A robot calibration point selection method, system and storage medium

Abstract

The application discloses a robot calibration point selection method and system and a storage medium. The robot calibration point selection method comprises the following steps: acquiring an initial point set; determining the weights of a plurality of evaluation elements according to a calibration requirement set by a user, and performing weighted calculation on the plurality of evaluation elements based on the weights to obtain a weighted evaluation index; screening a target calibration point set from the initial point set based on the weighted evaluation index; and calibrating the kinematic parameters of a robot by using the target calibration point set. Through a strategy combining a weighted comprehensive evaluation index and an intelligent point selection algorithm, the problems of incomplete selection of a single evaluation index, influence of an existing point selection method on a gross error and insufficient stability of calibration accuracy in space are solved.

Description

Technical Field

[0001] This invention relates to the field of robotics, and more specifically, to a method, system, and storage medium for selecting robot calibration points. Background Technology

[0002] The absolute positioning accuracy of industrial robots is crucial for them to perform high-precision tasks. Kinematic parameter calibration can effectively compensate for geometric parameter errors caused by manufacturing and assembly, and is a core means of improving the absolute positioning accuracy of robots. During the calibration process, the method of selecting calibration points directly affects the accuracy and stability of parameter identification.

[0003] In existing technologies, a single evaluation metric (such as the condition number or observability index) is typically used to select calibration points. However, while the condition number calculation method is simple, different scaling algorithms can yield different condition numbers when calibrating parameters with different dimensions, leading to inconsistent results. Although the observability index (such as the O1 exponent) remains invariant under any non-singular linear transformation, its calculation method is complex. Furthermore, there is no unified standard in the industry for which evaluation metric to use to select calibration points, resulting in subjectivity and limitations.

[0004] On the other hand, most existing point selection methods directly select from the initial point set, which may lead to points with large errors (i.e., coarse error points) being selected into the calibration point set. These "harmful" points will contaminate the calibration data, reduce the accuracy of parameter identification, and cause the calibrated robot to exhibit unstable positioning accuracy at different positions in the workspace. Patent CN109465831B discloses a method to improve the calibration accuracy of the tool coordinate system, which optimizes the "four-point method" through real-time accuracy evaluation and quaternion rotation angle verification, but its focus is on eliminating human error and does not involve comprehensive point selection based on weighted multi-index. Patent CN110421566B discloses a robot accuracy compensation method based on approximation weighted average interpolation, which uses approximation weighted average interpolation for model-free accuracy compensation, but its weighting idea is used for error prediction of neighborhood sample points, rather than for active screening and optimization of calibration points.

[0005] Therefore, there is an urgent need in this field for a method that can integrate multiple evaluation indicators, automatically eliminate gross error points, and select calibration points that make calibration parameters more sensitive to errors and more stable in accuracy across the entire workspace. Summary of the Invention

[0006] To address the aforementioned problems, this invention provides a robot calibration point selection method. The aim is to solve the issues of incomplete selection by a single evaluation index, susceptibility to gross errors in existing point selection methods, and insufficient spatial stability of calibration accuracy by combining a weighted comprehensive evaluation index with an intelligent point selection algorithm. The method includes the following steps:

[0007] S1. Obtain the initial set of points;

[0008] S2. Based on the calibration requirements set by the user, determine the weights of multiple evaluation elements, and perform weighted calculation on the multiple evaluation elements based on the weights to obtain a weighted evaluation index.

[0009] S3. Based on the weighted evaluation index, select a target calibration point set from the initial point set;

[0010] S4. The kinematic parameters of the robot are calibrated using the target calibration point set.

[0011] As a preferred technical solution, the present invention also provides a robot calibration point selection system, comprising:

[0012] The initial point generation module is used to obtain the initial point set;

[0013] The evaluation index calculation module is used to determine the weights of multiple evaluation elements according to the calibration requirements set by the user, and to perform weighted calculation on the multiple evaluation elements based on the weights to obtain the weighted evaluation index.

[0014] The point selection module is used to select a set of target calibration points from the initial point set based on the weighted evaluation index.

[0015] The parameter calibration module is used to calibrate the robot's kinematic parameters using the target calibration point set.

