Industrial robot visual positioning device, method and medium

Abstract

The present invention discloses an industrial robot visual positioning device, method and medium, comprising a visual acquisition module, an environment monitoring module, a central processing unit and a performance optimization module. The visual acquisition module is used to perform 3D visual positioning of a workpiece, thereby assisting in controlling a manipulator to perform fixed-point processing, and the environment monitoring module is used to measure the light of the workpiece processing environment, and then the central processing unit performs visual positioning impact analysis, evaluates the comprehensive impact of the workpiece's own motion changes and the workpiece processing environment light changes on the visual positioning accuracy, generates corresponding risk warning signals, and performs feedback refinement analysis through the risk warning signal level, analyzes the impact of the workpiece's own motion changes and the environment light changes on the visual positioning accuracy in turn, and generates a management signal group, and then receives and processes the signals through the performance optimization module, thereby improving the model application effect and realizing the adaptive growth of visual positioning performance.

Description

Technical Field

[0001] The present invention relates to the field of manipulator assistance technology, and in particular to an industrial robot visual positioning device, method and medium. Background Art

[0002] Modern manufacturing industry has increasing requirements for production efficiency and flexibility, especially in tasks such as assembly, welding, and painting, which require high-precision and high-speed processing capabilities. With the diversification of product types and miniaturization of batches, the working mode of traditional fixed-point robots is difficult to meet new needs. Therefore, intelligent industrial robot visual positioning devices are needed to assist in manipulating the manipulator.

[0003] The existing industrial robot visual positioning device has the defects of insufficient real-time and environmental adaptability. When the workpiece is in the process of dynamic transmission, the workpiece's own movement and data processing lag will cause the robot operation delay and lead to workpiece processing positioning deviation. When the workpiece processing environment has abnormal light interference, it will cause the workpiece surface reflection to affect the visual processing accuracy, and also cause the problem of poor workpiece positioning accuracy, affecting the reliability of the industrial robot manipulator control system. In view of the above technical defects, a solution is proposed. Summary of the invention

[0004] The purpose of the present invention is to solve the defects of insufficient real-time performance and environmental adaptability of existing industrial robot visual positioning devices. This scheme evaluates the comprehensive impact of the workpiece's own motion changes and the workpiece processing environment light changes on the visual positioning accuracy, generates corresponding risk warning signals and conducts feedback refinement analysis through the risk warning signal level, analyzes the impact of the workpiece's own motion changes and the ambient light changes on the visual positioning accuracy in turn and generates a management signal group, and then receives and processes the signals through the performance optimization module to improve the model application effect, realize the adaptive growth of visual positioning performance, and enhance the adaptability of the device in a dynamic environment, thereby improving the workpiece production quality.

[0005] In order to achieve the above object, the present invention adopts the following technical solutions:

[0006] A visual positioning method for an industrial robot comprises the following steps:

[0007] Step 1: Build a 3D stereo model of the workpiece through the visual acquisition module and perform visual positioning: obtain the 3D stereo model of the workpiece through the visual sensor and mark it as workpiece model A, obtain the velocity vector of workpiece model A in real time, and locate the feature points of workpiece model A. Perform fixed-point processing through the robot arm and mark the solid model of the workpiece as workpiece model B.

[0008] Step 2: Obtain the light parameters of the workpiece processing environment through the environment monitoring module: the light parameters of the workpiece processing environment include light intensity, luminous flux, illumination and brightness;

[0009] Step 3: Build a visual positioning impact model through the central processing unit: evaluate the visual positioning accuracy by analyzing workpiece model A and workpiece model B, and then analyze the change function between the velocity vector of workpiece model A and the visual positioning accuracy, as well as the change function between the light parameters of the workpiece processing environment and the visual positioning accuracy, so as to comprehensively evaluate the comprehensive impact of the workpiece's own motion changes and the light changes in the workpiece processing environment on the visual positioning accuracy, obtain the corresponding level of risk warning signals, and then feedback to generate a management signal group;

[0010] Step 4: Receive risk warning signals and management signal groups through the performance optimization module and perform corresponding processing.

[0011] Furthermore, the specific process of visual positioning by the visual acquisition module is as follows:

[0012] A1, obtain the 3D stereo model of the workpiece through the visual sensor and mark it as workpiece model A, locate the feature points of workpiece model A, mark any feature point as p, obtain the coordinates of point p, feed the coordinates of point p back to the robot for fixed-point processing, mark the solid model of the workpiece as workpiece model B, mark the positioning point of workpiece model B after fixed-point processing by the robot as q, and obtain the coordinates of point q;

[0013] A2, the visual acquisition module obtains the velocity vector of the workpiece model A in real time, and marks the velocity vector of the workpiece model A as Va, where the velocity vector Va includes the linear velocity υ, angular velocity ω, translation velocity v and translation direction r of the workpiece model A, and marks the velocity vector Va = <υ, ω, v, r>.

[0014] Furthermore, the specific process of the CPU constructing the visual positioning influence model is as follows:

[0015] B1, by analyzing the coordinate deviation between the feature point p of workpiece model A and the positioning point q of workpiece model B, the visual positioning accuracy coefficient Xjd is obtained to evaluate the initial accuracy of visual positioning;

[0016] B2, by analyzing the change function between the velocity vector of the workpiece model A and the visual positioning accuracy, the workpiece motion influence coefficient Xyd is obtained, thereby evaluating the first degree of influence of the workpiece's own motion change on the visual positioning accuracy;

[0017] B3, by analyzing the change function between the light parameters of the workpiece processing environment and the visual positioning accuracy, the environmental light influence index Zgx is obtained, so as to evaluate the second influence degree of the light change of the workpiece processing environment on the visual positioning accuracy;

[0018] B4, by combining the workpiece motion influence coefficient Xyd and the ambient light influence index Zgx, comprehensively obtain the visual positioning accuracy influence index Zsj, set the evaluation interval of the visual positioning accuracy influence index Zsj, and evaluate the comprehensive influence of the workpiece's own motion change and the workpiece processing environment light change on the visual positioning accuracy through interval comparison;

[0019] B5, obtain the corresponding level of risk warning signal through the visual positioning accuracy impact index, and then feed it back to the workpiece motion impact coefficient Xyd and the ambient light impact index Zgx for detailed comparison, and generate the corresponding level of compensation adjustment signal and reference adjustment signal respectively, and adjust and refresh the control parameters of the robot;

[0020] B6, by adjusting and refreshing, recalculating the visual positioning accuracy coefficient, thereby evaluating the correction accuracy of the visual positioning, and then setting the threshold V of the visual positioning accuracy coefficient. When the visual positioning accuracy coefficient is lower than the preset threshold V, a model optimization signal is generated to optimize the fitting function in the mathematical model;

[0021] The compensation adjustment signal, the reference adjustment signal and the model optimization signal are integrated and labeled as a management signal group.

