base area, which is used as the template position information; The part gripping and placing robot uses the vision subsystem to acquire complete images of the current position of the opposite positioning pin at preset time intervals; The complete image of the current location is preprocessed to retain the current appearance information of the circular base area, which is used as the current location information. The template feature vector, which extracts template position information, and the current feature vector, which extracts current position information, are extracted using a lightweight convolutional neural network. The cosine similarity between the template feature vector and the current feature vector is then calculated as the comparison result. The first threshold range is a similarity threshold range. If the cosine similarity is less than the minimum value of the similarity threshold range, it is determined that the comparison result exceeds the first threshold range and the current positioning pin position is abnormal.

[0013] As a further improvement of the present invention, if the comparison result exceeds the first threshold range, the current positioning pin position is determined to be abnormal, including: Two part-grabbing and placing robots on opposite sides of the production line periodically and simultaneously collect the position information of a fixed reference target, which is set between the two part-grabbing and placing robots. If the comparison result between the current position information of the opposite positioning pin collected by one of the part gripping and placing robots and the template position information of the positioning pin exceeds the first threshold range, and the deviation between the position information of the fixed reference target collected by the part gripping and placing robot and the actual position information of the fixed reference target is greater than the second threshold, then the vision subsystem of the part gripping and placing robot is determined to be abnormal. If the comparison result between the current position information of the opposite positioning pin collected by one of the part gripping and placing robots and the template position information of the positioning pin exceeds the first threshold range, and the deviation between the position information of the fixed reference target collected by the part gripping and placing robot and the actual position information of the fixed reference target is less than the second threshold, then the current positioning pin position is determined to be abnormal.

[0014] As a further improvement of the present invention, the detection method further includes: During the operation of the servo gripper, the comparison results for each positioning pin are acquired; Based on the comparison results, obtain the curve of the change of the comparison results of each positioning pin over time or the number of working cycles; Based on the change curve, predict the time or number of working cycles for each positioning pin to reach the alarm threshold based on the comparison result. Before the predicted time or number of working cycles is reached, generate a positioning pin position compensation command.

[0015] The present invention also provides a servo gripper position detection system, characterized in that: for the parts gripping and placing robots arranged on both sides of the production line, each parts gripping and placing robot has a gripping and placing actuator installed at the end of its robotic arm, and one or more servo grippers are installed on the gripping and placing actuator, each servo gripper is connected to a positioning pin, and each parts gripping and placing robot also has a vision subsystem installed at the end of its robotic arm. The visual range of the vision subsystem extends outward from the end of the robotic arm and covers the servo gripper and positioning pin of the part gripping and placing robot on the opposite side. The vision subsystem is used to collect template position information and current position information of each positioning pin; The vision subsystem is connected to a detection result determination subsystem, which is used to compare the current position information of the positioning pin with the template position information of the positioning pin, obtain the comparison result, and determine that the current positioning pin position is abnormal when the comparison result exceeds the first threshold range.

[0016] Compared with the prior art, the beneficial effects of the present invention are as follows: This invention utilizes the vision subsystem at the end effector of an existing part-grabbing robot on the production line for position detection of servo gripper positioning pins, eliminating the need for additional detection devices and reducing detection costs. This vision subsystem can periodically measure the position of opposite positioning pins at the same workstation online. Compared to existing technologies, it transforms offline measurement into online measurement, greatly improving measurement efficiency and avoiding the impact of offline measurement on production line capacity.

[0017] This invention enables online measurement of the positioning pins of a servo gripper, while avoiding positioning dimension deviations in batches of parts caused by pin size offsets, thus greatly reducing rework costs.

[0018] This invention achieves closed-loop detection of the online servo gripper's positioning position by acquiring the position data of the positioning pin through the end vision subsystem of the gripper robot.

[0019] This invention provides rich and searchable historical positioning data, dynamically tracks changes in position accuracy, and can promptly alarm and stop the machine when the positioning pin error reaches the alarm threshold, greatly reducing repair costs.

[0020] This invention breaks through the technical bottleneck of traditional servo grippers requiring offline measurement for positioning, expands the functional application range of vision subsystems on the production line, enables multiple uses with one machine, and provides macroscopically controllable positioning data, allowing for preventative maintenance and reducing downtime. Attached Figure Description

[0021] Figure 1 This is a schematic diagram of a part-grabbing and placing robot on a production line according to an embodiment of the present invention; Figure 2 This is a flowchart of a servo gripper position detection method disclosed in one embodiment of the present invention.

[0022] Explanation of reference numerals in the attached figures: 1. Parts gripping and placing robot; 2. Positioning pin; 3. Vision subsystem; 4. Gripping and placing actuator. Detailed Implementation

[0023] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0024] The present invention will now be described in further detail with reference to the accompanying drawings: like Figure 1As shown, a servo gripper position detection system includes: a part gripping robot 1 on each side of the production line, the part gripping robots 1 on both sides are arranged opposite to each other, a gripping actuator 4 is installed at the end of the robotic arm of each part gripping robot 1, one or more servo grippers are installed on the gripping actuator 4, the servo sliding mechanism of each servo gripper is connected to a positioning pin 2, and a vision subsystem 3 (including an industrial camera, a light source, and an image processing module) is also installed at the end of the robotic arm of each part gripping robot 1 via a bracket. The system also includes a historical database module and a predictive analysis module. The vision range of the vision subsystem 3 extends outward from the end of the robotic arm and covers the servo gripper and positioning pin 2 of the gripping mechanism of the part gripping robot 1 on the opposite side.

[0025] The vision subsystem 3 is used to visually guide the gripper 4 to grasp the part and transport it to the target position, as well as to capture complete images of the positioning pins 2 of each servo gripper on the opposite side. By analyzing the complete images, the position information of the positioning pins 2 is obtained, and online servo gripper position detection is realized.

