JPS60231208A - Control method of automatic machine - 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] [Technical Field of the Invention] This invention relates to control of an automatic machine such as a robot, in which the three-dimensional position, such as the shape, of an object to be worked on is recognized by a sensor and the machine is operated. It is about the method.

[Prior art]

First, page turning will be explained below as an example of the operation of a conventional automatic machine.

Figure 1 shows an example of a conventional page turning method. In this figure, 1.5 is a TV camera, 2 is an object to be worked on,
For example, a book (hereinafter referred to as water), 3 is a computer, 4 is a manipulator, a KTVTV camera is installed above the book 2, the image input from this nK is differentiated according to the instructions from the computer 3, and this t'L is Book 2 by analyzing
Since it is configured to recognize the position and shape of the book 2, information about the thickness of the book 2 and the curved surface of the paper surface cannot be obtained. Furthermore, if the contrast between the paper surface and the background is low, it is difficult to recognize the outline of the paper surface, and furthermore, it is difficult to distinguish between characters, etc. on the paper surface and the edges of the paper surface. Another way to solve this problem is to install the KTV camera next to the TV camera 2, but this requires two TV cameras. In addition, since a mechanism for driving the TV camera is required, there is a drawback that the cost is high.

[Summary of the Invention J] In order to eliminate these drawbacks, the present invention applies a sensor capable of measuring the inclination of the object to be worked on and the distance to the object to be worked on to an automatic machine. For example, in the case of a book, the shape of the book is 0, 0, thickness, 1, 0, and 0, so that the page turning operation can be performed reliably. The present invention will be described in detail below with reference to the drawings. Note that, in the following explanation, the page number will be used as an example, as in the conventional example.

[Embodiments of the invention]

FIG. 2 shows the configuration of an embodiment of the present invention. In this figure, 6 is a sensor that recognizes the shape and position tg of the rate 2. In order to operate this til, it is recognized by sensor 6? 1. Based on the shape 1 position information of the book 2, the manipulator 4 is controlled by instructions from the computer 3, and the pages of the book 2 are turned.

The sensor 6 measures the tip of the book 2 and the distance to the book 2.
The principle of the method described in No. 4 is shown in FIG.

In FIG. 3, T is a light emitting element that emits a source with high straightness such as a laser source, and 8 is a visual sensor such as a COD camera. In addition, a plurality of light emitting elements T arranged on the same circumference with radius R emit light to form an elliptical pattern on the surface of the book 2, and this image is inputted from the visual sensor 8 to create a curved surface. The shape of v
k: S analysis. Furthermore, k is the angle θ of the light emitted from the light emitting element 7.
By controlling and focusing the light on a single point V, the distance y1ml to the object from the angle θ at that time is 1=Raotθ (R is calculated from the center diameter of the circle where the light projecting element 7 is located. The thickness can be measured.

The actual configuration of the sensor 6 may be one in which the busy manipulator 4Vc is mounted as shown in FIG. 4, or one installed separately from the busy manipulator 4 as shown in FIG. Fourth
In the figure, 9 is the lever mechanism, 1o is the pole screw, P+I
Pa and Pm are pivot points, and in Fig. 5, 11 is X-
This is stage Y. The lever mechanism 9 moves up and down by the rotation of the ball screw 1o, and the light emitting element 1 is moved from the pivot point P from 7ttK.
It is rotated as the sY fulcrum to change the angle θ.

In the case of the JIIJ diagram, the sensor 6 is three-dimensionally driven by the movement of the manipulator 4. Since the relative position with respect to the book 2 to be recognized can be changed, it is suitable for detecting the shape of the book 2 and the thickness of the book 2.

#! In the system shown in FIG. 5, the sensor 6 is mounted on the X-Y stage 11, etc., and scanning is performed at appropriate intervals in row 5, which reduces the load on the manipulator 4 and enables highly accurate positioning.

Lever mechanism 9 that controls the angle θ shown in FIG. 3 using the method shown in FIG.
Since the superimposition becomes heavy when the
As shown in FIG.
..., etc., by combining multiple units, the weight t'f can be reduced.
I! : It is possible to perform almost the same functions without increasing the functionality. The following is an example of the case shown in Figure 5.
A'4I: Do it.

Page (in order to complete the page turning function of Rerobot, the size of book 2, the position of 2, the thickness of 2) wufla
Is there a need to do that? Furthermore, in order to read the information on the paper surface of the book 2 using this Jipot, K needs to scan the visual sensor 8 along the curvature of the paper surface. Therefore, it is possible to recognize one side of the book 2. The geometrical relationship regarding the sensor 6 is shown in FIG.

