JP3611457B2 - Ring-shaped object gripping method and alignment apparatus therefor - Google Patents

Description

[0001]
BACKGROUND OF THE INVENTION
The present invention relates to an apparatus for gripping and aligning a ring-shaped elastic material made of an example of a flexible elastic body and a gripping method thereof.
[0002]
[Prior art]
Conventionally, when attaching or inspecting a flexible ring-shaped object, such as a sealing made of synthetic rubber, it is necessary to work to uncover overlapping or tangled objects.
Automating these operations is an effective technique for system automation as well as the development of many flexible object handling techniques.
[0003]
[Problems to be solved by the invention]
However, a system for recognizing the placement state of such a ring-shaped object with a flexible and uncertain shape has not been established yet.
The object whose shape has been determined is subjected to pattern matching with a good image that has been taught in advance, and the shape of the object is recognized. It is a system that can not.
The present invention proposes an apparatus and method for aligning an object having an original shape such as the case shown in FIG. 6 by using a robot and image processing for a ring-shaped object including a belt-like object. .
[0004]
[Means to solve the problem]
The present invention is a gripping method for recognizing and gripping a state of a ring-shaped object having a flexible and uncertain shape , capturing the entire ring-shaped object with a camera, binarizing the image , performs labeling and allowed recognizing deformed state from the number and area of the label from the image and the distribution of the camera to determine the gripping position of the ring-like object, the gripping height and allowed detect the height at that location by another camera It is characterized in that it is determined and gripped.
[0005]
An alignment apparatus for recognizing and aligning the state of a ring-shaped object having a high flexibility and an indefinite shape, wherein the entire ring-shaped object is captured by a camera , the image is binarized , and labeling is performed. performed, and allowed to recognize the deformation from the number and area of the label from the image and the distribution of the camera to determine the gripping position of the ring-like object, determines the gripping height allowed detect the height at that location by another camera It is characterized by being gripped and aligned in a certain direction.
[0006]
DETAILED DESCRIPTION OF THE INVENTION
The preconditions for carrying out the present invention will be described below with reference to the drawings.
The apparatus shown in FIG. 1 is an image processing apparatus, and is composed of one 6-axis articulated robot and three cameras.
The camera C1 is disposed above the ring-shaped object and can capture the entire ring-shaped object.
The cameras C2 and C3 are attached to the robot hand 1 and can partially capture a ring-shaped object from an oblique direction.
[0007]
【Example】
Measurement Method First, the overlapping state is recognized from the area and distribution of the label from the image of the camera C1, and the gripping position where the robot hand 1 can be inserted is determined. The height at that position is detected by the two cameras C2 and C3 attached to the robot hand 1, and the gripping height is determined. Here, the overlap is resolved by lifting the ring-shaped object.
A recognition ring-like object having an overlapping pattern is captured by the camera C1, the image is binarized, and labeling is performed.
The number of labels LU1 with the number of pixels equal to or greater than TU is NU, the upper left coordinates of the ferret diameter are (FSXi, FSYi), the lower right coordinates of the ferret diameter are (FEXi, FEYi), (i = 1, 2, 3,... , NU).
When NU = 1, since the ring-shaped objects do not overlap, the ring-shaped objects, which are the next process, are aligned.
When NU ≧ 2, since the ring-shaped objects are overlapped, it is necessary to eliminate this overlap.
The label LU1 has the largest area LUMAX.
In order to determine the insertion position of the robot hand 1, overlapping patterns are divided.
[0008]
Pattern 1
When the label LUmax is not enclosed around the other label LU1 and pattern 1 as shown in FIG. 2, the case number 1 satisfy.
[0009]
[Expression 1]
[0010]
Pattern 2
When other than the pattern 1, a pattern 2 shown in FIG. 3 when surrounded around label LUmax is the other labels LU1, the case number 2 conditions are satisfied.
[0011]
[Expression 2]
[0012]
In the case of the method 1 for determining the gripping position of the ring-shaped object, the following processing is performed to detect the position where the robot hand 1 can be inserted. Contour line tracking processing is performed on the label LUmax.
The coordinates of each point on the contour line are (DXk, DYk), (k = 1, 2, 3,..., NC). Ring subject center of gravity (GX, GY) farthest contour of the point from (DXmax, DYmax) is calculated by the number 3.
[0013]
[Equation 3]
[0014]
