JPH06347257A - Shape processing method - Google Patents

Abstract

(57) [Abstract] [Purpose] To provide a shape processing method for accurately detecting a bending point position of an outline of an object. [Structure] For a contour point sequence having a long flat section, this is approximated by a straight line or a curve based on the least squares method, and for a contour point sequence having a short flat section, the point sequence of the plate obtained by the long contour point sequence part is used. Direction to project the end face with high accuracy,
As a result, the position of the bending point can be detected with high accuracy even if the shape of the laminated plate is thin.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a shape processing method for accurately detecting a bending point position of a contour from pattern data indicating the contour of an object, and more specifically, for example, a welding robot welds two welding joints. The present invention relates to a shape processing method used for highly accurately detecting the position of a groove that is a boundary between plate materials.

[0002]

2. Description of the Related Art Generally, in contour line data obtained by measuring the surface of an object, there are variations in the position of each contour point represented by the measurement error. Then, it is important how to accurately find the bending point position from the sequence of points having this variation. Conventionally, as this type of technique, the contour point sequence is divided into sections that can be approximated by a straight line or a curve by the least square method, A method is known in which the intersection of the obtained straight lines or curves is set as the bending point position (for example, Japanese Patent Application No.
-190796 "Pattern processor").

[0003]

However, in the method based on the method of least squares described above, accuracy can be expected when the number of points to be approximated by least squares is large to some extent. In the case of a thin plate stack having a small number of points measured on the end face portion of the upper plate in the sequence of contour points shown, there is a high risk that the detection error of the shape and position of the upper plate end face becomes large.

The present invention has been made in view of the above,
An object of the invention is to provide a shape processing method for detecting the bending point position of the contour of an object with high accuracy.

[0005]

In order to achieve the above object, the shape processing method of the present invention is a shape processing method for determining an overlapping end face position of two plate-like objects which are in contact with each other from the shape of sensing data. Measure the contour shape and convert it to a sequence of coordinate points, detect the bending point candidates in the sequence of contour points,
Using this as a temporary bending point, the interval between two adjacent temporary bending points is shorter than a given length, and the centrifugal direction with each of the temporary bending points as an end point when the interval is the central portion When there is no other temporary bending point in the section of a given length, it is determined that the given contour point sequence indicates the overlapping plate-shaped portion, and at the same time, the temporary bending points overlap each other. It is determined that the temporary upper plate corner point and the temporary upper and lower plate contact end points of the end face part of the plate are indicated, and the least square approximation straight line or curved temporary upper plate corner point passing through the temporary upper plate corner point obtained above With respect to a straight line having a slope equal to the slope at, the contour point sequence existing between the temporary upper plate angle point and the temporary upper and lower plate contact points is orthogonal to the angle calculated from the angle formed by the sensing surface and the measurement target surface. One-dimensional projection on a straight line, and the peak position in the distribution of the projected point sequence is optimal. It determines that gives a surface approximate line,
It is determined that the intersection of the end face approximation line and the temporary upper plate approximation line or curve is a true upper plate corner point, and the intersection of the end face approximation line and the lower plate approximation line or curve is a true upper and lower plate contact point. Use as a summary.

[0006]

In the shape processing method of the present invention, the characteristic that the angle between the plate surface and the end surface is often known is used, such that the end surface of the plate used for the laminated plate is cut at a right angle to the plate surface. Then, for a contour point sequence having a long flat section, this is approximated by a straight line or a curve based on the least squares method, and for a contour point sequence having a short flat section, this point sequence of the plate obtained by the long contour point sequence part Direction to project the end face with high accuracy,
As a result, the position of the bending point can be detected with high accuracy even if the shape of the laminated plate is thin.

[0007]

Embodiments of the present invention will be described below with reference to the drawings.

FIG. 1 is a flow chart showing an overall flow of processing by a shape processing method according to an embodiment of the present invention, and FIG. 2 is a schematic view showing a state of processing by the shape processing method shown in FIG. It is a figure. In FIG. 2, the points represent the positions of the contour point sequence, and the points P and Q represent the temporary bending points. Also L
₀ , L ₁ and L ₂ represent approximate straight lines, and C ₁ is L ₁ and L
The intersection of ₀ and C ₂ represents the intersection of L ₂ and L ₀ . Here, for example, with the horizontal axis as the x-axis and the vertical axis as the y-axis, the contour point sequence is point sequence data consisting of n points {(x _m , y _m ) | m =
1, 2, ..., N}, and the point sequence data are arranged in ascending order of the value of x (this is 3 of the cross-sectional contour line of the object in the laser range finder from the direction along the y-axis). Equivalent to measuring dimensional coordinates).

Next, the processing shown in FIG. 1 will be described with reference to FIG.

