A robot polishing offline positioning and recognition design system
By polishing the target module, verification module and positioning module of the offline positioning identification system, combining lighting information and grayscale processing, the problems of low positioning identification efficiency and unstable accuracy in the existing technology are solved, and more efficient and stable location recognition.
Patent Information
- Application Number
- CN202410467829.6
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- Filing Date
- 2024-04-18
- Publication Date
- 2024-11-08
- Estimated Expiration
- 2044-04-18
AI Technical Summary
The existing robot grinding technology is inefficient in offline positioning and recognition, and is susceptible to environmental changes and workpiece differences, making it difficult to ensure grinding accuracy and stability, especially for complex-shaped workpieces.
A robot polishing offline positioning recognition system is designed, including target module, calibration module and positioning module. The target positioning points are set through preset position evaluation model and lighting information analysis, and the grayscale processing and difference judgment model are checked. , ensure positioning accuracy.
Improve positioning accuracy, reduce environmental interference impact, reduce staff burden, improve work efficiency, and reduce positioning impact when light changes.
Smart Images

Figure CN118172411B_ABST
Abstract
Description
Technical Field
[0001] The invention belongs to the technical field of robot polishing positioning, and in particular is a robot polishing off-line positioning recognition design system. Background Art
[0002] With the rapid development of industrial automation, robotic polishing technology has become an indispensable part of modern manufacturing. However, the existing robotic polishing technology still has some problems in practical application, especially in offline positioning and recognition.
[0003] Traditional robot grinding systems usually rely on manual teaching or online detection for positioning and identification, which is not only inefficient but also easily affected by environmental changes and workpiece differences, making it difficult to ensure grinding accuracy and stability. In addition, for workpieces with complex shapes, traditional positioning and identification methods are often difficult to handle and cannot meet the needs of high-precision grinding.
[0004] Based on this, the present invention provides a robot polishing offline positioning and recognition design system. Summary of the invention
[0005] In order to solve the problems existing in the above-mentioned solutions, the present invention provides a robot polishing offline positioning and recognition design system.
[0006] The purpose of the present invention can be achieved through the following technical solutions:
[0007] A robot polishing offline positioning and recognition design system, including a target module, a verification module and a positioning module;
[0008] The target module is used to analyze the polishing target, establish a target model of the polishing target, and set each target positioning point based on the target model.
[0009] Furthermore, the method for setting the target positioning point includes:
[0010] The target model is analyzed by a preset positioning evaluation model to obtain each position to be selected; the illumination information corresponding to the polishing target is obtained, and the illumination information includes standard illumination information and each floating illumination information;
[0011] Performing simulation evaluation on each of the to-be-selected sites according to the illumination information to obtain a positioning value corresponding to each of the to-be-selected sites;
[0012] The sites to be selected are sorted in ascending order of positioning values, and the first A sites to be selected are marked as target positioning points, where A is a positive integer.
[0013] Furthermore, the expression of the positioning evaluation model is ;
[0014] Where: s is the input data; the output data is the positioning evaluation value 1 or 0.
[0015] Further, the method for simulating and evaluating each candidate positioning point according to the illumination information includes:
[0016] Obtain the acquisition images of each candidate positioning point in the grinding target under standard illumination conditions, and mark them as standard images; perform gray-scale processing on the standard images to obtain the reference gray-scale images corresponding to each candidate positioning point, identify the brightness corresponding to each pixel in the reference gray-scale images, and mark it as d(x, y), where (x, y) represents the pixel at the corresponding position;
[0017] Obtain the acquisition images of each candidate positioning point under each floating illumination condition, and mark them as floating images; perform gray-scale processing on the floating images to obtain the floating gray-scale images corresponding to each candidate positioning point, identify the brightness corresponding to each pixel in the floating gray-scale images, and mark it as d´(x, y)c, where c = 1, 2,..., v, and v is a positive integer; set the specific gravity values corresponding to each floating gray-scale image;
[0018] According to the formula Calculate the positioning values of each candidate positioning point;
[0019] Where: PD is the positioning value; δc is the corresponding specific gravity value; c = 1, 2,..., v, and v is a positive integer; U is the set of (x, y); D(x, y)c is the difference judgment model.
