An automated control method and system for setting up based on real-time image acquisition

Abstract

This invention relates to the field of machine vision control technology, specifically to an automated control method and system based on real-time image acquisition. The method includes the following steps: calling an image sequence, analyzing the structured light array spot, merging coordinates to connect the contour, fusing multi-view grayscale differences, constructing a reference path, analyzing load change trends, adjusting trajectory commands and rollback conditions, and generating angle rollback reference data. In this invention, through precise merging and analysis of contour pixel coordinates based on a structured light array, combined with multi-view grayscale region difference recognition, the accuracy of contour shape correction and the reliability of workpiece surface path construction are significantly improved. Furthermore, based on the correlation analysis of continuous load data and trajectory changes, the trajectory offset and rollback angle are adaptively adjusted, effectively overcoming contour recognition errors caused by a single viewpoint and the neglect of load changes. This achieves stable control and automatic correction of the workpiece trajectory, improving the stable operation capability of the control system.

Description

Technical Field

[0001] This invention relates to the field of machine vision control technology, and in particular to an automated control method and system for setting up based on real-time image acquisition. Background Technology

[0002] Machine vision control technology encompasses the acquisition of environmental image information through camera equipment, followed by analysis, recognition, and processing of these images to obtain the state parameters of target objects and feed them back to the control system. This enables automatic detection, recognition, positioning, and control during production processes. It includes image acquisition units, image processing algorithms, and coupling mechanisms between visual information and the control system. Applications include industrial automation, intelligent manufacturing, robot guidance, and detection control. Machine vision control technology typically acquires visual information of the target area through image sensors, and uses image processing methods such as edge recognition, template matching, and feature extraction to extract geometric, positional, and motion parameters of the object. This extracted information is then transmitted to the control unit, where control logic enables the automatic control of the equipment. Dynamic response and adjustment, among which, the setting automation control method based on real-time image acquisition refers to the control method that uses image acquisition equipment to acquire image data of the status of the object or equipment under test in real time, performs feature recognition and processing analysis on the images, and automatically adjusts the control parameters or operating settings of the target equipment. It covers real-time image acquisition, image processing, target recognition and parameter feedback adjustment, etc. Specifically, it includes configuring an image acquisition unit to acquire images of key parts of the system under adjustment in real time, using contour recognition and template comparison to process the acquired images, extracting control-related parameters such as displacement, angle, and position offset, and the control unit executes changes to the actuator settings according to the preset adjustment model to complete the automatic adjustment of the operating parameters of the target equipment.

[0003] Traditional automated control technology relies on a single perspective for image acquisition and contour analysis. In complex environments, it is prone to recognition errors caused by local lighting changes or perspective occlusion. Furthermore, it ignores the correlation between real-time load and trajectory changes, resulting in untimely correction of trajectory parameters. Specifically, this manifests as reduced accuracy in workpiece contour recognition, and the inability of the equipment to adaptively correct path coordinate deviations during operation, causing control to deviate from the target reference and affecting the overall accuracy of automated production and product consistency. Summary of the Invention

[0004] The purpose of this invention is to overcome the shortcomings of existing technologies and to propose an automated control method and system for setting up based on real-time image acquisition.

[0005] To achieve the above objectives, the present invention adopts the following technical solution: an automated control method for settings based on real-time image acquisition, comprising the following steps:

[0006] S1: Call the image acquisition sequence, analyze the image projected onto the workpiece surface by the structured light array, identify the pixel center position of each light spot, merge the horizontal and vertical coordinates according to the number, connect them into a contour line, compare the contour shape according to the correspondence between the contour and the light spot group, and generate continuous contour offset information.

[0007] S2: Extract the continuous offset information of the contour, acquire multi-camera images, divide the images into rectangular regions, locate the center pixel of the region, divide the gray level according to the gray value, merge the center coordinates of the multi-view, analyze the consistency of the numbering, filter the difference regions, and generate partition offset feature data.

[0008] S3: Call the contour continuous offset information and partition offset feature data to construct the workpiece surface reference path, analyze the path coordinates, compare the differences between adjacent coordinate directions, determine the relationship with the partition number, adjust the path coordinates, and establish the trajectory adjustment command result;

[0009] S4: Based on the trajectory adjustment command result, determine the load measurement data generated during the current execution unit's action, compare the position offset of the continuous periodic load measurement results, calculate the load change trend, determine the correlation between load change and trajectory adjustment, and generate load adjustment amplitude information.

[0010] As a further embodiment of the present invention, the continuous contour offset information includes the coordinate group corresponding to the spot number, the horizontal and vertical merging sequence, and the contour line segment splicing path; the partition offset feature data includes the gray level division result, the multi-view center point coordinate set, and the number consistency judgment label; the trajectory adjustment instruction result includes the path coordinate transformation information, the direction difference calculation value, and the path adjustment reference point set; and the load adjustment amplitude information includes the load growth trend value, the action amplitude limit parameter, and the trajectory offset response coefficient.

[0011] As a further aspect of the present invention, the step of obtaining the continuous contour offset information specifically includes:

[0012] S111: Call the image acquisition sequence, analyze the light spot image data formed by the structured light array on the workpiece surface, extract the pixel center coordinates corresponding to each light spot, and pair the horizontal and vertical adjacent numbers according to the light spot numbering order. Merge the paired coordinates according to the numbering order to establish a coordinate group list and generate the light spot numbering coordinate combination result.

[0013] S112: Based on the result of the combination of the light spot number coordinates, extract the horizontal and vertical coordinate point sequences, connect each coordinate point in the pairing order to form a contour path, perform matching judgment on the number grouping and the path contour, and calculate the structure fitting deviation value.

[0014] S113: Call the structure fitting deviation value, perform coordinate repositioning for each path segment, check the matching relationship between the corrected path number and the spot group, analyze the consistency between path continuity and spot number, correct the contour shape, and generate contour continuous offset information.

[0015] As a further aspect of the present invention, the step of obtaining the partition offset feature data specifically includes:

[0016] S211: Extract the continuous offset information of the contour, obtain images of the same workpiece area captured by multiple cameras, divide the image into multiple rectangular areas of equal area, extract the gray value of the center pixel of each area, divide the area into multiple gray level intervals according to the gray value, and generate gray level partition data.

