JPS5959394A - Device for estimating error of robot mechanism - Google Patents

Abstract

(57) [Summary] This bulletin contains application data before electronic filing, so abstract data is not recorded.

Description

DETAILED DESCRIPTION OF THE INVENTION The present invention relates to an apparatus for estimating mechanical errors for correction, which is suitable for correcting positioning errors of general equipment such as robots, which include mechanical parameters in a positioning control method.

A conventionally known example is G: '1', written by Susumu Yojo and one other person from Fujitsu Laboratories Ltd., 1985, Spring Conference Academic Lecture Collection, Volume 2, pages 587 to 589. There is a method for calibrating the coordinate system of an articulated robot, which was published in In this known example, the robot's mechanical parameters σ, arm length θ]y, and the difference between the two are estimated as follows. However, the mechanism parameters include the length of the vertical axis connecting the arms, the torsion angle (the angle at which the arms and the vertical axis are attached), and in known examples, the errors in these parameters are estimated. G" I haven't done it. Therefore, it has the disadvantage that +i cannot be applied to robots that include errors in the vertical axis or twist angle.

The purpose of the present invention is to estimate the amount of error in a robot mechanism, when errors are included in the arm length, the length of the vertical axis connecting the arms, and the torsion angle among the parameters constituting the mechanism. The purpose is to provide equipment.

In order to achieve the above object, an apparatus for estimating mechanical error of a robot according to the present invention includes a robot main body, a control device, and a robot position measuring device, which further includes a mechanical error calculation device, and a robot mechanical error estimation device according to the present invention. The amount of deviation between the position, position and direction commanded by the robot's control device and the position, position and direction in which the robot actually moved as measured by the position measuring device is measured, and the independence of the robot is determined. When the number of parameters is YM, the error of the parameters independent of the above deviation is 1
/Is the above measurement performed as many times as necessary to obtain an equation with a value of 3 (M-1) or more expressed by the IK formula? The key point is to have a function to repeatedly calculate the error value of mechanical parameters from design values.

The position and posture of the robot are determined once the mechanical parameters of the C40 bond and the rotation angles of each rotation axis are determined. Therefore, even if the same rotation angle is specified, the position and orientation will be different if the mechanism parameters are different. Now, the design values of mechanism parameters are: arm length S, vertical axis length A1, torsion angle α
, and the amount of error from those of the actual robot is ΔS
, ΔA, and Δα. Furthermore, the vector that collectively represents the commanded position and orientation of the robot is set to X, and the amount of deviation when the robot actually moves is not shown as ΔX. Here, → is a symbol representing a vector. X is S,
A.

Since it is a function of a vector P that is a collection of structural parameters such as α and a vector θ that is a collection of rotation angles of the shaft, it can be expressed as X=f(θ, P) (1). The functional form of f is unique for each robot and needs to be created each time.

Here, the elements of θ and P are determined by the mechanism and the degree of freedom of the robot. Now, assuming that the mechanical parameters include an error ΔP, the actual position and orientation of the robot is f(θ, P+ΔP) (2) in X. Regarding this equation P, a linear approximation T,6 is obtained. Here ΔX = X' −X (4)
Therefore, it is not within the range of linear approximation regarding P.

δX/δP1 can be calculated for each robot, and is also the deviation amount ΔX between the robot's command value and the actual movement value.
can be measured, equation (5) is a system of simultaneous equations in which the error amount ΔP of the mechanism parameter is used as a variable. Now, assuming that the number of parameters of an independent mechanism is M, in order to obtain M ΔP1 from equation (5), it is necessary to use M equations created from equation (5). For example, in a five-dimensional space, ma has six independent components of coordinate components and three direction cosines, so by measuring Δ at one point, we can obtain 6 from equation (5).
linear equations are obtained.

In addition, ■(In the case of an axis robot, the number of independent mechanical parameters is generally 8. Considering the six types of A and α, 3(K-1
), so for example, in the case of a 6-axis robot, there are 6 or M-
It is 15. In this case, therefore, 15 mechanical parameters have to be determined, 1 which requires position and orientation measurements at at least three points. Then, it is determined from three points.

By using 15 independent equations among the 018 equations, 15 ΔP1 can be found.

Also, when measuring one point, if the absolute coordinate value x +
If only Y*z were to be measured, measurements would have to be made at five points. However, it has the advantage that the direction cosine does not have to be measured.

In this way, for general equipment where the position to be moved is expressed as a linear equation of the mechanism/ζ parameter, by measuring the amount of deviation, a linear approximation of the mechanism error parameter of the correction angle can be obtained. , please refer to the attached diagram for examples. Although the present invention will be explained in more detail using the present invention, these are merely examples, and it goes without saying that various improvements and modifications may be made without going beyond the scope of the present invention. 1 is a diagram of an applied structure, which includes a robot 1, a control device 2, a robot position measurement device 3, and a mechanical error calculation device 4. The robot 1 moves according to command data stored in the control device 2. The position and posture of the robot are measured using a 6-bit position measuring device. The measured position and orientation data are sent to the mechanism error calculation device 4 to calculate the amount of mechanism error. According to this embodiment, the mechanical error amount included in the robot can be estimated in the following manner.