[0022] Furthermore, the specific process of evaluating the visual positioning accuracy is as follows:

[0023] B1-1, synchronously extract n0 feature points p of workpiece model A and n0 positioning points q of workpiece model B, mark the coordinates of feature point p as (Xp, Yp, Zp), and mark the coordinates of positioning point q as (Xq, Yq, Zq);

[0024] B1-2, by analyzing the coordinate deviation between the feature point p of the workpiece model A and the positioning point q of the workpiece model B, the visual positioning accuracy coefficient Xjd is obtained;

[0025] B1-3, set the evaluation interval of the visual positioning accuracy coefficient Xjd to evaluate the visual positioning accuracy.

[0026] Furthermore, the specific process of evaluating the first impact level is as follows:

[0027] B2-1, analyze the velocity vector and visual positioning accuracy of workpiece model A;

[0028] B2-101, feature point compensation positioning of workpiece model A is performed through velocity vector Va;

[0029] The specific process of compensation positioning is as follows: set the visual acquisition cycle Tc to perform 3D dynamic simulation on the workpiece model A, and obtain the motion trajectory Sp of the feature point p of the workpiece model A; mark the time difference between the robot receiving the instruction and the fixed-point processing as the delay cycle Ty, so as to obtain the compensation point b of the feature point p through the motion trajectory Sp; mark the coordinates of the compensation point b as (Xb, Yb, Zb), issue a control instruction to the robot through the coordinates of the compensation point b, and the robot performs fixed-point processing on the workpiece model B after the delay period Ty, and obtains the coordinates of the positioning point q of the workpiece model B;

[0030] B2-102, by compensating for the coordinate deviation of the feature point p and the positioning point q in the positioning operation, the corresponding visual positioning accuracy coefficient is obtained, and it is marked as the accuracy coefficient Xjd1 of the visual positioning compensation;

[0031] B2-103, through the velocity vector Va = <υ, ω, v, r>, fit and construct the variation function F1 between the velocity vector Va and the accuracy coefficient Xjd1 of the visual positioning compensation;

[0032] B2-2, obtain the workpiece motion influence coefficient Xyd through the change function F1 between the velocity vector Va and the accuracy coefficient Xjd1 of the visual positioning compensation;

[0033] B2-3, setting the evaluation interval of the workpiece motion influence coefficient Xyd, so as to evaluate the first influence degree of the workpiece's own motion change on the visual positioning accuracy, and generate a corresponding compensation adjustment signal.

[0034] Furthermore, the specific process of evaluating the second impact level is as follows:

[0035] B3-1, by analyzing the light parameters and visual positioning accuracy of the workpiece processing environment;

[0036] The light parameters of the workpiece processing environment include light intensity, luminous flux, illumination and brightness;

[0037] The workpiece processing environment is divided into N0 characteristic areas, any characteristic area is marked as z, and the light intensity, luminous flux, illumination and brightness of the characteristic area z are collected and marked as Gq, Gt, Zd, Ld respectively;

[0038] The ambient light influence coefficient Xgx of the characteristic area z is comprehensively obtained through the light intensity Gq, luminous flux Gt, illuminance Zd and brightness Ld of the characteristic area z;

[0039] A threshold Hg of the ambient light influence coefficient Xgx is set. When the ambient light influence coefficient Xgx of the feature area z exceeds the threshold Hg, the feature area z is determined to be an abnormal light area; otherwise, the feature area z is determined to be a normal light area;

[0040] When the feature point p is located in the normal light area, the robot is controlled to perform fixed-point processing according to the coordinates of the feature point p;

[0041] When the feature point p is located in the abnormal light area, the neighboring point c of the feature point p is extracted, and the workpiece model is referenced and positioned by the coordinates of the neighboring point c;

[0042] By referring to the coordinates of the feature point p in the positioning operation and the coordinate deviation of the positioning point q, the corresponding visual positioning accuracy coefficient is obtained and marked as the accuracy coefficient Xjd2 of the visual positioning reference;

[0043] Fitting constructs the variation function F2 between the ambient light influence coefficient Xgx and the accuracy coefficient Xjd2 of the visual positioning reference;

[0044] B3-2, comprehensively obtain the ambient light impact index Zgx of the workpiece processing environment through the change function F2 between the ambient light impact coefficient Xgx of N0 feature areas and the accuracy coefficient Xjd2 of their corresponding visual positioning reference;

[0045] B3-3, setting the evaluation interval of the ambient light impact index Zgx, thereby evaluating the second impact degree of the light change in the workpiece processing environment on the visual positioning accuracy, and generating a corresponding reference adjustment signal.

[0046] Furthermore, the specific process of referencing and positioning the workpiece model is as follows:

[0047] Mark the coordinates of the neighboring point c as (Xc, Yc, Zc), take the neighboring point c as the starting point and the feature point p as the target point, and obtain the displacement vector Through the coordinates of the neighboring point c and the displacement vector A control instruction is issued to the robot to obtain the target point coordinates of the workpiece model B. The robot performs fixed-point processing on the target point coordinates of the workpiece model B to obtain the coordinates of the positioning point q of the workpiece model B.

[0048] An industrial robot visual positioning device: comprising a visual acquisition module, an environmental monitoring module, a central processing unit and a performance optimization module, wherein the visual acquisition module, the environmental monitoring module, the central processing unit and the performance optimization module are communicatively connected, and the device applies the above-mentioned industrial robot visual positioning method.