In FIG. 1, first, the cross-sectional contour line of the object to be measured is measured by a laser range finder or the like, and contour point sequence coordinate data is acquired (step 110). Then, a temporary bending point is obtained for the contour point sequence measured in this way (step 120). At this time, the position of the contour point sequence varies due to a measurement error or the like, so that the position of the temporary bending point is also affected by the variation. There are various methods of obtaining the temporary bending point, but for example, for an appropriate value of natural number k

[Number 1] _{θ = arctan {(y m +} k -y m) / (x m + K -x m)} -arctan {(y mk -y m) / (x mk -x m)} (1) This value represents the difference between the forward and backward inclinations of the contour point (x _m , y _m ) along the contour line. For example, if the contour point is on a straight line, θ = 0, while If the contour point is a right-angled bending point, | θ | = 90 degrees. Therefore, with respect to a point where | θ | is larger than a certain threshold angle given in advance, a point giving the maximum value may be set as a temporary bending point. The sign of θ indicates whether the bending point is a concave portion or a convex portion of the figure. If θ <0, it is convex upward, and if θ> 0, it is convex downward. In the following, a point that is convex upward is called a mountain point, and a point that is convex downward is called a valley point. In Figure 2, the temporary bending point upward convex (Kariyamaten) P, a temporary bending points downward convex (Kariya point) Q is determined, respectively the coordinates of the two points (x _p, y _p) , (X
_q , y _q ) (p <q).

In the next step 130, it is determined from the combination of the detected temporary bending points whether the object to be measured is a thin plate stack or another shape. Specifically, from the obtained temporary bending point group, the mutual distance is within a predetermined length (amount that reflects the plate thickness), the signs of the respective angles are opposite, the other temporary bending point In the direction opposite to the direction, when there is a set of temporary bending points satisfying the condition that the third temporary bending point does not exist in the section of the predetermined length d ₁ , the measurement target portion is the thin plate overlapping portion. It is determined that In FIG. 2, the set of the temporary peak points P and the temporary valley points Q satisfies the above conditions. At this time, the section between the temporary mountain point P and the temporary valley point Q is regarded as the temporary end face section, the point sequence section extending from the temporary mountain point P in the direction opposite to the temporary valley point Q is regarded as the temporary upper plate section, and the temporary valley point is also considered. The point sequence section extending from Q in the direction opposite to the temporary mountain point P is regarded as the temporary lower plate section. In addition, in this determination process, when it is determined that the thin plate is not the overlapped part, the subsequent process is performed by a method described in Japanese Patent Application No. 4-190796 “Pattern processing device”, for example, as a target having another cross-sectional shape. It will be realized.

Next, for each of the temporary upper plate section and the temporary lower plate section obtained in step 130, the point sequence is set to a minimum of 2.
Approximation is performed with a straight line or a curve based on the multiplication method to obtain an upper plate approximate straight line or an upper plate approximate curve and a lower plate approximate straight line or a lower plate approximate curve (step 140). Regarding the linear approximation or the curved approximation, the linear approximation is effective when the laminated plate is a flat plate, but the curved approximation is necessary when the plates are curved, so follow this. And However, even in the case of the curve approximation, the curvature is not so large when the vicinity of the overlapped portion is locally seen, and therefore the approximation is performed by a parabola as shown in Japanese Patent Application No. 4-190796 “Pattern processing apparatus”. Is appropriate. FIG. 2 shows an example in which the upper plate approximate straight line L ₁ and the lower plate approximate straight line L ₂ are obtained.

Then, the point sequence of the temporary end face section is projected on the upper plate approximate straight line or approximate curve obtained in step 140 to obtain the approximate straight line (step 150). As a premise here, the end face to be detected is a part of the upper plate,
It is assumed that the angle formed by the flat portion of the upper plate is known (generally right angle). Specifically, the upper plate is approximately represented by a straight line y = b ₁ · x + c _1, and it is assumed that the angle formed by the straight line representing the upper plate surface portion and the straight line representing the end surface portion is a known value φ. Therefore, the slope b ₀ of the straight line representing the end face is

[Expression 2] b ₀ = (b ₁ · cotφ + 1) / (cotφ−b ₁ ) If the upper plate is approximated by a parabola y = a ₁ · x ² + b · x + c ₁ ,

