JPH052168B2 - - Google Patents

Description

DETAILED DESCRIPTION OF THE INVENTION A. Field of Industrial Application The present invention relates to a method for correcting the accuracy of a three-dimensional measuring robot (three-dimensional measuring device) that measures the three-dimensional shape of, for example, automobile parts or car bodies. The present invention relates to an accuracy correction method for a three-dimensional measurement robot in which a three-axis orthogonal cantilever-type measuring mechanism arm is provided with a multi-joint arm having two rotating axes and two bending axes and having a contact at the tip.

B. Prior Art Traditionally, gate-type or cantilever-type three-dimensional measuring devices with orthogonal three axes have been used to measure three-dimensional (three-dimensional shapes). The features of these two types of measuring devices are that the former basically measures from top to bottom, and the latter basically measures from the side, and has a wider measurable range than the former. Although it is widely used and highly effective, it is not suitable for measuring complex shapes.

Therefore, in recent years, a multi-jointed arm with two rotating axes and two bending axes has been installed on the arm of a cantilever type three-axis orthogonal measuring machine, and a contactor is provided at the tip of this multi-jointed arm, making it possible to measure complex shapes. 3D measurement robots are becoming a reality.

C. Problems to be Solved by the Invention However, since the above-mentioned improved three-dimensional measurement robot is equipped with a multi-joint arm, there are many movable parts, which inevitably leads to the possibility of large measurement errors. However, there was a problem in that it could not be put to practical use with very high accuracy unless appropriate correction means were taken. Conventionally, there has been no suitable correction means for this, and for this reason, an improved three-dimensional measurement robot equipped with a multi-joint arm has not been put into practical use.

The present invention was made in view of the current situation, and includes an improved three-dimensional measurement robot.
The purpose of this invention is to provide a method for correcting measurement errors caused by such structural factors and improving measurement accuracy.

D. Means for Solving the Problems In order to achieve the above object, a first aspect of the present invention provides a first aspect of the present invention, in which a first The object to be measured is measured by bringing a contactor provided at the tip of a multi-joint arm connected in this order to a first rotation axis, a first bending axis, a second rotation axis, and a second bending axis. In a measurement accuracy correction method for a three-dimensional measurement robot, the contactor is applied to arbitrary four points on the spherical surface of a reference jig installed at a predetermined position in a predetermined reference posture of the multi-jointed arm. A first step of determining the three-dimensional coordinate values of each of these points by making them contact, the three-dimensional coordinate values of each of the points obtained in the first step, and the equation of a sphere passing through each of the points. Therefore, a second step of determining the three-dimensional coordinate value and radius value indicating the center position of the sphere, the value of the radius of the sphere determined in the second step, and the radius of the sphere held in advance. When the error value between the value of a third step in which the second rotation axis and the second bending axis are held in predetermined postures, and the first rotation axis and the first bending axis are held in a predetermined posture; a fourth step of determining three-dimensional coordinate values of each of these points by bringing the contact into contact with four arbitrary points on the surface of the sphere while changing the rotation angle and the bending angle; a fifth step of determining a three-dimensional coordinate value indicating the center position of the sphere based on the three-dimensional coordinate value of each point obtained in the fourth step and an equation of a sphere passing through each point; a three-dimensional coordinate value indicating the center position of the sphere obtained in the fifth step;
A first correction is made by storing the difference between the three-dimensional coordinate value indicating the center position of the sphere held in the third step as a correction value on a matrix having the rotation angle and the bending angle as variables. In the sixth step of creating a matrix, the first rotation axis and the first bending axis are held in predetermined postures, and the rotation angle and the second bending axis are adjusted, respectively. a seventh step of determining three-dimensional coordinate values of each of these points by bringing the contact into contact with four arbitrary points on the surface of the sphere while changing the bending angle; an eighth step of determining a three-dimensional coordinate value indicating the center position of the sphere based on the three-dimensional coordinate value of each point obtained in the step and an equation of a sphere passing through each point; The rotation angle is determined by using the difference between the three-dimensional coordinate value indicating the center position of the sphere obtained in the step and the three-dimensional coordinate value indicating the center position of the sphere held in the third step as a correction value. and a ninth step of creating a second correction matrix by storing the bending angle on a matrix as variables; a tenth step of teaching and automatically measuring the measurement points based on this teaching to obtain a measured value; and a posture of the multi-jointed arm when obtaining the measured value in the tenth step. an eleventh step of retrieving approximate correction values from the first and second correction matrices, respectively, according to the
and coordinate transformation of either the first or second correction matrix to enable synthesis of the two correction values retrieved in the eleventh step, and and a twelfth step of correcting the measured value obtained in the tenth step by a composite correction value of the two correction values.

