Robot calibration system and calibration method
By using the data storage and parameter selection part to select mechanical error parameters in a multi-articular robot, combined with the repeated evaluation of the parameter identification and evaluation part, the problem of insufficient mechanical error parameter identification accuracy in the prior art is solved, and the front end of the robot is realized High precision calibration of position.
Patent Information
- Application Number
- CN201910863486.4
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- Priority Date
- 2018-09-11
- Filing Date
- 2019-09-09
- Publication Date
- 2024-12-31
- Estimated Expiration
- 2039-09-09
AI Technical Summary
The prior art cannot obtain the best accuracy when identifying the mechanical error parameters of multi-joint robots, resulting in insufficient absolute position accuracy at the front end of the robot.
The data storage unit is used to store the actual measured position information and the robot status information, and the parameter selection unit selects an appropriate parameter group from the mechanical error parameter group, and uses the parameter identification unit to identify it, and the identification result evaluation unit evaluates the results. Repeat selection and identification until the predetermined reference is met to ensure absolute accuracy of the robot's front-end position.
Even if the number of data is insufficient or pre-determined position information cannot be obtained, the appropriate number and type of mechanical error parameters can be identified, the absolute position accuracy of the robot front end can be improved, and the appropriate mechanical error parameters can be quickly set when the reference is not met. gather.
Smart Images
Figure CN110883771B_ABST
Abstract
Description
Technical Field
[0001] The invention relates to a robot calibration system and a robot calibration method. Background Art
[0002] For a long time, the following technology has been known: multiple mechanical error parameters are identified in a manner that minimizes the difference between actual measured position information obtained by measuring the front end position information of multiple postures of a multi-articulated robot using a three-dimensional position measuring instrument and the theoretical position of the front end calculated by kinematic forward transformation of the angle data of each rotating joint and the connecting rod length, thereby improving the absolute accuracy of the front end of the robot (for example, refer to patent document 1).
[0003] In the technology disclosed in Patent Document 1, mechanical error parameters to be identified are divided into three groups: posture parameters, linear axis movement direction parameters, and translation parameters, and all mechanical error parameters in the groups are identified in this order.
[0004] Prior art literature
[0005] Patent Literature
[0006] Patent Document 1: Japanese Patent Application Laid-Open No. 7-28514 Summary of the invention
[0007] Problem that the invention aims to solve
[0008] However, there is a problem that even if all the grouped mechanical error parameters are identified, the best accuracy cannot be obtained.
[0009] The object of the present invention is to provide a robot calibration system and calibration method, which can identify an appropriate number and type of mechanical error parameters and improve the absolute accuracy of the front end of the robot.
[0010] Solutions for solving problems
[0011] One embodiment of the present invention is a robot calibration system, which comprises: a data storage unit, which stores a plurality of data obtained by combining actual measurement position information obtained by actually measuring the front end position of the robot and information representing the state of the robot when the actual measurement position information is actually measured; a parameter selection unit, which selects a group of mechanical error parameters to be identified from a parameter group representing the mechanical error of the robot; a parameter identification unit, which uses the data stored in the data storage unit and the information of the mechanical error parameters selected by the parameter selection unit to identify each of the mechanical error parameters; and an identification result evaluation unit, which evaluates the identification result identified by the parameter identification unit, and when the evaluation performed by the identification result evaluation unit does not meet a predetermined benchmark, repeats the selection of different mechanical error parameters by the parameter selection unit, the identification by the parameter identification unit, and the evaluation by the identification result evaluation unit.
[0012] According to the present scheme, when a plurality of data obtained by combining the actual measured position information of the front end position of the robot actually measured and the information indicating the state of the robot when the actual measured position information is actually measured are stored in the data storage unit, a set of mechanical error parameters to be identified is selected from the parameter group indicating the mechanical error of the robot by the parameter selection unit. The parameter identification unit identifies the mechanical error parameters using the selected mechanical error parameters and the plurality of stored data, and the identification result is evaluated by the identification result evaluation unit. Furthermore, when the evaluation does not meet the predetermined criterion, a different mechanical error parameter is selected by the parameter selection unit, and the identification by the parameter identification unit and the evaluation by the identification result evaluation unit are repeated.
[0013] When the data storage unit can only store a predetermined number of actually measured position information of the predetermined front end position, the mechanical error parameters to be used for calculating the front end position can be determined when the robot is shipped. However, if the amount of data stored in the data storage unit is insufficient, or if the actually measured position information of the predetermined front end position cannot be obtained, even if the mechanical error parameters determined at the time of shipment are used, the absolute position accuracy of the front end of the robot may not be improved.
