JPH0950309A - Robot constant calibration calculation method and calibration calculation device - Google Patents

Abstract

(57) [Abstract] [PROBLEMS] Even if there is a large mounting error at the angle origin of the robot, a necessary calibration correction value including the error is obtained to realize a highly accurate operation of the robot. SOLUTION: The position data of the multi-joint robot is obtained by the robot angle / position conversion calculating means 3 from the angle data of the multi-joint robot and the calibration robot constant obtained by adding the calibration correction value to the robot constant, and the position data of the multi-joint robot is obtained. The correction value for the setting error of the robot constant is obtained from the position data of the articulated robot by the calibration correction value calculating means 4, and the value obtained by adding this correction value to the calibration correction value is updated as the calibration correction value by the calibration correction value storage means 5. The calibration correction value comparison means 7 compares the angle correction value of a specific joint of the multi-joint robot among the calibration correction values stored with a preset comparison value,
When the angle correction value is larger than the comparison value, the calibration correction value changing means 6 changes the remaining calibration correction value to perform a series of processes.
Repeat until the calibration correction values converge.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a robot constant calibration calculation method and a calibration calculation device for correcting a setting error which is a difference between a robot constant and an actual mechanism in a robot operated by a robot controller. The present invention relates to a calibration calculation method and a calibration calculation device for obtaining a correct calibration correction value even when the angle origin is installed in a state where it is largely deviated from the designed state.

[0002]

2. Description of the Related Art Generally, a robot has an error in setting a robot constant, which is caused by machining accuracy and assembly accuracy of constituent elements. If there is such a setting error, the tip position obtained by calculation is different from the tip position of the robot actually moving. Therefore, a position shift occurs when the posture is changed, and a position shift causes a shift when the robot is instructed to operate. A robot calibration method for eliminating these operational deviations is disclosed in, for example, Japanese Patent Laid-Open No. 61-1334.
No. 09 publication, a calibration correction value for a setting error of a robot is calculated.

[0003]

However, the setting error of the robot constant is caused not only at the time of manufacturing but also by the contact between the robot and the external environment during the operation of the robot. In such a case, if the motor replacement work or the like is performed due to the contact, it is necessary to perform the calibration correction calculation in the state where there is a large setting error of the angle origin. When the calibration correction value is calculated in the state where there is a large setting error at the origin of the joint angle, the following problems occur.

That is, in the conventional calibration method, the calibration correction value of the robot constant is approximately obtained by the least square method using the proportional relationship assumed between the minute displacement of the robot tip and the minute angle. There is. Therefore, when the existing error is small, there is a high possibility that an optimal solution will be obtained. However, when the calibration correction value is calculated in a state including a large error, there is a high possibility that a local optimum solution existing in the vicinity of the large error will be obtained. In this local optimum solution, a calibration correction value is set even for a portion where an error should not originally exist, and accurate calibration cannot be performed. Therefore, there is a problem that the operation accuracy of the robot cannot be ensured even though the calibration calculation is performed.

The present invention has been made in view of the above circumstances, and it is possible to obtain an optimum calibration correction value even when there is a large setting error in a part of the robot constant such as the angle origin. An object of the present invention is to provide a calibration calculation method and a calibration calculation device.

[0006]

According to another aspect of the present invention, there is provided a robot constant calibrating calculation method, wherein an angle origin data and a link for each joint of a multi-joint robot having a controller having a computer and a memory storing a program. Calibration of robot constant setting errors such as length and tool offset for each joint is performed. Therefore, the position data of the articulated robot is obtained from the angle data of the articulated robot and the calibration robot constant obtained by adding the calibration correction value to the robot constant, and the robot constant is calculated from the angle data of the articulated robot and the position data of the articulated robot. The correction value for the setting error is calculated, and the value obtained by adding this correction value to the calibration correction value is updated and stored as the calibration correction value, and the angle correction value of the specific joint of the articulated robot is preset in the calibration correction value. A series of processes of comparing with the comparison value and changing the remaining calibration correction values when the angle correction value is larger than the comparison value is repeated until the calibration correction values converge.