[0017] S212: Call the gray level partition data, extract the center pixel coordinates and gray values ​​of the same numbered area under multiple views, calculate the gray mean and coordinate mean of all numbers under all views, analyze the consistency of numbers under multiple views, calculate the combination difference score of each number, and compare it with the difference tolerance benchmark value to obtain the difference number list and obtain multi-view combination difference data.

[0018] S213: Based on the multi-view combined difference data, call the difference number list, extract the set of regional center coordinates of each difference number under multiple views, perform spatial order combination and continuity test on the coordinates of each difference number, mark the boundary of the difference interval, and establish partition offset feature data.

[0019] As a further aspect of the present invention, the step of obtaining the trajectory adjustment command result specifically includes:

[0020] S311: Call the continuous contour offset information and partition offset feature data to construct the workpiece surface reference path, extract the sequentially arranged coordinate points in the path, compare the positions of adjacent coordinate point pairs, calculate the directional difference between adjacent positions, and generate directional difference data.

[0021] S312: Based on the directional difference data, analyze the correspondence between directional differences and partition numbers, calculate the total metric of path adjustment direction change, and generate path adjustment change coefficient;

[0022] S313: Adjust the reference position of the path coordinates according to the path adjustment change coefficient, and generate the trajectory adjustment command result.

[0023] As a further aspect of the present invention, the step of obtaining the load adjustment range information specifically includes:

[0024] S411: Based on the trajectory adjustment instruction result, obtain the load measurement data of the current execution unit within the specified period, compare the position offset of the periodic load measurement results, and calculate the load position offset value;

[0025] S412: Based on the load position offset value, calculate the load change trend, obtain the load change value by comparing the changes in continuous period load measurement data, analyze the correlation between the load change value and the trajectory adjustment, and generate load change trend data;

[0026] S413: Based on the load change trend data, adjust the parameters of the action amplitude in the control command to generate load adjustment amplitude information.

[0027] As a further aspect of the present invention, the method further includes:

[0028] S5: Extract the load adjustment amplitude information, analyze the continuously collected workpiece trajectory change data, establish an ordered trajectory data sequence, construct a linear vector for the center coordinates of two consecutive frames of trajectory, calculate the direction difference, analyze the degree of offset, calculate the correlation in combination with the load adjustment amplitude, adjust the rollback conditions, and generate angle rollback reference data.

[0029] The angle rollback reference data specifically includes the trajectory offset angle sequence, rollback judgment reference index, and command rollback trigger signal.

[0030] As a further aspect of the present invention, the step of obtaining the angle rollback reference data specifically includes:

[0031] S511: Extract the load adjustment amplitude information, analyze the continuously collected workpiece trajectory change data, and establish an ordered trajectory data sequence;

[0032] S512: Based on the ordered trajectory data sequence, construct a linear vector for the center coordinates of the trajectory in every two frames, calculate the directional difference between the two vectors, analyze the degree of directional offset of each trajectory change, and generate a trajectory directional offset value.

[0033] S513: Using the trajectory direction offset value, combined with the load adjustment range, calculate the correlation index, determine the relationship between the load adjustment range and the trajectory direction change, adjust the rollback conditions, and generate angle rollback reference data.

[0034] An automated setting control system based on real-time image acquisition, wherein the automated setting control system based on real-time image acquisition is used to execute the above-described automated setting control method based on real-time image acquisition, the system comprising:

[0035] The spot merging and recognition module calls the image acquisition sequence, analyzes the image projected onto the workpiece surface by the structured light array, identifies the pixel center position of each spot, merges the horizontal and vertical coordinates according to the number, connects them into a contour line, compares the contour shape according to the correspondence between the contour and the spot grouping, and generates continuous contour offset information.

[0036] The multi-view partition fusion module extracts the continuous offset information of the contour, acquires images from multiple cameras, divides the images into rectangular regions, locates the center pixel of the region, divides gray levels according to gray values, merges the center coordinates of multiple views, analyzes the consistency of the numbering, filters out the difference regions, and generates partition offset feature data.

[0037] The path optimization comparison module calls the contour continuous offset information and partition offset feature data to construct the workpiece surface reference path, analyze the path coordinates, compare the differences between adjacent coordinate directions, determine the relationship with the partition number, adjust the path coordinates, and establish the trajectory adjustment command result.

[0038] The load trend analysis module determines the load measurement data generated during the current execution unit's operation based on the trajectory adjustment command result, compares the positional offset of the continuous period load measurement results, calculates the load change trend, and determines the correlation between load change and trajectory adjustment, generating load adjustment amplitude information.

[0039] The rollback reference setting module extracts the load adjustment amplitude information, analyzes the continuously collected workpiece trajectory change data, establishes an ordered trajectory data sequence, constructs a linear vector for the center coordinates of two consecutive frames of trajectories, calculates the directional difference, analyzes the degree of offset, calculates the correlation in combination with the load adjustment amplitude, adjusts the rollback conditions, and generates angle rollback reference data.

[0040] Compared with the prior art, the advantages and positive effects of the present invention are as follows:

[0041] In this invention, the accuracy of contour shape correction and the reliability of workpiece surface path construction are significantly improved by accurately merging and analyzing contour pixel coordinates based on structured light arrays and combining multi-view grayscale region difference recognition. Furthermore, based on the correlation analysis of continuous load data and trajectory changes, the trajectory offset and rollback angle are adaptively adjusted to effectively overcome the contour recognition error and load change neglect caused by a single viewpoint, thereby achieving stable control and automatic correction of the workpiece trajectory and improving the stable operation capability of the control system. Attached Figure Description

[0042] Figure 1 This is a schematic diagram of the workflow of the present invention;

[0043] Figure 2 This is a flowchart of the process for obtaining continuous contour offset information according to the present invention;

[0044] Figure 3 This is a flowchart of the partition offset feature data acquisition process of the present invention;

[0045] Figure 4 This is a flowchart of the trajectory adjustment command result acquisition process of the present invention;

[0046] Figure 5 This is a flowchart of the load adjustment range information acquisition process of the present invention;

[0047] Figure 6 This is a flowchart of the angle rollback reference data acquisition process of the present invention. Detailed Implementation

[0048] To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the invention.

[0049] In the description of this invention, it should be understood that the terms "length," "width," "upper," "lower," "front," "rear," "left," "right," "vertical," "horizontal," "top," "bottom," "inner," and "outer," etc., indicating orientation or positional relationships, are based on the orientation or positional relationships shown in the accompanying drawings and are only for the convenience of describing the invention and simplifying the description, and do not indicate or imply that the device or element referred to must have a specific orientation, or be constructed and operated in a specific orientation, and therefore should not be construed as a limitation of the invention. Furthermore, in the description of this invention, "a plurality of" means two or more, unless otherwise explicitly specified.

