A method and system for generating a configuration group for industrial robot calibration

Abstract

The present invention discloses a method and system for generating a configuration group for industrial robot calibration, belonging to the field of industrial robots, including: S1. Performing singular value decomposition on the coefficient matrix of the kinematic error model of the industrial robot to obtain an expression of the observability index of the configuration group for calibration, thereby establishing a generation model of the configuration group for calibration with the optimal observability index as the goal, and its constraint conditions include: the joint motion range constraint of the industrial robot, the motion space constraint and the acquisition constraint of the end of the industrial robot; the motion space constraint is used to constrain the end position of the industrial robot within its motion boundary range; the acquisition constraint is used to constrain the end position of the industrial robot within the measurement range of the measuring instrument; S2. Solving the generation model of the configuration group for calibration to generate an optimal configuration group for calibration. The present invention does not require prior actual data collection, nor does it require a large number of pose measurements in advance, and the calculation is fast, and it can directly generate a group of configuration groups with high calibration accuracy.

Description

Technical Field

[0001] The present invention belongs to the technical field of industrial robots, and more specifically, relates to a method and system for generating a configuration group for industrial robot calibration. Background Art

[0002] Kinematics parameter calibration is an effective means to improve the operation accuracy of a robotic arm. By using kinematics parameter calibration, the actual values of the kinematics parameters of the robotic arm can be obtained to achieve precise control of the robotic arm. The kinematics parameter calibration of an industrial robot consists of the following four steps: establishing a kinematics model and an error model of the kinematics model; collecting the actual poses of the configuration group for industrial robot calibration; solving the kinematics parameter error values according to an identification algorithm; and compensating the model parameters in the controller of the research object with the solved kinematics parameter error values. Among them, the actual pose collection in the second step will directly affect the final kinematics parameter calibration result.

[0003] The research on the existing optimization methods for the configuration group for industrial robot calibration mainly includes two directions: one is to propose a new observable index that can affect the calibration effect through mathematical theory derivation (i.e., a new optimization direction), and the second is to propose a new optimization algorithm for a certain optimization direction. The five observable indices proposed since the 1990s have been recognized by the academic community, so there are few new optimization directions proposed subsequently. Nowadays, most of the optimization algorithms are based on measuring more pose points than required for calibration calculation, and then selecting a better subset from them to form a configuration group for calibration. For example, iterative configuration search algorithms, improved sequential forward floating search algorithms based on random search techniques, improved optimal point set selection algorithms based on optimizing the initial configuration group, etc. are used to select a given number of optimal configuration groups from a limited configuration combination, which is prone to the problem of falling into local optimum, and these methods require a large number of pose measurements in advance, which is time-consuming and laborious; in addition, this method needs to select a subset of the configuration group for calibration from a measurement set with unstable quality, and the optimization effect is also relatively limited, and the calibration accuracy is low. Summary of the Invention

[0004] In view of the above defects or improvement requirements of the prior art, the present invention provides a method and system for generating a configuration group for industrial robot calibration to solve the problems that the prior art requires a large number of pose measurements in advance and has a low calibration accuracy.

[0005] To achieve the above object, the present invention provides a method for generating a configuration group for industrial robot calibration, including the following steps:

[0006] S1. Perform singular value decomposition on the coefficient matrix of the industrial robot kinematic error model to obtain the expression of the observability index of the calibration configuration set, thereby establishing a calibration configuration set generation model with the goal of optimizing the observability index. Among them, the constraint conditions of the calibration configuration set generation model include: the joint motion range constraint of the industrial robot, as well as the motion space constraint and acquisition constraint at the end of the industrial robot. Among them, the motion space constraint is used to constrain the end position of the industrial robot within its motion boundary range; the acquisition constraint is used to constrain the end position of the industrial robot within the measurement range of the measuring instrument.

[0007] S2. Solve the above calibration configuration set generation model to generate the optimal calibration configuration set for the industrial robot.

[0008] Further preferably, the above motion boundary range includes the Cartesian space motion range of the end of the industrial robot restricted by obstacles, specifically: the intersection of the Cartesian space motion range of the end of the industrial robot and the space range bounded by the obstacle position.

[0009] Further preferably, the measuring instrument is a non-contact measuring device for emitting light to the industrial robot; at this time, a target ball is arranged at the end position of the industrial robot.

[0010] The above acquisition constraint is used to constrain the angle between the light incident direction and the negative direction of the normal of the plane where the light receiving port of the target ball is located within half of the light receiving angle of the target ball; the negative direction of the above normal is the direction towards the inside of the target ball.

[0011] Further preferably, the above acquisition constraint is:

[0012]

[0013] Among them, is the vector corresponding to the light incident direction; is the vector corresponding to the negative direction of the normal of the plane where the light receiving port of the target ball is located; D is the light receiving angle of the target ball, and the unit is radian.