[0020] Further, the expression of the difference judgment model is:
[0021] ;
[0022] Where: X1 is the threshold; d(x, y) is the brightness of the corresponding pixel in the reference gray-scale image; d´(x, y)c is the brightness of the corresponding pixel in the floating gray-scale image.
[0023] The verification module is used to perform positioning verification on each target positioning point, determine whether each target positioning point meets the visual positioning accuracy requirements. When it is judged that the visual positioning accuracy is not affected, a normal verification result is obtained; when it is judged that the visual positioning accuracy is affected, an abnormal verification result is obtained;
[0024] Cancel the target positioning points with abnormal verification results; count the number B of target positioning points with normal verification results, and B is an integer; when B ≥ B´, mark the visual positioning method as the target positioning method; B´ is the preset verification value; when B < B´, mark the additional positioning method as the target positioning method.
[0025] Further, the method for determining whether each target positioning point meets the visual positioning accuracy requirements includes:
[0026] Obtain a captured image of the target positioning point, mark it as a monitoring image, perform grayscale processing on the monitoring image to obtain a monitoring grayscale image; obtain a reference grayscale image; perform comparative analysis based on the monitoring grayscale image and the reference grayscale image to obtain a brightness difference map;
[0027] Set a conversion model, perform binarization processing on the brightness difference map through the conversion model to obtain a brightness conversion map; divide the brightness conversion map into E unit areas in an equal manner; calculate the unit value of each unit area;
[0028] The check matrix R is set according to the cell values of each cell area, and the elements in the check matrix are marked as rij, i = 1, 2, ..., n, n is a positive integer; j = 1, 2, ..., m, m is a positive integer; then the check matrix ;
[0029] According to the formula Calculate the corresponding verification evaluation value;
[0030] Where: HP is the calibration evaluation value; R F represents the Frobenius norm of the calibration matrix R;
[0031] When the check evaluation value is greater than the threshold value X2, it is determined that the visual positioning accuracy is not affected;
[0032] When the verification evaluation value is not greater than the threshold value X2, it is determined that the visual positioning accuracy is affected.
[0033] Furthermore, the expression of the transformation model is ;
[0034] Where: q is the difference value corresponding to each pixel in the brightness difference map; the output data is the difference conversion value; X1 is the threshold.
[0035] Furthermore, the method for calculating the unit value of each unit area includes:
[0036] Identify the difference conversion value corresponding to each pixel in the unit area; mark the difference conversion value in the unit area as Yz, where z=1, 2, ..., g, and g is a positive integer;
[0037] According to the formula Calculate the cell value of each cell area;
[0038] Where: DY is the unit value.
[0039] The positioning module is used to locate the polishing target, obtain the corresponding target positioning method, and control the robot to perform positioning identification based on each target positioning point and the target positioning method.
[0040] Compared with the prior art, the present invention has the following beneficial effects:
[0041] Through the mutual cooperation among the target module, verification module and positioning module, the offline positioning of the robot grinding is realized, the positioning accuracy is improved, and the impact of environmental interference is reduced; by setting the target module, an intelligent model of the grinding target is realized, and the staff is assisted in establishing the corresponding target model based on the actual situation of the grinding target, which reduces the burden on the staff and improves work efficiency. In the process of setting the target positioning point, the influence of light is fully considered on the current screened benchmark, which reduces the impact of subsequent light changes on positioning and improves positioning accuracy. BRIEF DESCRIPTION OF THE DRAWINGS
[0042] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.
[0043] Figure 1 This is a principle block diagram of the present invention. DETAILED DESCRIPTION
[0044] The technical solution of the present invention will be clearly and completely described below in conjunction with the embodiments. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.
