Robot system and parameter identification method

Abstract

The invention relates to a robot system and a parameter identification method. Modeling parameters are easy to identify. The robot system includes: a robot having a degree of freedom of three or more axes; a control device that operates the robot on the basis of the operation plan; a measurement device that measures a trajectory of a control point of the robot when the robot operates on the basis of the operation plan; and a processor that estimates, on the basis of the operation plan and the trajectory of the control point, a setting error of a modeling parameter necessary for the control device to calculate the position of the control point of the robot, and identifies the modeling parameter.

Description

Technical Field

[0001] The present disclosure relates to a robot system and a parameter identification method. Background Art

[0002] Patent document 1 discloses a mechanical error derivation method, which can uniquely calculate the derived solution when deriving the geometric errors of robots and NC processing machines including linear axes, and can seek to speed up the calculation and improve the accuracy of the derived errors. The mechanical error derivation method first uses the Denavit-Hartenberg notation represented by various parameters to model the mechanism with the processing tool installed at the front end. The mechanical error derivation method then measures the position / posture of the processing tool when the mechanism is actuated, and uses the model to estimate the position / posture of the processing tool when the mechanism is actuated. The mechanical error derivation method then derives the mechanical error of the mechanism by comparing the measured value and the estimated value of the position / posture of the processing tool. In addition, in the mechanical error derivation method, only the parameters that change the coordinates of the drive axis of the mechanism are used in the model.

[0003] Prior art literature

[0004] Patent Literature

[0005] Patent Document 1: Japanese Patent Application Laid-Open No. 7-028514 Summary of the invention

[0006] The present disclosure has been made in view of the above-mentioned existing situation, and an object of the present disclosure is to facilitate identification of modeling parameters.

[0007] The present disclosure provides a robot system comprising: a robot having more than three axes of freedom; a control device for moving the robot based on an action plan; a measuring device for measuring the trajectory of the control point of the robot when the robot moves based on the action plan; and a processor for estimating, based on the action plan and the trajectory of the control point, the setting error of the modeling parameters required by the control device to calculate the position of the control point of the robot, and identifying the modeling parameters.

[0008] In addition, the present disclosure provides a robot system, comprising: a robot having more than three axes of freedom; a control device, which moves the robot based on an action plan; a measuring device, which measures the trajectory of the control points of the robot when the robot moves based on the action plan; a processor, which estimates the setting error of the modeling parameters based on the action plan and the trajectory of the control points and identifies the modeling parameters; and a display device, wherein the processor displays the identification result of the modeling parameters on the display device.

[0009] In addition, the present disclosure provides a robot system, comprising: a robot having more than three axes of freedom; a control device, which moves the robot based on an action plan; a measuring device, which measures the trajectory of the control points of the robot when the robot moves based on the action plan; a processor, which estimates the setting error of the modeling parameters based on the action plan and the trajectory of the control points, and identifies the modeling parameters; and a memory, which changes the stored modeling parameters to the modeling parameters identified by the processor.

[0010] In addition, the present disclosure provides a parameter identification method, which enables a robot with more than three axes of freedom to move based on an action plan, measures the trajectory of the control points of the robot when the robot moves based on the action plan, and estimates the setting errors of the modeling parameters required to calculate the positions of the control points of the robot based on the action plan and the trajectory of the control points, thereby identifying the modeling parameters.

[0011] In addition, these general or specific aspects may be implemented by a system, an apparatus, a method, an integrated circuit, a computer program, or a recording medium, or may be implemented by any combination of systems, apparatuses, methods, integrated circuits, computer programs, and recording media.

[0012] Effects of the Invention

[0013] According to the present disclosure, modeling parameters can be easily identified. BRIEF DESCRIPTION OF THE DRAWINGS

[0014] Figure 1 It is the system structure diagram of the robot system.

[0015] Figure 2 This is a flowchart showing an example of a method for identifying modeling parameters.

[0016] Figure 3 This is a schematic diagram of an example of a method for identifying modeling parameters.

[0017] Figure 4 This is a diagram showing an example of a robot.

[0018] Figure 5 1 is a diagram showing an example of data for correcting the trajectory of actually measured control points.

[0019] Fig. 6A 1 is a diagram showing an example of data for correcting the trajectory of the calculated control points.

[0020] Figure 6B 1 is a diagram showing an example of data for correcting the trajectory of the calculated control points.

[0021] Figure 6C 1 is a diagram showing an example of data for correcting the trajectory of the calculated control points.

[0022] Figure 7 It is a graph showing the residual sum of squares of each error amount.

[0023] Figure 8 It is a graph showing the numerical value of the residual sum of squares for each error amount.

[0024] Explanation of symbols

[0025] 1Robotic system

[0026] 10. Robot

[0027] 20Computing device

[0028] 21 Memory

[0029] 22 processors

[0030] 23Action Plan Evaluation Department

[0031] 24 Identification Department

[0032] 25 Identification and Evaluation Department

[0033] 26 communication I / F

[0034] 27 Display Devices

[0035] 30 Track Sensor System

[0036] 40 control devices. DETAILED DESCRIPTION

[0037] (Process of realizing this embodiment)

[0038] In order to make the robot move correctly or realize complete offline teaching, it is necessary to identify the modeled parameter errors contained in the robot system and calibrate them as needed. The so-called robot system is a system that includes a robot, a control device that controls the robot, and a sensor that measures the trajectory of the robot. The so-called modeled parameters refer to the parameters required by the control device to calculate the control point position of the robot. Modeled parameters are, for example, the position of the origin of each axis of the multi-joint robot, tool coordinates, tool center point, connecting rod length, rigidity value of each axis in the rotation direction, rigidity value of the axis peeling direction, flexural rigidity value, torsional error between axes, or positional relationship error between axes. The so-called tool coordinates are the coordinates of the tool control point in an orthogonal coordinate system with the front end of the multi-joint robot as the origin.

[0039] The theoretical position information of the control point position of the multi-joint robot is calculated by kinematic sequence transformation using modeling parameters such as the angle information detected by the encoder of each axis and the length of the connecting rod. However, due to processing errors and assembly errors of the components of the multi-joint robot, individual differences will occur in the modeling parameters such as the length of the connecting rod. In order to correctly calculate the control point position of the multi-joint robot, it is necessary to obtain the error of the modeling parameters of each multi-joint robot and calibrate them. Hereinafter, the error of the modeling parameters is referred to as the modeling parameter error.