[0026] The historical database module is used to continuously record the positional deviation of each positioning pin 2; The predictive analysis module is used to plot the change curves of each positioning pin 2 over time or the number of working cycles. Based on the change curves, it predicts when the deviation of the positioning pin 2 will reach the alarm threshold and performs maintenance in advance to avoid unplanned downtime.

[0027] In some embodiments, the vision subsystem 3 is connected to a detection result determination subsystem, which includes a data processing module, a feature extraction module, and a determination module. The data processing module preprocesses the complete image of the positioning pin 2 captured by the vision subsystem 3; The feature extraction module uses a lightweight convolutional neural network. After processing multiple images of the circular base of the positioning pin 2 that have passed the debugging, the images are input into the lightweight convolutional neural network to extract the high-dimensional latent feature vector of each image as the template feature vector. The image of the circular base of the positioning pin 2 that is currently being acquired is preprocessed and then input into a lightweight convolutional neural network to extract the current feature vector; The determination module calculates the cosine similarity between the current feature vector and the template feature vector, and determines whether the position of the servo gripper is abnormal based on the cosine similarity and the similarity threshold.

[0028] In some alternative embodiments, the detection result determination subsystem includes a data acquisition module and a data comparison module; The data acquisition module is used for: Obtain the circular base outline of each locating pin that has passed the debugging process, and calculate the standard center point position data of the circular base outline as the template position information; The parts gripping and placing robot acquires the current circular base outline of the opposite positioning pin through the vision subsystem at preset time intervals, calculates the current center point position data of the circular base outline, and uses it as the current position information. The processes for calculating the standard center point position data and the current center point position data both include: acquiring complete point cloud data of the opposite positioning pin through the vision subsystem; performing downsampling filtering, noise reduction filtering, and region segmentation on the point cloud to extract the point cloud clusters of the circular base; calculating the best fitting plane and normal vector of the point cloud; projecting the point cloud onto the fitting plane and fitting the two-dimensional circle center coordinates and radius using the least squares method; backprojecting the two-dimensional circle center onto 3D space to obtain the three-dimensional coordinates of the standard center point or the current center point; outputting the three-dimensional coordinates and verifying the radius deviation.

[0029] The data comparison module is used for: Based on the current center point coordinates (x1, y1, z1) and the standard center point coordinates (x0, y0, z0), calculate the positional deviation (Δx, Δy, Δz) and the comprehensive displacement deviation Δd between them, as the comparison result. The formula for the comprehensive displacement deviation Δd is as follows: Δx=x1-x0 Δy=y1-y0 Δz=z1-z0 The first threshold range is a preset comprehensive displacement threshold range. If the comprehensive displacement deviation Δd is greater than the maximum value of the preset comprehensive displacement threshold range, the comparison result is determined to exceed the first threshold range, and the current positioning pin position is abnormal.

[0030] Furthermore, the detection result determination subsystem is also connected to an alarm subsystem and a shutdown adjustment subsystem; The detection result determination subsystem sends alarm signals and shutdown adjustment signals to the alarm subsystem and shutdown adjustment subsystem, respectively. The shutdown adjustment subsystem generates a position compensation command for positioning pin 2 and sends it to the servo system. The servo motor of the servo system drives positioning pin 2 to move along the guide rail or lead screw, thus completing the position adjustment of positioning pin 2.

[0031] Specifically, the detection result determination subsystem is linked to the production line emergency stop function through the PLC control system. When the deviation exceeds the threshold, the PLC control system sends a stop signal to the stop adjustment subsystem.

[0032] This invention overcomes the technical bottlenecks of traditional offline measurement, which suffers from low efficiency, high cost, and slow response. By reusing existing vision resources, it achieves online closed-loop detection of the positioning accuracy of the servo gripper, expands the functional application range of the vision subsystem 3 (multi-purpose), and significantly improves the reliability and economy of the production line through real-time data monitoring and early warning mechanisms.

[0033] like Figure 2 As shown, the present invention provides a method for detecting the position of a servo gripper. A part-gripping robot 1 is located on each side of a production line, with the two robots facing each other. Each robot is equipped with a servo gripper and a vision subsystem 3. The servo gripper is equipped with a positioning pin 2 for positioning. The detection method includes: S1. Obtain the position information of each qualified positioning pin 2 as template position information; In one implementation, the 3D point cloud contour of the circular base of each qualified positioning pin 2 is obtained, and the standard center point position data is fitted as template position information. Furthermore, the positioning accuracy of the servo gripper can be verified through offline CMM. When the positioning accuracy meets the requirements (e.g., the positioning deviation does not exceed ±0.15), the servo gripper and the corresponding positioning pin are considered to be successfully debugged. The vision subsystem 3 (which can be a 3D camera) at the end of any part gripping and placing robot 1 acquires an image of the positioning pin 2 of the opposite servo gripper. The contour features of the circular base contour of the positioning pin 2 are extracted by image processing algorithms (such as 3D point cloud best fitting algorithms), the spatial coordinates of its standard center point are calculated, and it is used as the reference template position data and stored in the control system (such as PLC control system or host computer).