[0049] Furthermore, the visual acquisition module is used to construct a 3D stereo model of the workpiece and perform visual positioning;

[0050] The environment monitoring module is used to obtain the light parameters of the workpiece processing environment;

[0051] The central processing unit is used to build a visual positioning impact model, comprehensively evaluate the comprehensive impact of the workpiece's own motion changes and the light changes in the workpiece processing environment on the visual positioning accuracy, obtain the corresponding level of risk warning signals, and then feedback to generate a management signal group;

[0052] The performance optimization module is used to receive risk warning signals and management signal groups and perform corresponding processing.

[0053] A computer-readable storage medium stores a computer program, which, when executed by a central processing unit, applies the above-mentioned industrial robot visual positioning method.

[0054] In summary, due to the adoption of the above technical solution, the beneficial effects of the present invention are:

[0055] The present invention uses a visual acquisition module to perform 3D visual positioning of the workpiece, thereby assisting in controlling the manipulator to perform fixed-point processing, and uses an environmental monitoring module to measure the light of the workpiece processing environment, and then uses a central processing unit to perform visual positioning impact analysis, evaluate the comprehensive impact of the workpiece's own motion changes and the workpiece processing environment light changes on the visual positioning accuracy, ensure the integrity and comprehensiveness of data processing, generate corresponding risk warning signals, and quantify the risk level of deviation in visual positioning processing, which is convenient for subsequent feedback analysis and optimization management;

[0056] The present invention conducts feedback refinement analysis through risk warning signals, evaluates the effects of the workpiece's own motion changes and ambient light changes on the visual positioning accuracy in turn, generates a management signal group, and then receives and processes the signals through a performance optimization module, thereby improving the real-time performance of the industrial robot's visual positioning from the perspective of positioning compensation, ensuring the accuracy of visual positioning processing when the workpiece is in a dynamic transmission state, and performing optimization management from the perspective of positioning and adjustment with reference to adjacent points, so as to make up for the defect of insufficient environmental adaptability of the industrial robot's visual positioning, ensuring the visual positioning accuracy when the workpiece is in an environment with changing light, and then continuously improving the model application effect through model self-growth, thereby realizing adaptive growth of visual positioning performance. BRIEF DESCRIPTION OF THE DRAWINGS

[0057] Figure 1 A schematic diagram of a device module of the present invention is shown;

[0058] Figure 2 A schematic diagram showing the steps of the overall process of the present invention is shown;

[0059] Figure 3 A schematic flow chart of the visual positioning influence model of the present invention is shown. DETAILED DESCRIPTION

[0060] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. 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.

[0061] Embodiment 1:

[0062] like Figure 1-Figure 3 As shown, an industrial robot visual positioning device includes a visual acquisition module, an environmental monitoring module, a central processing unit and a performance optimization module. The visual acquisition module, the environmental monitoring module, the central processing unit and the performance optimization module are communicated and connected. The device applies an industrial robot visual positioning method as described below. The working steps are as follows:

[0063] S1, the visual acquisition module builds a 3D stereo model of the workpiece and performs visual positioning: the 3D stereo model of the workpiece is obtained by the visual sensor and marked as workpiece model A, the velocity vector of workpiece model A is obtained in real time, and the feature points of workpiece model A are located, and fixed-point processing is performed by the manipulator, and the solid model of the workpiece is marked as workpiece model B. The specific process is as follows:

[0064] A1, obtain the 3D stereo model of the workpiece through the visual sensor and mark it as workpiece model A, locate the feature points of workpiece model A, mark any feature point as p, obtain the coordinates of point p, feed the coordinates of point p back to the robot for fixed-point processing, mark the solid model of the workpiece as workpiece model B, mark the positioning point of workpiece model B after fixed-point processing by the robot as q, and obtain the coordinates of point q;

[0065] Among them, the work that requires fixed-point processing, such as drilling and spraying, can achieve visual-assisted positioning of the robot through the visual acquisition module;

[0066] A2, the visual acquisition module obtains the velocity vector of the workpiece model A in real time, and marks the velocity vector of the workpiece model A as Va; wherein the velocity vector Va includes the linear velocity υ, angular velocity ω, translation velocity v and translation direction r of the workpiece model A, that is, Va = <υ, ω, v, r>;

[0067] Linear velocity υ and angular velocity ω are used to monitor the dynamic changes of the workpiece during three-dimensional rotation;

[0068] The translation speed v and translation direction r are used to monitor the dynamic changes of the workpiece during planar movement;

[0069] Set the reference direction, and mark the angle between the workpiece moving direction and the reference direction as the translation direction r. When the workpiece moving direction is consistent with the reference direction, r = 0;

[0070] S2, the environment monitoring module obtains the light parameters of the workpiece processing environment: the light parameters of the workpiece processing environment include light intensity, luminous flux, illumination and brightness;

[0071] S3, the central processing unit builds a visual positioning impact model: by analyzing the workpiece model A and the workpiece model B to evaluate the visual positioning accuracy, and then by analyzing the change function between the velocity vector of the workpiece model A and the visual positioning accuracy, as well as the change function between the light parameters of the workpiece processing environment and the visual positioning accuracy, the comprehensive impact of the workpiece's own motion change and the light change of the workpiece processing environment on the visual positioning accuracy is comprehensively evaluated, and the corresponding level of risk warning signals is obtained, and then a management signal group is generated;

[0072] B1, by analyzing the coordinate deviation between the feature point p of workpiece model A and the positioning point q of workpiece model B, the visual positioning accuracy coefficient Xjd is obtained to evaluate the initial accuracy of visual positioning. The specific process is as follows:

[0073] B1-1, synchronously extract n0 feature points p of workpiece model A and n0 positioning points q of workpiece model B, mark the coordinates of feature point p as (Xp, Yp, Zp), and mark the coordinates of positioning point q as (Xq, Yq, Zq);

[0074] B1-2, by analyzing the coordinate deviation between the feature point p of workpiece model A and the positioning point q of workpiece model B, the visual positioning accuracy coefficient Xjd is obtained:

[0075]