[0014] Further preferably, the expression of the above configuration set observability index is:

[0015]

[0016] Among them, σ i is the i-th singular value of the above coefficient matrix, i = 1, 2,..., r; r is the number of kinematic parameters to be identified during kinematic parameter calibration; n is the number of pose points in the calibration configuration set to be generated.

[0017] Further preferably, the objective function of the above calibration configuration set generation model is:

[0018] min f(x) = 1 / O1

[0019] where x is the configuration group for calibration.

[0020] Further preferably, the SQP algorithm is used to solve the above-mentioned calibration configuration group generation model.

[0021] Further preferably, the WPSO-SQP algorithm is used to solve the above-mentioned calibration configuration group generation model to prevent falling into the local optimum problem.

[0022] Further preferably, the above-mentioned method for generating a calibration configuration group for an industrial robot further includes step S3 executed after step S2;

[0023] Step S3 includes: optimizing the configuration order in the optimal calibration configuration group by minimizing the sum of the distances between all adjacent two configurations in the optimal calibration configuration group;

[0024] where the distance between the l-th configuration and the (l + 1)-th configuration is:

[0025]

[0026] where N is the number of joint angles of the industrial robot; is the difference between the n-th joint angles of the l-th configuration and the (l + 1)-th configuration.

[0027] In a second aspect, a system for generating a calibration configuration group for an industrial robot includes:

[0028] A model construction module, which is used to perform singular value decomposition on the coefficient matrix of the kinematic error model of the industrial robot to obtain an expression of the observable index of the calibration configuration group, so as to establish a calibration configuration group generation model with the goal of optimizing the observable index; where the constraint conditions of the calibration configuration group generation model include: the joint motion range constraint of the industrial robot, and the motion space constraint and acquisition constraint of the end of the industrial robot; where the motion space constraint is used to constrain the end position of the industrial robot within its motion boundary range; the acquisition constraint is used to constrain the end position of the industrial robot within the measurement range of the measuring instrument;

[0029] A generation module, which is used to solve the above-mentioned calibration configuration group generation model to generate an optimal calibration configuration group for the industrial robot.

[0030] In a third aspect, the present invention provides a system for generating a calibration configuration group for an industrial robot, including: a memory and a processor, the memory stores a computer program, and when the processor executes the computer program, it executes the method for generating a calibration configuration group for an industrial robot provided in the first aspect of the present invention.

[0031] Fourthly, the present invention further provides a machine-readable storage medium storing machine-executable instructions, which, when called and executed by a processor, cause the processor to implement the method for generating a configuration group for industrial robot calibration provided in the first aspect of the present invention.

[0032] Generally speaking, through the above technical solutions conceived by the present invention, the following beneficial effects can be achieved:

[0033] 1. The present invention provides a method for generating a configuration group for industrial robot calibration, determines the joint motion range constraints in the kinematic parameter calibration process of the industrial robot, as well as the motion space constraints and acquisition constraints at the end of the industrial robot, and constructs a generation model of the calibration configuration group with the optimal observability index as the goal based on these constraints. By solving the generation model of the calibration configuration group, an optimal calibration configuration group that meets the Cartesian motion space range requirements of the industrial robot and the acquisition conditions of the measuring instrument can be freely generated; the present invention does not require actual data acquisition in advance, nor a large number of pose measurements in advance, and the calculation is fast, and a group of configuration groups with higher calibration accuracy can be directly generated.

[0034] 2. In the method for generating a configuration group for industrial robot calibration provided by the present invention, a hybrid WPSO-SQP algorithm is used to solve the above generation model of the calibration configuration group. This algorithm has both the global search optimal solution advantage of the WPSO algorithm and the local fast search advantage of the SQP algorithm that meets the constraints, and can avoid falling into the local optimal problem during the solution process, greatly improving the generation accuracy of the calibration configuration group.

[0035] 3. In the method for generating a configuration group for industrial robot calibration provided by the present invention, after obtaining the calibration configuration group, by minimizing the sum of the distances between all adjacent two configurations in the above optimal calibration configuration group, the measurement order of the calibration configuration group is further optimized, so that the attitude change is as small as possible when the industrial robot changes from one configuration to the next configuration, so as to avoid the phenomenon of frequent light interruption of the non-contact measurement device caused by an unreasonable order of measurement points.