[0034] In another embodiment, the parts gripping and placing robot 1 acquires a complete image of the properly tested and debugged opposite positioning pin 2 through the vision subsystem 3, ensuring that the image covers the entire circular base area and is clearly imaged. The image processing algorithm preprocesses the complete image, preserving the appearance information of the circular base area as template position information. Specifically, this includes: converting the acquired color image to a grayscale image to reduce the amount of data and simplify processing complexity; using a Gaussian filter to smooth the image, effectively suppressing random noise while preserving edge information as much as possible; and expanding the image's grayscale dynamic range through methods such as histogram equalization or adaptive grayscale transformation to improve the contrast between the circular base of the positioning pin 2 and the surrounding environment, making the target outline clearer. The process involves: converting the grayscale image into a binary image using Otsu's method or an adaptive threshold segmentation algorithm to separate the circular base region from the background; performing morphological opening operations (erosion followed by dilation) on the binary image to remove small noise points or burrs, followed by closing operations (dilation followed by erosion) to fill small holes in the target area and smooth the contour edges; and cropping the minimum bounding rectangle region containing the complete circular base based on the expected approximate position of the circular base of the positioning pin 2 or through preliminary contour detection results. The center point position data of this minimum bounding rectangle region is calculated, which serves as the standard center point position data and is used as template position information to reduce the amount of data and computation in subsequent processing, thereby improving processing speed and accuracy.

[0035] S2. Online periodic inspection: The part gripping and placing robot 1 on this side collects the current position information of the positioning pin 2 on the opposite side through the vision subsystem 3 at preset time intervals; In some embodiments, the trigger frequency for online periodic inspection is every N parts processed (N is a positive integer), or the preset time interval for online periodic inspection is every T hours, where T is the time length. The preset time interval is determined according to the production cycle, the design life of the servo mechanism, and the accuracy requirements of the parts. In typical scenarios, it is every 5-20 parts processed or every 0.5-2 hours. When the vibration is large or the accuracy requirement of the parts is higher than ±0.1mm, it is shortened to every 5 parts processed or every 0.5 hours, preferably every 10 parts processed or every 1 hour.

[0036] The parts gripping and placing robot 1 acquires the current circular base outline of the opposite side positioning pin 2 online, calculates the current center point position data of the circular base outline, and uses it as the current position information; Furthermore, based on the same image processing algorithm as in step S1, the center point coordinates of the circular bottom contour of the current positioning pin 2 are extracted in real time; In other embodiments, the process of directly acquiring standard center point position data and current center point position data (three-dimensional coordinates) using 3D point cloud processing technology includes: Step 1: Point Cloud Data Acquisition The parts gripping and placing robot acquires complete point cloud data of the opposite positioning pin through its end-effector vision subsystem (3D camera, such as structured light or LiDAR); it obtains the standard point cloud corresponding to the standard center point position data. After the positioning pin is properly adjusted (positioning deviation ≤ 0.15mm), it acquires the point cloud data of the circular base of the opposite positioning pin, ensuring coverage of the entire base area, and the point cloud density must meet the accuracy requirements (e.g., point spacing ≤ 0.1mm); it acquires the current point cloud corresponding to the current center point position data: the current point cloud data of the opposite positioning pin is acquired in real time at preset time intervals (e.g., every 10 parts processed). The point cloud should include 3D spatial coordinates (x, y, z) and intensity information, avoiding occlusion and noise, and ensuring the integrity of the point cloud in the base area.

[0037] Step 2: Point cloud preprocessing, including noise removal, downsampling to optimize data volume, and segmenting the point cloud subset of the circular base region, including: A voxel grid filter is used to reduce the number of point clouds and improve processing efficiency; the voxel size is set to 0.5–1.0 mm (adjusted according to camera accuracy) to balance detail preservation and computation speed; a statistical outlier removal algorithm is used to suppress noise, with the number of nearest neighbors k=50 and the standard deviation multiplier=1.5 (points with a distance mean exceeding 1.5σ are removed); based on geometric features, a plane is fitted and point cloud clusters belonging to the circular base are extracted using Random Sample Consensus (RANSAC) or region growing algorithms. The circular base point cloud is extracted, with the following settings: plane fitting threshold: distance tolerance ≤ 0.2 mm to ensure the accuracy of the base plane; clustering tolerance: point cloud spacing ≤ 1.0 mm, used to separate the base from the background; verification condition: the area of ​​the segmented point cloud must satisfy 0.8S theory ≤ S point cloud ≤ 1.2S theory (S theory is the standard area of ​​the base measured offline by CMM).

[0038] Step 3: Calculate the coordinates of the center point using 3D circle fitting. The three-dimensional coordinates (x, y, z) of the geometric center point of the circular base are calculated from the preprocessed point cloud using the best-fit algorithm.

[0039] Algorithm selection: The least squares method is used for 3D circle fitting, and the optimal circle parameters are solved directly based on the point cloud coordinates.

[0040] Mathematical model: Fit a spatial circle with center point coordinates (a,b,c), radius r, and normal vector of the circle plane (n_x,n_y,n_z).

[0041] Core parameters and calculation process: Input point set: Point cloud coordinates of the preprocessed circular base {(x_i,y_i,z_i)}, i=1…n.

[0042] Outlier removal: Based on residual analysis, points that deviate from the fitted circle by more than 2σ (σ is the standard deviation of the residual) are removed.

[0043] Fitting steps: (1) Determine the plane containing the circle by plane fitting Using principal component analysis (PCA) or least squares plane fitting, calculate the best-fit plane equation for the point cloud: Ax + By + Cz + D = 0.

[0044] Extract the plane normal vector (n_x,n_y,n_z) and project the point cloud onto the plane to obtain the two-dimensional projection point set {(u_i, v_i)}.

[0045] (2) Finding the center point and radius by fitting a two-dimensional circle On the projection plane, a circle is fitted using the geometric least squares method (similar to the original document's 2D algorithm, but based on the projection points).

[0046] Objective function: Minimize the sum of squares of the deviations between the distance from the projected point to the center of the circle and the radius. Iterative solution: Initial value setting: Use the centroid of the projection point set as the initial value of the circle center (a0=Σu_i / n, b0 =Σv_i / n).

[0047] Iterative update: The Newton-Raphson method is used to solve for the center (a,b) and radius r. The convergence conditions are |Δa|<0.01pixel, |Δb|<0.01pixel, and |Δr|<0.01pixel.