As described above, according to the method for calculating the calibration of the robot constant, when a calibration correction value for the robot constant is obtained, for example, a joint having a large error may have a large calibration correction value during the calculation. When it is obtained, the calibration correction values other than that joint are changed to, for example, 0 and the calculation is continued to avoid the local optimum solution, and the required optimum calibration correction value is obtained. With such a configuration, even if the joint of the robot has a large angle error, the correct calibration correction value can be determined, and the operation accuracy can be improved by the calibration calculation.

According to a second aspect of the present invention, there is provided a robot constant calibration calculation device for a multi-joint robot including a controller having a computer and a memory storing a program, such as angle origin data for each joint, link length and tool offset. Calibrate the setting error of the robot constant for each joint.
Therefore, the robot angle / position that repeatedly finds the position data of the articulated robot from the robot constant storage means that stores the robot constants, and the calibrated robot constant that is calibrated by adding the calibration correction value to the angle data of the articulated robot A conversion calculation unit, a calibration correction value calculation unit that repeatedly obtains a correction value for a setting error of a calibration robot constant from angle data of an articulated robot and position data obtained by the robot angle / position conversion calculation unit, and a calibration correction value calculation unit. Calibration correction value storage means for repeatedly updating and storing a value obtained by adding the correction value obtained by the means to the calibration correction value, and a calibration correction value storage means that has a comparison value stored in advance and stores this comparison value in the calibration correction value storage means. Calibration correction value comparison means for repeatedly comparing the angle correction value of a predetermined joint among the stored calibration correction values, and When the degree correction value is larger than the comparison value, in response to the output to the calibration correction value comparison means, the calibration correction values stored in the calibration correction value storage means other than the predetermined joint angle correction value are set to predetermined values, for example 0. A calibration correction value changing unit that repeatedly changes, and a calibration correction value obtained by the calibration correction value calculating unit are used to determine the convergence of the calculation, and at the time of convergence, a robot angle / position conversion calculating unit, a calibration correction value calculating unit, and a calibration correction value comparing unit. And calibration correction value convergence determining means for ending the operation of the calibration correction value changing means.

As described above, in this robot constant calibration calculation device, when the calibration correction value for the robot constant is obtained, the calibration correction value is stored in the calibration correction value storage means, and the calibration correction value comparison means is used for calibration. The angle correction value of a specific joint among the correction values is compared with the comparison value, and a large calibration correction value is obtained during calculation for a joint that may include a large error (the angle correction value is If it is larger than the comparison value), the calibration correction value changing means changes the calibration correction values other than that joint to, for example, 0 and continues the calculation, thereby avoiding the local optimum solution and obtaining the necessary optimum calibration correction value. I have to. With such a configuration, even when the robot joint has a large angle error,
The correct calibration correction value can be determined, and the calibration calculation can improve the operation accuracy.

According to a third aspect of the present invention, there is provided a robot constant calibration computing device according to the second aspect of the invention, wherein in the robot constant calibration computing device, a robot constant selection for designating a robot constant element to be compared by the calibration correction value comparing means. Further, means is provided for designating an element of a robot constant for making a comparison judgment of the angle correction values and performing a calibration calculation of the robot constant.

[0011]

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS A robot constant calibration calculation method and a calibration calculation device according to a first embodiment of the present invention will be described below with reference to FIGS. This robot constant calibration calculation method and calibration calculation device can be applied to a multi-joint robot equipped with a controller having a computer and a memory that stores a program, for each joint such as angle origin data for each joint, link length, and tool offset. The calibration calculation method and the calibration calculation device for calibrating the robot constant setting error.

The above "angle origin data for each joint" is
When representing the rotation angle of the motor that constitutes the joint portion of the robot, it is the compensation data of the reading angle for setting a certain reference position to 0 °. The "link length" is a member that constitutes the robot and is the length of a link that directly connects the joint axes. Further, the "tool offset" represents the tip position of the tool attached to the robot tip according to the work as the position as seen from the robot wrist.