[0076] Among them, α0 is the conversion coefficient of the coordinate deviation value, and α0 is greater than 0; through the coordinates between the feature point p of the workpiece model A and the positioning point q of the workpiece model B, the deviation value between n0 groups of corresponding points is obtained and the conversion coefficient α0 is assigned, so as to obtain the visual positioning accuracy coefficient Xjd. The conversion coefficient α0 refers to a constant value that converts the coordinate deviation into the visual positioning accuracy coefficient Xjd, and α0 is preset after a large amount of experimental data is measured. When the coordinate deviation value between the feature point p and the positioning point q is higher, the visual positioning accuracy coefficient Xjd is lower, indicating that the visual positioning accuracy is worse;

[0077] B1-3, set the evaluation interval of the visual positioning accuracy coefficient Xjd to evaluate the visual positioning accuracy;

[0078] There are M0 evaluation intervals of the preset visual positioning accuracy coefficient Xjd. Any evaluation interval is marked as Qm, where m is the serial number of the evaluation interval and 1≤m≤M0. When the visual positioning accuracy coefficient Xjd is in the evaluation interval Qm, the evaluated visual positioning accuracy is level m.

[0079] B2, by analyzing the change function between the velocity vector of the workpiece model A and the visual positioning accuracy, the workpiece motion influence coefficient Xyd is obtained, so as to evaluate the first degree of influence of the workpiece's own motion change on the visual positioning accuracy. The specific process is as follows:

[0080] B2-1, analyze the velocity vector and visual positioning accuracy of workpiece model A. The specific process is as follows:

[0081] The feature point compensation positioning of the workpiece model A is performed through the velocity vector Va:

[0082] Set the visual acquisition cycle Tc to perform 3D dynamic simulation on the workpiece model A, and obtain the motion trajectory Sp of the feature point p of the workpiece model A;

[0083] The time difference between the robot receiving the instruction and the fixed-point processing is marked as the delay period Ty, so as to obtain the compensation point b of the feature point p through the motion trajectory Sp, wherein the compensation point b and the feature point p are both located on the motion trajectory Sp, the time difference between the compensation point b and the feature point p is a delay period Ty, and the compensation point b is in front and the feature point p is in the back;

[0084] The coordinates of the compensation point b are marked as (Xb, Yb, Zb), and a control instruction is issued to the robot through the coordinates of the compensation point b. After the robot undergoes a delay period Ty, it performs fixed-point processing on the workpiece model B and obtains the coordinates of the positioning point q of the workpiece model B;

[0085] The feature point compensation positioning operation is used to ensure that the first degree of influence on the visual positioning accuracy caused by the change of the workpiece's own movement is compensated when the workpiece is in motion. At this time, the visual positioning accuracy after feature point compensation positioning may still have a certain deviation. Therefore, it is necessary to continuously adjust the coordinate position of the compensation point according to the visual positioning error after compensation positioning to improve the accuracy of fixed-point processing when the workpiece is in motion;

[0086] By compensating for the coordinate deviation of the feature point p and the positioning point q in the positioning operation, the corresponding visual positioning accuracy coefficient is obtained and marked as the accuracy coefficient Xjd1 of the visual positioning compensation;

[0087] Through the velocity vector Va=<υ,ω,v,r>, the variation function F1 between the velocity vector Va and the accuracy coefficient Xjd1 of the visual positioning compensation is constructed by fitting: F1(Va)=Xjd1=F1(υ,ω,v,r);

[0088] Substitute the velocity vector Va into the variation function F1, that is, substitute the linear velocity υ, angular velocity ω, translation velocity v and translation direction r of the workpiece model A into the variation function F1, and then obtain the corresponding precision coefficient Xjd1 of the visual positioning compensation; the fitting method of the variation function F1 is realized by the multivariate function fitting tool in the existing mathematical model, such as the multivariate function fitting database in the common Matlab for polynomial fitting;

[0089] B2-2, obtain the workpiece motion influence coefficient Xyd through the change function F1 between the velocity vector Va and the accuracy coefficient Xjd1 of the visual positioning compensation:

[0090] Among them, α1, α2, α3 and α4 are weight factor coefficients of linear velocity υ, angular velocity ω, translation velocity v and translation direction r, respectively, and α1, α2, α3 and α4 are all greater than 0; the weight factor coefficients are preset and obtained after a large amount of data measurement, when the linear velocity υ, angular velocity ω, translation velocity v and translation direction r of the workpiece model A are higher, the workpiece motion influence coefficient Xyd is higher, indicating that the first influence of the workpiece's own motion change on the visual positioning accuracy is higher;

[0091] B2-3, setting the evaluation interval of the workpiece motion influence coefficient Xyd, so as to evaluate the first influence degree of the workpiece's own motion change on the visual positioning accuracy;

[0092] The evaluation interval of the preset workpiece motion influence coefficient Xyd is [Qe1, Qe2];

[0093] When the workpiece motion influence coefficient Xyd is lower than Qe1, the first influence degree is evaluated as level I;

[0094] When the workpiece motion influence coefficient Xyd is in the evaluation interval [Qe1, Qe2], the first influence degree is evaluated as level II;

[0095] When the workpiece motion influence coefficient Xyd is higher than Qe2, the first influence degree is evaluated as level II;

[0096] B3, by analyzing the change function between the light parameters of the workpiece processing environment and the visual positioning accuracy, the ambient light impact index Zgx is obtained, so as to evaluate the second degree of influence of the light change of the workpiece processing environment on the visual positioning accuracy. The specific process is as follows:

[0097] B3-1, by analyzing the light parameters and visual positioning accuracy of the workpiece processing environment;

[0098] The light parameters of the workpiece processing environment include light intensity, luminous flux, illumination and brightness;

[0099] The workpiece processing environment is divided into N0 characteristic areas, any characteristic area is marked as z, and the light intensity, luminous flux, illumination and brightness of the characteristic area z are collected and marked as Gq, Gt, Zd, Ld respectively;

[0100] The ambient light influence coefficient Xgx of the characteristic area z is obtained comprehensively through the light intensity Gq, luminous flux Gt, illuminance Zd and brightness Ld of the characteristic area z: Xgx=β1*Gq+β2*Gt+β3*Zd+β4*Ld;