B ₀ = {(2a ₁ · x _p + b ₁ ) · cot φ + 1} /
It is expressed as {cot φ- (2a ₁ · x _p + b ₁ )}. Here, the angle φ is handled differently depending on the conditions regarding the optical arrangement when the contour point sequence is acquired in step 110. In one case, when the plane scanned by the laser beam used for measurement (or the plane formed by the laser slit light) is perpendicular to the upper surface of the upper plate, the upper surface of the laminated plate and the end surface are formed. The angle φ ₀ itself becomes φ. In the other case, the above optical arrangement relationship is not satisfied, and in that case, the laser beam scanning surface (or laser slit surface) is formed in the direction of the groove line of the upper plate flat surface and the upper and lower plate overlapping portions. Φ changes from φ ₀ depending on the angle. However, even in this case, for example, in welding, the welding work proceeds while always grasping the positional relationship between the upper plate surface of the laminated plate to be welded and the welding groove and the position of the sensor, so the angle φ is calculated each time. It is possible. Now, with respect to b ₀ obtained in this way, assuming that the point sequence of the temporary end face portion should originally lie on the straight line with the slope b ₀ , each point of the temporary end face portion {(x _r , y
_r ) | r = p, p + 1, p + 2, ..., q-1, q} satisfies y _r = b ₀ · x _r + c ₀ . Here, each point does not necessarily show the same straight line due to the influence of noise and the like, and the value of c ₀ varies, but this c ₀ is expected to vary around the true value, so it was obtained. the densest value by the value of c ₀ may be set to the optimum c _0.

However, in reality, when the value of b ₀ is large, the fluctuation of the value of c ₀ becomes large, and in such a case, it is difficult to obtain the dense value of c ₀ . Therefore, a straight line y = −x / that is orthogonal to the straight line having the slope b ₀ and passes through the origin.
A perpendicular line is drawn from the point (x _r , y _r ) with respect to b ₀ , and the dense position at the intersection is detected. That is,
The following simultaneous equations y = b ₀ · x + _cr , _cr = _yr −b ₀ · x _r y = −x / b ₀ are solved to find the intersection point, and the distance c from the origin of the intersection point
^If you ask for ^*

[Equation 4] Is obtained. Therefore, c ^* is obtained for each point and this dense position is detected. As a method of detecting a dense position, a histogram is created that counts the number of appearances of c ^{* in a} constant step interval, and c ^* that is the center value of the interval giving the peak of the frequency is obtained, and the optimum value is calculated by the following formula. Define c ₀ .

[0015]

[Equation 5] Note that this operation corresponds to one-dimensionally projecting the point sequence in the temporary end face section onto a straight line orthogonal to the straight line that the end face portion should make, and selecting the most frequent part of the projected image.

FIG. 2B shows a temporary end face section point sequence {(x _r , y
_r ) | r = p, p + 1, p + 2, ..., q-1, q} is schematically shown as being projected onto the upper plate approximate straight line L ₁ .

In the next step 160, step 140
An intersection of the approximate straight line or the approximate curve obtained in step 150 and the approximate straight line representing the end face obtained in step 150 is obtained. Specifically, as an intersection of the upper plate and the end face part, it is sufficient to solve a simultaneous equation of y = a ₁ · x ² + b ₁ · x + c ₁ y = _br · x + c ₀ (when the upper plate is a linear approximation , A ₁ = 0). Also, the simultaneous equation of y = a ₂ · x ² + b ₂ · x + c ₂ y = _br · x + c ₀ should be solved as the intersection of the lower plate and the end face portion (a ₂ = 0 when the lower plate is a linear approximation). Will be). In FIG. 2, C ₁ is the intersection of the upper plate approximate straight line L ₁ and the end face approximate straight line L ₀ , and the lower plate approximate straight line L
C ₂ is obtained as the intersection of ₂ and the end face approximation line L ₀ .

FIG. 3 is a block diagram showing the arrangement of an apparatus for carrying out the shape processing method according to the embodiment of the present invention.
FIG. 4 is a flow chart showing the operation of the apparatus of FIG.

The operation of the apparatus of FIG. 3 will be described with reference to the flowchart of FIG. In FIG. 3, first, the contour shape measuring unit 1 realized by a laser range finder or the like measures the cross-sectional contour shape of the target object (step 210 in FIG. 4) and stores the coordinate point sequence in the contour point sequence storage unit 2. Yes (Step 2
20). For the contour point string accumulated in the contour point string accumulating section 2, the temporary bending point detecting section 3 compares θ obtained by the calculation of the equation (1) with a preset threshold angle and detects the maximum and minimum points. Then, the position of the temporary bending point is obtained, and this is stored in the temporary bending point storage unit 4 (step 230). Then, the cross-sectional shape determination unit 5 checks whether the temporary bending points stored in the temporary bending point storage unit 4 satisfy the thin plate overlapping condition shown in Step 130 of FIG. 1 (Step 24).
0). If there are provisional bending points that satisfy the conditions, the contour point sequence accumulated in the contour point sequence accumulating unit 2 is used as the provisional upper plate corner point and the provisional upper and lower plate contact end points to determine the provisional upper plate section. A point sequence, a temporary end face section point sequence, and a temporary lower plate section point sequence are accumulated in the temporary upper plate section point sequence accumulation unit 6, the temporary end face section point sequence accumulation unit 7, and the temporary lower plate section point sequence accumulation unit 8, respectively ( Step 250). For each of these accumulated point sequences, the upper plate partial approximation unit 9 and the lower plate partial approximation unit 10 approximate the respective point sequences with a straight line or a curve based on the least squares method (steps 260 and 270), and the result is obtained. Approximate line group storage unit 1
Accumulate to 2. The end surface part approximating unit 11 is based on the inclination information of the upper plate obtained by the upper plate part approximating unit 9 and the angle between the straight line of the upper plate surface portion calculated by the angle calculation unit 13 and the straight line of the end surface portion. The process shown in step 150 of FIG. 1 is performed to perform the linear approximation of the end face portion, and the result is used as the approximate line group accumulating unit 12 (steps 280 and 290). Finally, the intersection detection unit 14 detects intersections with respect to the approximate straight line and the approximate curve accumulated in the approximate line group accumulating unit 12 (steps 300 to 320).