FIG. 1 is a block diagram of a robot constant calibration calculation device (device for implementing a calibration calculation method) according to a first embodiment of the present invention. In FIG. 1, 1
Is a robot angle data input means for reading angle data of each joint (each axis) of the robot. Reference numeral 2 is a robot constant storage means for storing angle origin data, link length and tool offset for each joint of the robot without calibration. Reference numeral 3 denotes a robot angle / position conversion calculation means, which uses known robot kinematics, and uses the robot angle input means 1 to input the angle data of each joint of the multi-joint robot and the robot constants stored in the robot constant storage means 2. The position of the robot tip is calculated from and.

Reference numeral 4 denotes a calibration correction value calculation means, which uses a known method such as a least-squares calculation from the robot angle obtained by the robot angle input means 1 and the position data obtained by the robot angle / position conversion means 3. , Calculate the correction value for the error of the robot constant. Reference numeral 5 denotes a calibration correction value storage means for temporarily storing a value obtained by adding the correction value obtained by the calibration correction value calculation means 4 to the previously obtained calibration correction value as the current calibration correction value. Reference numeral 6 denotes a calibration correction value comparison means, which compares, among the calibration correction values obtained by the calibration correction value calculation means 4, the angle correction data for a specific preset joint and the maximum value of the preset correction data. .

Reference numeral 7 denotes a calibration correction value changing means, which is one of the calibration correction values of a plurality of joints stored in the calibration correction value storage means 5.
The calibration correction values other than the angle correction value of the specific joint compared by the calibration correction value comparison means 6 are cleared, for example, changed to 0. 8
Is a calibration correction value convergence determination means, which compares the magnitude of the angle correction value obtained by the calibration correction value calculation means 4 with a set value which is a reference for convergence. It is determined that was obtained.

FIG. 2 is a flow chart showing the flow of calculation processing of the solution in the first embodiment of the robot constant calibration calculation apparatus of the present invention, in which the angular element of the m-th joint is the object of comparison judgment. First, all calibration correction values M of each joint are set to 0 (step 1). The contents of the calibration correction value M are the calibration correction values for all the calibration elements, and the correction values ΔL1 to ΔLn (n: the number of links to be calibrated) for each link length and the correction values Δθ1 to Δθn (n: for the angle origin). The number of angle origins to be calibrated) and the correction values ΔT1 to ΔTn (n: number of tool variables to be calibrated) for the tool portion.

Then, the reference value stored as the robot constant R and the calibration correction value M are added to calculate R + M as the corrected robot constant (step 2). The element of the robot constant R is the same element as the correction value M, and is a design value for each link length L1 to Ln (n: the number of links to be calibrated) and a value for each angle origin (all 0) θ1 to θn (n: calibration The number of angle origins to be used) and design values T1 to Tn (n: number of variables of the tool to be calibrated) regarding the tool portion.

Using the corrected robot constant (M + R), the angle data and the position data calculated by the robot angle / position conversion calculating means 3 are used, and the correction value obtained by one least-squares method is set as K. (Step 3). The calibration correction value M is the sum of all the values of the correction value K obtained by this calculation. Therefore, the K element also corresponds to the M element, and the difference between the correction values for each link length ΔL1 ′ to ΔLn ′ (n: the number of links to be calibrated) and the difference between the correction values for each angle origin Δθ1 ′ to Δθn ′ ( n: the number of angle origins to be calibrated) and the difference between the correction values for the tool portion ΔT1 ′ to ΔTn ′ (n: the number of variables of the tool to be calibrated).

When the calibration correction value M is updated with this correction value K, elements are added item by item to obtain a new calibration correction value M = M + K (step 4). ΔLi = ΔLi + ΔLi ′ (i = 1 to n) Δθi = Δθi + Δθi ′ (i = 1 to n) ΔTi = ΔTi + ΔTi ′ (i = 1 to n) Here, within the calibration correction value M of the joint number m for which the angle is determined. The absolute value of the correction value Δθm is compared with the judgment value V (step 5). V is a value larger than the design value, such as 0.5 °, and is stored in the calibration correction value comparison means 6 in advance.