[0050] Please see Figure 1 This invention provides a technical solution: an automated control method for settings based on real-time image acquisition, comprising the following steps:

[0051] S1: Call the image acquisition sequence, analyze the image projected onto the workpiece surface by the structured light array, identify the pixel center position of each light spot, merge the horizontal and vertical coordinates according to the number, connect them into a contour line, compare the contour shape according to the correspondence between the contour and the light spot group, and generate continuous contour offset information.

[0052] S2: Extract continuous contour offset information, acquire multi-camera images, segment the images into rectangular regions, locate the center pixel of the region, divide gray levels according to gray values, merge the center coordinates of multiple views, analyze the consistency of numbering, filter out the difference regions, and generate partition offset feature data.

[0053] S3: Call the contour continuous offset information and partition offset feature data to construct the workpiece surface reference path, analyze the path coordinates, compare the differences between adjacent coordinate directions, determine the relationship with the partition number, adjust the path coordinates, and establish the trajectory adjustment command result;

[0054] S4: Based on the trajectory adjustment command result, determine the load measurement data generated during the current execution unit's action, compare the position offset of the continuous cycle load measurement results, calculate the load change trend, determine the correlation between load change and trajectory adjustment, and generate load adjustment amplitude information.

[0055] S5: Extract load adjustment amplitude information, analyze continuously collected workpiece trajectory change data, establish an ordered trajectory data sequence, construct a linear vector for the center coordinates of two consecutive frames of trajectory, calculate the direction difference, analyze the degree of offset, calculate the correlation in combination with the load adjustment amplitude, adjust the rollback conditions, and generate angle rollback reference data.

[0056] The continuous contour offset information includes the coordinate group corresponding to the spot number, the horizontal and vertical merging sequence, and the contour line segment splicing path. The partition offset feature data includes the gray level classification result, the multi-view center point coordinate set, and the number consistency judgment label. The trajectory adjustment command result includes the path coordinate transformation information, the direction difference calculation value, and the path adjustment reference point set. The load adjustment amplitude information includes the load growth trend value, the action amplitude limit parameter, and the trajectory offset response coefficient. The angle rollback reference data specifically includes the trajectory offset angle sequence, the rollback judgment reference index, and the command rollback trigger signal.

[0057] Please see Figure 2 The specific steps for obtaining continuous contour offset information are as follows:

[0058] S111: Call the image acquisition sequence, analyze the light spot image data formed by the structured light array on the workpiece surface, extract the pixel center coordinates corresponding to each light spot, and pair the horizontal and vertical adjacent numbers according to the light spot numbering order. Merge the paired coordinates according to the numbering order to establish a coordinate group list and generate the light spot numbering coordinate combination result.

[0059] The process of invoking the image acquisition sequence first involves capturing frame-by-frame images of the light spot formed by the structured light array on the workpiece surface. During image acquisition, the workpiece size is used as a standard to control the projection of the structured light array onto the workpiece surface to ensure clear light spot image data is obtained. An industrial camera is then used to capture the workpiece surface (e.g., a flat aluminum alloy workpiece with dimensions of...). Take continuous image shots at a resolution set to [resolution setting]. For each pixel, after image binarization, the center coordinates of each light spot are extracted. Specifically, this involves binarizing the light spot region and calculating its centroid to obtain the pixel center coordinates. Assuming the structured light array is... There are a total of 16 light spots, numbered as follows: to The horizontal spot numbers are ordered from left to right, and the vertical numbers are ordered from top to bottom. Coordinate data is recorded and sorted by spot number. Then, coordinates of adjacent numbers are paired horizontally and vertically. For example... and , and Horizontal pairing, vertical pairing is and , and And so on. The paired coordinate sequences are merged in numerical order, for example, coordinates... Corresponding number ,coordinate Corresponding number The horizontal and vertical coordinates were paired and merged one by one to establish a coordinate group list, resulting in a final data table of 16 spot numbers and coordinate combinations, as shown in Table 1.

[0060] As shown in Table 1, the results of the combination of light spot number coordinates are obtained.

[0061] S112: Based on the spot number coordinate combination results, extract the horizontal and vertical coordinate point sequences, connect each coordinate point in pairing order to form a contour path, and perform matching judgment between the numbered grouping and the path contour, using the following formula:

[0062]

[0063] Calculate the structural fit deviation value;

[0064] in, For the first Segment path structure fitting deviation value, For the first The normalized value of the horizontal coordinate of each light spot is obtained by dividing the original pixel's horizontal coordinate by the image width. For the first The normalized value of the x-coordinate of each light spot. For the first The normalized value of the ordinate of each light spot is obtained by dividing the original pixel ordinate by the image height. For the first The normalized value of the ordinate of each light spot. For the first The corresponding number for each light spot is obtained directly through the sequence setting during structured light projection. For the first The number corresponding to each light spot For the first The polar coordinate angle normalization value of each light spot is obtained by converting the light spot coordinates to polar coordinates. arrive Range normalization yields, For the first The polar coordinate angle normalized value of each light spot. The penalty factor for numbering differences is set as an empirical constant, and its value range is set according to the system tolerance. arrive ;

[0065] For the acquired coordinate combination data, coordinates are connected in a numbered pairing order. Specifically, this involves connecting horizontally adjacent coordinates one pair at a time, such as... and Then, the vertical adjacent coordinates, such as and After connection, a contour path sequence is constructed. Based on the paired numbers, the corresponding coordinates are extracted. First, the pixel coordinates are normalized, for example, the horizontal coordinate. Image width The normalization result ; ordinate Image height ,but The normalization method for other coordinates is the same. Simultaneously, the normalized polar coordinate angle value for each coordinate point is calculated; for example, with the image center as the pole, the coordinates are calculated... Polar coordinate angle is , convert radians to After normalization The subsequent structural fitting deviation calculation process involves substituting the normalized values ​​of adjacent coordinates into the formula:

[0066] ;

[0067] Numbered by light spot and For example, set a penalty factor for number difference. (Experimental verification scope) arrive The values ​​between these ranges are appropriate and were selected through experiments. (Best), substituting the data, we get: ;

[0068] First, calculate the sum of squares of the coordinate differences:

[0069] ;

[0070] The path length is obtained by taking the square root again. ;