[0036] 4. Through simulation verification, it is found that the observability index O1 of the configuration group generated by using the method for generating a configuration group for industrial robot calibration provided by the present invention is much larger than the observability index O1 of the optimal configuration group randomly selected by using the configuration group selection method under the same conditions. It can be seen that the generation model of the calibration configuration group provided by the present invention is feasible and effective, and has good performance. Description of the Drawings

[0037] Figure 1Flow chart of the configuration group generation method for industrial robot calibration provided in Embodiment 1 of the present invention;

[0038] Figure 2 Schematic diagram of the Cartesian space motion range of the end of the HSR_BR616 industrial robot without considering obstacles provided in Embodiment 1 of the present invention;

[0039] Figure 3 Schematic diagram of the motion boundary range of the end position of the HSR_BR616 industrial robot provided in Embodiment 1 of the present invention;

[0040] Figure 4 Schematic diagram of the light tracker target ball provided in Embodiment 1 of the present invention;

[0041] Figure 5 Flow chart of the SQP algorithm provided in Embodiment 1 of the present invention;

[0042] Figure 6 Flow chart of the hybrid WPSO-SQP algorithm provided in Embodiment 1 of the present invention;

[0043] Figure 7 Flow chart of the optimization of the configuration group measurement order based on the simulated annealing algorithm provided in Embodiment 1 of the present invention;

[0044] Figure 8 Schematic diagram of the O1 value distribution of 10,000 groups of configuration groups randomly generated by the optimization method based on randomly selected configuration groups provided in Embodiment 1 of the present invention;

[0045] Figure 9 Variation process of the O1 value of the configuration group generated by the WPSO-SQP optimization algorithm provided in Embodiment 1 of the present invention;

[0046] Figure 10 Schematic diagram of the O1 value and constraint violation degree of Particle 1 after WPSO and SQP iterations respectively provided in Embodiment 1 of the present invention; among them, (a) is the schematic diagram of the O1 value of Particle 1 after WPSO and SQP iterations respectively; (b) is the schematic diagram of the constraint violation degree of Particle 1 after WPSO and SQP iterations respectively;

[0047] Figure 11 Schematic diagram of the O1 value and constraint violation degree of Particle 8 after WPSO and SQP iterations respectively provided in Embodiment 1 of the present invention; among them, (a) is the schematic diagram of the O1 value of Particle 8 after WPSO and SQP iterations respectively; (b) is the schematic diagram of the constraint violation degree of Particle 8 after WPSO and SQP iterations respectively. Detailed implementation manners

[0048] To make the objectives, technical solutions and advantages of the present invention clearer and more understandable, the present invention will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not used to limit the present invention. In addition, the technical features involved in the various embodiments of the present invention described below can be combined with each other as long as they do not conflict with each other.

[0049] Embodiment 1

[0050] In this embodiment, the HSR_BR616 type industrial robot of Foshan Huashu Robot Company is taken as the research object for the invention introduction. Specifically, as Figure 1 shown, the method for generating a configuration group for industrial robot calibration provided in this embodiment includes the following steps:

[0051] S1. Perform singular value decomposition on the coefficient matrix of the industrial robot kinematic error model to obtain the expression of the observable index of the calibration configuration group, thereby establishing a calibration configuration group generation model with the optimal observable index as the goal; wherein, the constraint conditions of the calibration configuration group generation model include: the joint motion range constraint of the industrial robot, as well as the motion space constraint and acquisition constraint of the industrial robot end; wherein, the motion space constraint is used to constrain the end position of the industrial robot within its motion boundary range; the acquisition constraint is used to constrain the end position of the industrial robot within the measurement range of the measuring instrument;

[0052] It should be noted that the kinematic error model of the industrial robot can be constructed based on the DH kinematic model or the MDH kinematic model of the industrial robot. In this embodiment, the DH kinematic model of the industrial robot is constructed to obtain the explicit analytical expressions of the forward and inverse kinematic solutions of the industrial robot; and based on the differential motion method, the kinematic error model of the industrial robot is constructed by integrating the errors between the base coordinate system and the tool coordinate system of the industrial robot; specifically:

[0053]

[0054] wherein, e is the kinematic error of the industrial robot; is the coefficient matrix; are the kinematic parameters to be identified. The end pose calculates the corresponding configuration through the inverse kinematic solution, and the configurations that meet the quantity are combined into a configuration group, and the configuration group generates the coefficient matrix and the observable index is obtained. The observable indices of the calibration configuration group include O1, O2, O3, O4 and O5. Specifically, wherein, σ i is the i-th singular value of the above coefficient matrix, i = 1, 2,..., r; σ1 > σ2 >... > σ r; r is the number of kinematic parameters to be identified during kinematic parameter calibration. In this embodiment, r = 32; n is the number of pose points in the calibration configuration group to be generated.

[0055] O2 is the reciprocal of the matrix condition number, specifically: Further, O3 = σ r ;

[0056] When the observability index is O1, a calibration configuration group generation model is established with the maximum of the observability index O1 as the goal. When the observability index is O2, a calibration configuration group generation model is established with the minimum of the observability index O2 as the goal. When the observability index is O3, a calibration configuration group generation model is established with the maximum of the observability index O3 as the goal. When the observability index is O4, a calibration configuration group generation model is established with the minimum of the observability index O4 as the goal. When the observability index is O5, a calibration configuration group generation model is established with the minimum of the observability index O5 as the goal.