[0040] The mechanical error derivation method in Patent Document 1 uses the error vector between the ideal coordinates of the control point in the robot coordinate system and the coordinates of the actually measured control point to identify the modeled parameter error. However, there is a problem that a large-scale measurement system including many sensors is required to measure the position of the control point in the robot coordinate system, which makes the measurement complicated. In addition, there is a problem that it is difficult to measure the position of the control point in the robot coordinate system with high accuracy.

[0041] In view of such a situation, the present disclosure provides a robot system that can easily identify modeling parameters.

[0042] Hereinafter, the implementation modes of the robot system and parameter identification method involved in the specific disclosure of the present disclosure will be described in detail with appropriate reference to the accompanying drawings. In particular, the necessary detailed description is sometimes omitted. For example, the detailed description of matters that are already well known and the repeated description of substantially the same structure are sometimes omitted. This is to avoid the following description from becoming unnecessarily lengthy and to make it easy for those skilled in the art to understand. In addition, the drawings and the following description are provided for those skilled in the art to fully understand the present disclosure, and are not intended to limit the subject matter recorded in the patent request.

[0043] First, refer to Figure 1 The system configuration of the robot system 1 will be described. Figure 1 2 is a system structure diagram of the robot system 1.

[0044] The robot system 1 includes a robot 10, a calculation device 20, a trajectory sensor system 30, and a control device 40. The robot system 1 is a system for setting modeling parameters used to calculate the positions of control points of the robot 10 (hereinafter referred to as "control points").

[0045] The robot 10 is, for example, a multi-joint robot having two or more axes. Figure 1 In FIG. 1 , the robot 10 represents a robot having six axes of freedom. The robot 10 and the control device 40 can be connected by wired or wireless communication, and the robot 10 operates based on a control signal from the control device 40. An encoder is installed on each axis of the robot 10, and the robot 10 sends the angle information of each axis detected by the encoder to the control device 40.

[0046] The computing device 20 is a device for estimating a modeling parameter error and identifying a modeling parameter. The computing device 20 is, for example, a personal computer or a server device. The computing device 20 includes a memory 21 , a processor 22 , a communication I / F 26 , and a display device 27 .

[0047] The memory 21 is composed of, for example, a random access memory (RAM) and a read only memory (ROM), and temporarily stores the program required for the operation of the computing device 20, and further temporarily stores the data generated during the operation. The RAM is, for example, a working memory used in the operation of the computing device 20. The ROM stores and stores, for example, a program for controlling the computing device 20 in advance. In addition, the memory 21 may also store the initial values ​​of the modeling parameters of the robot system 1.

[0048] The processor 22 is, for example, a central processing unit (CPU), a digital signal processor (DSP), a graphic processing unit (GPU), or a field programmable gate array (FPGA). The processor 22 functions as a controller that manages the overall operation of the computing device 20. The processor 22 performs control processing for the operation of each part of the computing device 20, data input and output processing between each part of the computing device 20, data calculation processing, and data storage processing. The processor 22 operates according to the program stored in the memory 21. The processor 22 uses the memory 21 when operating, and temporarily saves the data generated or obtained by the processor 22 in the memory 21. The processor 22 realizes the respective functions of the action plan evaluation unit 23, the identification unit 24, and the identification evaluation unit 25 by using the program and data stored in the memory 21.

[0049] The motion plan evaluation unit 23 evaluates whether the motion plan of the robot 10 (hereinafter referred to as the "corrected motion plan") obtained from the control device 40 via the communication I / F 26 satisfies a given condition when the modeled parameter error is estimated. In addition, the motion plan evaluation unit 23 can obtain the motion plan from the memory 21 or a memory (not shown) storing the corrected motion plan. The motion plan evaluation unit 23 can output the result of the evaluated corrected motion plan to the display device 27.

[0050] The identification unit 24 estimates the modeling parameter error and identifies the modeling parameter based on the trajectory of the control point obtained from the trajectory sensor system 30 via the communication I / F 26 (hereinafter referred to as "measured control point trajectory") and the correction action plan obtained from the control device 40 via the communication I / F 26. The identification unit 24 outputs the identified modeling parameter to the control device 40 via the communication I / F 26.

[0051] The identification evaluation section 25 evaluates the accuracy of the modeling parameter identified by the identification section 24. The identification evaluation section 25 may output the result of the evaluated modeling parameter to the display device 27.

[0052] The communication I / F 26 is an interface circuit for wireless or wired communication between the computing device 20 and the trajectory sensor system 30 and between the computing device 20 and the control device 40. The communication method based on the communication I / F 26 is wired communication or wireless communication. Specifically, the wireless communication can be mobile communication such as WAN, LAN, LTE, 4G, 5G, power line communication, short-range wireless communication (such as Bluetooth (registered trademark) communication) or communication for portable phones. In addition, the method of wireless communication is not limited to the above examples.

[0053] The display device 27 is a device that displays the recognized modeling parameters, the evaluation results of the correction action plan, the evaluation results of the modeling parameters, etc. The display device 27 is, for example, a display or the like.

[0054] The trajectory sensor system 30 is a system for measuring the measured control point trajectory of the robot 10 . The trajectory sensor system 30 is, for example, a three-dimensional laser or a stereo camera. The trajectory sensor system 30 outputs the measured measured control point trajectory of the robot 10 to the recognition unit 24 of the calculation device 20 .

[0055] The control device 40 is a controller that controls the control point position of the robot 10 based on the correction action plan. The control device 40 stores the correction action plan in a memory (not shown) or the like. In addition, the control device 40 stores the initial value of the modeling parameter in a memory (not shown) or the like. In addition, the computing device 20 and the control device 40 may be constituted by an integrated device. The control device 40 controls the robot 10 using the modeling parameter recognized by the computing device 20. That is, the control device 40 rewrites the pre-stored modeling parameter into the modeling parameter recognized by the computing device 20.