Further, in order to achieve the above object, a second aspect of the present invention provides a first rotary shaft, a first Measurement of a three-dimensional measurement robot that measures the object to be measured by bringing into contact a contactor provided at the tip of a multi-jointed arm that is connected in this order to a bending axis, a second rotation axis, and a second bending axis. In the accuracy correction method, in a predetermined reference posture of the multi-jointed arm, the contacts are brought into contact with arbitrary four points on the spherical surface of a reference jig installed at a predetermined position. a first step of determining the three-dimensional coordinates of each point; and a center of the sphere based on the three-dimensional coordinates of each point obtained in the first step and the equation of a sphere passing through each point. The second step is to obtain the three-dimensional coordinate value and radius value indicating the position.
When the error value between the value of the radius of the sphere obtained in the second process and the value of the radius of the sphere held in advance is within a predetermined tolerance range, the second process is performed. a third step of holding the three-dimensional coordinate values obtained in step 2 as error-free three-dimensional coordinate values indicating the center position of the sphere in the reference posture of the multi-jointed arm; With respect to the second bending axis, any four points on the surface of the sphere are held in a predetermined posture, and with respect to the first rotation axis and the first bending axis, the rotation angle and the bending angle are changed, respectively. a fourth step of determining the three-dimensional coordinate values of each of these points by bringing the contactor into contact with the object; a fifth step of determining a three-dimensional coordinate value indicating the center position of the sphere based on an equation of a sphere passing through the point; a three-dimensional coordinate value indicating the center position of the sphere determined in the fifth step; A first correction is made by storing the difference between the three-dimensional coordinate value indicating the center position of the sphere held in the third step as a correction value on a matrix having the rotation angle and the bending angle as variables. In the sixth step of creating a matrix, the first rotation axis and the first bending axis are held in predetermined postures, and the rotation angle and the second bending axis are adjusted, respectively. a seventh step of determining three-dimensional coordinate values of each of these points by bringing the contact into contact with four arbitrary points on the surface of the sphere while changing the bending angle; an eighth step of determining a three-dimensional coordinate value indicating the center position of the sphere based on the three-dimensional coordinate value of each point obtained in the step and an equation of a sphere passing through each point; The rotation angle is determined by using the difference between the three-dimensional coordinate value indicating the center position of the sphere obtained in the step and the three-dimensional coordinate value indicating the center position of the sphere held in the third step as a correction value. and a ninth step of creating a second correction matrix by storing the bending angle in a matrix as variables, and a control device that controls the three-dimensional measuring robot, which includes information necessary for controlling the three-dimensional measuring robot. a tenth input device for inputting data indicating the posture of the multi-jointed arm at the measurement point of the object and data indicating the path leading to the posture;
and a step of retrieving approximate correction values from the first and second correction matrices, respectively, according to the posture of the multi-joint arm at the measurement point related to the tenth step.
In step 11, one of the first and second correction matrices is subjected to coordinate transformation to enable synthesis of the two correction values retrieved in the eleventh step, and the coordinate transformation a 12th step of correcting the data input to the control device in the 10th step by a composite correction value of the two correction values; Part 13: Automatically measure points to obtain measured values
The structure consists of the following steps:

E. Function The first invention operates a three-dimensional measurement robot using a teaching playback method. The function is described below. First and second correction matrices are created by calculating the correction values for each posture with respect to the basic posture of the multi-jointed arm using a method of determining the center coordinates of the sphere as a reference, and these are stored. The method is stored in the device, and the three-dimensional measurement robot is taught the measurement point of the object to be measured and the route to the measurement point, and based on this teaching, the measurement point is automatically measured and the measurement value is obtained.
If the measured values are corrected using the stored first and second correction matrices, highly accurate measurement becomes possible.

Next, the second invention operates a three-dimensional measurement robot using a database method,
Similarly to the first invention, the first and second correction matrices are created in advance and stored in the storage device, and the control device determines the posture of the multi-jointed arm at the measurement point of the object to be measured and the resulting posture. By inputting data such as the route, correcting this data using the stored first and second correction matrices, and performing automatic measurement using this corrected data, high accuracy can be achieved. measurement becomes possible.