[0016] As a preferred technical solution, the present invention also provides a computer-readable storage medium having a computer program stored thereon, wherein the computer program, when executed by a processor, implements the above-described method.

[0017] Compared with the prior art, the beneficial effects of this invention are:

[0018] The robot calibration point selection method provided by this invention combines the advantages of multiple evaluation elements by introducing a weighted comprehensive evaluation index. Users can flexibly set the weights according to specific calibration requirements, overcoming the one-sidedness of single index selection, the scaling sensitivity problem when calibrating parameters with different dimensions, and the complexity of the objective function calculation of the point selection algorithm when calibrating parameters with the same dimensions. The point selection algorithm can reduce the influence of gross errors, select calibration points with stronger generalization ability, improve the sensitivity of calibration parameters to positioning errors, and enhance the stability of positioning accuracy of multiple sets of evaluation points in space. Attached Figure Description

[0019] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0020] Figure 1 This is a flowchart illustrating the robot calibration point selection method of the present invention;

[0021] Figure 2 This is a flowchart illustrating step S3 in the robot calibration point selection method of the present invention.

[0022] Figure 3 This is a schematic diagram of the robot calibration point selection system of the present invention. Detailed Implementation

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

[0024] It should be noted that the terms "first" and "second," etc., in the specification, claims, and accompanying drawings of this invention are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence. It should be understood that such data can be interchanged where appropriate so that the embodiments of the invention described herein can be implemented in orders other than those illustrated or described herein. Furthermore, the terms "comprising" and "having," and any variations thereof, are intended to cover a non-exclusive inclusion; for example, a process, method, system, product, or apparatus that comprises a series of steps or units is not necessarily limited to those steps or units explicitly listed, but may include other steps or units not explicitly listed or inherent to such processes, methods, products, or apparatus.

[0025] Figure 1 This is a flowchart of a robot calibration point selection method provided in an embodiment of the present invention, such as... Figure 1 As shown, it includes the following steps:

[0026] S1. Obtain the initial set of points;

[0027] S2. Based on the calibration requirements set by the user, determine the weights of multiple evaluation elements, and perform weighted calculation on the multiple evaluation elements based on the weights to obtain a weighted evaluation index.

[0028] S3. Based on the weighted evaluation index, select a target calibration point set from the initial point set;

[0029] S4. The kinematic parameters of the robot are calibrated using the target calibration point set.

[0030] In a preferred embodiment, obtaining the initial point set in step S1 includes: defining the size and location of the calibration space, dividing the space into multiple cubes, and taking 8 vertices and 4 internal points of each cube as initial points. These initial points together constitute the initial point set, which aims to cover the entire calibration space and have a certain degree of internal diversity.

[0031] In a preferred embodiment, in step S2, the user sets calibration requirements including, but not limited to, the type of calibration parameters (all angles, all lengths, or a mixture of both), whether the units of the calibration parameters are consistent, and specific requirements for calibration accuracy (such as high precision or standard precision). The multiple evaluation elements include at least observable indices and condition numbers. Through weighted calculation, these indicators with different dimensions and physical meanings are integrated into a comprehensive evaluation criterion.

[0032] In a preferred embodiment, in step S2, the formula for calculating the weighted evaluation index P is: in, The weights corresponding to each evaluation element, ; It is an observable index; It is a condition number.

[0033] It should be noted that the number of weights corresponding to each evaluation element is not specific, but is determined by the number of evaluation elements to be referenced. In this application, the evaluation elements include at least the observable index and the condition number. The observable index and the condition number characterize the sensitivity of the model error to the calibration parameters. The larger the observable index and the smaller the condition number, the higher the calibration accuracy and the higher the sensitivity to the calibration parameters. The observable index and the condition number are obtained by performing singular value decomposition on the Jacobian matrix of the error model and then calculating from the singular values. The observable index is one of the five observable indices commonly used in the field of robot calibration, and the condition number is one of the one commonly used in the field of robot calibration, which will not be described in detail here.

[0034] In a preferred embodiment, Figure 2 This is a flowchart illustrating step S3 in the robot calibration point selection method of the present invention, as shown below. Figure 2 As shown, step S3 includes:

[0035] S310. Perform error analysis on the points in the initial point set, and remove points with errors greater than a preset threshold based on the analysis results to form a candidate point set.