FIG. 5 is a block diagram showing the configuration of an apparatus for carrying out the shape processing method according to another embodiment of the present invention. The embodiment shown in the figure is obtained by removing the angle calculation unit 13 from the embodiment shown in FIG. 3, and other configurations and operations are the same as those shown in FIG. In the embodiment of FIG. 3, the contour point coordinates are acquired so that the angle formed by the laser beam scanning surface (or laser slit surface) is always constant with respect to the measurement target. Therefore, the angle calculator 13 shown in FIG. 3 is unnecessary.

[0021]

As described above, according to the present invention,
For a contour point sequence with a long flat section, this is approximated by a straight line or a curve based on the least-squares method. For a contour point sequence with a short flat section, this point sequence is projected in the direction of the plate obtained in the long contour point sequence section. By accurately approximating the end faces, and this makes it possible to detect the bending point position with high accuracy even in the case of thin laminated plates, it can be used, for example, for data processing of a visual sensor mounted on a welding robot for thin plates. If applied, the welding robot can perform high-quality welding without missing the groove points.

[Brief description of drawings]

FIG. 1 is a flowchart showing an overall flow of processing of a shape processing method according to an embodiment of the present invention.

FIG. 2 is an explanatory diagram schematically showing how the shape processing method shown in FIG. 1 performs processing.

FIG. 3 is a block diagram showing a configuration of an apparatus for carrying out a shape processing method according to an embodiment of the present invention.

4 is a flowchart showing the operation of the apparatus shown in FIG.

FIG. 5 is a block diagram showing a configuration of an apparatus for carrying out a shape processing method according to another embodiment of the present invention.

[Explanation of symbols]

 1 Contour shape measuring unit 2 Contour point sequence accumulating unit 3 Temporary bending point detecting unit 4 Temporary bending point accumulating unit 5 Cross section shape determining unit 6 Temporary upper plate section point sequence accumulating section 7 Temporary end face section point sequence accumulating section 8 Temporary lower plate section Point sequence accumulation section 9 Upper plate approximation section 10 Lower plate approximation section 11 End face section approximation section 12 Approximate line group accumulation section 13 Angle calculation section 14 Intersection detection section

Claims (2)

[Claims] 1. A shape processing method for determining an overlapping end face position of two plate-like objects that are in contact with each other from a shape of sensing data, the contour shape is measured, and this is converted into a coordinate point sequence. Is detected as a temporary bending point, the interval between two adjacent temporary bending points is shorter than a predetermined length, and the temporary bending is performed when the interval is the central portion. When there is no other temporary bending point in the section of the length given in advance for the centrifugal direction with each of the points as the end points, it is determined that the given contour point sequence indicates the laminated plate shape portion, At the same time, it is determined that the temporary bending points respectively indicate the temporary upper plate corner point and the temporary upper and lower plate contact end points of the end face portion of the laminated plate, and the least-squares approximation passing through the temporary upper plate corner points obtained above. A straight line or a straight line with a slope equal to the slope of the temporary upper plate corner On the other hand, the contour point sequence existing between the temporary upper plate corner point and the temporary upper and lower plate contact points is one-dimensionally projected on a straight line orthogonal to the angle calculated from the angle formed by the sensing surface and the measurement target surface, It is determined that the peak position in the distribution of the projected point sequence gives the optimum end face approximation straight line, and the intersection point of the end face approximation straight line and the temporary upper plate approximation straight line or curve is the true upper plate corner point, the end face approximation. A shape processing method characterized in that an intersection of a straight line and a lower plate approximate straight line or a curve is determined as a true upper and lower plate contact point. 2. The angle direction for one-dimensionally projecting a contour point sequence is set to a predetermined angle direction without calculating from the angle formed by the sensing surface and the measurement target surface. Shape processing method.