[0071] Next, calculate the correction term for the difference between the number and the angle:

[0072] The difference in number is The absolute value of the angle difference is The correction term is ;

[0073] Final structure fitting deviation value:

[0074] ;

[0075] The above method is used to calculate all combinations of spot numbers sequentially, generating a complete sequence of structure fitting deviation values. The structure fitting deviation value is a metric established during the spot contour reconstruction process for any two adjacent spot pairs, based on their coordinate differences, numbering sequence, and visual angle. It quantifies whether the two points possess actual structural continuity. A smaller value indicates greater consistency in position, numbering, and visual trajectory, suggesting they likely belong to the same contour segment. Significant fluctuations in this value may indicate abnormal numbering, local occlusion, or abrupt surface changes. Calculating this parameter effectively filters out erroneous connections in the early stages of contour construction, reducing contour fitting path errors and providing highly reliable path input for subsequent morphological correction and coordinate calibration. It is one of the most crucial control variables in the contour reconstruction process. The formula uses the difference in numbering... Differences in angle Scale factor The path deviation value is corrected to enhance the sensitivity to number misalignment and ensure the consistency and stability of the coordinate contour path and the structured light number pairing.

[0076] S113: Call the structural fitting deviation value, perform coordinate repositioning for each path segment, check the matching relationship between the corrected path number and the spot group, analyze the consistency between path continuity and spot number, correct the contour shape, and generate contour continuity offset information.

[0077] The structural fitting deviation data was retrieved, with a threshold value of 0.15 (experimental data showed that...). Using a significant deviation in the time path as a benchmark, determine whether the deviation value exceeds the limit. For the example above... Execute the corresponding coordinate point The repositioning involves finely adjusting its location along the contour normal direction. To reduce the value to within 0.15, for example, by moving approximately 3 pixels to the left along the contour normal and repeating the calculation, the result is obtained. The requirements are met, and the coordinate repositioning operation is completed. After repositioning, a number mapping check is performed. Specifically, the corrected coordinate points are re-matched with the spot numbers, and the coordinate continuity is compared sequentially. to Again By checking the coordinate differences of continuous points, it is determined whether the difference data increases continuously (for example, the horizontal coordinate increases by about 204 pixels ± 3 pixels each time, and the vertical coordinate increases by about 4 pixels ± 1 pixel each time). The continuity of the coordinate path and the consistency of the numbering order are confirmed. After multiple data verifications, the continuous offset data of the contour is updated to the set of paths after coordinate correction.

[0078] Please see Figure 3 The specific steps for obtaining partition offset feature data are as follows:

[0079] S211: Extract continuous contour offset information, obtain images of the same workpiece area captured by multiple cameras, divide the image into multiple equal-area rectangular areas, extract the gray value of the center pixel of each area, divide the area into multiple gray level intervals according to the gray value, and generate gray level partition data.

[0080] After extracting the coordinate values ​​of each point in the continuous offset data of the contour, three cameras were used to analyze the data from different perspectives (denoted as viewpoints). (Shoot the same area of ​​the workpiece, image size is) Pixels, with the standard of complete coverage of the workpiece (such as an aluminum alloy plate), uniformly divide the image into pixels. There are 16 rectangular regions of equal area, each with dimensions of [dimensions to be filled in]. Pixels, each labeled with a number to Locate the center coordinates of each rectangular region. For example, the center coordinates of region number 1 are: The center coordinates of region number 2 are Then, call the image processing program to obtain the grayscale value of the center pixel of each region, and then select the grayscale value range. Gray levels were divided into 5 levels using an interval value of 51 as the baseline value (Table 2).

[0081]

[0082] For example, if the gray value of the center pixel in region 1 is measured to be 120, it is classified as gray level 3. Similarly, if the gray value of the center pixel in region 2 is measured to be 180, it is classified as level 4. The gray values ​​of each region are substituted into the level classification standard for calculation and labeling to obtain the gray level partition data as shown in Table 3.

[0083] The above process yields grayscale level partitioned data.

[0084] S212: Call the grayscale level partitioned data, extract the center pixel coordinates and grayscale values ​​of regions with the same ID from multiple viewing angles, calculate the mean grayscale value and mean coordinates of all IDs from all viewing angles, and analyze the consistency of IDs across multiple viewing angles using the following formula:

[0085] ;

[0086] Calculate the combined difference score for each number and compare it with the difference tolerance baseline value to obtain a list of difference numbers and acquire multi-view combined difference data;

[0087] in, Number The combined difference score, Number From the perspective The grayscale normalized value of the lower region is obtained by acquiring the grayscale value of the center pixel of the numbered region in the original image and normalizing it according to the minimum and maximum grayscale range under each viewing angle. Number The average value of the grayscale normalized value under all views is calculated by... The arithmetic mean is obtained. Number From the perspective Lower region center point Orientation-normalized coordinates are obtained by dividing the original image coordinates by the image width. Number From the perspective Lower region center point Orientation-normalized coordinates are obtained by dividing the original image coordinates by the image height. Number All perspectives The average value of the normalized values ​​of the direction coordinates, through the... The average was calculated to obtain the result. Number All perspectives The average value of the normalized values ​​of the direction coordinates, through the... The average was calculated to obtain the result. Number The number of regions corresponding to each viewpoint is determined by counting and numbering. The number of times it appears in multi-camera images is obtained. For index of number, An index for the perspective;

[0088] After retrieving the grayscale level partitioning data, the center pixel coordinates and grayscale values ​​of each numbered region are extracted one by one from the three viewing angles. Taking region number 1 as an example, the center point of viewing angle 1... Grayscale value 120, viewpoint 2 corresponds to the center coordinates as follows: The grayscale value is 122, and the center coordinates corresponding to viewpoint 3 are... The grayscale value is 118. Then, coordinate and grayscale normalization operations are performed. The coordinate normalization process involves dividing the actual coordinate value by the image width or height. For example, the normalized coordinates of the center point of viewpoint 1... Perspective 2 is Perspective 3 is The grayscale normalization process involves linearly mapping the maximum and minimum grayscale values ​​of a single identifier across different viewing angles. For example, if region 1 has grayscale values ​​of 120, 122, and 118 at the three viewing angles, with a maximum of 122 and a minimum of 118, then the normalized grayscale value for viewing angle 1 is... Viewpoint 2 is 1.0, and viewpoint 3 is 0. Then, the mean grayscale value and the mean coordinate value are calculated. Taking number 1 as an example, the normalized mean grayscale value is... The mean of the coordinates X is The mean Y of the coordinates is Then substitute the values ​​into the formula to calculate the combined difference score. :

