JPS6344103A - Apparatus for confirming position and posture - Google Patents

Abstract

PURPOSE:To confirm the position and posture of an article of which image has to be picked up, by extracting a plurality of characteristic points of the article of which image has to be picked up and calculating the mutual position and distance between the characteristic points. CONSTITUTION:The image of a work from an image input means 10 is converted to a line image by a characteristic point extraction means 11 and the characteristic points thereof are subsequently calculated to be collated with the distance data between the characteristic point of the work stored in a distance data storing means 12 to select one set combination data specifying the work by a selection means 13. A conversion type forming means 15 calculates a conversion formula to the above mentioned one set of the combination data from the position data of each characteristic point stored in a position data storing means 14 and also calculates the max. error between the characteristic point in the position data and one set of the combination data. A comparing means 17 judges whether said max. error is within the allowable error range set to a allowable error storing means 16. When said error is within the allowable error range, the position and posture of the work are calculated on the basis of one set of the combination data specifying the work by a position/posture calculation means.

Description

DETAILED DESCRIPTION OF THE INVENTION (Field of Industrial Application) The present invention relates to a position and orientation recognition device that calculates the position and orientation of a wave original such as a work by performing image processing.

(Prior Art) In recent years, with the automation of production processes, for example, in assembly work, robots are used to grab workpieces that are automatically transported by automatic guided vehicles or the like and perform predetermined work.

In this case, the robot is equipped with a visual function such as a camera, and uses this visual function to recognize the posture and position of the transported workpiece, so that the robot always grasps the fixed position of the workpiece regardless of the transporting posture and position of the workpiece. I try to keep it that way.

Devices that recognize the position and orientation of such workpieces include
There are those shown in FIGS. 5 and 6.

The position and orientation recognition device shown in FIG. 5 includes a camera 1 that inputs image information of a workpiece as a subject, an image input unit 2 that converts the image information into digital data and stores this data as an image pattern, and a standard A taught pattern storage section 3 stores an image pattern of feature points to be drilled in the work, a pattern moving section 4 moves the image pattern stored in the taught pattern storage section 3, and an image input section 2 stores the image thus extracted. It consists of a pattern matching section 5 that matches the image patterns output from the pattern and pattern moving section 4.

The image of the workpiece imaged by the camera 1 is recorded as an image pattern in the image input section 2, and the image pattern stored in the taught pattern storage section 3 is transferred to the pattern moving section. 4, and the pattern matching unit 5 selects the one with the highest degree of matching, thereby calculating the position and orientation of the work.

The position and orientation recognition device shown in FIG. 6 also includes a camera 1 that inputs image information of a workpiece as a subject, and an image input unit 2 that converts the image information into digital data and stores this data as an image pattern. , a binarization processing unit 6 that binarizes the image pattern, and a binarization processing unit 6
A binarization level setting unit 7 sets a threshold value when binarizing by, and the center of gravity, area, circumference, and moment of the workpiece.

a teaching data storage section 8 that stores data such as the principal axis of inertia;
It consists of an arithmetic unit 9 that compares the data output from the binarization processing unit 6 with the data stored in the teaching data storage unit 8 and extracts the matched data.

Then, the image of the workpiece captured by the camera 1 is stored as an image pattern in the image input section 2, and this image pattern is converted by the binarization processing section 6 based on the setting level of the binarization level setting section 7. For the image that has been digitized and binarized by the calculation unit 9, its centroid 1 area 1 circumference length.

Moment, principal axis of inertia, etc. are calculated and the area of the center of gravity of the workpiece is stored in the teaching data storage section 8.

Data such as perimeter, moment, principal axis of inertia, etc. are compared,
The matching data is selected, and the position and orientation of the workpiece are calculated based on this data.

(Problems to be Solved by the Invention) However, the conventional position and orientation recognition device having a groove bottom as shown in FIGS. 5 and 6 has the following problems. .

In other words, in the case of the position and orientation recognition device having the groove bottom shown in FIG. Since the position and orientation of the workpiece are compared with the image pattern of the feature points on the workpiece and the image pattern with the highest matching degree is used to calculate the position and orientation of the workpiece, the pattern matching process takes time. There is a problem that high-speed detection of position and orientation is impossible, and it cannot be applied to a production line with extremely short takt time.Furthermore, in the case of the position and orientation recognition device having the groove bottom shown in Fig. 6, The image of the workpiece input from camera 1 is binarized using a predetermined threshold value, and the data of the center of gravity, area, perimeter, moment, principal axis, etc. of the binarized image are used to determine the image of the workpiece. Since the position and orientation of the workpiece are calculated, there is a problem in that when the external shape of the workpiece cannot be clearly detected due to the influence of the illumination conditions on the workpiece, an error occurs in the detection accuracy of the position and orientation.