[0048] (3) Back projection to 3D space The 2D fitted circle center (a, b) is back-projected onto the original 3D plane to obtain the 3D center point coordinates (x, y, z). Specifically: The three-dimensional center point is calculated using plane equations and projection transformations: if the projection plane is the XY plane, then z is derived from the plane equations; generally, it needs to be solved in conjunction with the normal vector.

[0049] Output: 3D coordinates of the center point (a, b, c) and radius r (for verification: if r deviates from the standard radius by more than 5%, the point cloud is considered abnormal).

[0050] Convergence conditions: Change in center coordinates Δ(a,b,c) < 0.01mm, change in radius Δr < 0.01mm.

[0051] (4) Result verification and output Standard center point location data: After successful debugging, the center point coordinates (x0, y0, z0) calculated through the above process are stored as template location information.

[0052] Current center point location data: During periodic detection, the center point coordinates (x1, y1, z1) are calculated in real time and used as the current location information.

[0053] Completeness check: The fitted radius r needs to be compared with the standard radius. If the deviation is >5%, the point cloud quality is deemed unqualified and re-acquisition is required. Optionally, the part gripping and placing robot 1 acquires a complete image of the current position of the opposite positioning pin 2 at preset time intervals through the vision subsystem 3. The complete image of the current location is preprocessed to retain the current appearance information of the circular base area as the current location information.

[0054] This invention does not rely on offline CMM equipment. It uses the end vision subsystem 3 of the existing part gripping robot 1 on the opposite workstation of the same production line (originally used to guide the gripping robot to accurately grip parts) to periodically detect the position of the positioning pin 2 of the servo gripper on the opposite side, reducing the single measurement time by more than 98% (from hours to minutes or even seconds).

[0055] S3. Compare the current position information of the positioning pin 2 with the template position information of the positioning pin 2, and obtain the comparison result; S4. If the comparison result exceeds the first threshold range, the current positioning pin 2 position is determined to be abnormal.

[0056] A comparison result exceeding the first threshold range includes both the maximum value greater than the first threshold range and the minimum value less than the first threshold range. For example, if the first threshold range is [x, y], then exceeding the first threshold range includes values ​​less than x and greater than y. If the first threshold range is greater than a, then exceeding the first threshold range means less than or equal to a. If the first threshold range is less than b, then exceeding the first threshold range means greater than or equal to b.

[0057] In some embodiments, the current center point position data is compared with the standard center point position data. That is, the positional deviation (Δx, Δy, Δz) and the comprehensive displacement deviation Δd between the two are calculated based on the current center point position coordinates (x1, y1, z1) and the standard center point position coordinates (x0, y0, z0), and the comparison result is given by the formula: Δx=x1-x0 Δy=y1-y0 Δz=z1-z0 The first threshold range is a preset comprehensive displacement threshold range, which is pre-set by the staff based on experience, historical data, or process requirements. For example, the preset comprehensive displacement threshold range can be 0-0.15mm, that is, not exceeding 0.15mm. When the deviation between the real-time measurement data and the template data exceeds 0.15mm, it is determined that the position of the positioning pin 2 has shifted abnormally; that is, if the comprehensive displacement deviation Δd between the two is greater than 0.15mm, it is determined that the position of the positioning pin 2 has shifted.

[0058] Optionally, in another implementation, the specific implementation steps of S1~S4 include: The parts gripping and placing robot 1 acquires a complete image of the properly tested and calibrated opposite positioning pin 2 through the vision subsystem 3, ensuring that the image covers the entire circular base area and is clearly visible. The image processing algorithm preprocesses the complete image, preserving the appearance information of the circular base area as template position information. Specifically, it converts the color image to grayscale to reduce data volume and simplify processing complexity; it uses a Gaussian filter (kernel size 3×3 or 5×5) to smooth the image, suppressing random noise while preserving edge information; and it expands the image's dynamic range through histogram equalization or adaptive grayscale transformation, enhancing the contrast between the circular base and the background and making the outline more defined. The image is clear; the Otsu algorithm or adaptive thresholding method is used to convert the grayscale image into a binary image, and the optimal threshold is automatically determined to separate the circular base area from the background; small noise points or burrs are removed by opening operation (erosion followed by dilation), and holes in the target area are filled by closing operation (dilation followed by erosion) to smooth the contour edges; according to the expected position of the circular base of the positioning pin or the preliminary contour detection results, the smallest bounding rectangle area containing the complete circular base is cropped to complete the preprocessing. The appearance information retained as template position information is the image pixel data of the cropped circular base area, which contains visual features such as the shape and texture of the base.

[0059] The part gripping and placing robot uses the vision subsystem to acquire complete images of the current position of the opposite positioning pin at preset time intervals; Based on the same image processing algorithm, the complete image of the current location is preprocessed to retain the current appearance information of the circular base area as the current location information; A lightweight convolutional neural network (CNN) extracts template feature vectors for template location information and current feature vectors for current location information. The cosine similarity between the template and current feature vectors is calculated as the comparison result. Here, a first threshold range is used to define the similarity threshold range. If the cosine similarity is less than the minimum value of the similarity threshold range, the comparison result is considered to exceed the first threshold range, indicating an anomaly in the current pin 2 position. Specifically, the core characteristics of the lightweight CNN are: ≤1 million parameters to ensure low computational load; inference speed ≥30fps to meet online real-time detection requirements; ≤10 network layers; and input image size uniformly adjusted to 224×224 pixels. The preprocessed template image and the current image are input into the trained lightweight CNN, which outputs high-dimensional feature vectors. The template feature vector is denoted as The current feature vector is denoted as The cosine similarity is used to measure the correlation between two feature vectors, and the formula is: Similarity = The calculation result ranges from [0,1], and the closer the value is to 1, the more similar the features are.