[0036] In a preferred embodiment, the error analysis in step S310 is performed by constructing a weight function matrix, wherein the error includes the standardized residual of the pose error.

[0037] Specifically, for each point in the initial set of points, the Jacobian matrix of the robot's kinematic model at that point is calculated, and a calibration error function is established. This function characterizes the linear mapping relationship between the robot's end-effector pose error and the kinematic parameter error to be calibrated (i.e., ΔY = J·Δθ, where ΔY is the pose error vector, J is the Jacobian matrix, and Δθ is the parameter error vector to be identified).

[0038] The actual pose error of each point is calculated through preliminary parameter estimation (such as using conventional least squares method) or by directly using measured and theoretical values, and then further converted into standardized residuals. The standardized residuals... It refers to the standardized residual of the j-th point, calculated based on the pose error p. It reflects the degree of deviation of the error of each point from the overall error distribution and is a key indicator for identifying outliers or gross error points.

[0039] Based on the standardized residuals at each point The weight function wj is calculated, and its design principle is that the larger the absolute value of the standardized residual, the smaller the corresponding weight, so as to reduce the impact of abnormal error data on subsequent processing.

[0040] Furthermore, using the aforementioned weight functions w1, w2, ..., w N Construct a diagonal weight function matrix W(v), which is expressed as:

[0041] in, The function represents the construction of a diagonal matrix, w1, w2, ..., w N The weight function is determined based on the standardized residuals of each point, where N is the number of points in the initial point set.

[0042] Furthermore, based on the pose error obtained from the weight function matrix, the points in the initial point set are divided into an effective information region g1, a favorable information front section region g2, a favorable information rear section region g3, and a harmful information region g4. Points in the harmful information region g4 are removed to prevent points with large gross errors from being selected as calibration points. The remaining points constitute a candidate point set containing N' points.

[0043] In a preferred embodiment, step S3 further includes:

[0044] S320. Using the weighted evaluation index as the fitness function, a search algorithm is used to search for the target calibration point set from the candidate point set.

[0045] In a preferred embodiment, the search algorithm is either a particle swarm optimization algorithm or a genetic algorithm.

[0046] In a preferred embodiment, the search algorithm is a particle swarm optimization algorithm, and the search algorithm is used to search for the target calibration point set from the candidate point set, including:

[0047] Initialize the particle population, using the weighted evaluation index as the fitness function;

[0048] Calculate the individual optimal solution and the population optimal solution for each particle;

[0049] Update the velocity and position of each particle, and repeat the particle velocity and position update steps until the evaluation index corresponding to the selected point reaches the maximum value. Stop the search and determine the set of calibration points as the target calibration point set.

[0050] Specifically, the particle swarm optimization algorithm updates the velocity and position of each particle i using the following formula:

[0051] in, and Let be the velocity and position of particle i in the t-th cycle, respectively; ω be the inertia weight; c1 and c2 be the learning factors; and r1 and r2 be random numbers in the range [0,1]. Let i be the individual optimal solution for particle i. This is the population-optimal solution for particle i.

[0052] It should be noted that the inertia weight ω is used to balance the algorithm's global exploration and local exploitation capabilities. A larger ω value is set in the early stages of iteration to enhance global search capabilities, while a smaller ω value is set in the later stages of iteration to improve local convergence accuracy. The learning factor c1 represents the tendency of particles to follow their own historical best position, and the learning factor c2 represents the tendency of particles to follow the group's historical best position. The values ​​of learning factors c1 and c2 range from 0 to 4. The values ​​of r1 and r2 range from 0 to 1 and are uniformly distributed random numbers. r1 and r2 are regenerated with each update, introducing randomness, which helps to escape local optima, avoids excessive determinism, and increases exploration capabilities.

[0053] In a preferred embodiment, calibrating the robot's kinematic parameters using the target calibration point set includes:

[0054] The target calibration point set is measured, an error model is established, and a fitting algorithm is used to calibrate the robot's kinematic parameter errors. The calibrated kinematic parameter errors are then compensated to the nominal parameters to obtain the calibrated kinematic parameters. The kinematic parameters include, but are not limited to, link offset, link length, link torsion angle, and joint angle.