As a result of the above comparison, when | Δθm | ≧ V, all the correction values other than Δθm in the calibration correction value M are set to 0.
(Step 6) and the calculation by the least squares method is performed again. If | Δθm | <V, all values of the correction value K of the present calculation are compared with the determination criterion value L of the convergence of the least square method (step 7). L is a small value close to 0,
For example, it is set to 0.001 or the like. If all the elements of K are smaller than L, it is determined that the solution has converged, and M is the final calibration correction value (step 8). If there is at least one element larger than L, it is determined that they have not converged, and the calculation by the least squares method is performed again.

The robot constant calibration calculation method will be briefly summarized below. The position data of the multi-joint robot is obtained from the angle data of the multi-joint robot and the calibration robot constant obtained by adding the calibration correction value to the robot constant, and the multi-joint robot is calculated. The correction value for the setting error of the robot constant is calculated from the robot angle data and the position data of the articulated robot, and the value obtained by adding this correction value to the calibration correction value is updated and stored as the calibration correction value. The calibration correction value converges through a series of processes for comparing the angle correction value of a specific joint of the multi-joint robot with a preset comparison value, and changing the remaining calibration correction value when the angle correction value is larger than the comparison value. It will be repeated until you do.

According to this embodiment, when a calibration correction value for a robot constant is obtained, a large calibration correction value is calculated during calculation for a joint that may contain a large error by the calibration correction value changing means 7. Then, the calibration correction values other than that joint are changed to 0, for example, and the calculation is continued to avoid the local optimum solution, and the required optimum calibration correction value is obtained. With such a configuration, even if the joint of the robot has a large angle error, the correct calibration correction value can be determined, and the operation accuracy can be improved by the calibration calculation.

A robot constant calibration calculation apparatus according to the second embodiment of the present invention will be described below with reference to FIGS. FIG. 3 is a block diagram of a robot constant calibration computing device (device for implementing the calibration computing method) according to the second embodiment of the present invention. In FIG. 3, reference numeral 1 is a robot angle data input means for reading angle data of each joint (each axis) of the robot. Reference numeral 2 is a robot constant storage means for storing angle origin data, link length and tool offset for each joint of the robot without calibration. Reference numeral 3 denotes a robot angle / position conversion calculation means, which uses known robot kinematics, and uses the robot angle input means 1 to input the angle data of each joint of the multi-joint robot and the robot constants stored in the robot constant storage means 2. The position of the robot tip is calculated from and.

Reference numeral 4 denotes a calibration correction value calculation means, which uses a known method such as least squares calculation from the robot angle obtained by the robot angle input means 1 and the position data obtained by the robot angle / position conversion means 3. , Calculate the correction value for the error of the robot constant. Reference numeral 5 denotes a calibration correction value storage means for temporarily storing a value obtained by adding the correction value obtained by the calibration correction value calculation means 4 to the previously obtained calibration correction value as the current calibration correction value. Reference numeral 6 denotes a calibration correction value comparison means for comparing the correction data of the angle with respect to a specific preset joint number and the maximum value of the preset correction data among the calibration correction values obtained by the calibration correction value calculation means 4. Compare.

Reference numeral 7 denotes a calibration correction value changing means, which is one of the calibration correction values of a plurality of joints stored in the calibration correction value storage means 5.
All the correction values not selected by the robot constant selection means 9 described later are changed to 0. Reference numeral 8 denotes a calibration correction value convergence determination means, which compares the magnitude of the angle correction value obtained by the calibration correction value calculation means 4 with a set value which is a reference for convergence, and when the value is less than the set value, the solution converges, and finally Judge that a solution has been obtained. Reference numeral 9 denotes a robot constant selection means, which has, for example, a storage area corresponding to the number of joints to be calibrated, and sets "1" to the area corresponding to the joints to be compared and judged, and "0" when not. By doing so, the numerical value is checked at the time of calculation and the necessary joint number is selected.