[0056] Next, refer to Figure 2 An example of a method for identifying modeling parameters will be described below. Figure 2 is a flowchart showing an example of a method for identifying modeling parameters. Figure 2 In this paper, the flow of the identification method of the modeling parameters is limited to explaining, and the details of each process are described using Figures 4 to 8 The following will explain.

[0057] The recognition unit 24 obtains the actually measured control point trajectory which is the trajectory of the control points of the robot 10 (step St101 ).

[0058] The motion plan evaluation unit 23 obtains the correction motion plan from the control device 40 (step St102 ).

[0059] The motion plan evaluation unit 23 determines whether the correction motion plan acquired in the process of step St102 satisfies a given condition (step St103 ). A specific example of the given condition will be described later.

[0060] When determining that the corrected motion plan does not satisfy the given condition (step St103 , No), the motion plan evaluation unit 23 ends the process related to the identification of the modeling parameters.

[0061] When the action plan evaluation unit 23 determines that the corrected action plan satisfies the given condition (step St103, yes), it outputs the corrected action plan to the identification unit 24. In addition, the process of step St101 may be performed after the processes of step St102 and step St103. In this case, when the action plan evaluation unit 23 determines that the corrected action plan satisfies the given condition, the process of step St101 is performed.

[0062] The identification unit 24 obtains the calibration action plan from the control device 40 . The identification unit 24 estimates the modeling parameter error using the measured control point trajectory and the calibration action plan obtained in the process of step St101 , and identifies the modeling parameter (step St104 ). The identification unit 24 outputs the identified modeling parameter to the identification evaluation unit 25 .

[0063] The identification evaluation unit 25 evaluates the reliability of the modeling parameters identified in the process of step St104 (step St105 ).

[0064] The identification and evaluation unit 25 determines whether the result of the evaluation in the process of step St105 satisfies a predetermined criterion (step St106 ).

[0065] When the identification and evaluation unit 25 determines that the evaluation result in the process of step St105 does not meet the predetermined criterion (step St106, No), the identification and evaluation unit 25 causes the identification unit 24 to execute the process of step St101 again. In addition, when the identification and evaluation unit 25 determines that the evaluation result does not meet the predetermined criterion (step St106, No), the identification and evaluation unit 25 causes the identification unit 24 to execute the process of step St101 again, and causes the display device 27 to display the evaluation result. In addition, when the identification and evaluation unit 25 determines that the evaluation result does not meet the criterion for a predetermined number of times, the evaluation result and a signal indicating that the modeling parameter cannot be identified are displayed on the display device 27, and the process ends.

[0066] When the identification and evaluation unit 25 determines that the result of the evaluation in the process of step St105 satisfies the predetermined benchmark (step St106, yes), the evaluation result is displayed on the display device 27 (step St107). In addition, when the identification and evaluation unit 25 determines that the result of the evaluation in the process of step St105 satisfies the predetermined benchmark (step St106, yes), it outputs a signal to the identification unit 24 indicating that the reliability of the identified modeling parameters is high. The so-called high reliability of the modeling parameters means that the control point of the robot 10 is controlled by the control device 40 using the identified modeling parameters, and the trajectory specified by the control device 40 and the measured control point trajectory of the robot 10 are consistent with high precision. In other words, it means that the trajectory specified by the control device 40 and the measured control point trajectory of the robot 10 have a high degree of similarity. Based on the evaluation result of the high reliability received from the action plan evaluation unit 23, the identification unit 24 outputs the identified modeling parameters to the control device 40 so that they are reflected in the action of the robot 10. That is, the control device 40 controls the robot 10 by programming the motion plan incorporating the identified modeled parameters.

[0067] The identification unit 24 outputs the identified modeling parameters to the memory 21 and changes the modeling parameters stored in advance to the identified modeling parameters (step St108 ). The memory 21 changes the modeling parameters stored in advance to the identified modeling parameters received from the identification unit 24 .

[0068] In addition, at least one of the processes of step St102, step St103, step St105, step St106, step St107, and step St108 may be omitted.

[0069] In addition, the computing device 20 sends the modeling parameters identified in the process of step Stl04 to the control device 40. The control device 40 rewrites the modeling parameters stored in advance into the modeling parameters received from the computing device 20. The control device 40 uses the rewritten modeling parameters to control the robot 10. Thus, the control device 40 controls the robot 10 by using the identified modeling parameters, and can control the robot 10 with higher accuracy than controlling the robot 10 using the modeling parameters before identification.

[0070] Through the above, the robot system 1 can identify various modeling parameters at a time. In addition, the robot system 1 evaluates the reliability of the identified modeling parameters, and when the reliability is low, the modeling parameters can be re-identified using the measured control point trajectory that causes the robot 10 to move on different tracks. Thus, the robot system 1 can identify the modeling parameters with high precision. In addition, when the modeling parameters are identified, the robot system 1 can change the values ​​of the modeling parameters stored in the memory 21. Thus, the robot system 1 can use the identified modeling parameters to control the control point positions of the robot 10 with high precision. In addition, the robot system 1 can display the identified results on the display device 27, and notify the user who is calibrating the robot 10 of the results. Thus, the robot system 1 can help the user recognize whether the identification result is reliable, and can assist in improving the user's calibration workability.

[0071] Next, refer to Figure 3 An overview of an example of a method for identifying modeling parameters will be described. Figure 3 This is a schematic diagram of an example of a method for identifying modeling parameters.

[0072] The basic principle of identification of the modeling parameters will be described: Sr is a trajectory of a control point actually measured in a coordinate system parallel to a coordinate system set for the robot 10 (hereinafter referred to as a “robot coordinate system”).

[0073] When the error amount of each modeling parameter of n (n: an integer greater than 1) identification objects is set to e (e1, ..., ei, ..., en, i = 1 to n), the control point trajectory of the robot coordinate system obtained by simulation analysis using the modeling parameters including the error amount e (hereinafter referred to as "calculated control point trajectory") is set to Sn.

[0074] The error amount e of the modeling parameters included in the robot system 1 is identified by obtaining a combination of the error amounts e of the modeling parameters when the shapes of the trajectories of the actually measured control point trajectory Sr and the calculated control point trajectory Sn are closest.