[0060] This invention was verified through 100 sets of comparative experiments. When the similarity threshold range is set to [0.9,1], the accuracy of positioning deviation ≤0.15mm reaches 99%, which meets the production precision requirements. Therefore, the similarity threshold range is set to [0.9,1]. If the calculated result is less than the minimum value of the similarity threshold range of 0.9, it is determined that the comparison result exceeds the first threshold range and the positioning pin position is abnormal.

[0061] Furthermore, if the current position of positioning pin 2 is abnormal, an alarm will be triggered; based on the alarm signal, the production line will be shut down. Furthermore, an alarm signal is triggered and linked to the PLC control system to immediately stop the production line and prevent continued processing from causing batches of parts to be out of size; at the same time, information such as the abnormal time, deviation value and corresponding servo gripper number is recorded.

[0062] In this invention, to achieve highly reliable detection of the servo gripper's position, a common reference benchmark can be introduced for collaborative analysis and diagnosis. This includes installing a high-precision fixed reference target between two part-grabbing robots positioned opposite each other on opposite sides of the production line, serving as a common reference benchmark. The two robots can periodically (e.g., per shift) simultaneously measure the target's position information and combine it with multi-source data for fusion analysis to collaboratively analyze the source of anomalies. Specifically, this includes: (1) Multi-source data acquisition Mutual inspection data: Part gripping robot A measures the current position information of the positioning pin of part gripping robot B and compares the results (referred to as A→B deviation). Part gripping robot B measures the current position information of the positioning pin of part gripping robot A and compares the results (referred to as B→A deviation).

[0063] Target measurement data: Two robots simultaneously measure the position information of a fixed reference target and calculate the target position deviation from the actual position information of the fixed reference target (the standard value calibrated during debugging).

[0064] (2) Diagnosis of the source of the abnormality The main sources of anomalies include: vision subsystem problems, positioning pin problems, and reference datum problems; Visual subsystem problem identification: Conditions: Only one part-grabbing robot's measurement comparison result exceeds the first threshold range, and the deviation between the position information of the fixed reference target acquired by the part-grabbing robot and the actual position information of the fixed reference target (target measurement deviation) is greater than the second threshold. Specifically, the actual position coordinates of the fixed reference target are calibrated using high-precision equipment and recorded as the reference coordinates (x0, y0, z0); the vision subsystem periodically acquires images of the fixed reference target and extracts the current coordinates (x1, y1, z1) of the target's center point; the deviation calculation formula is: Δx=x1-x0 Δy=y1-y0 Δz=z1-z0 Δd is the target measurement deviation, which is used to compare with the second threshold. By measuring the fixed reference target multiple times, the repetitive measurement error of the vision subsystem is statistically analyzed, and 2-3 times the repetitive measurement error is taken as the second threshold. In this application, the repetitive measurement error of the vision subsystem is found to be no more than ±0.05mm through multiple measurements, so the second threshold is set to ±0.1mm.

[0065] Diagnostic Conclusion: The fixed reference target is the target being tested and can be used for the calibration of the vision subsystem. If the deviation between the position information of the fixed reference target and its actual position information exceeds the second threshold, it indicates that the measurement accuracy of the vision subsystem is abnormal. Therefore, it can be determined that there is an anomaly in the vision subsystem of the part-grabbing robot, causing the measured position of the positioning pin to be abnormal. Therefore, it is necessary to rule out the anomaly in the vision subsystem and then re-determine the anomaly in the positioning pin position. Vision subsystem anomalies include calibration distortion and lens contamination. These anomalies can be resolved by recalibrating the vision subsystem, cleaning or replacing the lens, etc.

[0066] For example, if the deviation from A to B exceeds the first threshold range, and the deviation between the position information of the fixed reference target acquired by robot A and the actual position information of the fixed reference target is greater than the second threshold, then the vision subsystem of the part gripping and placing robot A is determined to be abnormal.

[0067] Determining the issue of the positioning pin: Condition: If the measurement comparison result of one of the part gripping and placing robots exceeds the first threshold range, and the target measurement deviation of the part gripping and placing robot is less than the second threshold.

[0068] Diagnostic Conclusion: Since the deviation between the position information of the fixed reference target and the actual position information of the fixed reference target is less than the second threshold, it is determined that the vision subsystem of the part gripping robot is not faulty, but the position of the positioning pin is abnormal. In this case, the abnormal positioning pin position may be due to deformation or other mechanical structural abnormalities. The problem can be resolved by replacing the positioning pin.

[0069] For example, if the deviation from A to B exceeds the first threshold range, and the deviation between the position information of the fixed reference target collected by robot A and the actual position information of the fixed reference target is less than the second threshold, then it is determined that the position of the positioning pin collected by the part gripping and placing robot A is abnormal.

[0070] Reference benchmark problem determination: Condition: The target position deviations measured by the two part-grabbing and placing robots are both greater than the second threshold.

[0071] Diagnostic conclusion: Since the target measurement deviation of the vision subsystems of the two part gripping and placing robots for the same fixed reference target is greater than the second threshold, it indicates that the basic reference system of the measurement system is distorted and cannot reliably distinguish between the positioning pin problem and the vision subsystem problem. This suggests that the system-level calibration has failed (the fixed target has moved slightly due to mechanical loosening or thermal deformation), and the common reference needs to be recalibrated.

[0072] S5. When the position of positioning pin 2 is abnormal, control the production line to stop and adjust the position of positioning pin 2 according to the comparison results.