[0089] ;

[0090] in, The specific formula is as follows:

[0091] Calculation of grayscale difference:

[0092] ;

[0093] Calculation of coordinate difference term:

[0094] ;

[0095] The calculation yields:

[0096] ;

[0097] Final combination difference score:

[0098] ;

[0099] Experiments verified that a tolerance benchmark of 0.0005 was set. If the calculated score exceeded 0.0005, the number was included in the difference number list. In this example, the combined difference score of number 1 was 0.000375, which did not exceed 0.0005, so it was not included in the difference number list. The above method was used to calculate the combined difference data of each of the 16 numbers, forming multi-view combined difference data. The combined difference score is an indicator used to comprehensively evaluate the consistency and stability of image regions. It combines two aspects of information: the grayscale consistency of the region under multiple views and the spatial stability of the region. Specifically, the combined difference score is measured in the following ways: grayscale difference reflects the brightness change of the region under different views. The larger the difference, the more unstable the brightness of the region is or the more it is affected by factors such as lighting and imaging angle; spatial difference reflects the position change of the region under different views. The larger the difference, the more unstable the position of the region is, which may be caused by imaging distortion or the geometric characteristics of the object itself. The effectiveness of this score lies in its ability to accurately distinguish the stability and consistency of image regions across multiple viewpoints. Especially when issues such as occlusion or reflection exist on the target surface, the combined difference score effectively quantifies these inconsistencies, aiding subsequent processing steps such as region selection and path adjustment in making informed decisions. This metric can identify anomalous regions in visual data, helping to optimize subsequent image processing and analysis workflows. The formula uses both grayscale difference and spatial coordinate difference to determine the combined difference score, reflecting the stability of regional data across multiple viewpoints.

[0100] S213: Based on multi-view combined difference data, call the difference number list, extract the set of regional center coordinates of each difference number under multiple views, perform spatial order combination and continuity test on the coordinates of each difference number, mark the boundary of the difference interval, and establish partition offset feature data.

[0101] After calling the multi-view combined difference data obtained in the previous step, extract the difference number list. For example, numbers 5, 7, and 12 are identified as difference numbers. Then, extract the set of center coordinates of the region corresponding to each difference number under each view. Taking number 5 as an example, the center coordinates of view 1 are... Viewpoint 2 Center Coordinates 3. Center coordinates of viewpoint The spatial order combination process involves arranging the coordinates according to the viewpoint number to form a coordinate sequence, and then performing a continuity check on each pair. The continuity check involves calculating the Euclidean distance between adjacent viewpoint coordinates, for example, the distance between viewpoint 1 and viewpoint 2. Pixels, distance between viewpoint 2 and viewpoint 3 In the experiment, the continuity tolerance benchmark value was set to 8 pixels. If the distance between continuous coordinates exceeds 8 pixels, it is determined to be discontinuous. The distance between viewpoints 2 and 3 in the above-mentioned area number 5 is 10 pixels, which exceeds the tolerance. It is determined that there is a discontinuity in the coordinate sequence. This is used to mark the spatial range boundary of the difference area. By performing the above operation on all difference numbers in sequence, the spatial boundary coordinate information of each difference area is obtained, and finally the partition offset feature data is established.

[0102] Please see Figure 4 The specific steps for obtaining the trajectory adjustment command result are as follows:

[0103] S311: Call the contour continuous offset information and partition offset feature data to construct the workpiece surface reference path, extract the sequentially arranged coordinate points in the path, compare the positions of adjacent coordinate point pairs, calculate the directional difference between adjacent positions, and generate directional difference data.

[0104] After calling the continuous contour offset data and the partition offset feature data, the coordinate points are extracted one by one from the continuous contour offset data in sequence, with the workpiece surface (aluminum alloy plate, size) as the reference. Taking this as an example, we select a coordinate sequence from the continuous offset data of the contour as the initial data. For example, the coordinates of two adjacent points in the sequence are respectively... and The operation compares the positions of adjacent coordinate points. Specifically, it converts each coordinate point to polar coordinates and sets the coordinates. Using the origin as a reference point, calculate the angle between the second point and the first point. Specifically, this involves calling the arctangent function to obtain the included angle. Calculate the angles between subsequent adjacent points in the same manner, such as the third point. Compared to the second point included angle This process is repeated for each pair of coordinate points, recording the difference in direction angle between each pair of adjacent coordinate points, thus forming a sequence of direction difference data. For example, the angle difference between the second pair of coordinate points is... A complete sequence of directional difference data is obtained by calculating multiple pairs of coordinates.

[0105] S312: Based on the directional difference data, analyze the correspondence between directional differences and partition numbers, using the formula:

[0106] ;

[0107] Calculate the total measure of the change in the direction of path adjustment and generate the path adjustment change coefficient;

[0108] in, This is the total metric for adjusting the direction of a path, representing the overall metric of the change in direction between adjacent coordinate points on the path. For the first The normalized direction angle of the nth location point represents the... The normalized value of the orientation angle of each position point relative to the maximum value of all angles. This is obtained by normalizing the angles of all position points. For the first The normalized direction angle of the nth location point represents the... Normalized values ​​of the orientation angles of each location point For the first The normalized path change at the i-th location point represents the... The normalized value of the path change between each location point and the previous location point relative to the maximum value of all path changes is obtained by normalizing all path changes. For the first The normalized path change at the i-th location point represents the... Normalized value of path change at each location point The index represents the number of each location point in the path, ranging from 1 to... integers, This represents the total number of locations in the path.