The present invention has been made in view of such conventional problems, and it is possible to identify the characteristic points of a workpiece (for example, the center point of a bolt hole, a circular outline, or It is an object of the present invention to provide a position/orientation recognition device that extracts the vertices of a sharp portion, etc., and determines the position and orientation of the workpiece based on the positions and distances between the feature points.

(Means for Solving the Problems) In order to achieve the above object, the present invention includes an image input means for inputting image information of an object to be imaged and outputting the image information as a digital value for each pixel; feature point extraction means for extracting a number of feature points in the object to be imaged based on the digital value; and distance data that stores in advance distance data between each feature point serving as a reference in the object to be imaged. The storage means and the distances between the feature points extracted by the feature point extraction means are sequentially compared based on the respective distance data stored in the distance data storage means, and the distances between the feature points extracted by the feature point extraction means are sequentially compared. a selection means for selecting one set of combination data in the R original; a position data storage means for storing position data of each feature point serving as a reference in the imaged object; A conversion formula is calculated to convert the position data of the target object into one coarse combination data selected by the selection means, and the position data of each feature point of the object to be imaged calculated by this conversion formula and the position data of the target object are calculated. conversion formula creation means for calculating the maximum error between position data of each feature point serving as a reference in the image carrier; position data of each feature point serving as a reference in the object to be imaged and each feature of the object to be imaged; The maximum error between the mutual position data of each feature point on the imaged object calculated by the tolerance storage means for storing the mutual tolerance of point position data and the conversion formula creation means is within the permissible error range. The present invention is characterized in that it includes a comparing means for comparing whether or not the object exists, and a position and orientation calculating means for calculating the position and orientation of the object to be imaged based on the conversion formula outputted through the comparing means.

(effect) Below, the operation of the present invention will be explained based on FIG.

First, the image of the workpiece inputted from the image input means 10 is converted by the feature point extraction means 11 into a line drawing image that is hardly affected by lighting conditions, and the number of feature points of the line drawing image is calculated. The feature points are selection means 1
3, a set of combination data that specifies the workpiece is selected by comparing it with the distance data between the feature points of the workpiece stored in the distance data storage means 12.

The conversion formula creation means 15 then converts the position data storage means 1
4. From the position data of each feature point stored in 4, calculate the conversion formula to the combination data of one silk described above, and calculate the maximum distance between each feature point between the position data and the set of combination data. The maximum error is the comparison means 1.
7, it is determined whether or not it is within the tolerance range set in the tolerance storage means 16. If this error is within the allowable range, the position and orientation calculation means 1
8, the position and orientation of the workpiece are calculated.

(Example) Below, an example of the present invention will be described in detail based on the drawings.

FIG. 2 shows a schematic configuration diagram of the position and orientation recognition device according to the present invention.The camera 1, which inputs an image of one node of the object to be imaged, converts an analog signal into a digital signal. to/D converter 20 is connected. In addition,
This camera 1 and A/D converter 20 constitute an image input means 10.

The A/D converter 20 is connected to the CPU 22 via a human output head 21. This CPU 22 is
Feature point extraction means 11. Selection means 13゜Conversion formula creation means 1
5. It also serves as the comparison means 17 and the position/orientation calculation means 18. In general, the CPU 22 has an RO in which a program for determining the position and orientation of the workpiece is written.
M23 and distance data storage means 122 position data +81 internal means 14 and RAM 24 as tolerance storage means 16
is connected.

The position and orientation recognition device configured as described above operates as follows based on the operation flowchart shown in FIG.

Below, this operation flowchart is shown in Figure 4 (a> to (
This will be explained with reference to C).

Step 1 First, the camera 1 captures an image of a workpiece W having a shape as shown in FIG. The digital data is then converted into digital data corresponding to the density value, and this digital data is sequentially input to the CPU 22 via the input/output device 21 and temporarily stored in t(A~124).

Step 2 Based on the digital data stored in the RAM 24, the CPU 22 determines characteristic points XAI, XA2, . XA3. ...
..., XBI, XB2. ......, XC2
,

Step 3 The CPU 22 calculates distances between all combinations of feature points based on the positions of the feature points set out in step 2. For example, between XAI and X81,
B1-XC2, XA2-X81, XA2-XC2, etc.), the distance between each feature point almost matches the distance between each feature point in any workpiece stored in advance in the RAM 24. A combination of subgroups (for example, one combination of XAl-XB2-XC2-X
C5> Select a candidate. These combination candidates are the fourth
This corresponds to point A, point 8, point 0, and point D shown in Figure (C), respectively.

The selection of this combination candidate is such that, for example, the minimum value within the tolerance range between XAl and X81 is δmin.