[0057] The selection of the observability index is related to the experimental purpose. Since O1 is applicable when the goal is to minimize the variance of the calibration result, and O1 is an observability index that is invariant under any non-singular linear transformation, it is therefore preferred to use O1 as the observability index.

[0058] Specifically, the objective function of the above calibration configuration group generation model is:

[0059] min f(x) = 1 / O1

[0060] Where x is the calibration configuration group. Specifically, from the above content, it can be seen that for any set of calibration configuration groups x, after calculating the extended Jacobian matrix H (i.e., the above coefficient matrix) and performing singular value decomposition to obtain the singular values of the coefficient matrix, the expression of the observability index O1 can be obtained, that is, the obtained f(x).

[0061] Further, the joint motion range constraint of the industrial robot is used to constrain the joint angles of the industrial robot within the joint motion range; specifically, in this embodiment, the joint motion range constraint of the HSR_BR616 type industrial robot is shown in Table 1.

[0062] Table 1

[0063] Joint Range of motion Joint 1 ±80° Joint 2 -120° / -30° Joint 3 +130° / +230° Joint 4 ±175° Joint 5 ±140° Joint 6 ±175°

[0064] Among them, the joint angle of joint 1 of the industrial robot is greater than or equal to -80° and less than or equal to 80°; the joint angle of joint 2 of the industrial robot is greater than or equal to -120° and less than or equal to -30°; the joint angle of joint 3 of the industrial robot is greater than or equal to 130° and less than or equal to 230°; the joint angle of joint 4 of the industrial robot is greater than or equal to -175° and less than or equal to 175°; the joint angle of joint 5 of the industrial robot is greater than or equal to -140° and less than or equal to 140°; the joint angle of joint 6 of the industrial robot is greater than or equal to -175° and less than or equal to 175°.

[0065] Furthermore, due to obstacles in the Cartesian motion space of the industrial robot, the end effector cannot reach this position, and when the position and pose do not meet the acquisition constraints of the measuring instrument, there is a situation where the end pose data generated by this configuration cannot be acquired. Therefore, the present invention introduces Cartesian space pose constraints on the end pose of the industrial robot, specifically including the motion space constraint and the acquisition constraint of the end of the industrial robot; the end pose of the industrial robot that meets this constraint does not require actual data acquisition in advance and is conducive to the calculation set simulation generation.

[0066] 1) For the motion space constraint:

[0067] The motion space constraint is used to constrain the end position of the industrial robot within its motion boundary range; among them, the motion boundary range includes the Cartesian space motion range of the end of the industrial robot restricted by obstacles, specifically: the intersection of the Cartesian space motion range of the end of the industrial robot and the space range bounded by the obstacle position. As Figure 2 shown is a schematic diagram of the Cartesian space motion range of the end of the HSR_BR616 type industrial robot without considering obstacles; if the motion space constraint is not added, the industrial robot may hit the ground or other obstacles. The normal working range of the industrial robot is relatively small. In this embodiment, the motion boundary range of the end position of the HSR_BR616 type industrial robot is a cuboid motion space in the front direction of the industrial robot, which is 700 mm in the z direction and 400 mm in the x direction of the base coordinate system {b} of the industrial robot, and is symmetrically distributed in the y direction. Its distribution and the set values of the length, width, and height are as Figure 3As shown; it should be noted that the measuring instrument can be a non-contact measuring device or a contact measuring device; among them, the non-contact measuring device can be a laser tracker, a dual-frequency laser rangefinder, etc.; the contact measuring instrument can be a ballbar, a wire-drawing distance measuring instrument, etc. In this embodiment, a non-contact measuring device is taken as an example for measurement. Specifically, the non-contact measuring device is used to emit light to the industrial robot; at this time, a target ball is arranged at the end position of the industrial robot. Taking the laser tracker as an example, its center is set on the positive x direction of the base coordinate system {b} of the industrial robot; a laser tracker target ball is arranged at the end position of the industrial robot; the light tracker target ball is a reflective target ball, and its schematic diagram is as Figure 4 shown. A tool coordinate system {t} is established for the target ball, denoted as the target ball coordinate system {t}. Specifically, the above-mentioned motion space constraints are:

[0068]

[0069] Among them, is the position of the origin of the target ball coordinate system {t} in the x direction of the industrial robot base coordinate system {b}; is the position of the origin of the target ball coordinate system {t} in the y direction of the industrial robot base coordinate system {b}; is the position of the origin of the target ball coordinate system {t} in the z direction of the industrial robot base coordinate system {b}.

[0070] 2) For the acquisition constraint:

[0071] The acquisition constraint is used to constrain the angle between the laser incident direction and the negative direction of the normal of the plane where the target ball light-receiving port is located (i.e., the target ball ring in Figure 4 ) within half of the target ball light-receiving angle; the negative direction of the above normal is the direction towards the inside of the target ball. Specifically, the above acquisition constraint is:

[0072]

[0073] Among them, is the vector corresponding to the laser incident direction; is the vector corresponding to the negative direction of the normal of the plane where the target ball light-receiving port is located; D is the target ball light-receiving angle, in radians.