[0075] Graphs Sim1, Sim2, Sim3, and Sim4 are calculated control point trajectories Sn obtained by simulation analysis by the identification unit 24. Graphs Sim1, Sim2, Sim3, and Sim4 are simulation analysis results when different error amounts e are included. The identification unit 24 arbitrarily sets the error amount e and calculates the calculated control point trajectories Sn as a plurality of simulation analysis results.

[0076] Graph Gr is the measured control point trajectory Sr.

[0077] In the graphs Sim1, Sim2, Sim3, Sim4, and Gr, the horizontal axis represents time and the vertical axis represents coordinates (i.e., positions of control points). The three lines shown in the graphs Sim1, Sim2, Sim3, Sim4, and Gr represent time changes of the XYZ coordinates of the three-dimensional coordinates.

[0078] The identification unit 24 compares the graphs Sim1, Sim2, Sim3, Sim4 and the graph Gr, and selects the calculated control point trajectory Sn that is the trajectory with the highest similarity to the graph Gr. The identification unit 24 estimates the error amount e used when calculating the selected calculated control point trajectory Sn as the modeling parameter error. In addition, the detailed calculation method of the modeling parameter error will be described later. The identification unit 24 identifies the modeling parameter based on the determined modeling parameter error.

[0079] (Example of a method for calculating modeling parameter errors)

[0080] Next, refer to Figures 4 to 8 An example of a method for estimating a modeling parameter error in the present embodiment will be described. Figure 4 It is a diagram showing an example of the robot 10 . Figure 5 1 is a diagram showing an example of data for correcting the trajectory of actually measured control points. Fig. 6A , Figure 6B as well as Figure 6C 1 is a diagram showing an example of data for correcting the trajectory of the calculated control points. Figure 7 It is a graph showing the residual sum of squares of each error amount. Figure 8 It is a graph showing the numerical value of the residual sum of squares for each error amount.

[0081] Figure 4 The robot 10 shown is a multi-joint robot with 6 axes of freedom. The robot 10 has a control point 11 at the front end. The numerical values ​​shown in the robot 10 represent the length between the arrows, and the unit is millimeter. In addition, Figure 4The lengths between the axes of the robot 10 shown are examples and are not limited. For the robot 10, the axis parallel to the gravity direction is set as the Z axis. The direction from the setting point of the robot relative to the ground toward the control point 11 is perpendicular to the Z axis and is set as the Y axis. The axis perpendicular to the Z axis and the Y axis is set as the X axis. The expressions related to these directions are used for convenience of explanation and are not intended to limit the posture of the structure when it is actually used.

[0082] First, explain Figure 2 The process of step St101 shown here corresponds to a method for measuring the actual control point trajectory. In order to measure the actual control point trajectory, the trajectory sensor system 30 is installed at a suitable position relative to the robot 1 .

[0083] The control device 40 sends the corrected motion plan to the robot 10 to move the robot 10. The trajectory sensor system 30 measures the trajectory of the control point of the robot 10 moving according to the corrected motion plan. The trajectory sensor system 30 sends the measured trajectory to the calculation device 20 as the measured control point trajectory Sr.

[0084] Here, the measured control point trajectory Sr needs to be measured in a coordinate system parallel to the robot coordinate system. For this reason, the above conditions can be ensured by configuring the coordinate system of the trajectory sensor system 30 to be parallel to the robot coordinate system. In addition, the coordinate system of the trajectory sensor system 30 can also be appropriately configured, and the computing device 20 obtains the correspondence between the coordinate system parallel to the robot coordinate system and the sensor coordinate system, and transforms the trajectory of the measured control point based on the correspondence, and calculates the measured control point trajectory Sr, thereby ensuring the above conditions.

[0085] The measured control point trajectory Sr is a series of control point position information obtained in a coordinate system parallel to the robot coordinate system at each measured posture, which can be three-dimensional information, or only one or two of the three dimensions can be used. When the measured control point trajectory Sr is the three-dimensional position information at each measured posture, the mathematical formula becomes formula (1).

[0086]

Mathematical formula 1

[0087]

[0088] Here, 1 represents the number of identification identifiers (hereinafter referred to as “measurement posture IDs”) of the measurement posture, and is an integer greater than or equal to 1.

[0089] Through the above, the calculation device 20 obtains the measured control point trajectory Sr.

[0090] Next, explain Figure 2An example of an identification method of a modeling parameter corresponding to the process of step St104 shown in FIG.

[0091] First, the identification unit 24 sets the combination of modeling parameters of the robot 10 as E (E1, E2, ..., Ei, ..., En), and sets the combination of the error amount of each modeling parameter as e i (ei l , e i2 ,……,ei j ,……,ei m , j = 1 ~ m, m: the number of combinations). E is arbitrarily set in accordance with the robot 10. For example, when the tool offset parameter of the robot 10 is used as the identification object, it becomes E = (X component of the tool offset parameter, Y component of the tool offset parameter, Z component of the tool offset parameter), n = 3. Here, the so-called tool offset parameter refers to the position of the tool action point and the tool control point of the robot 10 relative to the coordinates of the front end axis of the robot 10. E and e do not need to be determined after the measurement of the actual control point trajectory Sr, and can also be determined in advance.

[0092] Next, the identification unit 24 assumes that the robot 10 contains a combination of arbitrary error amounts e. i The correction action plan for the case of is to obtain the calculated control point trajectory through numerical calculation. For the measured control point trajectory Sr and the calculated control point trajectory Sn, the initial position will be aligned with the arbitrary coordinates (X0, Y0, Z o ) are set as the corrected measured control point trajectory Sr' and the corrected calculated control point trajectory Sn'. The calculated control point trajectory Sn is represented by equation (2), the corrected measured control point trajectory Sr' is represented by equation (3), and the corrected calculated control point trajectory Sn' is represented by equation (4).

[0093]

Mathematical formula 2

[0094]

[0095]

Mathematical formula 3

[0096] Sr′=Sr-(sr1-X0 sr2-Y0 sr3-Z0) (3)

[0097]

Mathematical formula 4

[0098] Sn′=Sn-(sn1-X0 sn2-Y0 sn3-Z0) (4)

[0099] In addition, the calculated control point trajectory Sn and the corrected calculated control point trajectory Sn' do not need to be calculated after the measured control point trajectory Sr is measured. Since they can be calculated when the correction action plan or the setting value of the robot 10, which are the conditions for obtaining the measured control point trajectory Sr, are known in advance, they can also be obtained in advance.