[0073] The adjustment of the position of positioning pin 2 based on the comparison results includes: Based on the comparison results, a position compensation instruction for positioning pin 2 is generated, including the target position, which is the template position information (e.g., standard center point position data). Send the target position compensation command to the servo system; The servo system drives the motor to move the positioning pin 2 along the guide rail or lead screw to the target position, achieving position compensation for the positioning pin 2. The servo motor's speed is set to a low-speed adjustment mode, defaulting to 5-10 mm / s (reducing to 2-5 mm / s when approaching the target position) to avoid overshoot; the acceleration is 0.5-1 m / s². 2 (S-shaped acceleration and deceleration curve) to reduce mechanical shock; Positioning tolerance: ±0.05mm (stricter than the 0.15mm abnormal threshold to ensure adequate compensation); Repeat positioning accuracy: ±0.02mm (inherent parameter of the servo system to ensure consistency in multiple adjustments); Motion mode: Point-to-point control (PTP) to precisely drive the positioning pin 2 to move along the guide rail / lead screw to the target position (reference template coordinates).

[0074] Furthermore, in this invention, the position deviation (Δx, Δy, Δz) can be directly converted into compensation commands for the servo system, and the position of the positioning pin 2 can be automatically fine-tuned in the next production cycle or at a specific idle time, so as to realize a seamless closed loop of "detection-compensation" and truly eliminate the cumulative error of accuracy.

[0075] The compensation logic of this invention is as follows: a PID algorithm or feedforward compensation is adopted, and a compensation strategy considering the backlash of mechanical transmission is used; a "post-compensation retest and verification" step is added to form a sub-closed loop of "detection-compensation-verification" to ensure the effectiveness of compensation.

[0076] S6. Store the location information of the positioning pin 2 (including timestamp, center coordinates of positioning pin 2, and deviation value) of each online measurement in a local database or cloud to form a queryable historical record.

[0077] When storing the location information of the positioning pin 2, the stored content includes the positioning pin 2 number (such as #1, #2, ...), the center coordinates (X, Y, Z) of the positioning pin 2, the acquisition time, camera parameters, calibration parameter version, and other information, which facilitates traceability and reuse.

[0078] This invention records the position data of the center point of the positioning pin 2 during each online measurement (compared with the position data of the template that passed the first debugging), forming a complete historical database, supporting dynamic tracking of accuracy change trends, and providing a basis for preventive maintenance.

[0079] Furthermore, to avoid unplanned downtime, the present invention also includes advance compensation for the position of the positioning pin.

[0080] S7. Based on the historical position information sequence of positioning pin 2, predict the future accuracy change trend of positioning pin 2. According to the prediction result, before reaching the alarm threshold, generate a positioning pin position compensation command, which includes the target position. Send the target position compensation command to the servo system. The servo system drives the motor to move the positioning pin along the guide rail or lead screw according to the target position to realize the positioning pin position compensation.

[0081] In some embodiments, Based on the historical position information sequence of positioning pin 2, the future accuracy change trend of positioning pin 2 is predicted, including: continuously recording the comprehensive displacement deviation Δd of each positioning pin 2 and plotting its change curve over time or number of work cycles; based on the change curve, using algorithms such as linear regression and exponential smoothing, predicting when the deviation will reach the alarm threshold. The alarm threshold can be preset by staff based on experience, historical data, process requirements, or production needs.

[0082] Based on the prediction results, before reaching the alarm threshold, a positioning pin position compensation command is generated, including: Based on the change curve, predict the time or number of work cycles for each positioning pin to reach the alarm threshold. Before the predicted time or number of work cycles is reached, select an appropriate time to generate a positioning pin position compensation instruction in conjunction with the production plan, such as when the production line is in an idle or low-load state.

[0083] In other embodiments, Based on the historical position information sequence of positioning pin 2, predict the future accuracy change trend of positioning pin 2, including: based on the recent deviation value, trend slope and fluctuation range of the comprehensive displacement deviation Δd of each positioning pin 2, calculate a comprehensive health index for each servo gripper, compare the health index with the preset health threshold, and arrange maintenance when the health index exceeds the preset health threshold to avoid unplanned downtime.

[0084] Specifically, The methods for calculating the health index include: (1) Retrieve the continuous measurement records of each positioning pin 2 (numbered as #1, #2, ...) from the database, sort them by time to form a sequence {(ti, Δdi)}, where ti is the timestamp and Δdi is the comprehensive displacement deviation; (2) Obtain the recent deviation value (Δd) avg ): Take the arithmetic mean of the deviations of the most recent N measurements to reflect the short-term accuracy status, where N is preferably 10; (3) Calculate the trend slope (k): Fit the time linear regression model Δd=k⋅t+b of the deviation sequence using the least squares method. The slope k represents the rate of change of the deviation (unit: mm / hour). A positive value indicates accelerated degradation of accuracy.

[0085] (4) Calculate the fluctuation range (σ): Calculate the standard deviation of the most recent N deviations and assess the stability.

[0086] (5) Health Index H (range 0-100, higher score indicates better health) Based on the above characteristics, a weighted scoring model is adopted: H=w1 f(Δd avg )+w2 g(k)+w3 h(σ) in: f(Δd avg ) represents an inverse proportional function, where Δd avg Mapping to a subrange of [0,100] ensures that the larger the deviation, the lower the score; g(k) represents the exponential decay function, which maps the value of k to a score. The score is highest when k=0, and decreases as k>0. h(σ) represents a logarithmic function that maps σ to a fraction; the smaller the fluctuation, the higher the score. Weighting (can be adjusted according to production needs): w1=0.5 (recent deviation has the highest weight and is directly related to current risk). w2=0.3 (Trend slope weight, pay attention to long-term risks). w3=0.2 (fluctuation range weight, reflecting stability).

[0087] This invention detects the decrease in accuracy caused by mechanical wear and part deformation in advance by comparing the position deviation of the positioning pin 2 in real time, thus avoiding the situation of batch parts having out-of-tolerance dimensions.