[0101] Among them, β1, β2, β3 and β4 are weight factor coefficients of light intensity Gq, luminous flux Gt, illuminance Zd and brightness Ld, respectively, and β1, β2, β3 and β4 are all greater than 0; when the light intensity Gq, luminous flux Gt, illuminance Zd and brightness Ld are higher, the ambient light influence coefficient Xgx is higher, indicating that the light parameter influence degree of the feature area is higher;

[0102] A threshold Hg of the ambient light influence coefficient Xgx is set. When the ambient light influence coefficient Xgx of the feature area z exceeds the threshold Hg, the feature area z is determined to be an abnormal light area; otherwise, the feature area z is determined to be a normal light area;

[0103] When the feature point p is located in the normal light area, the robot is controlled to perform fixed-point processing according to the coordinates of the feature point p. At this time, the coordinate position of the feature point p is relatively clear.

[0104] When the feature point p is located in the abnormal light area, the workpiece model is referenced for feature point positioning. At this time, the coordinate position of the feature point p is caused by mirror reflection due to the strong light, which leads to unclear visual acquisition. Therefore, the neighboring point c of the feature point p is extracted, and the reference positioning is performed through the coordinates of the neighboring point c. The neighboring point c is collected and extracted through the adjacent normal light area;

[0105] Mark the coordinates of the neighboring point c as (Xc, Yc, Zc), take the neighboring point c as the starting point and the feature point p as the target point, and obtain the displacement vector Among them, the displacement vector Including the distance and direction from the neighboring point c to the feature point p;

[0106] Through the coordinates of the neighboring point c and the displacement vector The robot sends a control command to the robot. The robot first determines the coordinate position of the reference point C of the workpiece model B corresponding to the adjacent point c, and then takes the reference point C as the starting point and uses the displacement vector Perform reference positioning operation to obtain the target point coordinates of workpiece model B, and the manipulator performs fixed-point processing on the target point coordinates of workpiece model B to obtain the coordinates of positioning point q of workpiece model B;

[0107] By referring to the coordinates of the feature point p in the positioning operation and the coordinate deviation of the positioning point q, the corresponding visual positioning accuracy coefficient is obtained and marked as the accuracy coefficient Xjd2 of the visual positioning reference;

[0108] Construct a change curve between the ambient light influence coefficient Xgx and the accuracy coefficient Xjd2 of the visual positioning reference and fit to generate a change function F2: F2(Xgx)=Xjd2;

[0109] Substituting the ambient light influence coefficient Xgx into the variation function F2, the corresponding visual positioning reference accuracy coefficient Xjd2 is obtained; the fitting method of the variation function F2 is to first establish a variation curve between the independent variable and the dependent variable, and then use the function fitting tool in the existing mathematical model to perform linear analysis on the variation curve;

[0110] B3-2, through the change function F2 between the ambient light influence coefficient Xgx of N0 feature areas and the accuracy coefficient Xjd2 of their corresponding visual positioning reference, the ambient light influence index Zgx of the workpiece processing environment is comprehensively obtained:

[0111] Among them, γz is the weight coefficient of the ambient light influence coefficient Xgx of the feature area z, and γz is greater than 0; when the ambient light influence coefficient Xgx of the N0 feature areas is higher, the ambient light influence index Zgx of the workpiece processing environment is higher, indicating that the second degree of influence of the light change in the workpiece processing environment on the visual positioning accuracy is higher;

[0112] B3-3, setting the evaluation interval of the ambient light impact index Zgx, so as to evaluate the second degree of influence of the light change in the workpiece processing environment on the visual positioning accuracy;

[0113] The evaluation interval of the preset ambient light impact index Zgx is [Qe3, Qe4];

[0114] When the ambient light impact index Zgx is lower than Qe3, the second impact level is assessed as Level I;

[0115] When the ambient light impact index Zgx is in the evaluation interval [Qe3, Qe4], the second impact level is evaluated as Level II;

[0116] When the ambient light impact index Zgx is higher than Qe4, the second impact level is assessed as Level III;

[0117] B4, by combining the workpiece motion influence coefficient Xyd and the ambient light influence index Zgx, the visual positioning accuracy influence index is comprehensively obtained, so as to evaluate the comprehensive influence of the workpiece's own motion change and the light change of the workpiece processing environment on the visual positioning accuracy. The specific process is as follows:

[0118] B4-1, by combining the workpiece motion influence coefficient Xyd and the ambient light influence index Zgx, the visual positioning accuracy influence index Zsj is comprehensively obtained: Zsj = δ1*Xyd+δ2*Zgx;

[0119] Among them, δ1 and δ2 are weight factor coefficients of the workpiece motion influence coefficient Xyd and the ambient light influence index Zgx, respectively, and both δ1 and δ2 are greater than 0; when the workpiece motion influence coefficient Xyd and the ambient light influence index Zgx are higher, the visual positioning accuracy influence index Zsj is higher;

[0120] B4-3, set the evaluation interval of the visual positioning accuracy impact index Zsj, and evaluate the comprehensive impact of the workpiece's own motion changes and the light changes in the workpiece processing environment on the visual positioning accuracy through interval comparison;

[0121] B5, obtains the corresponding level of risk warning signal through the visual positioning accuracy impact index, and then feeds back to the workpiece motion impact coefficient Xyd and the ambient light impact index Zgx for detailed comparison, and generates corresponding levels of compensation adjustment signals and reference adjustment signals respectively, and adjusts and refreshes the control parameters of the manipulator, so that the positioning point q of the workpiece model B constantly coincides with the target point, thereby improving the visual positioning accuracy;

[0122] B5-1, there are M1 evaluation intervals for the preset visual positioning accuracy impact index Zsj, and any evaluation interval is marked as Qu, where u is the serial number of the evaluation interval and 1≤u≤M1. When the visual positioning accuracy impact index Zsj is in the evaluation interval Qu, the comprehensive impact of the workpiece's own motion change and the workpiece processing environment's light change on the visual positioning accuracy is u-level, and a u-level risk warning signal is generated;