[0109] Based on the directional difference data sequence, the relationship between the directional differences between coordinate points and the corresponding partition numbers is analyzed one by one. The process involves normalizing the coordinates of the center of each partition region, with the normalization method set as dividing the directional angle value by the maximum value among all angle values. It is assumed that the experimentally measured maximum directional difference angle is... Then the normalized direction angle of position 1 is Position number 2 Simultaneously, path change data is retrieved, with the Euclidean distance of each path segment representing the path change. For example, the path change between position 2 and position 1. Let the maximum path change be as verified by experiments. Then the normalized path change Substituting the normalized direction angle and normalized path change into the formula:

[0110]

[0111] The parameters are explained below: The total metric for changes in the direction of path adjustment. Indicates the first The normalized direction angle value for each coordinate point is obtained by dividing the direction angle value by the maximum value of all angle values. Indicates the first The normalized value of the path change at each coordinate point is calculated by dividing the total path change by the maximum path change. Index of coordinate points, This represents the total number of coordinate points. An actual calculation example is as follows:

[0112] By number Substitute the coordinate data into the formula,

[0113] Calculate the direction angle difference term: ;

[0114] Calculate the difference in path variation: ;

[0115] The score for number 2 is: ,

[0116] The calculations are then performed for the other numbers in sequence. For example, for number 3: the directional difference is 0, the path change difference is 0, and the score is 0. Therefore, the total score is... After completing the calculations for all numbers, the path adjustment change coefficient is obtained. The total metric for path adjustment direction change quantifies the angular changes between points on the path. It integrates the directional changes of adjacent points and considers the impact of path change, reflecting the path's directional adjustment status. A weighted approach is used to account for the influence of path change on directional change; when path change is small, the impact of directional change is amplified; when path change is large, the impact of directional change is weakened. This metric helps the path adjustment system identify areas requiring finer adjustments, ensuring a smooth path transition and avoiding excessive angular changes, thereby improving the overall path adjustment effect. The formula accurately quantifies the stability of path adjustment by normalizing the ratio of directional difference to path change difference.

[0117] S313: Adjust the reference position of the path coordinates according to the path adjustment change coefficient, and generate the trajectory adjustment command result;

[0118] After adjusting the change coefficient according to the path, adjust the change coefficient according to the path. The magnitude of the value is used to adjust the path coordinate reference position one by one. Specifically, the experimental verification reference value of 0.5 for the path adjustment change coefficient is used as the judgment threshold. If the path is found to have a significant deviation, the baseline position needs to be adjusted and numbered accordingly. Taking location as an example, if the calculated local path adjustment coefficient is 0.6, exceeding the threshold of 0.5, then the original coordinate data will be called. At this position, fine-tuning is performed along the path perpendicularly. Specifically, this fine-tuning involves calculating the path angle formed by the preceding and following positions, and adjusting the target angle to the average of the original angles to reduce the local path variation coefficient. Let's assume the angles between the preceding and following positions are... and Then the included angle of the target is The trigonometric functions are called to calculate the new coordinate position, assuming the adjusted coordinate position... Complete numbering The positions are adjusted, and this method is used to check all the positions in the path one by one. The region is fine-tuned to obtain an updated set of path coordinate reference positions, forming trajectory adjustment instruction data.

[0119] Please see Figure 5 The specific steps for obtaining load adjustment range information are as follows:

[0120] S411: Based on the trajectory adjustment instruction result, obtain the load measurement data of the current execution unit within the specified period, compare the position offset of the periodic load measurement results, and calculate the load position offset value;

[0121] After calling the trajectory adjustment command data, the actual workpiece (aluminum alloy plate size is...) is used. Using a force sensor as a reference, the load data of the actuator is acquired within three consecutive specified cycles (cycle length 5 seconds). The unit of load measurement is Newtons (N). Specifically, the load value is recorded every second to obtain the load data corresponding to the position. For example, the load data measured at point 10 on the workpiece path in cycle 1 is 12.0N, the measured value at the same point in cycle 2 is 12.5N, and the measured value in cycle 3 is 13.1N. Detailed measurement data are shown in Table 4.

[0122]

[0123] Based on the data in Table 4, a position offset comparison was performed. The position offset calculation involved calculating the numerical difference between the load measurement data for each consecutive cycle. For example, the load position offset value for cycle 2 relative to cycle 1 was [value missing]. The load position offset of period 3 relative to period 2 is... This method is used to complete the data calculation for all measurement locations, ultimately generating a complete sequence of load position offset values.

[0124] S412: Based on the load position offset value, calculate the load change trend, obtain the load change value by comparing the changes in continuous period load measurement data, analyze the correlation with trajectory adjustment, and generate load change trend data;

[0125] After calling the load position offset value, the load change trend is calculated one by one based on the calculated data. Specifically, the difference between the load position offset values ​​of two consecutive cycles is used to quantify the load trend change. For example, the difference between the position offset values ​​of cycle 3 and cycle 2 at path position 10 is... Then, the load trend direction is judged based on this value. If the position offset value of the next cycle is greater than that of the previous cycle, the load trend is judged to be an upward trend; otherwise, it is a downward trend. For example, if the value of cycle 3 relative to cycle 2 is 0.6, which is greater than the value of cycle 2 relative to cycle 1 is 0.5, it is judged to be an upward trend in load, and the trend direction is recorded as positive (+). Otherwise, it is recorded as negative (-). After the above process is completed, the load change value and trend sign are obtained. The correlation between the trend change value and the trajectory adjustment is further calculated. For example, the adjustment amplitude value of 0.8mm at the corresponding position (number 10) in the trajectory adjustment instruction data is called. The calculation action is to compare the load trend value of 0.1N with the trajectory adjustment amplitude of 0.8mm. Specifically, the ratio of the two is calculated as 0.1N / 0.8mm = 0.125N / mm. If the ratio is greater than the threshold of 0.1N / mm determined by experimental verification, it is determined that there is a significant correlation between load change and trajectory adjustment. The above method is used to traverse all position data to generate complete load change trend data.

[0126] S413: Based on load change trend data, adjust the parameters of the action amplitude in the control command to generate load adjustment amplitude information;

[0127] After retrieving the load change trend data, the motion amplitude parameters are adjusted one by one based on the calculated load change trend values. The adjustment is performed by incrementing or decrementing the original motion amplitude value set in the control command, combined with the load change trend value. For example, if the initial motion amplitude parameter at trajectory position 10 is 0.8mm and the load change trend value is 0.125N / mm, an adjustment coefficient for the motion amplitude parameter is set. To verify the reasonable range obtained by experiment The value between these two ranges is taken here. Adjust the calculation action to The value is updated to the new action amplitude parameter value at that position. Using this action as the standard, the action amplitude adjustment of all path positions is calculated one by one to complete the action amplitude parameter update of the entire path control command and obtain complete load adjustment amplitude data.