[0045] like Figure 1 As shown, a robot polishing offline positioning and recognition design system includes a target module, a verification module and a positioning module;
[0046] The target module is used to analyze the polishing structures, components, etc. that need to be polished, mark the structures, components, etc. that need to be polished as polishing targets, collect relevant information such as the size, material, shape, etc. of the polishing targets, mark them as target part information, establish a three-dimensional model of the polishing target based on the obtained target part information, and mark the three-dimensional model of the polishing target as the target model;
[0047] A large amount of positioning point selection data is obtained from the current positioning field, that is, the relevant data for selecting positioning points on various polishing targets under the same conditions as this visual system. A corresponding positioning evaluation model is established based on the obtained positioning point selection data to evaluate whether each position on the polishing target can be used as a positioning point, that is, whether it meets the positioning point setting requirements, such as whether the position information can be collected during operation, whether it is affected by work, whether it is different from the surrounding area, whether it is easy to identify and other related requirements. The specific positioning requirements need to be set according to the actual polishing target and polishing robot, and the establishment is combined with the isolated forest algorithm. Points that do not meet the positioning requirements are regarded as abnormal data, and the size of the specific points is set by professionals; the expression of the positioning evaluation model is , s is the input data, i.e. the model information at the corresponding position of the target model, such as shape, material, color and other related data; the output data is the positioning evaluation value 1 or 0;
[0048] Each position in the target model is evaluated by the positioning evaluation model to obtain a positioning evaluation value of each position, and a position with a positioning evaluation value of 0 is marked as a position to be selected;
[0049] Obtain the lighting information corresponding to the polishing target during subsequent polishing, that is, the lighting type data of various lighting conditions that may appear in the working environment, mark the lighting condition with the largest proportion as standard lighting information, and mark other lighting conditions as floating lighting information;
[0050] Under standard lighting conditions, the collected images corresponding to each to-be-selected site in the polishing target are collected and marked as standard images; grayscale processing is performed on the standard images to obtain reference grayscale images corresponding to each to-be-selected site, and the brightness corresponding to each pixel in the reference grayscale image is identified and marked as d(x, y), where (x, y) represents the pixel at the corresponding position;
[0051] Under floating illumination conditions, the collected images corresponding to each site to be selected in the polishing target are collected and marked as floating images; grayscale processing is performed on the floating images to obtain floating grayscale images corresponding to each site to be selected, and the brightness corresponding to each pixel in the floating grayscale image is identified and marked as d´(x, y)c, where c=1, 2, ..., v, and v is a positive integer; because there are many floating illumination conditions, a site to be selected will have multiple floating grayscale images; the weight value corresponding to each floating grayscale image is set, that is, the weight value of the floating illumination condition, and is set according to its occurrence probability;
[0052] According to the formula Calculate the positioning value of each site to be selected; where: PD is the positioning value; δc is the corresponding specific gravity value; c=1, 2, ..., v, v is a positive integer; U is the set of (x, y), that is, the value range corresponding to (x, y); D(x, y)c is the difference judgment model; the expression of the difference judgment model is ; where: X1 is a threshold value, which is set by professionals according to the influence of light changes on visual positioning;
[0053] Sort each candidate positioning point in ascending order of the positioning value, and select the first A candidate positioning points before sorting as the target positioning points; A is a positive integer, which is set according to the number of target positioning points required.
[0054] By setting the target module, an intelligent model for the grinding target is realized. Based on the actual situation of the grinding target, it helps the staff to establish the corresponding target model, reduces the burden on the staff, improves work efficiency, and fully considers the influence of light during the setting process of the target positioning point, reduces the influence of subsequent light changes on positioning, and improves positioning accuracy.
[0055] The verification module is used to verify the positioning of the target positioning points, and judge whether the current light environment affects the visual positioning accuracy. When it is judged that it does not affect the visual positioning accuracy, it means that the influence degree is within the allowable range and is regarded as no influence; obtain a normal verification result; when it is judged that it affects the visual positioning accuracy, obtain an abnormal verification result; cancel the target positioning points with an abnormal verification result, that is, it cannot be used as a positioning point under the current conditions until the verification is normal; count the number B of target positioning points with a normal verification result, B is an integer; when B≥B´, mark the visual positioning method as the target positioning method; B´ is a preset verification value, which specifically needs to be set according to the number of positioning points required for visual positioning, and is generally set to 2 or a positive integer greater than 2; when B<B´, mark the additional positioning method as the target positioning method.