FIG. 2 is a flow sheet showing the flow of calculating the mechanical error amount.

(1) In block 5, six different command values are prepared.

(2) In block B, take out one command value,
Find each joint angle and move.

(b) Block 7 measures the amount of deviation between the moving point and the command value, and sets it as ΔX; (i=1 to 3).

(4) Block 6.7g Repeat 3 times.

(5) In block 9, the mechanism error amount ΔP, (j-1 to 15) is determined using the mechanism error amount calculation device 4.

With the above steps, the estimation of the mechanism error is completed. Next, when actually operating the robot, ΔP calculated using equation (5)
Using j, substitute into , find Δma, add it to the command value ma, and add it to XN=X+
All you have to do is issue a command to move to XN as ΔX.

According to the present invention, the amount of mechanical error included in the robot can be estimated, resulting in the following effects.

(1) By substituting into the mechanical error formula M, the amount of positional and orientation deviation at any given position and orientation can be determined.

(2) By correcting this amount of deviation to the position and orientation determined from the design values, a robot with high absolute position accuracy can be realized.

[Brief explanation of the drawing]

FIG. 1 is a perspective view of a robot structure equipped with a robot mechanical error estimation device according to the present invention, and FIG. 2 is a sheet 70 showing the flow of calculations performed in the device shown in FIG. . DESCRIPTION OF SYMBOLS 1... Robot, 2... Robot control device, 6... Position measuring device, 4... Mechanical error amount calculation device. Representative Patent Attorney Susuki 1) Yuki Tori - 1st review! No. ? Figure incident table Jri Showa 57 Il"!, '+-H (≦1i (16th
8.15 Title of the invention Device for estimating mechanical errors in robots
Works 1 ・ ・＜ Knee (++
Contents of the amendment by Shigeyo Katsu (1) The entire specification has been amended as shown in the attached document. Description 1, Title of the invention Robot mechanical error θ) estimation device 2,
Claims: A structure consisting of a robot main body, a control device, and a position measuring position of the robot, and further includes a mechanism error calculation device, which allows the robot to move during operation and calculates the position commanded by the robot control device. Alternatively, when measuring the amount of deviation between the position and direction and the position or position and direction σ) where the robot actually moved as measured by a fixed device, and setting the harameter V to M at the independent 1 of the robot, Express the above deviation amount as a linear equation with respect to the independent parameter σ) error. Obtain the Fuller equation on the 1'4'-M individual clothes. Repeat the measurements as many times as necessary for σ), and calculate from the design value. An apparatus for estimating a mechanical error of a robot, characterized by having a function of calculating an error of a mechanical parameter σ. DETAILED DESCRIPTION OF THE INVENTION The present invention relates to an apparatus for estimating a mechanism error for correction, which is suitable for positioning error correction of general equipment, such as a robot σ), in which a positioning control formula 77 includes a mechanism parameter. As a conventionally known example, Susumu Yojo and one other person from Fujitsu Laboratories Ltd. published the 1985 Spring Conference of the Japan Society of Precision Machinery Engineers, Volume 2, pages 587 to 589. There is a "coordinate system σ) calibration method for articulated robots". In this known example, the amount of error in the arm length among the mechanical parameters σ) of the robot is estimated. However, in addition to the arm length, the mechanism parameters include whether the arms are attached at an angle, and in the known example, the attachment is an angle parameter σ) error estimation is not performed. Therefore, the disadvantage was that the trick could only be applied to robots that encounter errors in corners. The object of the present invention σ) is that in a robot y'a structure, among the parameters constituting the mechanism, arm θ) length and arm-to-arm attachment include errors in the angles. The idea is to provide a device to estimate the amount. In order to achieve the above object, the device for estimating the mechanical error of the mouth pot according to the present invention is composed of a robot body, a control device, a position determining device for the robot, and a mechanical error calculation device. When the robot is actually operated, the position or position and direction commanded by the control device and the position or position and direction actually moved by the robot measured by the position aiming device are determined. Measure the amount of deviation between M
Then, the above deviation amount is expressed by a linear equation with respect to the independent parameter σ) error, and the above deviation is repeated as many times as necessary to obtain the equation, and the mechanism parameter 0) error from the design value is The gist is that it has the ability to calculate . The position and posture of Robonote are determined once the mechanical parameters of the robot and the rotation angles of each rotation axis are determined. Therefore, even if the same rotation angle is specified, the position and orientation will be different if the mechanism parameters are different. Now, the design value of the mechanism parameter σ) is the arm length S, and the x, y, and z components of the angle α for arm-to-arm attachment are ΔSx, ΔSy, and ΔSz.
and ΔαX, Δαy, Δα2. In addition, let the vector that collectively represents the commanded position and orientation of the robot be σ)