[0128] Please see Figure 6 The specific steps for obtaining the angle rollback baseline data are as follows:

[0129] S511: Extract load adjustment amplitude information, analyze continuously collected workpiece trajectory change data, and establish an ordered trajectory data sequence;

[0130] The load adjustment amplitude data is retrieved, and the corresponding motion adjustment amplitude of each path point is extracted one by one. The continuous motion trajectory data of the center position of the aluminum alloy workpiece surface is collected using an industrial vision system. The collected trajectory center coordinate points are sorted in chronological order to form an ordered trajectory data sequence. During the acquisition process, trajectory data is captured at 1-second intervals, and the trajectory center coordinate data of 5 time points are captured continuously, namely the coordinate point (200, 300) at the 1st second, the coordinate point (202, 304) at the 2nd second, the coordinate point (205, 307) at the 3rd second, the coordinate point (207, 311) at the 4th second, and the coordinate point (210, 315) at the 5th second, forming a complete and ordered trajectory data sequence.

[0131] S512: Based on the ordered trajectory data sequence, construct a linear vector for the center coordinates of the trajectory in every two frames, calculate the directional difference between the two vectors, analyze the degree of directional offset of each trajectory change, and generate the trajectory directional offset value.

[0132] The aforementioned ordered trajectory data sequence is used to extract the coordinates of adjacent trajectory center points sequentially. Trajectory direction vectors are then constructed based on the differences in coordinate positions. For example, the difference between the trajectory center coordinates in the 1st and 2nd seconds is 2 pixels horizontally and 4 pixels vertically, resulting in a trajectory direction vector of (2,4). The same process is then applied to the trajectory center coordinates in the 2nd and 3rd seconds, yielding a trajectory vector of (3,3). Next, the direction differences between each pair of consecutive trajectory direction vectors are compared. Specifically, the angle difference between the two trajectory vectors is calculated. For example, the angle difference between trajectory vectors (2,4) and (3,3) is approximately 9.74 degrees. This process is repeated for all consecutive trajectory vectors to obtain the direction offset angle data for each trajectory change, ultimately forming a sequence of trajectory direction offset values.

[0133] S513: Utilize the trajectory direction offset value and combine it with the load adjustment range to calculate the correlation index, determine the relationship between the load adjustment range and the trajectory direction change, adjust the rollback conditions, and generate angle rollback baseline data.

[0134] The trajectory direction offset data is retrieved and sequentially correlated with the load adjustment amplitude at the corresponding positions. The correlation between the trajectory direction offset angle and the load adjustment amplitude is determined by calculating the ratio between the two. For example, for trajectory position number 1 with a direction offset angle of 9.74 degrees and a corresponding load adjustment amplitude of 0.925 mm, the calculated correlation index is approximately 10.53 degrees per millimeter. The correlation threshold verified through experiments is set to 8.0 degrees per millimeter. The above index is compared with the threshold. If the correlation index is greater than the threshold, it is determined that there is a significant correlation between the two, and the rollback angle threshold in the rollback condition needs to be lowered. For example, if the original rollback threshold is 10 degrees, after calculating the difference through the correlation index, the rollback threshold is lowered to 7.47 degrees. For trajectory positions where the correlation index does not reach the threshold, the original rollback threshold is retained without adjustment. The above operation is performed one by one to complete the correction of the rollback conditions for all trajectory positions and generate the final angle rollback reference data.

[0135] An automated setting control system based on real-time image acquisition is provided. This system executes the aforementioned automated setting control method based on real-time image acquisition. The system includes:

[0136] The spot merging and recognition module calls the image acquisition sequence, analyzes the image projected onto the workpiece surface by the structured light array, identifies the pixel center position of each spot, merges the horizontal and vertical coordinates according to the number, connects them into a contour line, compares the contour shape according to the correspondence between the contour and the spot grouping, and generates continuous contour offset information.

[0137] The multi-view partition fusion module extracts continuous contour offset information, acquires multi-camera images, divides the images into rectangular regions, locates the center pixel of each region, classifies gray levels based on gray values, merges the center coordinates of multiple views, analyzes the consistency of the numbering, filters out the difference regions, and generates partition offset feature data.

[0138] The path optimization and comparison module calls the contour continuous offset information and partition offset feature data to construct the workpiece surface reference path, analyze the path coordinates, compare the differences between adjacent coordinate directions, determine the relationship with the partition number, adjust the path coordinates, and establish the trajectory adjustment command result.

[0139] The load trend analysis module determines the load measurement data generated during the current execution unit's action based on the trajectory adjustment command result, compares the positional offset of the continuous period load measurement results, calculates the load change trend, and determines the correlation between load change and trajectory adjustment, generating load adjustment amplitude information.

[0140] The rollback reference setting module extracts load adjustment amplitude information, analyzes continuously collected workpiece trajectory change data, establishes an ordered trajectory data sequence, constructs a linear vector for the center coordinates of two consecutive frames of trajectories, calculates the directional difference, analyzes the degree of offset, calculates the correlation in combination with the load adjustment amplitude, adjusts the rollback conditions, and generates angle rollback reference data.

[0141] The above are merely preferred embodiments of the present invention and are not intended to limit the present invention in any other way. Any person skilled in the art may make changes or modifications to the above-disclosed technical content to create equivalent embodiments that can be applied to other fields. However, any simple modifications, equivalent changes, and modifications 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 protection scope of the present invention.

Claims

1. A method for setting automation control based on real-time image acquisition, characterized in that, The method comprises the following steps: S1: calling an image acquisition sequence, analyzing an image of a structured light array projected onto a workpiece surface, identifying a pixel center position of each light spot, merging horizontal and vertical coordinates according to a number, connecting as a contour line, comparing contour shapes according to a corresponding relationship between the contour and the light spot grouping, and generating contour continuous offset information; S2: extracting the contour continuous offset information, acquiring multi-camera images, dividing the images into rectangular regions, positioning a central pixel point of the region, dividing gray levels according to gray values, merging multi-view central coordinates, analyzing number consistency, screening a difference region, and generating partition offset feature data; S3: calling the contour continuous offset information and the partition offset feature data, constructing a workpiece surface reference path, analyzing path coordinates, comparing direction differences of adjacent coordinates, judging a relationship with a partition number, adjusting path coordinates, and establishing a trajectory adjustment instruction result; S4: according to the trajectory adjustment instruction result, judging load measurement data generated in a current execution unit action process, comparing position offset of continuous period load measurement results, calculating a load change trend, and judging relevance between load change and trajectory adjustment, and generating load adjustment amplitude information; The acquisition step of the load adjustment amplitude information is specifically: S411: according to the trajectory adjustment instruction result, using a force sensor to acquire load measurement data of the current execution unit in a specified period, comparing position offset of period load measurement results, and calculating a load position offset value; S412: based on the load position offset value, calculating a load change trend, acquiring a load change value by comparing changes of continuous period load measurement data, and analyzing relevance between the load change and trajectory adjustment, and generating load change trend data; S413: according to the load change trend data, adjusting a parameter of an action amplitude in a control instruction, and generating load adjustment amplitude information; S5: extracting the load adjustment amplitude information, analyzing continuous acquisition workpiece trajectory change data, establishing an ordered trajectory data sequence, constructing a linear vector for continuous two frames of trajectory center coordinates, calculating a direction difference, analyzing an offset degree, combining load adjustment amplitude to calculate relevance, adjusting a rollback condition, and generating angle rollback reference data.