[0056] The visual positioning method is set according to the current visual positioning technology.
[0057] The additional positioning method is other existing positioning methods, but the efficiency may not be as high as the visual positioning method; specifically preset the corresponding additional positioning method according to the actual situation.
[0058] The method for judging whether each target positioning point meets the visual positioning accuracy requirements includes:
[0059] Obtain the acquisition image of the target positioning point, mark it as the monitoring image, perform gray processing on the monitoring image to obtain the monitoring gray image; obtain the reference gray image; perform comparative analysis based on the monitoring gray image and the reference gray image to obtain the brightness difference map, that is, align the monitoring gray image with the reference gray image, which can be aligned in combination with the corresponding gray values, calculate the absolute value of the brightness difference between each pixel as the difference value at the corresponding position of each pixel in the brightness difference map, and form the brightness difference map; set the conversion model, and the expression of the conversion model is , where: q is the difference value corresponding to each pixel in the brightness difference map; the output data is the difference conversion value; X1 is the threshold, which is the same as X1 in the difference judgment model;
[0060] The brightness difference map is binarized by the conversion model to obtain a brightness conversion map; the brightness conversion map is divided into E unit areas, where E is the area according to 2 2 , 3 2 , 4 2 , ..., which is set manually, but the maximum value is not greater than 100; divide it into equal parts; identify the difference conversion value corresponding to each pixel in the unit area; mark the difference conversion value in the unit area as Yz, z=1, 2, ..., g, g is a positive integer; according to the formula Calculate the unit value of each unit area; where: DY is the unit value;
[0061] The check matrix R is set according to the cell values of each unit area, that is, the matrix template is preset according to the E value, and the cell values corresponding to each unit area are subsequently substituted; the elements in the check matrix are marked as rij, i=1, 2, ..., n, n is a positive integer; j=1, 2, ..., m, m is a positive integer; then the check matrix ;
[0062] According to the formula Calculate the corresponding calibration evaluation value; where: HP is the calibration evaluation value; R F represents the Frobenius norm of the calibration matrix R;
[0063] When the check evaluation value is greater than the threshold value X2, it is determined that the visual positioning accuracy is not affected;
[0064] When the verification evaluation value is not greater than the threshold value X2, it is determined that the visual positioning accuracy is affected.
[0065] The positioning module is used to locate the polishing target, obtain the corresponding target positioning method, and control the robot to perform positioning identification based on each target positioning point and the target positioning method.
[0066] The above formulas are all calculated by removing dimensions and taking numerical values. The formula is a formula that is closest to the actual situation obtained by collecting a large amount of data and performing software simulation. The preset parameters and preset thresholds in the formula are set by technical personnel in this field according to actual conditions or obtained by simulating a large amount of data.
[0067] The above embodiments are only used to illustrate the technical method of the present invention rather than to limit it. Although the present invention has been described in detail with reference to the preferred embodiments, those skilled in the art should understand that the technical method of the present invention may be modified or replaced by equivalents without departing from the spirit and scope of the technical method of the present invention.