[0088] The lightweight convolutional neural network described in this invention refers to a convolutional neural network with ≤1 million parameters, inference speed ≥30fps, and ≤10 layers. Its difference from conventional convolutional neural networks lies in reducing computational load through depthwise separable convolutions, meeting the requirements for online real-time detection. It preferably adopts the MobileNetV2 architecture, with an input layer size of 224×224. The minimum value of the similarity threshold can be 0.85-0.95. In specific implementation, the minimum value of the similarity threshold can be 0.9, i.e., the similarity threshold range is [0.9, 1]. When the cosine similarity is lower than 0.9, the corresponding positioning deviation is ≥0.15mm, indicating an abnormal position.

[0089] Example 1: like Figure 1 , 2 As shown, the present invention provides a method and system for detecting the position of a servo gripper. The circular bottom contour of the positioning pin 2 is used as a feature recognized by the vision subsystem 3 to perform online detection of the position of the positioning pin 2 on the production line, ensuring the positioning function of the servo gripper and avoiding production capacity loss and rework. The specific process includes: Step 1, as follows Figure 2 As shown, each servo gripper has two positioning pins 2. The two part gripping and placing robots 1 on opposite sides of station A collect the initial circular base contours of the two positioning pins 2 after initial debugging, calculate the initial center point position data, and store it. Ptemplate=(Xt,Yt,Zt) ; Step 2: One hour later, the two opposing part-grabbing and placing robots 1 at workstation A again collect the current circular base contours of the two positioning pins 2 on the opposite sides, calculate the current center point position data, and store it. Pcurrent=(Xc,Yc, Zc) ; Step 3: Compare the current center point position data of each positioning pin 2 with the initial center point position data, and calculate the position deviation. ΔX=Xt-Xc,ΔY=Yt-Yc,ΔZ=Zt-Zc ; One of the positioning pins is 2. ΔX =0.08mm, ΔY =0.06mm, ΔZ =0.02mm, the comparison result is Δd= 0.1mm; another locating pin 2 ΔX =0.05mm, ΔY =0.08mm, ΔZ =0.10mm, the comparison result is Δd= 0.14mm; Step 4: The preset alarm threshold is 0.15mm. If the comparison result is less than the alarm threshold, the production line will continue to operate. Step 5: After another hour, the two opposing part gripping and placing robots 1 at workstation A will again collect the current circular base contours of the two positioning pins 2 on the opposite side, calculate the current center point position data, and store it. Step 6: Compare the current center point position data of each positioning pin 2 with the initial center point position data. The comparison result for one positioning pin 2 of the left-side part gripping robot 1 is as follows: ΔX =0.07mm, ΔY =0.16mm, ΔZ =0.02mm, the comparison result is Δd= 0.18mm; Step 7: If the comparison result is greater than the alarm threshold, an alarm will be triggered; Step 8: The PLC stops the production line based on the alarm signal; Step 9: After stopping the machine, the controller sends the target position command to the servo system responsible for driving the positioning pin 2 based on the comparison results. Step 10: The servo driver performs closed-loop control based on the target position and the current feedback position. The drive motor drives the positioning pin 2 to move along the guide rail or lead screw in the X, Y, and Z directions respectively to compensate for the deviation. During the movement, the servo system provides real-time feedback on the current actual position until the error between the actual position and the target position is less than the set tolerance (e.g., ±0.01mm). At this point, the adjustment is considered to be in place and the qualified standard is met.

[0090] Step 11: After the adjustment is completed, the position of the positioning pin 2 can be quickly retested through the vision subsystem 3 to confirm whether the actual position has returned to the allowable range of the standard template position.

[0091] The closed-loop detection described in this invention refers to a complete process that includes location information acquisition, comparison with a benchmark, deviation determination, and triggering at least one of the following subsequent actions: alarm, shutdown, or location compensation based on the determination result.

[0092] Advantages of this invention: This invention utilizes the vision subsystem of an existing part-grabbing robot end effector on the production line for position detection of servo gripper positioning pins, eliminating the need for additional detection devices and reducing detection costs. This vision subsystem can periodically measure the position of opposite positioning pins at the same workstation online. Compared to existing technologies, it transforms offline measurement into online measurement, greatly improving measurement efficiency and avoiding the impact of offline measurement on production line capacity.

[0093] This invention enables online measurement of the positioning pins of a servo gripper, while avoiding positioning dimension deviations in batches of parts caused by pin size offsets, thus greatly reducing rework costs.

[0094] This invention achieves closed-loop detection of the positioning position of the online servo gripper by acquiring the position data of the positioning pin through the end vision subsystem of the gripping robot.

[0095] This invention provides rich and searchable historical positioning data, dynamically tracks changes in position accuracy, and can promptly alarm and stop the machine when the positioning pin error reaches the alarm threshold, greatly reducing repair costs.

[0096] This invention breaks through the technical bottleneck of traditional servo grippers requiring offline measurement for positioning, expands the functional application range of vision subsystems on the production line, enables multiple uses with one machine, and provides macroscopically controllable positioning data, allowing for preventative maintenance and reducing downtime.

[0097] The above are merely preferred embodiments of the present invention and are not intended to limit the present invention. Various modifications and variations can be made to the present invention by those skilled in the art. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the scope of protection of the present invention.

Claims

1. A method for detecting the position of a servo gripper, characterized in that, A part-grabbing robot is located on each side of the production line, facing each other. Each robot is equipped with a servo gripper and a vision subsystem. The servo gripper has positioning pins for positioning. The detection method includes: Obtain the position information of each locating pin that has passed the debugging process, and use it as template position information; The parts gripping and placing robot collects the current position information of the opposite positioning pin through the vision subsystem at preset time intervals; The current position information of the positioning pin is compared with the template position information of the positioning pin to obtain the comparison result; If the comparison result exceeds the first threshold range, the current positioning pin position is determined to be abnormal.