2. The real-time image acquisition based setup automation control method of claim 1, wherein, The contour continuous offset information includes a light spot number corresponding coordinate group, a horizontal and vertical merging sequence, and a contour line segment splicing path. The partition offset feature data includes a gray level division result, a multi-view central point coordinate set, and a number consistency judgment label. The trajectory adjustment instruction result includes path coordinate transformation information, a direction difference calculation value, and a path adjustment reference point set. The load adjustment amplitude information includes a load growth trend value, an action amplitude limitation parameter, and a trajectory offset response coefficient. The angle rollback reference data is specifically a trajectory offset angle sequence, a rollback judgment reference index, and an instruction rollback trigger signal.

3. The real-time image acquisition based setup automation control method of claim 1, wherein, The acquisition step of the contour continuous offset information is specifically: S111: Call the image acquisition sequence, analyze the spot image data formed by the structured light array on the workpiece surface, extract the pixel center coordinates corresponding to each spot, and pair the transverse and longitudinal adjacent numbers in sequence, merge the paired coordinates in sequence to establish a coordinate group list, and generate a spot number coordinate combination result; S112: According to the spot number coordinate combination result, extract the transverse and longitudinal coordinate point sequence, connect each coordinate point in sequence to form a contour path, perform matching judgment on the number grouping and path contour, and calculate the structure fitting deviation value; S113: Call the structure fitting deviation value, perform coordinate repositioning on each path segment, check the mapping relationship between the corrected path number and the spot group, analyze the path continuity and spot number consistency, correct the contour shape, and generate contour continuous offset information.

4. The real-time image acquisition based setup automation control method of claim 3, wherein, The acquisition step of the partition offset feature data is specifically: S211: Extract the contour continuous offset information, obtain images of the same workpiece area shot by multiple cameras, divide the screen into multiple equal-area rectangular regions, extract the center pixel point gray value of each region, divide the region into multiple gray level intervals according to the gray value, and generate gray level partition data; S212: Call the gray level partition data, extract the center pixel coordinates and gray values of the same number region under multiple viewing angles, calculate the gray mean value and coordinate mean value of all numbers under all viewing angles, analyze the consistency of the number under multiple viewing angles, calculate the combination difference score of each number, and compare it with the difference tolerance reference value to obtain a difference number list and obtain multi-view combination difference data; S213: Based on the multi-view combination difference data, call the difference number list, extract the region center coordinate set of each difference number under multiple viewing angles, perform spatial sequence combination and continuity test on the coordinates of each difference number, demarcate the difference interval boundary, and establish partition offset feature data.

5. The real-time image acquisition based setup automation control method of claim 4, wherein, The acquisition step of the trajectory adjustment instruction result is specifically: S311: Call the contour continuous offset information and partition offset feature data, construct a workpiece surface reference path, and extract the sequentially arranged coordinate points in the path, compare the positions of adjacent coordinate points, calculate the direction difference between adjacent positions, and generate direction difference data; S312: According to the direction difference data, analyze the corresponding relationship between the direction difference and the partition number, calculate the total measure of the path adjustment direction change, and generate a path adjustment change coefficient; S313: According to the path adjustment change coefficient, adjust the reference position of the path coordinate, and generate a trajectory adjustment instruction result.

6. The real-time image acquisition based setup automation control method of claim 1, wherein, The acquisition step of the angle rollback reference data is specifically: S511: Extract the load adjustment amplitude information, analyze the continuously collected workpiece trajectory change data, and establish an ordered trajectory data sequence; S512: Based on the ordered trajectory data sequence, construct a linear vector for every two frames of trajectory center coordinates, calculate the direction difference of the two vectors, analyze the direction offset degree of each change of the trajectory, and generate a trajectory direction offset value; S513: Using the trajectory direction offset value, the correlation index is calculated in combination with the load adjustment amplitude, the relationship between the load adjustment amplitude and the trajectory direction change is judged, the rollback condition is adjusted, and the angle rollback reference data is generated.

7. A real-time image acquisition based setting automation control system, characterized in that, The system is used to realize the setting automatic control method based on real-time image acquisition in any one of claims 1-6, and the system comprises: The light spot merging recognition module calls the image acquisition sequence, analyzes the image of the structured light array projected on the workpiece surface, identifies the pixel center position of each light spot, merges the horizontal and vertical coordinates according to the number, connects as a contour line, compares the contour shape according to the corresponding relationship between the contour and the light spot grouping, and generates contour continuous offset information; The multi-view partition fusion module extracts the contour continuous offset information, obtains multi-camera images, divides the images into rectangular regions, locates the center pixel points of the regions, divides the gray levels according to the gray values, merges the multi-view center coordinates, analyzes the consistency of the numbers, filters the difference regions, and generates partition offset feature data; The path optimization comparison module calls the contour continuous offset information and the partition offset feature data, constructs a workpiece surface reference path, analyzes the path coordinates, compares the direction difference of adjacent coordinates, judges the relationship with the partition number, adjusts the path coordinates, and establishes a trajectory adjustment instruction result; The load trend analysis module judges the load measurement data generated in the current execution unit action process according to the trajectory adjustment instruction result, compares the position offset of the continuous period load measurement results, calculates the load change trend, and judges the relevance of the load change and the trajectory adjustment, and generates load adjustment amplitude information; The rollback reference setting module extracts the load adjustment amplitude information, analyzes the continuously collected workpiece trajectory change data, establishes an ordered trajectory data sequence, constructs a linear vector for the center coordinates of two consecutive trajectories, calculates the direction difference, analyzes the offset degree, calculates the correlation in combination with the load adjustment amplitude, adjusts the rollback condition, and generates angle rollback reference data.

Images