[0014] According to the present scheme, even when the amount of data stored in the data storage unit is insufficient or when the actual measured position information of the predetermined front end position cannot be obtained, a set of mechanical error parameters selected from the parameter group can be identified and the identification result can be evaluated, and when the predetermined reference is not satisfied, a different set of mechanical error parameters can be selected, identified and evaluated, so that mechanical error parameters close to the reference can be set. Thus, the absolute position accuracy of the front end position of the robot calculated using the information indicating the state of the robot and the mechanical error parameters can be improved.
[0015] In the above aspect, the information indicating the state of the robot may include an angle of each joint of the robot at the time when the actual measured position information is actually measured.
[0016] According to this configuration, it is possible to improve the absolute position accuracy of the tip position of the robot calculated using the angles of each joint of the robot and the mechanical error parameters.
[0017] Furthermore, in the above aspect, the information indicating the state of the robot includes torque acting on each joint of the robot at the time when the actual measurement position information is actually measured.
[0018] According to this configuration, it is possible to improve the absolute position accuracy of the tip position of the robot calculated using the torque acting on each joint of the robot and the mechanical error parameter.
[0019] Furthermore, in the above aspect, when the evaluation performed by the identification result evaluation unit does not satisfy the criterion within a predetermined time, the mechanical error parameter with the highest evaluation obtained so far is set as the mechanical error parameter to be used.
[0020] According to this configuration, even when the selection of the mechanical error parameters is switched and the identification result does not satisfy the reference, it is possible to set an appropriate set of mechanical error parameters without taking a lot of time.
[0021] In the above aspect, if the evaluation by the identification result evaluation unit does not satisfy the criterion even after the mechanical error parameter is selected a predetermined number of times, the mechanical error parameter with the highest evaluation thus far is set as the mechanical error parameter to be used.
[0022] According to this configuration, even if the selection of the mechanical error parameters is switched and the identification result does not satisfy the reference, a more appropriate set of mechanical error parameters can be set without taking too much time.
[0023] In addition, another embodiment of the present invention is a robot calibration method, which includes the following steps: storing a plurality of data that is a combination of actual measurement position information obtained by actually measuring the front end position of the robot and information representing the state of the robot when the actual measurement position information is actually measured; selecting a group of mechanical error parameters to be identified from a parameter group representing the mechanical error of the robot; using the stored data and the selected information of the mechanical error parameters to identify each of the mechanical error parameters; evaluating the identification results; and if the evaluation does not meet a predetermined benchmark, repeating the selection, identification, and evaluation of different mechanical error parameters.
[0024] Effects of the Invention
[0025] According to the present invention, there is an effect that an appropriate number and type of mechanical error parameters can be identified, and the absolute accuracy of the tip of the robot can be improved. BRIEF DESCRIPTION OF THE DRAWINGS
[0026] Figure 1 1 is an overall configuration diagram of a robot system including a calibration system according to an embodiment of the present invention.
[0027] Figure 2 Yes means Figure 1 Block diagram of the calibration system.
[0028] Figure 3 It is due to Figure 2 A diagram illustrating the selection of a set of mechanical error parameters performed by a parameter selection unit of a calibration system.
[0029] Figure 4 It is to use Figure 2 A flow chart illustrating a calibration method of a calibration system.
[0030] Figure 5 Yes means Figure 4 Flowchart of a first variant of the calibration method.
[0031] Figure 6 Yes means Figure 4 Flowchart of a second variant of the calibration method.
[0032] Description of Reference Numerals
[0033] 1: Calibrate the system
[0034] 2: Robot
[0035] 5: Data storage unit
[0036] 6: Parameter storage unit (parameter selection unit)
[0037] 7: Selection section (parameter selection section)
[0038] 8: Parameter identification unit
[0039] 9: Identification result evaluation department
[0040] A: Front position DETAILED DESCRIPTION
[0041] A robot calibration system 1 according to an embodiment of the present invention will be described below with reference to the drawings.
[0042] like Figure 1As shown, the calibration system 1 of this embodiment is connected to a system having a robot 2, a control device 3, and a three-dimensional position measuring instrument 4, and is a system set to calculate mechanical error parameters of the front end position in the control device 3, wherein the control device 3 is connected to the robot 2, and the three-dimensional position measuring instrument 4 actually measures the front end position A of the robot 2 from the outside.
[0043] The robot 2 is, for example, an articulated robot.
[0044] The control device 3 converts the angle information and the link length information detected by the encoders of the respective rotating joints (joints) of the robot 2, which is a multi-articulated robot, into theoretical position information (hereinafter referred to as the theoretical position) of the front end position of the robot 2 in a fixed three-dimensional orthogonal coordinate system through kinematic forward transformation. However, due to the processing errors and assembly errors of the components of the robot 2, the link length and the like have individual differences.