In this way, in the present invention, even if the three-dimensional measuring machine is equipped with a multi-joint arm that tends to amplify measurement errors, it is possible to cover the movable range of the multi-joint arm and store correction values for each posture in advance. Then, in the teaching playback method, the specific stored correction value is extracted from the actual measured value, and in the database method, the stored specific correction value is applied to the input data to the control device, and the correction process is performed by the amount of this correction value. This enables highly accurate measurement.

F. Embodiments Hereinafter, embodiments of the present invention will be described with reference to the drawings.

First of all, FIGS. 1 and 2 illustrate an improved three-dimensional measuring robot according to the present invention, where 1 is a measuring device, 2 is a rail (or a groove), and 3 is this robot. A base that moves along the rail 2 (in the z-axis direction), 4 a column that stands vertically on the base 3, 5 a head that moves along the column 4 (in the y-axis direction), and 6 a rail. 2 and column 4 in the horizontal direction (x-axis direction), and slides in the head 5 in this axial direction. The configuration up to this point is the same as the conventional one, and the conventional one had a contact at the tip of the arm 6.

Measuring device 1 for a three-dimensional measuring robot according to the present invention
In addition to the basic configuration of a cantilever type with three orthogonal axes, the arm 6 further includes a multi-jointed arm 7 at the tip, and a contactor is provided at the tip of the multi-jointed arm 7. That is, in the multi-jointed arm 7, 8 is a first rotating shaft that can rotate coaxially with the arm shaft 6, and 9 is rotatably supported at the tip of the first rotating shaft 8. A first bending axis rotatable on a plane perpendicular to the central axis of the 9, 10 a second rotation axis coaxially rotatable with respect to the first bending axis 9;
is pivotally supported on the tip of the second rotating shaft 10, and the first rotating shaft 1
A second bending shaft 12 is a contact provided at the tip of the second bending shaft 11, which is rotatable on a plane perpendicular to the central axis of the second bending shaft 11.

Note that each part of the multi-jointed arm 7 is not necessarily rotatable within a 360° range. first rotating shaft 8,
The first bending axis 9, the second rotation axis 10, and the second rotation axis 1
If the angles of 1 are denoted by θ ₁ , θ ₂ , θ ₃ , and θ ₄ ,
−90°≦θ ₁ ≦90°, −190°≦θ ₂ ≦10°, −90°≦θ ₃
≦90°,
-100°≦θ ₄ ≦100° (However, when the state shown in Figures 1 and 2 is the origin, that is, the reference position.)
It is.

Further, 13 is a control device consisting of a computer for measuring the three-dimensional measuring robot.

In FIG. 1, reference numeral 14 indicates a surface plate on which the object to be measured is placed, and reference jig 15 includes a perfect circular sphere 16.

A feature of the present invention is that the error time (this is used as a correction value) at each posture in the movable range of each joint part of the multi-joint arm 7 is stored in advance, and these correction values are used during measurement. This is to correct the value. Therefore, in the present invention, what is called a correction matrix is created.

There are two correction matrices, a first and a second, and the first correction matrix is for the first rotation axis 8.
and the first bending axis 9, and the second correction matrix corresponds to the second rotation axis 10 and the second bending axis 11. This will be explained below.

First, in creating a correction matrix, the origin (reference) posture of the multi-joint arm 7 is determined. This reference posture should be maintained as much as possible with the multi-joint arm 7.
Considering the structure of the sensor, it is preferable to use a posture that does not easily cause measurement errors, and for example, the postures shown in FIGS. 1 and 2 correspond to this. In this case, θ ₁ = θ ₂ =
θ ₃ =θ ₄ =0°.

Note that the reference posture is not necessarily limited to this, and may be any other posture.

In this reference posture, the perfect circular sphere 16 of the reference jig 15 placed at a fixed position on the surface plate 14
Contactor 1 provided at the tip of multi-joint arm 7 on the surface of
2 are brought into contact with each other to find the center coordinates of this sphere 16.

The center coordinates of the sphere 16 can be easily determined from the coordinates of four points on the sphere. The formula below has 4 points (x ₁ ,
y ₁ , z ₁ ), (x ₂ , y ₂ , z ₂ ), (x ₃ , y ₃ , z ₃ ), (x ₄ , y ₄
，
This is a general solution using the determinant of the equation of the sphere passing through z ₄ ).

When this determinant is expanded and further transformed, the following formula is obtained.

(x-a) ² + (y-b) ² + (z-c) ² = R ² ... Here, a, b, c are the center coordinate values of the sphere (a,
b, c), and R is constant at the radius of the sphere.