When the sizes of the robot hand 1 and the captured image correspond to each other, the opening width from the center of the robot hand 1 is PA pixels, and the vertical and horizontal sizes of the robot hand 1 are PB pixels and PC pixels. The normal of the contour line at the point (DXmax, DYmax) is obtained. The presence / absence of an object within the rectangle of the vertical PB pixel and the horizontal PC pixel is examined from the normal direction and the point of the PA pixel.
If the object is confirmed on the left side from the center, the normal line at that point is searched for by moving one pixel to the right on the contour line.
Similarly, search is also performed in the direction outside the normal.
A point on the contour line where the object is not detected on both the inside and the outside is defined as a gripping position. The normal direction is the gripping direction. (See Figure 4)
[0015]
In the case of pattern 2, contour line tracking is performed on the label LUmax and the outer periphery of the ring-shaped object.
Each coordinate (DIXm, DIYm), (DOXn , DOYn), the total number of pixels and NI, NO, and closest point of two points (DOXmin, DOYmin) was calculated from the number 4, and the grip position of this point.
[0016]
[Expression 4]
[0017]
The normal line of the contour line at the point (DOXmin, DOYmin) is obtained, and this is taken as the gripping direction.
[0018]
Calculation of the height of the gripping position The height of the ring-shaped object at the gripping position determined using the cameras C2 and C3 is calculated, and only the overlapping part can be gripped even if the ring-shaped objects overlap. Determine the gripping height. As shown in FIG. 5, when the image of the camera C2 appears to be shifted by D2 than when the ring-shaped object is viewed from directly above, if the height of the ring-shaped object is H and the width is W, Equation 5 It becomes like this.
[0019]
[Equation 5]
H = CH · D2 / (CW + D2)
The right end and left end coordinates of the ring-like object as viewed from the camera C2 R2, and L2, similarly expressed as the ones viewed from the camera C3 R3, L3 to the D2 number 6.
[0020]
[Formula 6]
D2 = R3-R2
The number 7 and substituting the number 6 to number 5.
[0021]
[Expression 7]
H = CH. (R3-R2) / (CW + R3-R2)
[0022]
At the gripping position and gripping height calculated for gripping and overlapping, the robot hand 1 grips the ring-shaped object and lifts it upward until the ring-shaped object is separated from the workbench. As shown in FIG. It is possible to cancel the overlap of the object and return to the original shape.
When the ring-shaped object is lowered from this state, it is lowered and aligned in a diagonally lower plane so as not to overlap again.
[0023]
In order to accurately align the ring-shaped objects, it is possible to determine a feature portion of the ring-shaped object by accurately specifying a gripping position. A gripping position is determined and aligned by creating a database and matching feature portions.
[0024]
【The invention's effect】
In the present invention, it is possible to reliably hold a ring-shaped object having flexibility.
The overlapping state of the ring-shaped objects can be recognized using the area and arrangement of the labels.
By recognizing the height of the ring-shaped object in each state, only the uppermost part can be gripped even when the ring-shaped objects are overlapped, and the overlap can be eliminated and aligned.
[Brief description of the drawings]
FIG. 1 is a schematic diagram showing an image processing apparatus of the present invention.
FIG. 2 is a plan view of a pattern 1;
FIG. 3 is a plan view of a pattern 2;
FIG. 4 is a plan view showing a gripping position of a pattern 1;
FIG. 5 is an explanatory diagram showing a positional relationship for calculating a height.
FIG. 6 is a plan view showing an original shape (that is, an aligned shape) of a ring-shaped object.
[Explanation of symbols]
1 Robot Hand C1 Camera C2 Camera C3 Camera

Claims (2)

In a gripping method for recognizing and gripping the state of a ring-shaped object with a flexible and uncertain shape,
The uptake by the ring-shaped object across the camera, the image is binarized, performs labeling, the gripping position of the ring-shaped object made to recognize the deformation from the number and area of the label from the image and the distribution of the camera A method of gripping a ring-shaped object, characterized in that the ring-shaped object is determined and determined by detecting the height at the position by another camera and determining the grip height . In an alignment device that recognizes the state of a ring-shaped object that is flexible and has an indeterminate shape, and grips and aligns it,
The uptake by the ring-shaped object across the camera, the image is binarized, performs labeling, the gripping position of the ring-shaped object made to recognize the deformation from the number and area of the label from the image and the distribution of the camera An apparatus for aligning a ring-shaped object, characterized in that the height at that position is detected by another camera, the gripping height is determined , gripped and aligned in a certain direction.

Images