[0123] When a risk warning signal is generated, the corresponding risk warning text is edited and displayed immediately, thereby prompting the device management personnel to perform visual positioning optimization management;

[0124] B5-2, the detailed comparison process of risk warning signal feedback to workpiece motion influence coefficient Xyd is as follows:

[0125] When the first impact level is assessed to be level I, a level I compensation adjustment signal is generated;

[0126] When the first impact level is assessed to be level II, a level II compensation adjustment signal is generated;

[0127] When the first impact level is assessed to be level II, a level III compensation adjustment signal is generated;

[0128] The coordinates of the compensation point b are compensated and adjusted according to the compensation adjustment signal of the corresponding level, and the specific adjustment range is set according to the actual situation;

[0129] B5-3, the detailed comparison process of risk warning signal feedback to ambient light impact index Zgx is as follows:

[0130] When the second impact level is assessed to be level I, a level I reference adjustment signal is generated;

[0131] When the second impact level is assessed to be level II, a level II reference adjustment signal is generated;

[0132] When the second impact level is assessed to be level III, a level III reference adjustment signal is generated;

[0133] According to the corresponding level of reference adjustment signal, the displacement vector of the reference point C Make corresponding adjustments, and the specific adjustment range is set according to the actual situation;

[0134] B6, by adjusting and refreshing, recalculating the visual positioning accuracy coefficient, so as to evaluate the correction accuracy of the visual positioning, and then setting the threshold V of the visual positioning accuracy coefficient. When the visual positioning accuracy coefficient is lower than the preset threshold V, a model optimization signal is generated to optimize the fitting function in the visual positioning influence model, and continuously improve the fitting accuracy of the change function, thereby improving the positioning performance of the model;

[0135] Among them, if the correction accuracy of the visual positioning after adjustment and refresh continues to improve, it means that the adjustment and refresh process is effective, and the function is to improve the visual positioning accuracy; on the contrary, if the visual positioning accuracy after adjustment and refresh is too low, it means that the adjustment and refresh process is invalid. When the visual positioning accuracy coefficient is lower than the preset threshold V, it means that there is still an abnormality in the visual positioning after the refresh adjustment, which further indicates that there is an abnormality in the visual positioning influence model itself. Therefore, it is necessary to continuously optimize the fitting function in the visual positioning influence model, for example, by conducting multiple visual positioning experiments with industrial robots, continuously increasing the amount of sample data, and incorporating fitting tools into curve fitting, so as to continuously improve the fitting accuracy of the change function;

[0136] The compensation adjustment signal, the reference adjustment signal and the model optimization signal are integrated and marked as a management signal group;

[0137] S4, the performance optimization module receives the risk warning signal and the management signal group and performs corresponding processing to improve the model application effect and realize the adaptive growth of visual positioning performance;

[0138] S4-1, according to the corresponding level of compensation adjustment signal, the coordinates of the compensation point b are compensated and adjusted, and the specific adjustment range is set according to the actual situation; for example, the level I compensation adjustment signal corresponds to the level I compensation adjustment range Ge1%, the level II compensation adjustment signal corresponds to the level II compensation adjustment range Ge2%, and the level III compensation adjustment signal corresponds to the level III compensation adjustment range Ge3%;

[0139] By compensating and adjusting the coordinates of the compensation point b, the coordinates (Xb1, Yb1, Zb1) of the refreshed compensation point b1 are obtained, where: Ge% refers to the compensation adjustment margin;

[0140] When Xp>Xq, then Xb1=Xb*(1+Ge%); when Xp<Xq, then Xb1=Xb*(1-Ge%);

[0141] The horizontal coordinate analysis process: the feature point p is the target point position of the robot fixed-point processing, and the positioning point q is the actual point position of the robot fixed-point processing. When the horizontal coordinate of the feature point p is greater than the horizontal coordinate of the positioning point q, it means that the horizontal coordinate position of the positioning point q is too small. At this time, the horizontal coordinate position of the compensation point needs to be increased, so that the horizontal coordinate of the positioning point of the robot fixed-point processing increases, so that the positioning point q is constantly approaching the horizontal coordinate position of the feature point p. The same is true for the opposite reason, which will not be repeated.

[0142] When Yp>Yq, then Yb1=Yb*(1+Ge%); when Yp<Yq, then Yb1=Yb*(1-Ge%);

[0143] When Zp>Zq, then Zb1=Zb*(1+Ge%); when Zp<Zq, then Zb1=Zb*(1-Ge%);

[0144] The analysis process of the ordinate and vertical coordinates is the same as that of the horizontal coordinates, so we will not elaborate on it here. By adjusting the compensation points, the accuracy of feature point positioning compensation can be continuously improved.

[0145] S4-2, according to the corresponding level of reference adjustment signal, the displacement vector of the reference point C Make corresponding adjustments, and the specific adjustment range is set according to the actual situation; for example, when the level I reference adjustment signal corresponds to setting the level I reference adjustment range He1%; the level II reference adjustment signal sets the level II reference adjustment range He2%; the level III reference adjustment signal sets the level III reference adjustment range He3%;

[0146] Displacement vector of neighboring point c Including the distance and direction from the neighboring point c to the feature point p, the preset distance is Lx, the direction angle is θ; the displacement vector of the reference point C after reference adjustment is marked as The displacement vector is Including the distance Lx1 and direction θ1 from the reference point C to the target point. The target point is the feature point p. After reference adjustment, fixed-point processing is performed to obtain the coordinates of the positioning point q of the workpiece model B.

[0147] Then Lx1=Lx*(1±He%), θ1=θ*(1±He%), where the displacement vector after adjustment is The distance and direction are slightly increased or decreased according to the reference adjustment range. The specific increase or decrease analysis process is the same as the analysis process of positioning compensation, that is, by comparing the coordinates of the positioning point q with the feature point p, the positioning point q and the feature point p after positioning processing are finally made to approach each other continuously, avoiding the diffuse interference of the regional light intensity of the target point on the positioning processing.

[0148] A computer-readable storage medium stores a computer program, wherein the computer program implements the above-mentioned industrial robot visual positioning method when executed by a central processing unit.