Therefore, if the contactor 12 is brought into contact with four points on the spherical surface of the sphere 16, the coordinate values of these four points can be read, and the center coordinates and radius of the sphere 16 can be determined from the coordinate values of these four points. It can be easily calculated. Here, the center coordinate value of the sphere 16 is required, and the radius value is not necessarily required, but the radius value obtained by this calculation is based on the four points on the surface of the sphere 16 when determining the center coordinates of the sphere 16. It can be used to verify whether measurements were performed correctly. That is, in reality, the radius R of the sphere 16 is known, so if the error value of the calculated radius value with respect to the known radius value is within a certain tolerance range, the series of measurements of the sphere 16 is correct. If this is not the case, it is determined that the measurement was inappropriate and the measurement is performed again.

In addition, in order to obtain only the center coordinate value of the sphere 16, it is sufficient to measure at least three points on the surface of the sphere 16, and such a method may be used.

In this way, the center coordinates (x ₀₀ , y ₀₀ , z ₀₀ ) of the sphere 16 are determined with the reference posture of the multi-joint arm 7 .
Then, using the center coordinates measured in the basic posture as a reference, and assuming that the error at that time is 0, (0, 0,
0) is determined.

Hereinafter, the center coordinates and error values of the sphere 16 will be determined in the same manner for each posture of the multi-joint arm 7. In this case, the first rotation axis 8 and the first bending axis 9 will be considered as one set, and the second rotation axis 10 will be and the second bending axis 11 are set as one set. That is, first, the second rotating shaft 10 and the second
The bending axis 11 is held at θ ₃ =θ ₄ =0 (or constant), and the first rotation axis 8 and/or the first bending axis 9 are set at an arbitrary angle, and the center of the sphere 16 is adjusted as in the above basic posture. Find the coordinates. For example, θ ₁ = α ₁ , θ ₂
When = β ₁ , the center coordinates are (x ₀₁ , y ₀₁ , z ₀₁ ).
Further, the error values (Δx ₀₁ , Δy ₀₁ , Δz ₀₁ ) are calculated from this center coordinate and the reference center coordinate, and
The opposite sign value is used as the correction value. Here, Δx ₀₁ =
x ₀₁ −x ₀₀ , Δy ₀₁ =y ₀₁ −y ₀₀ , Δz ₀₁ =z ₀₁ −z ₀₀ .

In addition, when measuring this spherical coordinate, the multi-jointed arm 7 is held in a predetermined posture and measured at four points. can be measured.

In this way, θ ₁ and θ ₂ can be changed to small angle pitch α°,
Take the above measurements over its entire range at β°,
The computer of the control device 13 performs the process of determining the correction value, and the result is stored in a storage device (not shown).

FIG. 3 schematically shows the first correction matrix created by these series of operations and stored in the storage device.

As shown in FIG. 3, correction values are stored on a matrix with θ ₁ and θ ₂ as variables.

Next, in the same way, θ ₁ = θ ₂ = 0 (or constant)
By changing θ ₃ and θ ₄ by appropriate small angle pitches γ° and δ°, a second correction matrix schematically shown in FIG. 4 is created.

In this way, two correction matrices are created.

Note that the correction values of the correction matrix do not necessarily have to have opposite signs.

Automatic measurement by three-dimensional measuring robots includes a teaching playback method and a database method.

Therefore, first, automatic measurement using a teaching playback method using a correction matrix will be explained.

In this method, automatic measurement is performed after the measurement robot is taught the measurement point and route of the object to be measured. The measured values obtained by this automatic measurement still contain errors.

At the time when the posture of the multi-joint part of the measurement robot is determined, the angle data of each joint (θ ₁ , θ ₂ , θ ₃ , θ ₄ ) is input to the control device. As a result, the control device extracts the posture that most closely approximates the posture, that is, the correction value data of the approximate posture, from the first and second correction matrices.

However, in this system, the extraction method is a two-part search.

Using this method, we search for two correction matrices (first and second matrices created from θ ₁ , θ ₂ and θ ₃ , θ ₄ ), obtain correction value data for each, and then Synthesize. However, the second
Since the correction value data obtained from the first correction matrix has a different coordinate system from the first correction matrix,
It is not possible to simply add and synthesize, but first unify the coordinate system and then synthesize. For example, the correction value data obtained from the first correction matrix is (Δx _0o ,
Δy _0o , Δz _0o ), and the correction value data obtained from the second correction matrix are (Δx _0o ′, Δy _0o ′,
Δz _0o ′), then the data obtained from this second correction matrix is transformed into (Δx′ _0o ′,
Δy′ _0o ′, Δz′ _0o ′), and the synthesized correction value data is (Δx _0o ′+
Δx′ _0o ′, Δy _0o ′+Δy′ _0o ′, Δz _0o ′+Δz′ _0o ′
).