[0074] In this embodiment, the adopted target ball light-receiving angle is 60°, and it is required to satisfy that the angle between the vector from the laser tracker to the target ball position and the z-direction vector of the target ball coordinate system {t} (the negative direction of the above normal) is less than 30°, otherwise acquisition is not possible. Specifically, in this embodiment, the above acquisition constraint is:

[0075]

[0076] By integrating the joint motion range constraints of the general industrial robot, the motion space constraints of the end of the industrial robot, and the acquisition constraints, a configuration that meets the requirements of the Cartesian motion space range of the industrial robot and the acquisition conditions of the measuring instrument can be freely generated.

[0077] S2. Solve the above-mentioned calibration configuration group generation model to generate the optimal calibration configuration group of the industrial robot.

[0078] The solution problem of the above-mentioned calibration configuration group generation model is:

[0079] minf(x)=1 / O1

[0080] s.t.g(i)≤0

[0081] Where x represents a group of calibration configurations; i = 1, 2,..., 7; in this embodiment, 50 configurations are set as a group of configurations. Each configuration satisfies the above constraints; therefore, a total of 350 constraints are included.

[0082] Use Taylor expansion to simplify f(x) to a quadratic function at the iteration point x k The resulting is a quadratic programming problem:

[0083]

[0084] Equation (1) is an approximate problem of the original problem, but its solution is not necessarily a feasible solution of the original problem. Therefore, let S = x - x k , and substitute it into Equation (1) to generate a problem about the S variable:

[0085]

[0086] Let:

[0087]

[0088] Therefore, Equation (2) is transformed into a general quadratic programming problem:

[0089]

[0090] The solution method for this problem is to use the optimal solution S * as the next search direction S k of the original problem, and perform the original problem constraint search to obtain an approximate solution x k+1 . Repeating the whole process can obtain the optimal solution. The key above lies in how to obtain the H matrix. The commonly used method is the BFGS formula:

[0091]

[0092] Where, Δx k =xk+1 -x k , Δq k = f(x k+1 ) - f(x k ).

[0093] Find the iterative point x in the quadratic programming problem with inequality constraints k The active constraints form a new equality-constrained quadratic programming problem:

[0094]

[0095] Its Lagrangian function can be expressed as:

[0096]

[0097] The extreme value condition of a multivariate function is its derivative

[0098] The extreme value condition of a multivariate function is its derivative

[0099]

[0100] According to linear algebra knowledge, if equation (8) has a solution, its unique solution can be obtained by using elimination transformation. According to the KKT conditions, if the λ in the solution k+1 is not all 0, then S k+1 is the optimal solution S of equation (6) * , if the multipliers corresponding to the active inequality constraint conditions are not less than 0, then S k+1 is the optimal solution S of equation (4) * .

[0101] In an alternative embodiment, the sequential quadratic programming algorithm SQP is used to solve the above calibration configuration group generation model. The SQP algorithm constructs an augmented objective function through the objective function and equality and inequality constraint conditions, thereby transforming the constrained optimization problem into an ordinary unconstrained optimization problem. Specifically, in each iteration process, a quadratic programming problem is solved, the second-order approximation of the Lagrangian function is expanded as the objective function, and the constraint conditions are transformed into approximate linear optimal conditions of the constraints, which is a quasi-Newton method. Theoretically, the SQP algorithm has the local fast convergence speed of the Newton method. The flowchart of the SQP algorithm is as Figure 5 shown.

[0102] The SQP algorithm can quickly solve the optimal problem, but the condition for obtaining the global optimal solution is that the objective function is a convex function, and the objective function of the calibration configuration group generation model based on the Cartesian space pose constraint at the end of the industrial robot proposed in the present invention is not a convex function. Therefore, the obtained solution is the same as the initial value x 0The relevant local optimal solutions rather than the global optimal solution. Although the SQP algorithm has global convergence while maintaining the speed of local superlinear convergence, it will fall into the local optimal solution. To further improve the accuracy of the solution, the WPSO algorithm can be used to solve it to search for the global optimal solution. However, it is extremely difficult to use the WPSO algorithm to solve this high-dimensional optimization problem. The search space grows exponentially, and the "curse of dimensionality" requires an extremely large number of particles to possibly solve it, making it difficult to implement in practice. To solve the above problems, in another alternative implementation, the present invention uses a hybrid WPSO-SQP algorithm to solve the configuration group optimization objective function, which combines the respective advantages of the WPSO algorithm and the SQP algorithm, utilizes the WPSO algorithm to increase the possibility of the global optimal solution, and the local fast searchability of the SQP algorithm will pull the particles that violate the constraints back within the constraints. Specifically, the flowchart of the hybrid WPSO-SQP algorithm is as Figure 6 shown. After updating the particle velocity and position in each iteration of the WPSO algorithm, the SQP algorithm is used to pull the particles that violate the constraints back within the boundary and perform local optimal calculation. Among them, the initialization of the initial value of the particle position is generated by the inverse kinematic solution that meets the constraint conditions, avoiding the problem of rapid loss of population diversity caused by the SQP algorithm optimizing the initial infeasible particles to the local optimal solution at a few points on the boundary.