[0100] Next, the search for the similar trajectory Sn' that is most similar to the corrected measured control point trajectory Sr' is described. MostSimilar , calculate the similar trajectory Sn' MostSimilar The combination of error quantities used MostSimilar The method of estimating the error of the modeling parameters. The similarity between the trajectories can be evaluated by, for example, cosine similarity, correlation coefficient, or residual sum of squares. In addition, as a method for obtaining the similar trajectory Sn' MostSimilar The method of full search is described as an example.

[0101] The identification unit 24 sets a combination e of error amounts with appropriate intervals within the assumed range for each element of E. For example, if the X component of the tool offset parameter is set to ±0.2 [mm], and if the appropriate interval is set to 0.01 [mm], the combination e of error amounts becomes [-0.2, -0.19, ..., 0.18, 0.19, 0.2]. The combination e of error amounts for each element of E is evaluated. i (i=1~m) The similarity between the calculated corrected measured control point trajectory Sr' and the corrected calculated control point trajectory Sn', and the corrected calculated control point trajectory Sn' with the highest similarity is selected as the similar trajectory Sn' MostSimilarr The combination of error amounts used when calculating the correction control point trajectory Sn' with the highest similarity is determined as e MostSimilar .

[0102] Through the above, the identification unit 24 can infer Figure 2 The processing of step St104 shown is to model the parameter error and identify the modeling parameter.

[0103] Next, a method for evaluating the similarity between the corrected measured control point trajectory Sr' and the corrected calculated control point trajectory Sn' using the residual sum of squares is described. The method for evaluating the similarity between the corrected measured control point trajectory Sr' and the corrected calculated control point trajectory Sn' described below is Figure 2 An example of the processing of step St105 is shown.

[0104] exist Figure 4In the robot 10 shown, the tool offset setting value is [185, 0, 0]. [185, 0, 0] represent the values ​​of the XYZ coordinates in the tool coordinate system, respectively, and the unit is millimeter. The correction action plan is set as follows: the control point 11 is fixed to the initial value [0, 1150, 265], and the initial posture of the tool is a posture rotated -70 degrees around the Y axis from the posture where the vector from the tool coordinate origin to the control point is equal to the negative direction of the Z axis. In the correction action, a so-called pivot rotation action is specified to rotate the tool angle 140 degrees around the Y axis. In addition, the measured control point trajectory Sr is set to only the Y component in the three-dimensional coordinates, namely the XYZ components, and is set to (X0, Y0, Z0) = (0, 0, 0).

[0105] Figure 5 The graph shown is the corrected measured control point trajectory Sr' when the robot 10 performs the action according to the correction action plan. Since (X0, Y0, Z0) = (0, 0, 0), the vertical axis represents the displacement [mm] from the starting point of the trajectory. The horizontal axis represents the measured posture ID associated with the posture information of the robot 10 when each data in the measured control point trajectory Sr is measured. Figure 5 In the graph shown, the differences in the amounts of rotation about the Y axis between the respective data are set to be at equal intervals.

[0106] In addition, when the robot 10 performs the above-mentioned pivoting motion, it is expected that the modeled parameter errors exist only in the X component and the Z component of the tool offset parameter. The error amount of the X component of the tool offset parameter is set to Tx, and the error amount of the Z component is set to Tz. If the error amount Tx is set to [-0.1, 0, 0.1] and the error amount Tz is set to [-0.1, 0, 0.1], the combination e of the error amount is 3×3=9, and there are 9 combinations (m=9). Fig. 6A , Figure 6B as well as Figure 6C The graph shown represents each corrected calculated control point trajectory Sn′ calculated when the robot system 1 is assumed to include combinations e1 to e9 of error amounts. Fig. 6A , Figure 6B as well as Figure 6C The horizontal axis of the graph shown is the same measurement posture ID as the measured control point trajectory Sr, and each component is the same as the measured control point trajectory Sr, only the Y component of the XYZ component. Fig. 6A , Figure 6B as well as Figure 6C The title of the graph represents the value of the error amount Tx and the error amount Tz. For example, Fig. 6A The "Tx-0.1Tz-0.1" in FIG. 1 represents that the error amount Tx is -0.1 and the error amount Tz is -0.1. Each graph represents the correction calculation control point trajectory Sn' when the error amount Tx and the error amount Tz shown in the title are set.

[0107] Figure 7 The graph shown represents the residual sum of squares for each combination of the error amount Tx and the error amount Tz. The residual sum of squares is calculated using equation (5).

[0108]

Mathematical formula 5

[0109]

[0110] When the identification unit 24 evaluates the similarity using the residual sum of squares, it is assumed that the trajectory with the smallest evaluation value has the highest similarity. Figure 8 As shown, in the example of the present embodiment, the residual sum of squares is 0.03 when the error amount Tx is 0.1 and the error amount Tz is -0.1, which is the minimum. Therefore, the identification unit 24 estimates that the error amount Tx of the tool offset parameter is 0.1 [mm] and the error amount Tz is -0.1 [mm] as the modeled parameter error.

[0111] The identification evaluation unit 25 may, for example, use a preset threshold value for the value of the residual sum of squares of the estimated modeled parameter errors to determine whether the estimation result satisfies the benchmark. For example, when the value of the residual sum of squares of the estimated modeled parameter errors is greater than the threshold value, the identification evaluation unit 25 determines that the estimation result does not satisfy the benchmark. In addition, when the value of the residual sum of squares of the estimated modeled parameter errors is less than the threshold value, the identification evaluation unit 25 determines that the estimation result satisfies the benchmark.

[0112] (Evaluation of Corrective Action Plan)

[0113] Next, explain Figure 2 In the processing of step St103, the action plan evaluation unit 23 evaluates whether the correction action plan is a condition used when the action plan is suitable for identifying the modeled parameters. By setting the correction action plan to an action plan that satisfies the three conditions described below, the calculation device 20 can more easily calculate the identification result of the modeled parameters with higher reliability.