[0149] In summary, the present invention uses a visual acquisition module to perform 3D visual positioning of the workpiece, thereby assisting in controlling the manipulator to perform fixed-point processing, and uses an environmental monitoring module to measure the light of the workpiece processing environment, and then uses a central processing unit to perform visual positioning impact analysis, evaluate the comprehensive impact of the workpiece's own motion changes and the workpiece processing environment light changes on the visual positioning accuracy, ensure the integrity and comprehensiveness of data processing, generate corresponding risk warning signals, and thus quantify the deviation risk degree of visual positioning processing and perform subsequent optimization management;

[0150] The present invention conducts feedback refinement analysis through risk warning signals, evaluates the effects of the workpiece's own motion changes and ambient light changes on the visual positioning accuracy in turn, generates a management signal group, and then receives and processes the signals through a performance optimization module, thereby improving the real-time performance of the industrial robot's visual positioning from the perspective of positioning compensation, ensuring the accuracy of visual positioning processing when the workpiece is in a dynamic transmission state, and performing optimization management from the perspective of positioning and adjustment with reference to adjacent points, so as to make up for the defect of insufficient environmental adaptability of the industrial robot's visual positioning, ensuring the visual positioning accuracy when the workpiece is in an environment with changing light, and then continuously improving the model application effect through model self-growth, thereby realizing adaptive growth of visual positioning performance.

[0151] The size of the interval and threshold is set to facilitate comparison. The size of the threshold depends on the amount of sample data and the number of bases set by technical personnel in this field for each group of sample data; as long as it does not affect the proportional relationship between the parameter and the quantized value.

[0152] The above formulas are all dimensionless and numerical calculations. The formula is a formula obtained by collecting a large amount of data and performing software simulation to obtain the most recent real situation. The preset parameters in the formula are set by technicians in this field according to actual conditions.

[0153] The above description is only a preferred specific implementation manner of the present invention, but the protection scope of the present invention is not limited thereto. Any technician familiar with the technical field can make equivalent replacements or changes according to the technical scheme and inventive concept of the present invention within the technical scope disclosed by the present invention, which should be covered by the protection scope of the present invention.

Claims

1. A visual positioning method for an industrial robot, characterized in that: The following steps are involved: Step 1: Build a 3D stereo model of the workpiece through the visual acquisition module and perform visual positioning: obtain the 3D stereo model of the workpiece through the visual sensor and mark it as workpiece model A, obtain the velocity vector of workpiece model A in real time, and locate the feature points of workpiece model A. Perform fixed-point processing through the robot arm and mark the solid model of the workpiece as workpiece model B. Step 2: Obtain the light parameters of the workpiece processing environment through the environment monitoring module: the light parameters of the workpiece processing environment include light intensity, luminous flux, illumination and brightness; Step 3: Build a visual positioning impact model through the central processing unit: evaluate the visual positioning accuracy by analyzing workpiece model A and workpiece model B, and then analyze the change function between the velocity vector of workpiece model A and the visual positioning accuracy, as well as the change function between the light parameters of the workpiece processing environment and the visual positioning accuracy, so as to comprehensively evaluate the comprehensive impact of the workpiece's own motion changes and the light changes in the workpiece processing environment on the visual positioning accuracy, obtain the corresponding level of risk warning signals, and then feedback to generate a management signal group; Step 4: Receive risk warning signals and management signal groups through the performance optimization module and perform corresponding processing.

2. The industrial robot visual positioning method according to claim 1, characterized in that: The specific process of visual positioning by the visual acquisition module is as follows: A1, obtain the 3D stereo model of the workpiece through the visual sensor and mark it as workpiece model A, locate the feature points of workpiece model A, mark any feature point as p, obtain the coordinates of point p, feed the coordinates of point p back to the robot for fixed-point processing, mark the solid model of the workpiece as workpiece model B, mark the positioning point of workpiece model B after fixed-point processing by the robot as q, and obtain the coordinates of point q; A2, the visual acquisition module obtains the velocity vector of the workpiece model A in real time, and marks the velocity vector of the workpiece model A as Va, where the velocity vector Va includes the linear velocity υ, angular velocity ω, translation velocity v and translation direction r of the workpiece model A, and marks the velocity vector Va = <υ, ω, v, r>.

3. The industrial robot visual positioning method according to claim 2, characterized in that: The specific process of the CPU building the visual positioning impact model is as follows: B1, by analyzing the coordinate deviation between the feature point p of workpiece model A and the positioning point q of workpiece model B, the visual positioning accuracy coefficient Xjd is obtained to evaluate the initial accuracy of visual positioning; B2, by analyzing the change function between the velocity vector of the workpiece model A and the visual positioning accuracy, the workpiece motion influence coefficient Xyd is obtained, thereby evaluating the first degree of influence of the workpiece's own motion change on the visual positioning accuracy; B3, by analyzing the change function between the light parameters of the workpiece processing environment and the visual positioning accuracy, the environmental light influence index Zgx is obtained, so as to evaluate the second influence degree of the light change of the workpiece processing environment on the visual positioning accuracy; B4, by combining the workpiece motion influence coefficient Xyd and the ambient light influence index Zgx, comprehensively obtain the visual positioning accuracy influence index Zsj, set the evaluation interval of the visual positioning accuracy influence index Zsj, and evaluate the comprehensive influence of the workpiece's own motion change and the workpiece processing environment light change on the visual positioning accuracy through interval comparison; B5, obtain the corresponding level of risk warning signal through the visual positioning accuracy impact index, and then feed it back to the workpiece motion impact coefficient Xyd and the ambient light impact index Zgx for detailed comparison, and generate the corresponding level of compensation adjustment signal and reference adjustment signal respectively, and adjust and refresh the control parameters of the robot; B6, by adjusting and refreshing, recalculating the visual positioning accuracy coefficient, thereby evaluating the correction accuracy of the visual positioning, and then setting the threshold V of the visual positioning accuracy coefficient. When the visual positioning accuracy coefficient is lower than the preset threshold V, a model optimization signal is generated to optimize the fitting function in the mathematical model; The compensation adjustment signal, the reference adjustment signal and the model optimization signal are integrated and labeled as a management signal group.