[0103] Further, in an alternative implementation mode three, the above method for generating a configuration group for industrial robot calibration further includes step S3 executed after step S2;

[0104] After obtaining the optimal calibration configuration group, an unreasonable measurement order of the configuration group will affect the smoothness of the acquisition process: for contact measurement devices such as Dynalog, the wire may be wound around the robot body, resulting in invalid measurement data; for non-contact measurement devices such as laser trackers, there may be a phenomenon of light interruption during the configuration conversion process, which requires manual adjustment of the measurement device and measurement software, resulting in low calibration efficiency. Therefore, optimizing the measurement order of the configurations in the already calculated configuration group is a guarantee for the smooth progress of the measurement process and also improves the calibration efficiency.

[0105] First, the optimization objective function needs to be determined. In this embodiment, taking a laser tracker as an example, in order to improve the measurement efficiency and ensure that there is no light interruption during the measurement process as much as possible, it is required that the attitude change of the industrial robot be as small as possible during the process of changing from one configuration to the next. Specifically, the above step S3 includes: optimizing the configuration order in the optimal calibration configuration group by minimizing the sum of the distances between all adjacent two configurations in the optimal calibration configuration group;

[0106] Among them, the definition of distance can be weighted and calculated according to the influence degree of each joint of the industrial robot on the end pose; specifically, the distance between the l-th configuration and the (l + 1)-th configuration is:

[0107]

[0108] where N is the number of joint angles of the industrial robot, and in this embodiment, N is taken as 6; is the difference between the n-th joint angles of the l-th configuration and the (l + 1)-th configuration.

[0109] Making the sum of the distances between all adjacent two configurations in the configuration group for calibration the shortest is the objective function Dist(w) of this problem, where w is the configuration order. This problem is similar to the Traveling Salesman Problem (TSP), except that the total distance does not need to consider the distance of returning to the first configuration after measuring the last configuration. TSP is a Non-Deterministic Polynomial (NP) problem, which means it can be solved by a non-deterministic machine in polynomial time. The main methods for solving TSP are enumeration method, backtracking method, branch and bound method, and heuristic algorithm. As the problem scale increases, exact brute-force methods such as enumeration will cause the solution time of this problem to be infinitely extended, and the problem cannot be solved within an acceptable time, so it cannot be applied to the industrial site. Therefore, the present invention focuses on heuristic or approximate algorithms. In this embodiment, the simulated annealing method is used to optimize the measurement order of the configuration group. Specifically, the simulated annealing algorithm is used to simulate the annealing phenomenon in thermodynamics: the system energy of an object will gradually decrease as the temperature of the object slowly decreases. When the temperature reaches a sufficiently low level, the liquid will condense and crystallize, and at this time, the overall system energy of the object will reach the lowest; if the temperature is not slowly decreased, it cannot be guaranteed that the system will finally be in the lowest energy state. Therefore, TSP can use this method to solve the problem of reducing the total distance. Initially, a relatively high temperature is set. When the temperature is relatively high, the solution of the problem is likely to change, and it has a higher global search ability to jump out of the local optimal solution. As the temperature decreases, the solution in the algorithm also tends to be stable, the changeability decreases, the search speed decreases, and the accuracy increases. Specifically, the flowchart of the optimization of the measurement order of the configuration group based on the simulated annealing algorithm is as Figure 7 shown.

[0110] To further illustrate the advantages of the method for generating a configuration group for calibrating an industrial robot provided by the present invention, the following will be described in detail in combination with specific experiments:

[0111] For the joint motion range constraints, the motion space constraints and the acquisition constraints of the end of the industrial robot constructed by the present invention, if only the calibration configuration group optimization is carried out based on the above constraints, simulating the actual acquisition pose, 500 poses that meet the constraints are randomly generated by a computer, and the configurations are generated by inverse kinematics. Using the optimization method of randomly selected configuration groups, 50 configurations are randomly selected to form a configuration group, and a total of 10,000 configuration groups are generated. The O1 values are as Figure 8 shown. From Figure 8 it can be seen that if the random selection method or a certain search optimization method is used, only the configuration groups can be extracted and formed from a large number of existing configurations, and the effect is limited by the samples of a large number of configurations.