[0045] Therefore, in order to calculate the theoretical position information, in addition to the information representing the state of the robot 2, such as the angle information, multiple mechanical error parameters including the DH parameter of the connecting rod length and the deflection caused by the load torque of the rotating joint are used for modeling to identify the mechanical error parameters so as to minimize the difference between the actual measured position and the theoretical position, thereby improving the absolute position accuracy of the calculated front end position.
[0046] like Figure 2 As shown, the calibration system 1 comprises: a data storage unit 5, which stores actual measurement position information actually measured by the three-dimensional position measuring instrument 4, a plurality of data corresponding to the information representing the state of the robot 2 including the angle information of each joint of the robot 2 at the moment when the actual measurement position information is actually measured; a parameter storage unit (parameter selection unit) 6, which stores mechanical error parameter information representing the mechanical error of the robot 2; and a selection unit (parameter selection unit) 7, which selects a set of mechanical error parameters to be identified from the mechanical error parameters stored in the parameter storage unit 6.
[0047] Furthermore, the calibration system 1 further includes a parameter identification unit 8 for identifying the selected mechanical error parameter and an identification result evaluation unit 9 for evaluating the identification result.
[0048] The data storage unit 5 and the parameter storage unit 6 are, for example, memories. The selection unit 7, the parameter identification unit 8, and the identification result evaluation unit 9 are constituted by, for example, a processor.
[0049] The parameter storage unit 6 stores a plurality of candidates that can be selected as the mechanical error parameter. Figure 3 In the example shown, identification information of 60 mechanical error parameter candidates from P1 to P60 is stored.
[0050] The selection unit 7 selects a set of multiple candidates including the candidates stored in the parameter storage unit 6. The initial selection may use a predetermined set or an arbitrarily selected set. Figure 3 In the example shown, a set including, for example, 21 mechanical error parameters out of 60 mechanical error parameters stored in the parameter storage unit 6 is selected.
[0051] The second and subsequent selections can be made by randomly adding mechanical error parameters that have not been selected so far, or randomly deleting any mechanical error parameter that has been selected. Addition or deletion can be done one by one, or in any number. Figure 3 In the example shown, the mechanical error parameter P1 is added in the second selection, and the mechanical error parameter P3 is added in the third selection.
[0052] The parameter identification unit 8 uses the plurality of data stored in the data storage unit 5 and the set of mechanical error parameters selected by the selection unit 7, and identifies the mechanical error parameters included in the set using a known method. For example, an error function between the actual measured position of the tip position of the robot 2 stored in the data storage unit 5 and the theoretical position calculated using the actual measured position and the mechanical error parameters is created, and the mechanical error parameters are identified using the least square method.
[0053] The identification result evaluation unit 9 evaluates the results identified by the parameter identification unit 8 according to a predetermined criterion. For example, for all actual measurement positions, the error between the theoretical position and the actual measurement position is calculated, and it is sufficient to determine whether the average value satisfies the evaluation criterion (for example, less than 0.3 mm), wherein the theoretical position is calculated using the set of identified mechanical error parameters.
[0054] Next, a calibration method using the calibration system 1 of the robot 2 of the present embodiment configured as described above will be described.
[0055] In order to perform calibration using the calibration system 1 of this embodiment, as shown in FIG. Figure 4 As shown, first, each joint of the robot 2, which is a multi-joint robot, is moved to arrange the front end at a plurality of positions, and the front end position is actually measured by the three-dimensional position measuring device 4 at each position (step S1). In addition, the angle information of each joint of the robot 2 at the time of the actual measurement is obtained (step S2), and the obtained angle information is stored in the data storage unit 5 in correspondence with the measured actual measurement position (step S3).
[0056] Next, the selection unit 7 selects a set of mechanical error parameters from the candidates of the mechanical error parameters stored in the parameter storage unit 6 (step S4 ).
[0057] Then, an error function between the theoretical position calculated using the selected set of mechanical error parameters and the angle information stored in the data storage unit 5 and the actual measured position is created, and the mechanical error parameters are identified by the least square method (step S5).
[0058] Then, the identification result evaluation unit 9 uses the identified mechanical error parameters to evaluate whether the identification result satisfies a predetermined criterion (step S6 ).
[0059] If the criterion is not satisfied, the process returns to step S4 and the process from step S4 to step S6 is repeated to select a set of mechanical error parameters different from the set already selected.
[0060] If the criterion is satisfied in step S6 , the mechanical error parameter identified at that time is output to the control device 3 (step S7 ).
[0061] Thus, according to the calibration system 1 and the calibration method of the robot 2 of the present embodiment, since they do not merely identify predetermined mechanical error parameters but search for a set of mechanical error parameters whose identification results satisfy predetermined benchmarks, they have the following advantages: they are able to identify an appropriate number and types of mechanical error parameters, and are able to improve the absolute accuracy of the front end of the robot 2.