FIG. 4 is a flow chart showing the flow of calculation processing of the solution in the second embodiment of the robot constant calibration calculation device of the present invention. First, all the calibration correction values M are set to 0 (step 1), and a value is set in the selection area N in the robot constant selection means 9 corresponding to the joint number for determining the angle correction value (step 2). When 6 joints are targeted as joints, for example, 6 areas are secured as N as shown in [Table 1] below.

[0027]

[Table 1]

When the joints 4 and 5 are to be compared and judged in [Table 1], as shown in [Table 2], the joint 4
The target to be selected is designated by setting the portion corresponding to 5 and 5 to "1" and the other portions to "0".

[0029]

[Table 2]

Here, the contents of the robot constant R and
The contents of the calibration correction values M and K are the same as in the first embodiment. First, similarly to the first embodiment, the reference value stored as the robot constant R and the calibration correction value M are added to calculate R + M as the corrected robot constant (step 3). Then, using the corrected robot constant (M + R), the correction value obtained by one least-squares method is set to K (step 4). With this correction value K, the calibration correction value M is updated with a new calibration correction value M = M + K (step 5).

Here, the number m of the joint to be judged is stored as a set numerical value in the selection area N at the start of processing as described above. Therefore, selection is made by calling the contents of the area numbers 1 to 6 of the storage area N and checking the number in which 1 is stored (step 6). In the above example, it can be seen that two joint numbers 4 and 5 are selected. Therefore, the number of selections n = 2, and each joint number selected as il = 4 and i2 = 5 is shown.

Here, the selected n Δθij and the judgment value V stored in the calibration value comparison means 6 are compared with each other, and | Δθij | ≧ V of the selected n angle correction values.
If there is even one, the calibration correction value M
Δθi selected by the robot constant selection means 9 among
All the correction values other than j (j = 1 to n) are set to 0 (step 7). In the above example, the calibration correction value M other than Δθ4 and Δθ5 is set to zero. All selected Δθij (j =
1 to n), if | Δθij | <V, all values of the correction value K of this time are set to the criterion value L for the convergence of the least squares method.
(Step 8). L is a value close to 0 and is set to, for example, 0.001. All K elements are L
If smaller, it is determined that the solution has converged, and M is the final calibration correction value (step 9). If there is even one element larger than L, it is determined that the element has not converged, and the calculation by the least square method is performed again using M of this time.

According to the calibration calculation device of this embodiment,
When the angle correction value of any one or more joints of the multi-joint robot exceeds the maximum correction value, the remaining calibration correction value can be set to zero. Even when there is a large angle error in a plurality of joints, the correct calibration correction value can be determined, and the operation accuracy can be improved by the calibration calculation.

FIG. 5 is a front view showing an example of the overall configuration of a robot to which the method for calculating and calibrating the robot constant of the present invention is applied. In FIG. 5, reference numeral 10 is a robot controller, and 11 is a teaching pendant, which directs an operation to the robot, displays position data and angle data, and inputs position data and angle data. Reference numeral 12 denotes a robot body, which performs actual work. A robot tool 13 is, for example, a welding torch attached to the tip of the robot body 12.

[0035]

According to the method for calculating and calibrating the robot constants of claim 1, the necessary calibration correction values are selected by repeating the calculation after clearing the correction values other than the set robot constants. Therefore, it is possible to obtain a calibration correction value that corrects the error correctly, and it is possible to calibrate the robot mechanism even when there is a large setting error, and it is possible to achieve an accurate operation, which is an excellent effect. Is.

According to another aspect of the present invention, there is provided a calibration correction value storage means, a calibration correction value comparison means, and a calibration correction value changing means to clear correction values other than the set robot constant. It is possible to select the required calibration correction value by repeating the above calculation and obtain the calibration correction value that corrects the error correctly, and calibrate the robot mechanism even when there is a large setting error. It is possible to achieve this, and it is possible to achieve an excellent effect of realizing a highly accurate motion of the robot.

According to the third aspect of the present invention, the robot constant selecting device further includes the robot constant selecting means, so that the angle correction value of any one or more joints of the multi-joint robot exceeds the maximum correction value. In this case, the robot mechanism can be calibrated even when there is a large angle error in a plurality of joints, and an excellent effect of realizing a highly accurate motion of the robot is achieved.