[0112] The present invention constructs a calibration configuration group generation model based on Cartesian space pose constraints, and solves the calibration configuration group generation model, which can directly generate a group of calibration configuration groups, so that the computer can freely generate configurations that meet the requirements of the Cartesian motion space range of the industrial robot and the acquisition conditions of the measuring instrument, without the need for actual data acquisition in advance. Specifically, taking the example of using the hybrid WPSO-SQP algorithm to generate the optimal configuration group, the particle data is set to 10, the number of WPSO iterations is 50 times, the number of SQP iterations is 40 times, and the iteration step size is not less than 5e -3 , and the constraint violation tolerance is 2000. The change process of the O1 value of the configuration group generated in each iteration process is as Figure 9 shown. It can be seen from the figure that the configuration group generated by the first SQP optimization has increased the O1 value from 0.0271 to 0.0365, an increase of 34.69%. Subsequently, the iterative process of the WPSO-SQP search for the global optimal solution has increased the O1 value from 0.0365 to 0.0415, an increase of 13.70%. Compared with the initial random configuration group, the O1 value has increased by 53.14%. Compared with the maximum O1 value of randomly generating 10,000 configuration groups, it has increased by 45.10%. Therefore, it proves the effectiveness of the WPSO-SQP algorithm in improving the observable index. The change of the global optimal particle number during the iteration process is shown in Table 2:

[0113] Table 2

[0114] Number of iterations 1 1 2 3 4 10 18 22 27 39 Particle number 8 1 6 1 5 1 5 4 9 8

[0115] Among them, the particle numbered 8 in the first iteration process is the initialized optimal particle. It can be seen from the continuous change of the particle numbers in the table that WPSO is active in searching for the global optimum. Figure 10 and Figure 11Taking particle 1 (with the most occurrences of the optimal number) and the global optimal particle 8 as examples respectively, the O1 values and the degree of constraint violation after WPSO and SQP iterations are given, demonstrating the adjustment ability of the SQP algorithm to the constraint violation of WPSO (abbreviated as PSO in the figure for convenience of adjustment); among them, Figure 10 Figure (a) in it is a schematic diagram of the O1 values of particle 1 after WPSO and SQP iterations respectively; Figure 10 Figure (b) in it is a schematic diagram of the degree of constraint violation of particle 1 after WPSO and SQP iterations respectively; Figure 11 Figure (a) in it is a schematic diagram of the O1 values of particle 8 after WPSO and SQP iterations respectively; Figure 11 Figure (b) in it is a schematic diagram of the degree of constraint violation of particle 8 after WPSO and SQP iterations respectively; it can be seen from the figure that the degree of constraint violation of the particle after SQP calculation all drops to 0.

[0116] To sum up, the present invention first determines the Cartesian motion space position constraint of the end effector of the industrial robot and the attitude constraint of the measuring device on the tool coordinate system during the kinematic parameter calibration process of the industrial robot, and constructs a calibration configuration group generation model based on these constraints, avoiding the acquisition of actual points and providing the basis for the configuration group generation method. In addition, the present invention also proposes a hybrid WPSO-SQP algorithm that combines the global search optimal solution advantage of WPSO and the local fast search advantage of SQP that satisfies constraints to achieve the solution of the optimization objective function of the calibration configuration group. Again, to avoid the phenomenon of frequent light interruption of the non-contact measuring device caused by the unreasonable order of the measuring points, the present invention further optimizes the measuring order of the calibration configuration group. Finally, through simulation verification, it is found that the configuration group obtained by the configuration group generation model proposed by the present invention has an observability index O1 that is much larger than the observability index O1 of the configuration group obtained by using the configuration group selection method under the same conditions, indicating the feasibility and effectiveness of the calibration configuration group generation model provided by the present invention.

[0117] Example 2

[0118] An industrial robot calibration configuration group generation system, comprising:

[0119] A model construction module, used to perform singular value decomposition on the coefficient matrix of the industrial robot kinematic error model to obtain an expression of the observability index of the calibration configuration group, thereby establishing a calibration configuration group generation model with the optimal observability index as the goal; among them, the constraint conditions of the calibration configuration group generation model include: the joint motion range constraint of the industrial robot, and the motion space constraint and acquisition constraint of the end of the industrial robot; among them, the motion space constraint is used to constrain the end position of the industrial robot within its motion boundary range; the acquisition constraint is used to constrain the end position of the industrial robot within the measurement range of the measuring instrument;

[0120] A generation module, configured to solve the above-mentioned calibration configuration group generation model, and generate an optimal calibration configuration group for an industrial robot.

[0121] The related technical solutions are the same as those in Embodiment 1, and will not be elaborated here.

[0122] Embodiment 3

[0123] A calibration configuration group generation system for an industrial robot, comprising: a memory and a processor, where the memory stores a computer program, and when the processor executes the computer program, it executes the industrial robot calibration configuration group generation method provided in Embodiment 1 of the present invention.

[0124] The related technical solutions are the same as those in Embodiment 1, and will not be elaborated here.