[0055] Specifically, to improve calibration accuracy, a step-by-step calibration strategy can be adopted. The specific process for calibrating kinematic parameters is as follows: if the calibration accuracy requirement is high, first calibrate and compensate for the error of the joint angle, and then calibrate and compensate for the error of the link offset and link length; if the calibration accuracy requirement is conventional accuracy, directly calibrate and compensate for the errors of the link offset, link length, link torsion angle and joint angle.

[0056] In a preferred embodiment, the method further includes:

[0057] S5. Select W sets of evaluation points in evaluation spaces of different locations and sizes, measure and calculate the position accuracy and attitude accuracy of each set of points, and statistically analyze the average, maximum and root mean square of the W sets of evaluation points to verify the stability of the calibration accuracy throughout the entire workspace.

[0058] Specifically, the methods for selecting evaluation points include the five-point method and the random method; the five-point method involves taking the vertices and center points of the diagonal face of the evaluation space as evaluation points, with 5 evaluation points selected for each evaluation space; the random method involves randomly selecting 50 evaluation points within the evaluation space.

[0059] It should be noted that the method described in this invention is not only applicable to robot kinematic parameter calibration, but can also be extended to any scenario requiring optimization of calibration point selection, such as dynamic parameter calibration, and has good universality.

[0060] This invention also provides a robot calibration point selection system. Figure 3 This is a schematic diagram of the robot calibration point selection system of the present invention, as shown below. Figure 3 As shown, it includes:

[0061] The initial point generation module is used to obtain the initial point set;

[0062] The evaluation index calculation module is used to determine the weights of multiple evaluation elements according to the calibration requirements set by the user, and to perform weighted calculation on the multiple evaluation elements based on the weights to obtain the weighted evaluation index.

[0063] The point selection module is used to select a set of target calibration points from the initial point set based on the weighted evaluation index.

[0064] The parameter calibration module is used to calibrate the robot's kinematic parameters using the target calibration point set.

[0065] In a preferred embodiment, the point selection module includes:

[0066] The gross error elimination unit is used to perform error analysis on the points in the initial point set, and eliminate points with errors greater than a preset threshold based on the analysis results to form a candidate point set.

[0067] The intelligent search unit is used to search for the target calibration point set from the candidate point set using the weighted evaluation index as the fitness function and a search algorithm.

[0068] In a preferred embodiment, the robot calibration point selection system further includes: an accuracy evaluation module, which selects W sets of evaluation points in evaluation spaces of different positions and sizes, measures and calculates the position accuracy and attitude accuracy of each set of points, and calculates the average value, maximum value and root mean square of the W sets of evaluation points to verify the stability of calibration accuracy.

[0069] The present invention also provides a computer-readable storage medium having a computer program stored thereon, wherein the computer program, when executed by a processor, implements the above-described robot calibration point selection method.

[0070] It should be understood that the computer-readable storage medium is any data storage device capable of storing data or programs that can subsequently be read by a computer system. Examples of computer-readable storage media include read-only memory, random access memory, CD-ROM, HDD, DVD, magnetic tape, and optical data storage devices. Computer-readable storage media can also be distributed across network-coupled computer systems, enabling computer-readable code to be stored and executed in a distributed manner. The program code contained on the computer-readable storage medium can be transmitted using any suitable medium, including but not limited to: wireless, wired, optical fiber, radio frequency (RF), or any suitable combination thereof.

[0071] In some implementations, the computer-readable storage medium may be non-transitory.

[0072] In summary, the robot calibration point selection method provided by this invention combines the advantages of multiple evaluation elements by introducing a weighted comprehensive evaluation index. Users can flexibly set the weights according to specific calibration requirements, overcoming the one-sidedness of single index selection, the scaling sensitivity problem when calibrating parameters with different dimensions, and the complexity of the objective function calculation of the point selection algorithm when calibrating parameters with the same dimensions. The point selection algorithm can reduce the influence of gross errors, select calibration points with stronger generalization ability, improve the sensitivity of calibration parameters to positioning errors, and enhance the stability of positioning accuracy of multiple sets of evaluation points in space.