[Brief description of drawings]

FIG. 1 is a diagram showing a first method of calibrating and calculating a robot constant according to the present invention.
3 is a block diagram showing the configuration of a calibration calculation device for carrying out the embodiment of FIG.

FIG. 2 is a diagram showing a first method of calibrating and calculating a robot constant according to the present invention.
3 is a flowchart of a solution calculation process in the embodiment.

FIG. 3 is a second method of calibrating and calculating a robot constant according to the present invention.
3 is a block diagram showing the configuration of a calibration calculation device for carrying out the embodiment of FIG.

FIG. 4 is a second method of calibrating and calculating a robot constant according to the present invention.
3 is a flowchart of a solution calculation process in the embodiment.

FIG. 5 is a front view showing a configuration example of a robot to which the robot constant calibration calculation method of the present invention is applied.

[Explanation of symbols]

 1 robot angle data input means 2 robot constant storage means 3 robot angle / position conversion calculation means 4 calibration correction value calculation means 5 calibration correction value storage means 6 calibration correction value comparison means 7 calibration correction value change means 8 calibration calculation correction value convergence judgment Means 9 Robot constant selection means 10 Robot controller 11 Teaching pendant 12 Robot body 13 Welding torch

Claims (3)

[Claims] 1. A multi-joint robot equipped with a controller having a computer and a memory storing a program, the calibration of an error in setting a robot constant for each joint, such as angle origin data for each joint, link length, and tool offset. A method for calibrating robot constants, wherein position data of the multi-joint robot is obtained from angle data of the multi-joint robot and a calibration robot constant obtained by adding a calibration correction value to the robot constant, A correction value for the setting error of the robot constant is obtained from the angle data and the position data of the articulated robot, and a value obtained by adding the correction value to the calibration correction value is updated and stored as the calibration correction value, and the calibration correction value is stored. The angle correction value of a specific joint of the articulated robot is compared with a preset comparison value in the A series of processing, calibration operation method of a robot constant and repeating until the calibration correction value converges to change the remaining calibration correction value when the value is greater than the comparison value. 2. A multi-joint robot equipped with a controller having a computer and a memory storing a program, and calibration of an error of robot constant setting for each joint such as angle origin data for each joint, link length and tool offset. A robot constant calibration calculation device for performing, wherein robot constant storage means for storing the robot constant,
Robot angle / position conversion calculating means for repeatedly obtaining position data of the articulated robot from angle data of the articulated robot and a calibration robot constant calibrated by adding a calibration correction value to the robot constant, and an angle of the articulated robot Calibration correction value calculation means for repeatedly obtaining a correction value for the setting error of the calibration robot constant from the data and the position data obtained by the robot angle / position conversion calculation means, and the correction value obtained by the calibration correction value calculation means. Is added to the calibration correction value, the calibration correction value storage means for repeatedly updating and storing the calibration correction value as the calibration correction value, and the comparison value stored in advance in the calibration correction value storage means. Calibration correction value comparison means for repeatedly comparing the angle correction value of a predetermined joint among the calibration correction values, and the angle correction value of the predetermined joint Repeated to a predetermined value other than the angle correction value of the predetermined joint among the calibration correction values stored in the calibration correction value storage means in response to the output to the calibration correction value comparison means when the comparison value is larger than the comparison value. The calibration correction value changing means to be changed and the calibration correction value obtained by the calibration correction value calculating means are used to determine the convergence of the calculation, and at the time of convergence, the robot angle / position conversion calculating means, the calibration correction value calculating means, the calibration A robot constant calibration operation device comprising a correction value comparison means and a calibration correction value convergence determination means for ending the operations of the calibration correction value changing means. 3. A robot constant selection means for designating an element of a robot constant to be compared in the calibration correction value comparison means, wherein a robot constant element for making a comparison judgment of the angle correction value is designated to make a robot. 3. The robot constant calibration calculation device according to claim 2, wherein a constant calibration calculation is performed.