2. The method for detecting the position of a servo gripper according to claim 1, characterized in that: When the position of the positioning pin is determined to be abnormal, the production line is stopped, and the position of the positioning pin is adjusted according to the comparison results.

3. The method for detecting the position of a servo gripper according to claim 2, characterized in that: Adjust the position of the positioning pins according to the comparison results, including: Based on the comparison results, a positioning pin position compensation command is generated, including the target position; Send the target position compensation command to the servo system; The servo system drives the motor to move the positioning pin along the guide rail or lead screw according to the target position, thereby achieving positioning pin position compensation.

4. The method for detecting the position of a servo gripper according to claim 1, characterized in that: Obtain the circular base outline of each locating pin that has passed the debugging process, and calculate the standard center point position data of the circular base outline as the template position information; The parts gripping and placing robot acquires the current circular base outline of the opposite positioning pin through the vision subsystem at preset time intervals, calculates the current center point position data of the circular base outline, and uses it as the current position information. Compare the current center point location data with the standard center point location data to obtain the comparison results.

5. The method for detecting the position of a servo gripper according to claim 4, characterized in that: The processes for calculating the standard center point position data and the current center point position data both include: Complete point cloud data of the opposite positioning pin is acquired through the vision subsystem; The point cloud is subjected to downsampling filtering, noise reduction filtering, and region segmentation to extract the point cloud clusters of the circular base. Calculate the best-fit plane and normal vector of the point cloud; The point cloud is projected onto the fitting plane, and the coordinates of the two-dimensional circle center and radius are fitted using the least squares method. By back-projecting the two-dimensional center of the circle onto 3D space, the three-dimensional coordinates of the standard center point or the current center point can be obtained. Output the 3D coordinates and verify the radius deviation.

6. The method for detecting the position of a servo gripper according to claim 4, characterized in that: Obtain the comparison results, including: Based on the current center point coordinates (x1, y1, z1) and the standard center point coordinates (x0, y0, z0), calculate the positional deviation (Δx, Δy, Δz) and the comprehensive displacement deviation Δd between them, as the comparison result. The formula for the comprehensive displacement deviation Δd is as follows: Δx = x1 - x0 Δy = y1 - y0 Δz = z1 - z0 The first threshold range is a preset comprehensive displacement threshold range. If the comprehensive displacement deviation Δd is greater than the maximum value of the preset comprehensive displacement threshold range, it is determined that the comparison result exceeds the first threshold range and the current positioning pin position is abnormal.

7. The method for detecting the position of a servo gripper according to claim 1, characterized in that: The part gripping and placing robot acquires a complete image of the qualified opposite side positioning pin through the vision subsystem; The complete image is preprocessed to retain the appearance information of the circular base area, which is used as the template position information; The part gripping and placing robot uses the vision subsystem to acquire complete images of the current position of the opposite positioning pin at preset time intervals; The complete image of the current location is preprocessed to retain the current appearance information of the circular base area, which is used as the current location information. The template feature vector, which extracts template position information, and the current feature vector, which extracts current position information, are extracted using a lightweight convolutional neural network. The cosine similarity between the template feature vector and the current feature vector is then calculated as the comparison result. The first threshold range is a similarity threshold range. If the cosine similarity is less than the minimum value of the similarity threshold range, it is determined that the comparison result exceeds the first threshold range and the current positioning pin position is abnormal.

8. The method for detecting the position of a servo gripper according to any one of claims 1-7, characterized in that: If the comparison result exceeds the first threshold range, the current positioning pin position is determined to be abnormal, including: Two part-grabbing and placing robots on opposite sides of the production line periodically and simultaneously collect position information of a fixed reference target, which is set between the two part-grabbing and placing robots. If the comparison result between the current position information of the opposite positioning pin collected by one of the part gripping and placing robots and the template position information of the positioning pin exceeds the first threshold range, and the deviation between the position information of the fixed reference target collected by the part gripping and placing robot and the actual position information of the fixed reference target is greater than the second threshold, then the vision subsystem of the part gripping and placing robot is determined to be abnormal. If the comparison result between the current position information of the opposite positioning pin collected by one of the part gripping and placing robots and the template position information of the positioning pin exceeds the first threshold range, and the deviation between the position information of the fixed reference target collected by the part gripping and placing robot and the actual position information of the fixed reference target is less than the second threshold, then the current positioning pin position is determined to be abnormal.

9. The method for detecting the position of a servo gripper according to any one of claims 1-6, characterized in that, The detection method further includes: During the operation of the servo gripper, the comparison results for each positioning pin are acquired; Based on the comparison results, obtain the curve of the change of the comparison results of each positioning pin over time or the number of working cycles; Based on the change curve, predict the time or number of working cycles for each positioning pin to reach the alarm threshold based on the comparison result. Before the predicted time or number of working cycles is reached, generate a positioning pin position compensation command.

10. A servo gripper position detection system, characterized in that: The parts gripping and placing robots are positioned on both sides of the production line. Each parts gripping and placing robot has a gripping actuator installed at the end of its robotic arm. One or more servo grippers are installed on the gripping actuator, and each servo gripper is connected to a positioning pin. Each parts gripping and placing robot also has a vision subsystem installed at the end of its robotic arm. The visual range of the vision subsystem extends outward from the end of the robotic arm and covers the servo gripper and positioning pin of the part gripping and placing robot on the opposite side. The vision subsystem is used to collect template position information and current position information of each positioning pin; The vision subsystem is connected to a detection result determination subsystem, which is used to compare the current position information of the positioning pin with the template position information of the positioning pin, obtain the comparison result, and determine that the current positioning pin position is abnormal when the comparison result exceeds the first threshold range.

Images