Then, the combined correction value data is added to the corresponding measurement value (x _o , _yo , zo ₎ , the measurement value is corrected, and this corrected measurement value is output.

Next, the case of the database method will be explained. This method is basically the same as the teaching playback method. In other words, even in the database method, first and second correction matrices are created in advance, data on the posture and path at the measurement points of the multi-joint arm are input into the control device of the measurement robot using the database, and from this data Correction value data is searched for in the first and second correction matrices, and further, these data are synthesized and the corresponding data are corrected based on the data, and then automatic measurement is performed to obtain measured values.

Incidentally, FIG. 5 is a flowchart showing the correction matrix creation procedure, and FIGS. 6 and 7 are flowcharts showing the automatic measurement and correction procedure.

G. Effects of the Invention As explained above, according to the first aspect of the present invention, in the eleventh step, the first correction matrix created in the sixth step and the second correction matrix created in the ninth step are From the correction matrix and the posture of the multi-joint arm when obtaining the measured value in the 10th step,
The correction value is approximately searched for each, and in the twelfth step, the measured value obtained in the tenth step is corrected by the composite correction value of the two correction values, and the second aspect of the present invention
According to this aspect, in the eleventh step, the first correction matrix created in the sixth step, the second correction matrix created in the ninth step, and the multiplier in the measured value related to the tenth step are combined. From the posture of the jointed arm,
Each approximate correction value is searched, and in the 12th step, the data input to the control device in the 10th step is corrected by the composite correction value of the two correction values, and in the 13th step,
Since we decided to automatically measure the measurement points and obtain the measured values using the data corrected in the 12th step, we equipped it with a multi-jointed arm that has multiple joints and is prone to measurement errors. Three-dimensional measurement robots can correct measurement errors and perform highly accurate measurements, making it possible to measure objects with complex shapes.

[Brief explanation of the drawing]

Fig. 1 is an explanatory diagram of the three-dimensional measurement robot, Fig. 2 is an explanatory diagram of the multi-joint arm, Fig. 3 is a schematic explanatory diagram of the first correction matrix, and Fig. 4 is a schematic illustration of the second correction matrix. 5 is a flowchart showing the correction matrix creation procedure, FIG. 6 is a flowchart showing the automatic measurement and correction procedure using the teaching playback method, and FIG. 7 is the automatic measurement and correction procedure using the database method. This is a flowchart representing the following.

Claims (1)