#I7 In the diagram, A is the center of the circle, S is 'yt, M is the point where light is focused on the center line of the circle, B and C are the points at both ends of the long sleeve of the ellipse formed on the object surface, and φ is the point of the circle. The angle between the center line and the ray, α, is the inclination of the 20th surface of the book. Now, let us say that the distance between point B and π is vb, and the distance between points C and π is vc. Since b and C only need to know the length ratio,
The distance to the book 2 is shortened as described above by converging the light from the bright spot on the image inputted from the visual sensor 8 to one point.

In order to recognize the curved surface of a paper surface, the sensor 6 is scanned at several points on the paper surface, and the distance and inclination of each point are determined.

The shape of any smooth curved surface as these t data is expressed by a curved surface batch as shown in FIG. By setting the curvature of the abrasive surface using a curved Batcher representation and an appropriate offset, the document information formula Q can be transformed into two parameters U.
, represents the tangent vector of the curve in the W direction. The curved surface batch P (u * w) is expressed as follows.

P(u,w)=(Hoju) Ho,+(ul H+,
o(u) La(u))y, = r L, salt. (tl=
(t-1)"(2t+1)HJt)=t" (3-
2 H1,. (t) = (t-1)'', t Hx4(t) = (t -1) t'', and t is U
Mak is wY hail.

The position vector Q is calculated from the distance ah to the book 2 and the amount of scanning in the XY direction. Tangent vector Q,,Q.

As shown in Figure 9, the normal vector δ is δ= (cooαcosβcos(X slnβwin
α) 1, so δ・Q, = 0, δ・Q, = O
and U.

It is difficult to match the W direction with the vX*7 direction. k
, is a one-word constant.

In the above manner, the curved surface of the paper surface can be recognized, and the edges of the paper surface can be recognized based on the gradation image. For example, as shown in Fig. 10, when the contour candidate is detected from the differential of the input image and the sensor 6 is scanned as shown by the dotted line, and the tilt and position of the object are measured +1 at each grid point, the scan and the contour candidate intersect. Find out whether or not. When the grid points intersect, distances are measured at fine pitches between the grid points to obtain graphs such as those shown in FIGS. 11(a) and 11(b). This t is used to determine whether it is a true outline or information on paper such as characters.

In the case of a contour, it cannot be approximated by a curved surface batch, so it is expressed by creating batches on both sides of this point (divide A-b, b-BK in Figure 11). □If we approximate, we can calculate the size of 2 as shown in this paper.

Depending on the size of the book 2, the edges of the book 2 may be uneven as shown in Figure 12, which may impede the page turning operation, but if you measure the distance using the method shown in Figure 4, the unevenness of the edges can be detected. We meet by measuring.

Next, the operation of the present invention will be explained with reference to the flowchart shown in FIG. Note that (1) to (13) in the figure indicate each step.

First, import n images included in book 2 as grayscale images (
1), After differential processing '&mLk (2), binarization is performed using an appropriate threshold (3). Based on this result, check whether the scan M (dotted line) shown in Figure 1O intersects with the contour candidate (pixel whose differential value is larger than the threshold value) (4), and if it intersects, The distance is measured by the sensor 6 (5), a graph like that shown in Fig. 11 is obtained, and it is necessary to judge whether it is a contour or not.
), memorize (7). If this process is performed on the entire screen, the size and location of book 2 can be recognized by tracing the outline (9). Based on this result, perform a page turning operation to turn the page (1
0).

Next, by projecting a spot as shown in FIG. 7 onto a location corresponding to the paper surface of Book 2, the above-mentioned curved surface can be requested with surface parameters Q, Q, . . . Q=.

Since the three-dimensional shape of the paper surface can be determined from the results of the entire paper surface, the information on the paper surface can be inputted without having to scan the high-resolution input head, etc. along the curvature of the paper surface ( 12). When performing a busy page turning operation (13), the process begins again with the input of a grayscale image.

FIG. 14 is a graphic diagram showing the configuration of an embodiment of the present invention, in which 101 is an image memory;
103 is a differentiation circuit, 103 is a binarization circuit, 104 is a binary image memory, 105 is an address command circuit, 106 is a buffer,
107 is a judgment circuit, 108 is a distance measurement command circuit, 10g
is a controller, 110 is a laser light transmission filter, 11
1 is a motor, 112 is a memory, 113 is a determination circuit, 11
4 is memory, 115 is the position of the book, shape i&! ! ! 116 is a page turning operation instruction circuit, 111 is a curved surface recognition instruction circuit, 118 is a memory, 119 is an input head instruction circuit, 120 is an input head, and 2.4,
6.8 is the same as FIGS. 4 and 5.