Claims
1. A robot polishing offline positioning and recognition design system, characterized in that: It includes a target module, a verification module, and a positioning module; The target module is used to analyze the grinding target, establish a target model of the grinding target, and set each target positioning point based on the target model; The verification module is used to perform positioning verification on each target positioning point, determine whether each target positioning point meets the visual positioning accuracy requirements. When it is judged that the visual positioning accuracy is not affected, a normal verification result is obtained; when it is judged that the visual positioning accuracy is affected, an abnormal verification result is obtained; Cancel the target positioning points with an abnormal verification result; Count the number B of target positioning points with a normal verification result, where B is an integer; when B≥B´, mark the visual positioning method as the target positioning method; B´ is a preset verification value; when B<B´, mark the additional positioning method as the target positioning method, and the additional positioning method is a positioning method other than the visual positioning method; The positioning module is used to position the grinding target, obtain the corresponding target positioning method, and control the robot to perform positioning recognition based on each target positioning point and the target positioning method; The method for setting target positioning points includes: Analyze the target model through a preset positioning evaluation model to obtain each candidate positioning point; obtain the illumination information corresponding to the grinding target, and the illumination information includes standard illumination information and each floating illumination information; the standard illumination information is the illumination information with the largest proportion when the grinding target is ground subsequently, and the non-standard illumination information is marked as floating illumination information; Perform simulation evaluation on each candidate positioning point according to the illumination information to obtain the positioning value corresponding to each candidate positioning point; Sort each candidate positioning point in ascending order of the positioning value, and mark the first A candidate positioning points as target positioning points, where A is a positive integer; The method for performing simulation evaluation on each candidate positioning point according to the illumination information includes: Obtain the acquisition image of each candidate positioning point in the grinding target under the standard illumination condition and mark it as the standard image; the standard illumination condition is the illumination condition corresponding to the standard illumination information; perform gray processing on the standard image to obtain the reference gray image corresponding to each candidate positioning point, and identify the brightness of each pixel in the reference gray image and mark it as d(x, y), where (x, y) represents the pixel at the corresponding position; Obtain the acquisition image of each candidate positioning point under each floating illumination condition and mark it as the floating image; the floating illumination condition is the illumination condition corresponding to the floating illumination information; perform gray processing on the floating image to obtain the floating gray image corresponding to each candidate positioning point, and identify the brightness of each pixel in the floating gray image and mark it as d´(x, y)c, where c = 1, 2,..., v and v is a positive integer; set the proportion value corresponding to each floating gray image; According to the formula Calculate the positioning value of each site to be selected; In the formula: PD is the positioning value; δc is the corresponding proportion value; c = 1, 2,..., v and v is a positive integer; U is the set of (x, y); D(x, y)c is the difference judgment model; The expression of the difference judgment model is: ; In the formula: X1 is the threshold; d(x, y) is the brightness of the corresponding pixel in the reference gray image; d´(x, y)c is the brightness of the corresponding pixel in the floating gray image; Methods for determining whether each target positioning point meets the visual positioning accuracy requirements include: Obtain a captured image of the target positioning point, mark it as a monitoring image, perform grayscale processing on the monitoring image to obtain a monitoring grayscale image; obtain a reference grayscale image; perform comparative analysis based on the monitoring grayscale image and the reference grayscale image to obtain a brightness difference map; Set a conversion model, perform binarization processing on the brightness difference map through the conversion model to obtain a brightness conversion map; divide the brightness conversion map into E unit areas in an equal manner; calculate the unit value of each unit area; The check matrix R is set according to the cell values of each cell area, and the elements in the check matrix are marked as rij, i = 1, 2, ..., n, n is a positive integer; j = 1, 2, ..., m, m is a positive integer; then the check matrix ; According to the formula Calculate the corresponding verification evaluation value; Where: HP is the calibration evaluation value; ||R|| F represents the Frobenius norm of the calibration matrix R; When the check evaluation value is greater than the threshold value X2, it is determined that the visual positioning accuracy is not affected; When the check evaluation value is not greater than the threshold value X2, it is determined that the visual positioning accuracy is affected; Methods for calculating the cell value of each cell area include: Identify the difference conversion value corresponding to each pixel in the unit area; mark the difference conversion value in the unit area as Yz, where z=1, 2, ..., g, and g is a positive integer; According to the formula Calculate the cell value of each cell area; Where: DY is the unit value.
2. A robot polishing offline positioning recognition design system according to claim 1, characterized in that: The expression of the positioning evaluation model is: ; Where: s is the input data; the output data is the positioning evaluation value 1 or 0.
3. The robot polishing offline positioning recognition design system according to claim 1 is characterized in that: The expression of the transformation model is ; Where: q is the difference value corresponding to each pixel in the brightness difference map; the output data is the difference conversion value; X1 is the threshold.