If the maximum value is set as δma

Step 4 As a result of the processing in Step 3, the CPU 22 selects a set of combinations in which the distance between each feature point is considered to be approximately the same as the distance between the feature points in any workpiece stored in advance in the RAM 24. Determine whether a candidate exists. As a result of this judgment, if a combination candidate exists, the process proceeds to step 5, and if it does not exist, the process proceeds to step 6.

Step 5 The CI-'U 22 converts this position data into position data of the set of combination candidates obtained in Step 3, based on the upright data at the feature points of arbitrary works stored in advance in the RAM 24. The conversion coefficients in the conversion formula and each feature point (X to 1.
XBI, XC2, XC3) and each feature point (xal, xbl, XC2) stored in advance in the RAM 24
, XC3), and stores this value in the RAM 24.

This conversion coefficient is determined as described below.

conversion formula X=Axo10B......(1), △ that minimizes △ in the expression given by .

Find the value of each B. The values of A and 8 at this time are the values of the conversion coefficients.

Further, the maximum error δ is obtained by finding the maximum value among the values of each n on the right side of equation (2).

Step 6 In Step 4, it is determined that there is no candidate combination that seems to approximately match the distance between 1h feature points on any workpiece, so the CPU 22 controls the input/output device 21. A signal indicating that there is no identification is output to an external device (not shown).

Step 7 CPtJ22 is the value of the maximum error δ calculated in step 5 and the allowable error δ preset in the RAM 24.
Compare the values of O, and if δ≦δO, proceed to step 8, δ〉
If δO, proceed to step 3, respectively.

Step 8 The CPU 22 calculates the conversion coefficient △ calculated in step 5.
, the rotation angle θ and the X axis from B to the reference coordinate system of the workpiece,
Y@! The amount of parallel movement in one direction is calculated using the following formula.

Rotation angle θ=TaTl-1a12/all parallel movement foot X
=bl y=b2 As described above, the position and orientation of the workpiece can be determined. Note that the process in step 3 is originally unnecessary for the purpose of determining the position and orientation of the workpiece, but if the process in step 5 is performed without performing this process,
This process is performed because the long processing time of step 5 is very disadvantageous in terms of reducing the processing time.

(Effects of the Invention) As is clear from the above description, according to the present invention, a plurality of feature points in an imaged object are extracted, and the position of the wave original is determined based on the positions and distances between the feature points. Since the system recognizes the image and posture, it is less affected by lighting conditions, and as a result, it is possible to have highly accurate recognition ability, and extract the true feature points of the object to be imaged. Since front edge processing is performed, high-speed processing is possible. Furthermore, by changing the position and distance data to those corresponding to the imaged object, the present invention can be applied to many types of objects, and has excellent effects such as being highly versatile.

[Brief explanation of the drawing]

FIG. 1 is a block diagram of the position and orientation recognition device according to the present invention, FIG. 2 is a schematic configuration diagram of the position and orientation recognition device according to the present invention, and FIG. 3 is the position and orientation recognition device shown in FIG. The operation flowchart of FIG. 4 (a> to (C) is a diagram for explaining the operation flowchart shown in FIG.
6 are schematic configuration diagrams of a conventional position and orientation recognition device. 1... Camera (image input means), 20... A/D converter (image input means), 22.
...CPU (candidate point extraction means, selection means, conversion formula creation means, comparison means, position and orientation exclusion means), 24...RAM (distance data storage means, position data storage means, tolerance storage means), W ...Work (object to be imaged). Patent applicant Nissan Motor Co., Ltd. agent Patent attorney 8 1) Mikio (and 2 others) Figure 3 Figure 4 (b)

Claims (1)

[Scope of Claims] Image input means for inputting image information of an object to be imaged and outputting the image information as a digital value for each pixel; a feature point extracting means for extracting a feature point, a distance data storage means for storing in advance distance data between each feature point serving as a reference in the imaged object, and a distance data storage means for storing in advance distance data between each feature point serving as a reference in the object to be imaged; a selection means for sequentially comparing distances between respective feature points extracted by the feature point extracting means based on the data, and selecting one set of combination data of the imaged object from the comparison results; a position data storage means for storing position data of each feature point serving as a reference in an object; and a position data storage means for converting the position data stored in the position data storage means into a set of combination data selected by the selection means. A transformation that calculates a power conversion formula and calculates the maximum error between the position data of each feature point in the imaged object calculated by this conversion formula and the position data of each feature point serving as a reference in the imaged object. formula creation means; tolerance storage means for storing mutual tolerance between position data of each feature point serving as a reference in the object to be imaged and position data of each feature point in the object to be imaged; a comparison means for comparing whether a maximum error between the positional data of each feature point of the imaged object calculated by the means is within the permissible error range; and the conversion outputted via the comparison means. A position/orientation recognition device comprising: position/orientation calculation means for determining the position and orientation of the object to be photographed using an equation.