[0062] That is, even if the user cannot prepare data including the required number or actual measured positions and theoretical positions about the required positions, the absolute accuracy of the front end of the robot 2 can be improved by searching for a set of mechanical error parameters whose identification results meet a predetermined benchmark.
[0063] In addition, in this embodiment, as information representing the state of the robot 2, the angle information of each rotating joint detected by the encoder is exemplified, but instead of or in addition to this, other information such as torque or external force detected by a sensor or calculated based on other information may be used.
[0064] In addition, in the present embodiment, the selection, identification and evaluation of the mechanical error parameter are repeated until the identification result satisfies the predetermined criterion, but instead, Figure 5 As shown, the time is counted when the set of mechanical error parameters is selected (step S8), and if the identification result does not meet the reference, it is determined whether a predetermined time has passed (step S9).
[0065] That is, if the predetermined time has not passed, the process returns to step S4, and the selection, identification, and evaluation of the mechanical error parameters are repeated. If the predetermined time has passed, the mechanical error parameters indicating the identification result closest to the reference at that time are output (step S7). In this way, when the evaluation reference is strictly set, it takes a long time to meet the reference. Therefore, by limiting the time for processing, the mechanical error parameters indicating the identification result close to the reference can be set in a short time.
[0066] Alternatively, instead of limiting the time to process, Figure 6 As shown in FIG. 1 , when the number of selections k of the set of mechanical error parameters reaches the predetermined number of times, the mechanical error parameters representing the identification result closest to the benchmark at that moment are output. In the figure, the number of selections k is reset before the selection starts (step S10) and incremented (step S11). Moreover, when the identification result does not meet the benchmark, it is determined whether the predetermined number of selections kO times have been performed (step S12). If the predetermined number of selections kO times have not been performed, the process returns to step S4, and the selection, identification, and evaluation of the mechanical error parameters are repeated.
[0067] Thus, when the evaluation criterion is strictly set, it takes a long time to satisfy the criterion. Therefore, by limiting the number of repetitions and performing the process, it is possible to set a mechanical error parameter indicating an identification result close to the criterion in a short time.
Claims
1. A robot calibration system, characterized in that: have: a data storage unit storing a plurality of data in which actual measurement position information obtained by actually measuring the position of the front end of the robot and information indicating the state of the robot when the actual measurement position information is actually measured are associated with each other; a parameter storage unit storing a plurality of parameters representing the mechanical error of the robot and serving as candidates for mechanical error parameters used by the robot; a parameter selection unit, which selects a set of parameters from the plurality of stored parameters as a set of mechanical error parameters; a parameter identification unit that uses the stored plurality of data and the information of the selected set of mechanical error parameters to identify the selected set of mechanical error parameters; as well as an identification result evaluation unit that evaluates the identification result identified by the parameter identification unit, When the evaluation performed by the identification result evaluation unit does not meet the predetermined benchmark, the parameter selection unit selects a set of parameters that is a different combination from the set of parameters selected as the set of mechanical error parameters from the stored multiple parameters as a new set of mechanical error parameters, and repeats the identification performed by the parameter identification unit and the evaluation performed by the identification result evaluation unit.
2. The robot calibration system according to claim 1, characterized in that: The information indicating the state of the robot includes an angle of each joint of the robot at a time when the actual measured position information is actually measured.
3. The robot calibration system according to claim 1, characterized in that: The information indicating the state of the robot includes torque acting on each joint of the robot at the time when the actual measured position information is actually measured.
4. The robot calibration system according to any one of claims 1 to 3, characterized in that: When the evaluation performed by the recognition result evaluation unit does not satisfy the reference within a predetermined time, the set of mechanical error parameters with the highest evaluation obtained so far is set as the mechanical error parameters used by the robot.
5. The robot calibration system according to any one of claims 1 to 3, characterized in that: If the evaluation by the identification result evaluation unit does not satisfy the criterion even after the mechanical error parameters are selected a predetermined number of times, the set of mechanical error parameters with the highest evaluation obtained so far is set as the mechanical error parameters used by the robot.
6. A robot calibration method, characterized in that: The following steps are involved: storing a plurality of data in which actual measurement position information obtained by actually measuring the front end position of the robot and information indicating the state of the robot when the actual measurement position information is actually measured are associated with each other, and storing a plurality of parameters indicating the mechanical error of the robot and serving as candidates for mechanical error parameters used by the robot; Selecting a set of parameters from the stored plurality of parameters as a set of mechanical error parameters; identifying the selected set of mechanical error parameters using the stored plurality of data and information about the selected set of mechanical error parameters; Evaluate the identification results; as well as When the evaluation does not satisfy a predetermined criterion, a set of parameters different from the set of parameters selected as the set of mechanical error parameters is selected from the plurality of stored parameters as a new set of mechanical error parameters, and identification and evaluation are repeated.