[0125] Embodiment 4

[0126] A machine-readable storage medium stores machine-executable instructions, and when the machine-executable instructions are called and executed by a processor, the machine-executable instructions cause the processor to implement the industrial robot calibration configuration group generation method provided in Embodiment 1 of the present invention.

[0127] The related technical solutions are the same as those in Embodiment 1, and will not be elaborated here.

[0128] Those skilled in the art can easily understand that the above are only preferred embodiments of the present invention and are not intended to limit the present invention. Any modifications, equivalent replacements, and improvements made within the spirit and principle of the present invention should be included in the protection scope of the present invention.

Claims

1. A method for generating a configuration group for industrial robot calibration, characterized in that It includes the following steps: S1. Perform singular value decomposition on the coefficient matrix of the industrial robot kinematic error model to obtain the expression of the observability index of the calibration configuration group, thereby establishing a calibration configuration group generation model with the goal of optimal observability index. Among them, the constraint conditions of the calibration configuration group generation model include: the joint motion range constraint of the industrial robot, as well as the motion space constraint and acquisition constraint of the end of the industrial robot. The motion space constraint is used to constrain the end position of the industrial robot within its motion boundary range. The acquisition constraint is used to constrain the end position of the industrial robot within the measurement range of the measuring instrument. S2. Solve the calibration configuration group generation model to generate the optimal calibration configuration group for the industrial robot. Optimize the configuration order in the optimal calibration configuration group by minimizing the sum of the distances between all adjacent two configurations in the optimal calibration configuration group. Among them, the distance between the l-th configuration and the (l + 1)-th configuration in the optimal calibration configuration group is: Where N is the number of joint angles of the industrial robot; is the difference between the n-th joint angles of the l-th configuration and the (l + 1)-th configuration.

2. The method for generating a configuration group for industrial robot calibration according to claim 1, wherein The motion boundary range includes the Cartesian space motion range of the end of the industrial robot restricted by obstacles, specifically: the intersection of the Cartesian space motion range of the end of the industrial robot and the space range bounded by the obstacle position.

3. The method for generating a configuration group for industrial robot calibration according to claim 1, wherein The measuring instrument is a non-contact measuring device for emitting light to the industrial robot. At this time, a target ball is arranged at the end position of the industrial robot. The acquisition constraint is used to constrain the angle between the light incident direction and the negative direction of the normal of the plane where the light receiving port of the target ball is located within half of the light receiving angle of the target ball. The negative direction of the normal is the direction towards the inside of the target ball.

4. The method for generating a configuration group for industrial robot calibration according to claim 3, wherein The acquisition constraint is: wherein, is the vector corresponding to the light incident direction; is the vector corresponding to the negative direction of the normal of the plane where the light receiving port of the target ball is located; D is the light receiving angle of the target ball, in radians.

5. The method for generating a configuration group for industrial robot calibration according to any one of claims 1-4, characterized in that The objective function of the calibration configuration group generation model is: minf(x) = 1 / O1 where x is the configuration group for calibration; O1 is the observable index of the configuration group; σ i is the i-th singular value of the coefficient matrix, i = 1, 2, …, r; r is the number of kinematic parameters to be identified during kinematic parameter calibration; n is the number of pose points in the configuration group for calibration to be generated.

6. The method for generating a configuration group for industrial robot calibration according to any one of claims 1-4, characterized in that, Use the SQP algorithm to solve the calibration configuration group generation model.

7. The method for generating a configuration group for industrial robot calibration according to any one of claims 1-4, characterized in that Use the WPSO-SQP algorithm to solve the calibration configuration group generation model.

8. A configuration group generation system for industrial robot calibration, characterized in that, It includes: A model construction module for performing singular value decomposition on the coefficient matrix of the industrial robot kinematic error model to obtain the expression of the observability index of the calibration configuration group, thereby establishing a calibration configuration group generation model with the goal of optimal observability index. Among them, the constraint conditions of the calibration configuration group generation model include: the joint motion range constraint of the industrial robot, as well as the motion space constraint and acquisition constraint of the end of the industrial robot. The motion space constraint is used to constrain the end position of the industrial robot within its motion boundary range. The acquisition constraint is used to constrain the end position of the industrial robot within the measurement range of the measuring instrument. A generation module for solving the calibration configuration group generation model to generate the optimal calibration configuration group for the industrial robot. Optimize the configuration order in the optimal calibration configuration group by minimizing the sum of the distances between all adjacent two configurations in the optimal calibration configuration group. Among them, the distance between the l-th configuration and the (l + 1)-th configuration in the optimal calibration configuration group is: Where N is the number of joint angles of the industrial robot; is the difference between the n-th joint angles of the l-th configuration and the (l + 1)-th configuration.

9. A machine-readable storage medium, characterized in that, The machine-readable storage medium stores machine-executable instructions, and when the machine-executable instructions are called and executed by a processor, the machine-executable instructions cause the processor to implement the method for generating a configuration group for industrial robot calibration according to any one of claims 1-7.

Images