The amount of deviation is expressed as ΔX. In this N, → is a symbol representing a vector. Since X is a function of the vector P that summarizes the mechanical parameters such as S and α, and the vector θ that summarizes the rotation angle of the shaft, it can be expressed as 1-. The functional form of f is unique for each robot and needs to be created each time. Determined by my friend. Now, if the mechanical parameters include the error ΔP, then the actual position and posture of the robot are as follows. When this equation is linearly approximated with respect to P, it becomes. Here, since 'f' is within the range of linear approximation regarding P. aX/δP1 can be calculated for each robot, and is also the deviation amount Δ between the robot's command value, actual movement value, and σ).
Since X is measurable, equation (5) is expressed as a simultaneous-order equation in which the error amount ΔP of the mechanism parameter is a 4' number. now,
If the number of independent mechanism parameters is M and f, then from equation (5), N
In order to find ΔPi, it is necessary to use M equations created from L equation (5). For example, in three-dimensional space,
has 6 coordinate components and 6 independent components of 3 direction cosines, so by measuring ΔX at -1 point, we can obtain 6 from equation (5).
One linear equation can be obtained. In addition, in the case of a K-axis robot, the error amount of independent mechanical parameters is generally ΔSx,
ΔSy, ΔSz, ΔαXΔα)・, Δα2, 6 and 6, so for example, in the case of a 6-axis robot, M-36
It is. Therefore, in this case, it is necessary to determine the error amount of 36 mechanical parameters, which requires measuring the position and direction at at least six points. Then, using the six equations that can be found from the six points, 56 ΔPi can be found. Also, if in the measurement of one point, the absolute coordinates θ) x, y
, z, it would take 1f to measure 12 points in a row. However, there is an eleventh point that it is not necessary to measure the direction cosine. For general equipment, where the position to be moved is expressed as a function of a mechanical parameter, by measuring the amount of deviation, a first-order approximation of the mechanical error parameter for correction can be obtained. Hereinafter, the present invention will be explained in more detail using examples with reference to the accompanying drawings, but they are merely illustrative, and various improvements and modifications can be made without going beyond the scope of the present invention. Of course, this is possible. Fig. 1 is a diagram of a structure to which the present invention is applied, and the structure includes a robot 1, a control device 2, a robot position measuring device, and a mechanical error calculation device. The robot 1 moves according to the command data stored in the control position 2. The position and orientation of the robot are measured using the position measuring device 3. The measured position and orientation data are used to calculate mechanical errors. Send it to device 4 and calculate the amount of mechanical error 1. According to this embodiment, the amount of mechanical error included in the robot can be estimated as follows. Figure 2 shows the calculation of the amount of mechanical error of a 6-axis robot. This is a flow sheet showing the flow. (1) In block 5, prepare 6 different command values. (2) In block 5, take out 1 command value,
Find each joint angle and move. (3) In block 7, the amount of deviation between the moving point and the command value is measured, and ΔX1 (l=1 to 6) and f are calculated. (4) Repeat block 6.7 six times. (5) In block 9, use the mechanism error calculation device 4 to find the mechanism error amount ΔPj (j=1 to 36). With the above steps, the estimation of the mechanism error is completed. Next, if you want the robot to operate according to the command value, use formula (5)
It is sufficient to issue a command to correct the command value X using ΔPj obtained in and move to XN. According to the present invention, the amount of mechanical error included in the robot can be estimated, so the following effects can be achieved. (1) By substituting the mechanical error amount into equation (5),
The amount of positional and orientation deviation can be determined at any position and orientation. (2) By correcting the machine error amount to the commanded position and orientation, a high degree of absolute position accuracy can be achieved. 4. Brief description of the drawings Fig. 1 shows a diagram of a robot structure equipped with a mechanical error σ) estimating device according to the present invention;
This is a flow route showing the flow of calculations σ) carried out in the apparatus shown in the figure. 1 Robonote, 2...Robonote control device, position measurement g-j mechanism error calculation device. Representative Patent Attorney Susuki 1) Toshiyuki

Claims (1)

[Claims] The structure consisting of the robot body, a control device, and a robot position measuring device further has a mechanism calculation device Rw, which calculates the position or position IA and direction commanded by the robot control device when the robot actually operates. , the actual position G of the robot as measured by the position measuring device
When the number of independent parameters of the robot is M, the error of the independent parameters is expressed by a linear equation 3.
A function that repeats the above measurement as many times as necessary to obtain (M-1) or more equations and calculates the error of the mechanical parameters from the design values? A device for estimating a mechanical error of a robot, which is characterized by the ability to be prepared.