[0114] First, as the first condition, the independence condition can be cited. The so-called independence condition is a condition that the calculated control point trajectories Sn are not similar to each other. When the correction action plan satisfies the independence condition, the computing device 20 can obtain the identification result of the modeling parameters with high reliability. For example, when the computing device 20 estimates the modeling parameter error and identifies the modeling parameter by the above-mentioned full search method, consider the situation where the similarity between the calculated control point trajectory Sn1 and the calculated control point trajectory Sn2 in the calculated control point trajectory Sn numerically calculated according to the combination e of all error amounts is high. When the measured control point trajectory Sr is highly similar to the two trajectories of the calculated control point trajectory Sn1 and the calculated control point trajectory Sn2, it is difficult to determine which one is the true identification result. For this reason, in order to improve the reliability of the identification result of the modeling parameter, the correction action plan needs to meet the independence condition. When the correction action plan meets the independence condition, the obtained identification result is highly reliable, and the computing device 20 can reduce the process of confirming whether the identification result is a correct result.

[0115] Here, a specific example of a method for evaluating the independence condition is described. For example, when the computing device 20 estimates the modeling parameter error by the full search method and performs identification of the modeling parameter, it numerically calculates the control point trajectory Sn for the combination e of all error amounts that can be considered in advance. The action plan evaluation unit 23 of the computing device 20 calculates the similarity of each calculated control point trajectory Sn, and determines whether all evaluation values ​​are below a predetermined reference value. The reference value is a value set according to the identification system or the modeling parameter of the identification object, and is a value that can uniquely obtain the degree of the identification result. Thus, the computing device 20 can determine whether the correction action plan meets the independence condition.

[0116] Next, as the second condition, there is a condition of no mechanical play. The so-called condition of no mechanical play refers to a condition in which the backlash, idling, or other unique mechanical looseness or friction of the reducer used in the robot 10 does not affect the trajectory of the control point when the robot 10 is performing the action according to the correction action plan. When the correction action plan satisfies the condition of no mechanical play, the calculation device 20 does not need to model the above-mentioned influence and consider it in the numerical calculation when estimating the modeled parameter error, so it can simply make an estimate. This is because it is difficult to correctly perform numerical calculations on the influence of backlash, idling, or other unique mechanical looseness or friction on the trajectory of the control point.

[0117] Here, a specific example of a method for evaluating the condition of no mechanical play is described. Backlash or lost motion occurs when the direction of action of the torque applied to a mechanical component is reversed in the direction in which the backlash or lost motion occurs. When the robot 10 performs an action according to the correction action plan, the calculation device 20 obtains the torque generated by the mechanical component that is concerned about backlash or lost motion through numerical calculation. When the calculation device 20 does not produce a reversal in the direction of action of the torque during the action of the robot 10, the evaluation of the condition of no mechanical play is judged to be good (i.e., the condition of no mechanical play is satisfied), and when a reversal in the direction of action of the torque occurs, the evaluation of the condition of no mechanical play is judged to be poor (i.e., the condition of no mechanical play is not satisfied). In addition, the same is true for other specific mechanical looseness or friction. When the torque or force applied to the mechanical component is reversed in the direction in which the looseness or friction occurs, the looseness or friction occurs. When the robot 10 performs an action according to the correction action plan, the calculation device 20 obtains the force or torque applied to the mechanical component through numerical calculation. The calculation device 20 determines that the condition without mechanical backlash is good when the direction of action of torque or force is not reversed during the operation of the robot 10 , and determines that the condition without mechanical backlash is poor when the direction of action of torque is reversed.

[0118] Next, as a third condition, a low acceleration condition can be cited. The low acceleration condition is a condition in which the influence of the measured control point trajectory Sr due to the dynamic deflection of the robot 10 generated during the acceleration and deceleration of the robot 10 is such that the estimated accuracy of the modeled parameter error is not affected. When the correction action plan satisfies the low acceleration condition, the calculation device 20 can easily identify the influence because it does not need to model the influence when estimating the modeled parameter error and take it into account in the numerical calculation. This is because it is difficult to correctly calculate the influence of the dynamic deflection accompanying the acceleration and deceleration action on the control point trajectory.

[0119] (Evaluation method of identification results of modeling parameters)

[0120] Next, explain as Figure 2 The method of evaluating the reliability of the identification result of the modeling parameter is an example of the processing of step St105. As a method for evaluating the identification result of the modeling parameter, it is possible to cite the method of determining the similar trajectory Sn' MostSimilarA method for determining whether the similarity with the corrected measured control point trajectory Sr' satisfies a given benchmark. Here, the given benchmark depends on the calculation method of the similarity or the identification accuracy of the obtained modeling parameters, and is appropriately determined in conjunction with the identification conditions. For example, a predetermined threshold is used as a given benchmark. When the similarity satisfies the given benchmark, that is, when the identification evaluation unit 25 determines that the similarity is above the threshold, the control point trajectory can be explained by a model obtained by considering the modeling parameter error of the correction object in the benchmark model, and the reliability of the identification result can be evaluated as high. On the other hand, when the similarity does not satisfy the given benchmark, that is, when the identification evaluation unit 25 determines that the similarity is less than the threshold, the reliability of the identification result can be evaluated as low considering the possibility that the robot 10 contains a modeling parameter error outside the correction object, or the possibility that the control point trajectory cannot be correctly obtained. In addition, the method for evaluating the identification result of the modeling parameter is not limited to the above example.