[0073] It should be noted that the above embodiments are only used to illustrate specific implementations of the present invention and are not intended to limit the present invention. For example, the elements constituting the weighted evaluation index are not limited to the observability index and condition number, but can also be other indicators that can reflect observability and sensitivity, and the number can be increased or decreased. The method for eliminating gross errors is not limited to the weight function matrix, but can also use other statistical methods such as the 3σ criterion. The search algorithm is not limited to the particle swarm optimization algorithm, but can also use global optimization algorithms such as genetic algorithms and simulated annealing algorithms. These simple substitutions and modifications based on the core concept of the present invention should all be considered to fall within the protection scope of the present invention.

[0074] It should be understood that the various forms of processes shown above can be used, with steps reordered, added, or deleted. For example, the steps described in this invention can be executed in parallel, sequentially, or in different orders, as long as the desired result of the technical solution of this invention can be achieved, and no limitation is imposed herein.

[0075] The specific embodiments described above do not constitute a limitation on the scope of protection of this invention. Those skilled in the art should understand that various modifications, combinations, sub-combinations, and substitutions can be made according to design requirements and other factors. Any modifications, equivalent substitutions, and improvements made within the spirit and principles of this invention should be included within the scope of protection of this invention.

Claims

1. A robot calibration point selection method, characterized by, The method comprises the following steps: S1, obtaining an initial point set; S2, determining the weights of a plurality of evaluation elements according to a user-set calibration requirement, and performing weighted calculation on the plurality of evaluation elements based on the weights to obtain a weighted evaluation index; the plurality of evaluation elements at least include an observable index and a condition number; S3, screening a target calibration point set from the initial point set based on the weighted evaluation index; S4, calibrating kinematic parameters of a robot by using the target calibration point set; The step S3 comprises: S310, performing error analysis on the points in the initial point set, and eliminating points with error greater than a preset threshold based on the analysis result to form a candidate point set; S320, using the weighted evaluation index as a fitness function, and searching the target calibration point set from the candidate point set by using a search algorithm.

2. The robot calibration point selection method of claim 1, wherein, In step S2, the calculation formula of the weighted evaluation index P is: wherein, is a weight corresponding to each evaluation element, ; is an observable index; is a condition number.

3. The robot calibration point selection method of claim 1, wherein, The error analysis is performed by constructing a weight function matrix, and the error includes a standardized residual error of a pose error.

4. The robot calibration point selection method of claim 1, wherein, The search algorithm is any one of a particle swarm optimization algorithm and a genetic algorithm.

5. The robot calibration point selection method of claim 4, wherein, The search algorithm is a particle swarm optimization algorithm, and the searching of the target calibration point set from the candidate point set by using the search algorithm comprises: initializing a particle population, and using the weighted evaluation index as a fitness function; calculating an individual optimal solution and a population optimal solution of each particle; updating the speed and position of each particle, and cyclically performing the particle speed and position updating step until the evaluation index corresponding to the screened point reaches a maximum value, stopping the search, and determining the calibration point set as the target calibration point set.

6. A robot calibration point selection system, comprising: The method comprises: an initial point generation module configured to obtain an initial point set; an evaluation index calculation module configured to determine the weights of a plurality of evaluation elements according to a user-set calibration requirement, and perform weighted calculation on the plurality of evaluation elements based on the weights to obtain a weighted evaluation index; the plurality of evaluation elements at least include an observable index and a condition number; a point selection module configured to screen a target calibration point set from the initial point set based on the weighted evaluation index; a parameter calibration module configured to calibrate kinematic parameters of a robot by using the target calibration point set; The point selection module comprises: a gross error elimination unit configured to perform error analysis on the points in the initial point set, and eliminate points with error greater than a preset threshold based on the analysis result to form a candidate point set; an intelligent search unit configured to use the weighted evaluation index as a fitness function, and search the target calibration point set from the candidate point set by using a search algorithm.

7. A computer-readable storage medium having stored thereon a computer program, characterized in that The computer program, when executed by a processor, implements the robot calibration point selection method of any one of claims 1-5.

Images