4. The industrial robot visual positioning method according to claim 3, characterized in that: The specific process of evaluating visual positioning accuracy is as follows: B1-1, synchronously extract n0 feature points p of workpiece model A and n0 positioning points q of workpiece model B, mark the coordinates of feature point p as (Xp, Yp, Zp), and mark the coordinates of positioning point q as (Xq, Yq, Zq); B1-2, by analyzing the coordinate deviation between the feature point p of the workpiece model A and the positioning point q of the workpiece model B, the visual positioning accuracy coefficient Xjd is obtained; B1-3, set the evaluation interval of the visual positioning accuracy coefficient Xjd to evaluate the visual positioning accuracy.

5. The industrial robot visual positioning method according to claim 4, characterized in that: The specific process of assessing the first impact level is as follows: B2-1, analyze the velocity vector and visual positioning accuracy of workpiece model A; B2-101, feature point compensation positioning of workpiece model A is performed through velocity vector Va; The specific process of compensation positioning is as follows: set the visual acquisition cycle Tc to perform 3D dynamic simulation on the workpiece model A, and obtain the motion trajectory Sp of the feature point p of the workpiece model A; mark the time difference between the robot receiving the instruction and the fixed-point processing as the delay cycle Ty, so as to obtain the compensation point b of the feature point p through the motion trajectory Sp; mark the coordinates of the compensation point b as (Xb, Yb, Zb), issue a control instruction to the robot through the coordinates of the compensation point b, and the robot performs fixed-point processing on the workpiece model B after the delay period Ty, and obtains the coordinates of the positioning point q of the workpiece model B; B2-102, by compensating for the coordinate deviation of the feature point p and the positioning point q in the positioning operation, the corresponding visual positioning accuracy coefficient is obtained, and it is marked as the accuracy coefficient Xjd1 of the visual positioning compensation; B2-103, through the velocity vector Va = <υ, ω, v, r>, fit and construct the variation function F1 between the velocity vector Va and the accuracy coefficient Xjd1 of the visual positioning compensation; B2-2, obtain the workpiece motion influence coefficient Xyd through the change function F1 between the velocity vector Va and the accuracy coefficient Xjd1 of the visual positioning compensation; B2-3, setting the evaluation interval of the workpiece motion influence coefficient Xyd, so as to evaluate the first influence degree of the workpiece's own motion change on the visual positioning accuracy, and generate a corresponding compensation adjustment signal.

6. The industrial robot visual positioning method according to claim 5, characterized in that: The specific process of assessing the second level of impact is as follows: B3-1, by analyzing the light parameters and visual positioning accuracy of the workpiece processing environment; The light parameters of the workpiece processing environment include light intensity, luminous flux, illumination and brightness; The workpiece processing environment is divided into N0 characteristic areas, any characteristic area is marked as z, and the light intensity, luminous flux, illumination and brightness of the characteristic area z are collected and marked as Gq, Gt, Zd, Ld respectively; The ambient light influence coefficient Xgx of the characteristic area z is comprehensively obtained through the light intensity Gq, luminous flux Gt, illuminance Zd and brightness Ld of the characteristic area z; Set a threshold Hg of the ambient light influence coefficient Xgx, and when the ambient light influence coefficient Xgx of the feature area z exceeds the threshold Hg, the feature area z is determined to be an abnormal light area; Otherwise, the characteristic area z is determined to be a normal light area; When the feature point p is located in the normal light area, the robot is controlled to perform fixed-point processing according to the coordinates of the feature point p; When the feature point p is located in the abnormal light area, the neighboring point c of the feature point p is extracted, and the workpiece model is referenced and positioned by the coordinates of the neighboring point c; By referring to the coordinates of the feature point p in the positioning operation and the coordinate deviation of the positioning point q, the corresponding visual positioning accuracy coefficient is obtained and marked as the accuracy coefficient Xjd2 of the visual positioning reference; Fitting constructs the variation function F2 between the ambient light influence coefficient Xgx and the accuracy coefficient Xjd2 of the visual positioning reference; B3-2, comprehensively obtain the ambient light impact index Zgx of the workpiece processing environment through the change function F2 between the ambient light impact coefficient Xgx of N0 feature areas and the accuracy coefficient Xjd2 of their corresponding visual positioning reference; B3-3, setting the evaluation interval of the ambient light impact index Zgx, thereby evaluating the second impact degree of the light change in the workpiece processing environment on the visual positioning accuracy, and generating a corresponding reference adjustment signal.

7. The industrial robot visual positioning method according to claim 6, characterized in that: The specific process of reference positioning of the workpiece model is as follows: Mark the coordinates of the neighboring point c as (Xc, Yc, Zc), take the neighboring point c as the starting point and the feature point p as the target point, and obtain the displacement vector Through the coordinates of the neighboring point c and the displacement vector A control instruction is issued to the robot to obtain the target point coordinates of the workpiece model B. The robot performs fixed-point processing on the target point coordinates of the workpiece model B to obtain the coordinates of the positioning point q of the workpiece model B.

8. An industrial robot visual positioning device, characterized in that: The device comprises a visual acquisition module, an environmental monitoring module, a central processing unit and a performance optimization module, wherein the visual acquisition module, the environmental monitoring module, the central processing unit and the performance optimization module are communicatively connected with each other; the device applies an industrial robot visual positioning method as described in any one of claims 1 to 7 above.

9. The industrial robot visual positioning device according to claim 8, characterized in that: The visual acquisition module is used to build a 3D model of the workpiece and perform visual positioning; The environment monitoring module is used to obtain the light parameters of the workpiece processing environment; The central processing unit is used to build a visual positioning impact model, comprehensively evaluate the comprehensive impact of the workpiece's own motion changes and the light changes in the workpiece processing environment on the visual positioning accuracy, obtain the corresponding level of risk warning signals, and then feedback to generate a management signal group; The performance optimization module is used to receive risk warning signals and management signal groups and perform corresponding processing.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a central processing unit, the computer program implements an industrial robot visual positioning method as described in any one of claims 1 to 7.

Images