[0121] (About the robot's corrective action)

[0122] The trajectory sensor system 30 needs to measure the measured control point trajectory in the sensor coordinate system that is parallel to the robot coordinate system. In order to make the robot coordinate system and the sensor coordinate system parallel, it is necessary to configure the sensor coordinate system of the trajectory sensor system 30 in parallel with the robot coordinate system, or to obtain a coordinate transformation matrix for transforming from the sensor coordinate system to the robot coordinate system to perform coordinate transformation. However, when an angle deviation occurs in the parallel relationship, a difference will occur between the shape of the measured control point trajectory and the calculated control point trajectory, thereby there is a possibility that the estimation accuracy of the modeling parameter error will be reduced. In addition, it is difficult to obtain the coordinate transformation matrix for transforming from the sensor coordinate system to the robot coordinate system with high accuracy. Here, the robot system 1 uses a correction plan action in which the posture of the robot 10 changes when the displacement of the three-dimensional position of the control point of the robot 10 is small, as a correction action plan. As a result, the robot system 1 can suppress the influence of a slight angle deviation of the parallel relationship between the sensor coordinate system and the robot coordinate system to measure the measured control point trajectory. As a correction action plan, for example, an action plan in which the displacement of the control point of the robot 10 from the initial position is less than 300 mm is set. When the robot system 1 measures the control point trajectory of the robot 10 that executes the correction plan action in the robot coordinate system, if the displacement of the trajectory from the initial position is small, the coordinate transformation error can be reduced, and the appropriate identification accuracy of the modeled parameters can be obtained. It can be seen from the experiment that although it also depends on the desired identification accuracy, the characteristics of the robot or the type of modeled parameters of the identification object, when the control point trajectory contains an error of more than 0.05mm, the appropriate identification accuracy cannot be obtained. In addition, it can be seen from the experiment that in the case of the method of measuring the robot coordinate system by making the robot actually move, the error of the coordinate transformation produces 0.01deg in the uncorrected robot. In addition, assuming that the error of the coordinate transformation contains 0.01deg and the correction action plan is set to an action of the displacement of the control point from the initial position to be less than 300mm for measurement, it can be calculated that the measured trajectory contains a maximum error of about 0.05mm. Through the above, when the correction action plan is set to an action of the displacement of the control point from the initial position to be 300mm, the robot system 1 is easy to obtain appropriate identification accuracy. Therefore, the robot system 1 does not need a large-scale measuring device because the displacement of the control point is small, and the actual control point trajectory can be measured by a small-scale measuring device. In addition, by setting the so-called pivoting action as the correction plan action, the influence of the error in the parallel relationship between the sensor coordinate system and the robot coordinate system on the identification of the modeling parameters can be further reduced, wherein the pivoting action is the following action plan: the displacement of the three-dimensional position of the control point of the robot 10 is suppressed to less than 0.05mm, the degree of calculation error when calculating the angle plan of each axis from the action plan is suppressed, and the control point is deliberately not moved.

[0123] (Summary of the present embodiment)

[0124] The following technology is disclosed through the description of the present embodiment above.

[0125] <Technology 1>

[0126] The robot system involved in this embodiment includes: a robot having more than three axes of freedom; a control device that moves the robot based on an action plan; a measuring device that measures the trajectory of the robot's control points when the robot moves based on the action plan; and a processor that estimates the setting errors of the modeling parameters required by the control device to calculate the positions of the robot's control points based on the action plan and the trajectory of the control points, and identifies the modeling parameters.

[0127] Thus, the robot system according to the present embodiment does not need to align the origins of the coordinate system set in the robot and the coordinate system set in the measuring device by using the trajectory of the control point, and can estimate the setting error of the modeling parameter and identify the modeling parameter by simply aligning the parallel relationship. Thus, the robot system does not need a large-scale device when measuring the trajectory of the control point, and can easily identify the modeling parameter.

[0128] <Technology 2>

[0129] In the robot system described in technology 1, the processor performs the following processing: obtaining multiple calculated control point trajectories, wherein the calculated control point trajectories are obtained by calculating the trajectories of the control points using the action plan and modeling parameters including arbitrary setting errors, calculating the similarity between the trajectories of the control points measured by the measuring device and the calculated control point trajectories, and estimating the setting error used when calculating the calculated control point trajectory with the highest similarity as the setting error of the modeling parameters.

[0130] Thus, the robot system according to the present embodiment can estimate setting errors of various modeling parameters at once.

[0131] <Technology 3>

[0132] In the robot system described in technology 1 or technology 2, the processor performs the following processing: evaluating the reliability of the identified modeling parameters based on whether the similarity is above a threshold value, and when it is determined that the similarity is less than the threshold value, causing the measuring device to measure the trajectory of the control point again, and based on the action plan and the trajectory of the re-measured control point, estimating the setting error of the modeling parameters, and identifying the modeling parameters.

[0133] Thus, the robot system according to the present embodiment can evaluate the reliability of the identified modeling parameters, and if the reliability is low, re-measure the trajectory of the control point and perform identification of the modeling parameters. That is, the robot system can identify the modeling parameters whose reliability satisfies a given benchmark. Thus, the robot system can calibrate the robot with high accuracy.

[0134] <Technique 4>

[0135] The robot system according to any one of techniques 1 to 3 further includes a display device, and when the processor determines that the similarity is equal to or greater than a threshold value, the processor displays the evaluation result on the display device.

[0136] Thus, the robot system according to the present embodiment can notify the user who manages the robot of the result of the reliability of the identified modeling parameters. In other words, the robot system can assist the user in the work of calibrating the robot.

[0137] <Technique 5>

[0138] In the robot system described in any one of Techniques 1 to 4, the control device changes the operation of controlling the robot based on a setting error of a modeling parameter.

[0139] Therefore, the robot system according to the present embodiment can control the position of the robot with high accuracy by using the identified modeling parameters.

[0140] <Technique 6>

[0141] In the robot system described in any one of Techniques 1 to 5, the processor rewrites the modeling parameters of the control device into the identified modeling parameters.

[0142] Therefore, the robot system according to the present embodiment can control the robot with high accuracy by using the identified modeling parameters.

[0143] <Technique 7>

[0144] In the robot system described in any one of the techniques 1 to 6, the motion plan is an motion plan characterized by a displacement of the control point from an initial position being 300 mm or less.

[0145] Thus, the robot system according to the present embodiment can measure the position of the control point using a small-scale measuring device by using an action with a small position displacement of the control point as an action plan. Thus, the robot system does not need a large-scale measuring device and can easily obtain the setting error of the modeling parameter using a small-scale measuring device.

[0146] <Technology 8>

[0147] In the robot system according to any one of techniques 1 to 7, the motion plan is an motion plan in which the posture of the robot changes when the displacement of the three-dimensional position of the control point is 0.05 mm or less.