[Claims] 1. A first rotation axis, a first bending axis, a second rotation axis, and a second bending axis supported by arms of a measuring device mechanism having three axes perpendicular to the surface of the object to be measured. In a measurement accuracy correction method for a three-dimensional measurement robot that measures the object to be measured by bringing into contact a contactor provided at the tip of a multi-joint arm connected in the order of axes, the method comprises: A first step of determining three-dimensional coordinate values of each point by bringing the contact into contact with four arbitrary points on the spherical surface of a reference jig installed at a predetermined position in a predetermined reference posture. a three-dimensional coordinate value and a radius value indicating the center position of the sphere based on the three-dimensional coordinate value of each point obtained in the first step and the equation of a sphere passing through each point; a second step of calculating; and when an error value between the value of the radius of the sphere calculated in the second step and the value of the radius of the sphere held in advance is within a predetermined tolerance range. a third step of holding the three-dimensional coordinate values obtained in the second step as error-free three-dimensional coordinate values indicating the center position of the sphere in the reference posture of the multi-jointed arm; The rotation axis and the second bending axis are held in predetermined postures, and the rotation angle and bending angle of the first rotation axis and the first bending axis are changed, respectively. a fourth step of determining three-dimensional coordinate values of each of the points by bringing the contactor into contact with four arbitrary points; and three-dimensional coordinate values of each of the points obtained in the fourth step. and a fifth step of determining a three-dimensional coordinate value indicating the center position of the sphere based on the equation of a sphere passing through each point; and a three-dimensional coordinate value indicating the center position of the sphere determined in the fifth step. The difference between the coordinate value and the three-dimensional coordinate value indicating the center position of the sphere held in the third step is stored as a correction value on a matrix having the rotation angle and the bending angle as variables. a sixth step of creating a first correction matrix; the first rotation axis and the first bending axis are held in predetermined postures, and the second rotation axis and the second bending axis are respectively a seventh step of determining three-dimensional coordinate values of each of the four arbitrary points on the surface of the sphere by bringing the contact into contact with the four arbitrary points on the surface of the sphere while changing the rotation angle and the bending angle; an eighth step of determining a three-dimensional coordinate value indicating the center position of the sphere based on the three-dimensional coordinate value of each point obtained in the seventh step and an equation of a sphere passing through each point; Using the difference between the three-dimensional coordinate value indicating the center position of the sphere obtained in the eighth step and the three-dimensional coordinate value indicating the center position of the sphere held in the third step as a correction value, a ninth step of creating a second correction matrix by storing the rotation angle and the bending angle on a matrix as variables; a tenth step of teaching a route leading to the point and automatically measuring the measurement point based on this teaching to obtain a measured value; an eleventh step of retrieving correction values approximately from the first and second correction matrices, respectively, according to the orientation of the object; Coordinate transformation is performed to enable synthesis of the two correction values retrieved in step 2, and the measured value obtained in the tenth step is corrected by the combined correction value of the two correction values obtained by this coordinate transformation. A measurement accuracy correction method for a three-dimensional measurement robot, characterized by comprising a twelfth step. 2 A first rotation axis, a first bending axis, a second rotation axis, and a second bending axis are connected in this order to the surface of the object to be measured, and are supported by arms of a measuring instrument mechanism with three orthogonal axes. In a measurement accuracy correction method for a three-dimensional measuring robot that measures the object to be measured by bringing a contactor provided at the tip of a multi-joint arm into contact with the object, the method comprises: , a first step of determining three-dimensional coordinate values of each of these points by bringing the contact into contact with four arbitrary points on the spherical surface of a reference jig installed at a predetermined position; A second step of determining the three-dimensional coordinate value and radius value indicating the center position of the sphere using the three-dimensional coordinate value of each point obtained in step 1 and the equation of a sphere passing through each point. and when the error value between the value of the radius of the sphere obtained in the second step and the value of the radius of the sphere held in advance is within a predetermined tolerance range, the second step is performed. a third step of holding the three-dimensional coordinate values obtained in the process as error-free three-dimensional coordinate values indicating the center position of the sphere in the reference posture of the multi-jointed arm; With respect to the bending axis, the rotation angle and bending angle of the first rotation axis and the first bending axis are changed, respectively, while the bending axis is held in a predetermined posture, and the rotation angle and the bending angle are changed respectively. a fourth step of determining three-dimensional coordinate values of each of these points by bringing the contacts into contact with each other; a fifth step of determining a three-dimensional coordinate value indicating the center position of the sphere based on the equation of a passing sphere; a three-dimensional coordinate value indicating the center position of the sphere determined in the fifth step; A first correction matrix is created by storing the difference between the three-dimensional coordinate value indicating the center position of the sphere held in step 3 as a correction value on a matrix having the rotation angle and bending angle as variables. a sixth step of creating the first rotation axis and the first bending axis while holding them in predetermined postures, and adjusting the rotation angle and bending angle of the second rotation axis and the second bending axis, respectively; a seventh step of determining three-dimensional coordinate values of each of these points by bringing the contact into contact with four arbitrary points on the surface of the sphere while changing the spherical surface; an eighth step of determining a three-dimensional coordinate value indicating the center position of the sphere based on the obtained three-dimensional coordinate value of each point and an equation of a sphere passing through each point; The rotation angle and bending are performed using the difference between the obtained three-dimensional coordinate value indicating the center position of the sphere and the three-dimensional coordinate value indicating the center position of the sphere held in the third step as a correction value. a ninth step of creating a second correction matrix by storing angles on a matrix as variables; a tenth step of inputting data indicating the posture of the multi-joint arm at the measurement point of the measurement object and data indicating a path to the posture; an eleventh step of searching for approximate correction values from the first and second correction matrices, respectively; searching either one of the first and second correction matrices in the eleventh step; Coordinate transformation is performed so that the two correction values can be combined, and the data input to the control device in the tenth step is corrected by the combined correction value of the two correction values obtained by this coordinate conversion. A three-dimensional measuring robot comprising: a twelfth step; and a thirteenth step of automatically measuring the measurement point using the data corrected in the twelfth step to obtain a measured value. Measurement accuracy correction method.