Next, the operation will be explained. First, input the grayscale image of book 2 through the visual sensor 8 of the TV left camera, and take an image memo! J
Store tOIK. Next, this image data is subjected to differentiation processing in the differentiation circuit 1G2, and binarized in the binarization circuit 103 using an appropriate threshold value.
Store in 4. According to the command from the address command circuit 105,
Data sound buffer 106 on the scanning line (dotted line in Fig. 1θ)
The determination circuit 101 determines whether or not there is a contour candidate pixel. If there is, the distance measurement command circuit 10
8 is given to the hunt roller 109.

In the controller 109, first, the laser transmission filter 110 is set in front of the visual sensor 8.

Furthermore, by controlling the motor 111 of the sensor 6, the light spot is focused on one point. Motor 11 at this time
The distance is longer than the rotation angle in 1, so take a note! 711
Store in 2. Memo the distance for a unit section! After being stored in 7112, the determination circuit 113 determines whether or not it is truly a contour, and if it is a contour, its coordinate values are stored in the memory 114.

The address command circuit 105 gives a signal to the book position and shape recognition circuit 115 and the memory 114 at the end of scanning the entire screen, and based on the contour coordinates read from the memory 114, the book position and shape are recognized. A circuit 115 calculates the position and shape of book 2. Based on the result, the page turning operation command circuit 116 determines a page turning trajectory suitable for the book 2 recognized, and drives the manipulator 4 to turn the page. When the page turning operation is completed, the curved surface recognition command circuit 117 A spot pattern is scanned at a position corresponding to the surface of Book 20, and the curved surface is defined by the surface parameters Q, Q,
, Q, is calculated and stored in memory 118. Accordingly, the input head command circuit 119 generates a trajectory for the input head 120, and the input head 120 follows the curvature of the book 2).
By driving k, information on the paper is input.

In addition, in the above case, the case of book 2 is taken as the work target object,
Although the page turning operation has been described, it goes without saying that the present invention is not limited to nVc and can be applied to any control.

〔Effect of the invention〕

As explained in detail above, this invention equips an automatic machine with a sensor that detects the three-dimensional position of an object to be worked on, and uses this sensor to recognize the shape, size, and inclination of the object to be worked on. Since the required operations are made to be performed, accurate operations can be performed at all times, which has the advantage of being widely applicable to controlling automatic machines in each building, including page turning.

[Brief explanation of the drawing]

Fig. 1 is an explanatory diagram of a page turning method as an example of a conventional automatic machine control method, Fig. 2 is a diagram showing a schematic configuration of a page turning method as an embodiment of the present invention, and Fig. 3 is a work target. FIG. 4 is an explanatory diagram of the operating principle of a sensor that recognizes the shape and position of an object. Fig. 5 is a perspective view showing an example of the configuration of a manibunter-mounted type, and Figs. 6(a) and (b) show other examples of the door sensor. An explanatory diagram for explanation and a diagram of the person's arrow direction, Figure 7 is a diagram showing the geometric relationship of each part in Figure 6, Figure 8 is an explanatory diagram of a curved punch, and Figure 9 is a tangent vector. Explanatory diagram of how to put it on, 1st
The θ diagram is a diagram showing an example where there is a sudden density change part, Figures 1.1 (a) and (b) are explanatory diagrams of distance measurement examples, and Figure 12 is a main part showing an example of uneven edges of a book. FIG. 13 is a flow chart for explaining the operation of the present invention, and FIG. 14 is a block diagram showing the configuration of an example of an automatic machine implementing the present invention. In the figure, 1.5 is the TV left camera 2 is a book, 3 is a computer, 4 is a manipulator, 6 is a sensor, 7°7! +7b+
7c is a light projecting element, 8 is a visual sensor, 9 is a lever mechanism, 1
0 is a pole screw, 11 is an X-Y stage, 101 is an image memory, 102 is a differentiation circuit, 103 is a binarization circuit, 10
4 is a binary image memory, 105 is an address command circuit, 10
6 is a buffer, 107 and 113 are judgment circuits, 108 is a distance measurement command circuit, 109 is a controller, 110 is a laser transmission filter, 111 is a motor, 112° 114,
118 is a memory, 115 is a book position and shape recognition circuit, 1
16 is a page turning operation instruction circuit, 117 is a curved surface recognition instruction circuit, 119 is an input head instruction circuit, and 120 is an input head. Figure 1 Figure 2 Figure 3 Figure 4 Figure 5 Figure 6 Figure 8 Figure 7 Figure 9 Figure 11 Figure 12 Figure 13

Claims (1)

[Claims] An automatic machine that remotely controls an object to be worked on is equipped with a sensor that detects the three-dimensional position of the object to be worked on, and uses this sensor to recognize the shape, size, and inclination of the object to be worked on to control the object to be worked on. A method for controlling an automatic machine, characterized by causing an object to be manipulated.