[0148] Therefore, the robot system involved in this embodiment can reduce the influence of the error in the parallel relationship between the coordinate system set in the measuring device and the coordinate system set in the robot on the identification of the setting error of the modeling parameter because the displacement of the position of the control point is extremely small, less than 0.05 mm. Therefore, the robot system can estimate the setting error of the modeling parameter with high accuracy and identify the modeling parameter.

[0149] <Technique 9>

[0150] In the robot system according to any one of the techniques 1 to 8, the processor determines whether the action plan satisfies a condition that the calculated control point trajectories are not similar to each other.

[0151] Therefore, the robot system according to the present embodiment can uniquely estimate the setting error of the modeling parameter and identify the modeling parameter, thereby obtaining a highly reliable identification result.

[0152] <Technique 10>

[0153] In the robot system recorded in any one of Technology 1 to Technology 9, the modeling parameters are any combination of the origin of each axis, tool offset, stiffness value of each axis in the rotation direction, stiffness value of the axis peeling direction, torsional error between axes, positional relationship error between axes, and connecting rod length.

[0154] Thus, the robot system according to the present embodiment can collectively identify setting errors of various modeling parameters.

[0155] <Technology 11>

[0156] In the robot system described in any one of techniques 1 to 10, when the control device causes the robot to move based on the motion plan and then causes the robot to move again based on the motion plan, the control device causes the robot to move in a different motion from that performed immediately before.

[0157] Thus, when re-identifying the setting error of the modeling parameters, the robot system according to the present embodiment can use the trajectory data of the control point different from the behavior of the robot just before. That is, the robot system can estimate the setting error of the modeling parameters with high accuracy and identify the modeling parameters.

[0158] The above embodiments are described with reference to the accompanying drawings, but the present disclosure is not limited to the relevant examples. As long as one is skilled in the art, it is obvious that various changes, corrections, substitutions, additions, deletions, and equivalents can be thought of within the scope of the scope of the patent claim, and it should be understood that these also belong to the technical scope of the present disclosure. In addition, the various constituent elements in the above embodiments can also be arbitrarily combined within the scope of the main purpose of the invention.

[0159] Industrial Applicability

[0160] The technology disclosed in the present invention is useful as a robot system and a parameter error estimation method that can easily identify modeled parameters.

Claims

1. A robot system comprising: Robots with more than 3 axes of freedom; A control device for causing the robot to move based on the motion plan; a measuring device for measuring a trajectory of a control point of the robot measured in a coordinate system different from a robot coordinate system when the robot moves based on the motion plan; and The processor estimates the setting error of the modeling parameter required by the control device to calculate the position of the control point of the robot based on the action plan and the trajectory of the control point actually measured in the coordinate system different from the robot coordinate system, and identifies the modeling parameter.

2. The robot system according to claim 1, wherein: The processor performs the following processing: obtaining a plurality of calculated control point trajectories, the calculated control point trajectories being obtained by calculating the trajectories of the control points using the motion plan and modeled parameters including any set errors, calculating the similarity between the trajectory of the control point measured by the measuring device and the trajectory of the calculated control point, The setting error used when calculating the trajectory of the control point with the highest similarity is estimated as the setting error of the modeling parameter.

3. The robot system according to claim 2, wherein: The processor performs the following processing: evaluating the reliability of the identified modeling parameters based on the determination of whether the similarity is above a threshold, When it is determined that the similarity is less than a threshold value, the measuring device is caused to measure the trajectory of the control point again. Based on the motion plan and the re-measured trajectory of the control point, a setting error of the modeling parameter is estimated, and the modeling parameter is identified.

4. The robot system according to claim 3, wherein: The robot system also includes a display device. The processor displays the evaluation result on the display device when it is determined that the similarity is equal to or greater than a threshold value.

5. The robot system according to claim 1, wherein: The control device changes and controls the operation of the robot based on a setting error of the modeling parameter.

6. The robot system according to claim 1, wherein: The processor rewrites the modeled parameters of the control device into the identified modeled parameters.

7. The robot system according to claim 1, wherein: The motion plan is an motion plan characterized in that the displacement of the control point from the initial position is 300 mm or less.

8. The robot system according to claim 7, wherein: The motion plan is a motion plan in which the posture of the robot changes in a state in which the displacement of the three-dimensional position of the control point is 0.05 mm or less.

9. The robot system according to claim 2, wherein: The processor determines whether the action plan satisfies a condition that the calculated control point trajectories are not similar to each other.

10. The robot system according to claim 1, wherein: The modeling parameters are any combination of the origin of each axis, tool offset, rigidity value of each axis in the rotation direction, rigidity value of the axis peeling direction, torsional error between axes, positional relationship error between axes, and link length.

11. The robot system according to claim 1, wherein: The control device operates the robot in a different behavior from the behavior that the robot performed immediately before, when the control device operates the robot again based on the motion plan after operating the robot based on the motion plan.

12. A robot system comprising: Robots with more than 3 axes of freedom; A control device for causing the robot to move based on the motion plan; a measuring device for measuring a trajectory of a control point of the robot measured in a coordinate system different from a robot coordinate system when the robot moves based on the motion plan; and a processor that estimates a setting error of a modeling parameter based on the motion plan and a trajectory of the control point actually measured in the coordinate system different from the robot coordinate system, and identifies the modeling parameter; and Display devices, The processor causes the identification result of the modeling parameter to be displayed on the display device.

13. A robot system comprising: Robots with more than 3 axes of freedom; A control device for causing the robot to move based on the motion plan; a measuring device for measuring a trajectory of a control point of the robot measured in a coordinate system different from a robot coordinate system when the robot moves based on the motion plan; a processor that estimates a setting error of a modeling parameter based on the motion plan and a trajectory of the control point actually measured in the coordinate system different from the robot coordinate system, and identifies the modeling parameter; and The memory changes the stored modeling parameters to the modeling parameters recognized by the processor.

14. A parameter identification method, Based on the motion plan, the robot with more than 3 axes of freedom moves. measuring a trajectory of a control point of the robot measured in a coordinate system different from the robot coordinate system when the robot moves based on the motion plan, Based on the motion plan and the trajectory of the control point actually measured in the coordinate system different from the robot coordinate system, setting errors of modeling parameters required to calculate the positions of the control points of the robot are estimated, and the modeling parameters are identified.

Images